This action might not be possible to undo. Are you sure you want to continue?
THE ANALYSIS OF THE COEXISTENCE OF POINTTOMULTIPOINT FWS CELLS
IN THE 3.4  3.8 GHz BAND
Revised in Vilamoura, February 2006
Electronic Communication Committee (ECC)
within the European Conference of Postal and Telecommunications Administrations (CEPT)
ECC REPORT 33
Page 2
EXECUTIVE SUMMARY AND CONCLUSIONS
Summary
The scope of this ECC Report is to provide uptodate guidelines for efficient, technology independent deployment of
3.5 GHz (or 3.7 GHz) PointtoMultipoint (PMP) Fixed Wireless Systems (FWS), traditionally referred to as Fixed
Wireless Access (FWA) systems.
The Report recognises that the current technology for FWA in bands around 3.5 GHz is in continuous extensive
evolution since first ERC Recommendations 1403 and 1208 were developed. A detailed study on the coexistence of
various technologies was needed in order to provide guidance to Administrations that wish to adopt an efficient and
technology neutral approach to the deployment rules in these bands.
It is also noted that ETSI ENs in these bands are not presently designed for a technology neutral deployment (this is
done only in the 40 GHz Multimedia Wireless Systems in EN 301 997) therefore do not contain system controlling
parameters, in terms of EIRP, useful for the desired “technology neutral” and “uncoordinated” deployment. Not having
any ECC harmonised guidance for such deployment, the ENs are still bound to a cellbycell “coordinated
deployment” concept actually not used in most of the licensing regimes. This report might generate future feedback
actions in revising also ETSI ENs accordingly.
Aspects that relate to sharing issues with PointtoPoint FS, FSS, radiolocation (in adjacent band) and ENG/OB are not
considered in this Report. However they should be taken into account when applying any method of deployment
suggested in this document.
The applicability limits of the current Report are as follows:
 Application is mostly devoted to “block assignment” licensing methods, rather than “channel assignment”
method.
 The guidelines presented have been maintained, as far as possible, independent from the access methods
described in the ETSI ENs (e.g. EN 302 326
1
).
 MPMP (MESH) architectures are not yet considered. In order to include MESH architectures, a number of
assumptions on “typical” application (e.g. on the omnidirectional/directional antenna use) need to be defined
in order to devise the typical intraoperators, mixed MPMP/PMP interference scenarios for which simulations
would have to be carried.
 Channel sizes and modulation schemes are also not specifically considered unless for defining “typical” system
parameters. It should be noted that high state modulations (e.g. 64/128 QAM) have not been specifically
addressed in the typical system parameters; nevertheless they would not change the general framework of this
report. This may be considered during future update.
 FDD/TDD, symmetric/asymmetric deployments are considered.
 Additionally, system independent, EIRP density limits and/or guardbands at the edge of deployed region (pfd
boundary conditions) as well as at the edge of assigned spectrum (block edge boundary conditions) are
considered as licensing conditions for neighbouring operators’ coexistence (similarly to the latest principles in
ERC Recommendation 0104 in the 40 GHz band).
Presently, the spectrum blocks assigned to an operator vary widely from country to country  from 10 MHz up to
28MHz (single or duplex) blocks have been typically assigned. The block allocation size and the frequency reuse plan
employed by the operator to achieve a multicell and multisector deployment drives the channel bandwidth of the
systems presently on the market to be typically no greater than 7MHz. Conversely, the requirement for higher data
throughputs is driving the need for wider channel widths (e.g. up to ~28 MHz) and therefore correspondingly wider
spectrum blocks assignment in the future.
Therefore, system channel bandwidths and block sizes are not fixed, even if typical data for current technologies are
used for feasibility analysis of the “blockedge” constraints.
The report considers two different aspects of deployment scenarios for two operators:
1. Operating in the same or partly overlapping area with adjacent bands assignment
1
The approved ETSI EN 302 326 will supersede equipment EN 301 021, EN 301 124, EN 301 744, EN 301 080, EN 301 253,
antenna EN 302 085 and related HEN 301 753; technical characteristics in EN 302 326 are identical, or very similar, to those present
in the ENs listed.
ECC REPORT 33
Page 3
2. Operating in adjacent or nearby areas and reusing the same band assignment.
A number of different methods have been used to assess the severity of interference. These are:
 Worst Case (WC) (generally used for Central Station (CS) to CS interference) and for PFD limits at
geographical boundary for frequency (block) reuse
 Interference Scenario Occurrence Probability (ISOP) (for CS to Terminal Stations (TS) interference between
adjacent blocks in rural LoS situations)
 Monte Carlo simulations for CDF (cumulative distribution function) vs. C/I (for CS to TS interference between
adjacent blocks in urban mostly NLoS situations and for TS to CS interference between adjacent blocks).
For the above methods it has been possible to estimate the probability of interference between FWA systems. From
these results, estimates have been made of the frequency and/or geographical spacing needed between these systems in
order to reduce the level of interference to an acceptably low level. Absolute recommendations cannot be made because
some system parameters are not defined by the available standards and because the effects of buildings and terrain are
very difficult to model. The report therefore gives guidelines that will lead to acceptably low levels or low probability
of interference in most cases.
For the above methods that might be described as:
 The Worst Case (WC) method derives system deployment parameters to ensure that interference is always
below a set threshold for all cases.
 The Interference Scenario Occurrence Probability (ISOP) is defined as the probability that an operator
places at least one terminal in the IA. ISOP is related to the number of terminals deployed by the operator, and
possibly to the cell planning methodology. The ISOP method evaluates the NFD or the outofblock rejection
required in order to meet an interference probability lower than a certain value.
 The Cumulative Distribution Function (CDF), derived from Monte Carlo simulation of large number of
“trial” TSs with a certain equipment/antenna/propagation model, is defined as the probability that a certain
percentage of those trials would result in a C/I of the victim CS exceeding a predefined target limit.
The Report derives the following alternative parameters, useful for defining an “uncoordinated technology
independent” deployment:
 The Interference protection factor (IPF) and associated guardband method used to define the amount of
isolation required from the interfering station to victim receivers in adjacent frequency block in terms of Net
Filter Discrimination (NFD), obtained also by frequency separation (guard bands) and EIRP limitation.
 The Block Edge EIRP Density Mask (BEM) method is used for directly limiting the EIRP density in the
adjacent block, and for assessing the CS to CS worst case interference, the CS to TS interference through
acceptable ISOP value and the TS to CS through acceptable probability of exceeding a limit C/I to the victim
CS.
An important finding of this Report is that stringent protection requirement (e.g. in terms of BEM or NFD) is required
only for CS emissions; the protection factor for TS is far less stringent and reduces as the antenna directivity is
improved.
Another important conclusion is a significant impact of CS antenna height on coordination distance for the frequency
block reuse; due to the low LoS attenuation with distance, sensible size of coordination distance and associated PFD
value are obtained only considering spherical diffraction attenuation. If the CS antenna height is not limited (or a down
tilt angle is required) as a licensing parameter, it is nearly impossible to tell how far away the block may be reused.
The example presented, made with typical system values, led to examples of BEM coherent with a “technology neutral”
deployment of different systems in adjacent blocks. Receiver filters are assumed to be stringent enough to maintain the
potential NFD implicit in the BEM (i.e. have sufficient outofblock selectivity for avoiding non linear distortion in the
RX frontend chain).
In annexes to the report, some further technical background and studies for related issues are also reported. They
include urban obstructed propagation (nearNLoS) models and examples of practical application of RF filtering for
easing the CS absolute EIRP BEM fulfilment when using equipmentgeneric relative spectrum masks defined by ETSI.
ECC REPORT 33
Page 4
Conclusions
This Report has considered a number of facts as initial considerations for deriving the coexistence study:
1. Presently ERC Recommendations 1403 and 1208 for the bands 3.6 GHz and 3.8 GHz do not give
harmonised and detailed suggestion to administration for implementing FWA (such as those produced for
26, 28 and 40 GHz). Those ERC Recommendations offer only channel arrangements.
2. The band is limited and wasted guardbands might drastically reduce the number of licensed operators,
limiting the potential competition for new services.
3. Legacy FWA (WLL) systems have been already licensed in these bands in some countries, and their co
existence with new FWA systems, assigned in accordance with provisions recommended in this study, may
need additional check on a casebycase basis, using the same methodology as described in this report, but
with the parameters corresponding to those of legacy systems.
4. Sharing issues with PP FS systems, FSS, radiolocation (in adjacent band), ENG/OB were not considered in
this report, but are subject of a different CEPT study.
5. At least for CSs, ETSI ENs in these bands are not presently designed for a technology neutral deployment
(this is done only in the 40 GHz MWS EN 301 997) therefore do not contain system controlling parameters,
in terms of EIRP, which would be useful for the desired “technology neutral” and “uncoordinated”
deployment
6. The study suggested that the block edge mask together with the contiguous assignment of frequency blocks
should be the main means for avoiding interference between neighbouring frequency blocks. Since block
edge mask parameters are linked to the size of the block, it is also recommended that the neighbouring
blocks should be of a similar size.
7. It is also shown that, for PMP TSs, the antenna RPE plays a fundamental role in the coexistence; the more
directive is the antenna of TSs, the less demanding might be their NFD (or the EIRP density BEM) required
(offering a flexible tradeoff to the market). That is why the resulting BEM limits outside the block are
described in terms of transmitter output power, allowing operators to make practical use of this
phenomenon.
8. In addition, basic rules has been set for the coordination distance and PFD boundary levels between
operators reusing the same block in adjacent geographical areas. In this field, the importance of limiting CS
antenna height (or downtilt angle) as possible licensing parameter is highlighted in order of having sensible
coordination distances (i.e. limited by spherical diffraction attenuation).
9. MPMP (MESH) architectures have not been considered in this Report. It is recognised that whilst some of
the results in this report might also be applicable to mixed PMP and MESH deployment, others may clearly
need additional work. These studies might be carried on out in due time if needed and when manufacturers
will be in a position to offer the necessary information.
ECC REPORT 33
Page 5
INDEX TABLE
1 INTRODUCTION........................................................................................................................................................ 7
1.1 SCOPE...................................................................................................................................................................... 7
1.2 THE FREQUENCY LICENSING POLICY AND THE POSSIBLE APPROACHES.................................................................. 8
1.2.1 The Worst Case deployment scenario (derived from ERC Report 99)......................................................... 8
1.2.2 The “predefined guard band deployment” ................................................................................................... 8
1.2.3 The "guided unplanned deployment" (Block edge emission density mask).................................................. 9
2 “SAME AREA  ADJACENT FREQUENCY BLOCKS” INTERFERENCE SCENARIO.............................. 9
2.1 ANALYSIS OF THE COEXISTENCE OF TWO FWA CELLS IN THE 3.4  3.8 GHZ BAND............................................... 9
2.1.1 Typical System Parameters............................................................................................................................ 9
2.1.2 Cell coverage. .............................................................................................................................................. 11
2.1.2.1 Rural LOS scenario .................................................................................................................................................... 12
2.1.2.2 Urban NLOS scenario ................................................................................................................................................ 14
2.1.3 Interference protection factor (IPF)............................................................................................................ 15
2.1.3.1 Channel arrangements ................................................................................................................................................ 16
2.2 CSTOCS INTERFERENCE .................................................................................................................................... 17
2.3 CSTOTS INTERFERENCE..................................................................................................................................... 19
2.3.1 Rural LOS scenario...................................................................................................................................... 19
2.3.2 Urban NLOS scenario ................................................................................................................................. 22
2.3.2.1 Methodology............................................................................................................................................................... 26
2.3.2.2 Results: ....................................................................................................................................................................... 27
2.3.2.2.1 C/I statistics .......................................................................................................................................................... 27
2.3.2.2.2 Interblock protection evaluation......................................................................................................................... 29
2.3.3 Conclusions for CSTS protection in urban NLOS environment ............................................................. 30
2.4 TS TO CS INTERFERENCE...................................................................................................................................... 31
2.4.1 Interference evaluation................................................................................................................................ 31
2.4.2 Conclusions for the TS to CS interference study......................................................................................... 32
2.5 TERMINAL STATION TO TERMINAL STATION ....................................................................................................... 32
3 ”ADJACENT AREA  SAME FREQUENCY BLOCK” INTERFERENCE SCENARIO.............................. 32
3.1 POWER FLUX DENSITY LIMITS FOR ADJACENT FWS SERVICE AREAS ................................................................. 32
3.1.1 Assumptions.................................................................................................................................................. 33
3.1.2 Methodology................................................................................................................................................. 35
3.1.3 Central Station to Central Station............................................................................................................... 35
3.1.3.1 Worst case single interferer scenario: 3.5 GHz calculations ..................................................................................... 35
3.1.3.2 Antenna height............................................................................................................................................................ 39
3.1.3.3 Transmit EIRP............................................................................................................................................................ 39
3.1.3.4 Conclusions and possible selfregulation method for CSs coordination distance................................................... 40
3.1.4 Terminal Station to Central Station ............................................................................................................ 40
3.1.4.1 ATPC impact .............................................................................................................................................................. 40
3.1.4.2 Worst case single interferer scenario, 3.5 GHz calculations ..................................................................................... 40
3.1.4.3 Examples: ................................................................................................................................................................... 41
3.1.4.4 TS to CS Conclusions................................................................................................................................................. 42
3.1.5 Terminal Station to Terminal Station.......................................................................................................... 43
3.2 CONCLUSIONS ON ADJACENT SERVICE AREAS BOUNDARY BLOCK REUSE COORDINATION................................. 43
4 CONCLUSIONS OF THE REPORT...................................................................................................................... 44
ANNEX 1: AVERAGE CELL SIZE EVALUATION................................................................................................... 45
A1.A) SIMULATION MODEL ........................................................................................................................................ 45
A1.B) MEAN EXCESS LOSS (MEL) ONLY ................................................................................................................... 45
A1.C) MEAN EXCESS LOSS AND LOGNORMAL SHADOWING.................................................................................... 46
A1.D) MEAN EXCESS, LOGNORMAL SHADOWING AND RICIAN FADING.................................................................. 47
A1.E) SIMULATION CAVEAT....................................................................................................................................... 48
A1.F) URBAN SCENARIOS, DIRECTIONAL OUTDOOR VERSUS OMNIDIRECTIONAL INDOOR TS.................................. 49
A1.f.1 Coverage and availability objectives: ......................................................................................................... 49
A1.f.2 Methodology................................................................................................................................................. 49
A1.f.3 Cell radius evaluation.................................................................................................................................. 50
A1.G) CONCLUSIONS OF THE ANNEX........................................................................................................................... 53
ECC REPORT 33
Page 6
ANNEX 2: TS TO CS INTERFERENCE EVALUATION........................................................................................... 54
A2.1 DIRECTIONAL OUTDOOR TS ............................................................................................................................. 54
A2.1.1 Rural scenario.............................................................................................................................................. 54
A2.1.1.1 System Model and Simulation Methodology........................................................................................................ 54
A2.1.1.2 Unfaded Simulation Results .................................................................................................................................. 56
A2.1.1.3 Rayleigh Faded Simulation Results....................................................................................................................... 59
A2.11..4 Conclusions............................................................................................................................................................ 61
A2.1.2 Urban Scenario............................................................................................................................................ 62
A21..2.1 Simulation Methodology ....................................................................................................................................... 62
A2.1.2.2 Simulation Results ................................................................................................................................................. 62
A2.1.2.2.1 Unfaded ............................................................................................................................................................. 62
A2.1.2.2.2 Rician Faded...................................................................................................................................................... 69
A2.1.2.3 Conclusions, directional outdoor antennas............................................................................................................ 69
A.2.2 OMNIDIRECTIONAL INDOOR TS....................................................................................................................... 69
A.2.2.1 Input parameters and models .................................................................................................................. 69
A.2.2.1.1. Objectives............................................................................................................................................................... 69
A.2.2.1.2 CS data (same used for CSTS evaluation): ....................................................................................................... 70
A.2.2.1.3 TS omnidirectional indoor out of block eirp:....................................................................................................... 70
A.2.2.1.4 TS activity factor and density: ............................................................................................................................... 71
A.3.2.1 Shared traffic, VDSL like...................................................................................................................................... 72
A.2.2.2 Summary of TS density and OOB eirp reference values: ....................................................................... 72
A.2.2.3 Scenarios.................................................................................................................................................. 73
A.2.2.4 Propagation models:................................................................................................................................ 73
A.2.2.5 Simulation results..................................................................................................................................... 73
A.2.2.5.1 h
CS
= 30 m.............................................................................................................................................................. 74
A.2.2.5.2 h
CS
= 40 m.............................................................................................................................................................. 74
A.2.2.6 Conclusions.............................................................................................................................................. 76
APPENDIX 1 TO ANNEX 2: MODEL BASIC DESCRIPTION................................................................................. 77
APPENDIX 2 TO ANNEX 2: ACCEPTANCEREJECTION METHOD.................................................................. 83
ANNEX 3: EXAMPLES FOR MANAGING A CS BLOCKEDGE MASK.............................................................. 84
ANNEX 4: EVALUATION OF THE C/I RATIO EXCEEDED FOR MORE THAN 99% (AND 97% FOR
OMNIDIRECTIONAL CASE) OF PROBABILITY. .................................................................................................. 88
REFERENCES:.................................................................................................................................................................. 96
ECC REPORT 33
Page 7
THE ANALYSIS OF THE COEXISTENCE OF POINTTOMULTIPOINT FWS CELLS
IN THE 3.4  3.8 GHz BAND
1 INTRODUCTION
1.1 Scope
The scope of this report is to investigate the coexistence of PointtoMultipoint (PMP) Fixed Wireless Systems (FWS),
traditionally referred to as Fixed Wireless Access (FWA). These systems are developed in accordance with the ETSI
EN 301 021, EN 301 080, EN 301 124, EN 301 253 and EN 301 744. In conjunction with the CEPT channel plan
defined by the ERC Recommendations 1403 (sections A1 and B1) and 1208 (sections B2.1.1 and B2.2.1).
During the development of this report, the account was taken of the latest developments of FWA technologies, in
particular developments with ETSI BRAN (HIPERMAN) and IEEE 802.16 standardisation activities, and new market
realities calling for wide spread use of flexibly deployed user terminals with nondirectional antennas.
Systems, owned by different operators, should be able to be deployed without mutual interference when operating in:
a) adjacent frequency blocks in the same area or,
b) the same frequency block(s) in adjacent areas.
This report aims to assist the administrations in the assignment of frequency blocks to the operators who operate FWA
systems in the available bands between 3.4 GHz to 3.8 GHz.
These bands were subject to the previous ERC Recommendation 1403, on harmonised radio frequency arrangements
for Multipoint systems. Nowadays more experience has been gained from recent studies for the 26 and 28 GHz bands,
finalised by ERC Report 99 and Recommendations 0005 and 0103, and most of all for the 42 GHz MWS band,
finalised by ECC Recommendation 0104.
ERC Report 97 qualitatively summarised requirements for modern licensing process and has also been taken into
account in developing this report.
This report incorporates and enriches the information in earlier reports and recommendations.
In order to cater for the mix of technologies and services to be delivered it is most appropriate that a block (or blocks) of
spectrum is made available to a potential operator in a manner consistent with the technology and market that the
operator may wish to address.
It should be noted that the current ETSI masks for PMP equipment in this band were not designed with the block
assignment coordination in mind; therefore they were not specifically considered as a starting point in this study.
Taking into account the final conclusions of this study, CEPT may wish to liaise with ETSI in order to align CEPT and
ETSI deliverables on this issue.
A key principle of the assignment guidelines is that even though a technology specific channelisation scheme is
expected to operate within an assigned block this channelisation is not the basis for the assignment process. Operators
are free to subdivide the assigned frequency block in the most efficient way for deploying or redeploying the selected
technology.
Due to the flexibility required in newly deployed services, it is important that the block assignment process supports
systems for both symmetric and asymmetric traffic as well as systems that employ FDD and TDD techniques.
In principle no assumption has been made regarding the architecture of any FWA network; however MPMP (MESH)
architectures have not been considered in detail in this Report. Other ECC work has reported and concluded on MESH
systems in higher frequency millimetric bands. It is recognised that whilst some of the results in this report might also
be applicable to mixed PMP and MESH architectures, others may clearly need additional work. These studies might be
carried out in due time if needed and when manufacturers will be in a position to offer the necessary information.
Measures are suggested in this report for dealing with the issue of interoperator coexistence both between adjacent
frequency blocks and between neighbouring geographic areas. The basis for these measures is to allow deployment with
the minimum of coordination although more detailed coordination is encouraged as an interoperator issue.
ECC REPORT 33
Page 8
In order to cope with the oftenconflicting requirements of a number of technologies in terms of efficient and
appropriate block assignments, some compromise is suggested in order to develop reasonable assignment guidelines,
which balance constraints as far as possible on any specific technology.
1.2 The frequency licensing policy and the possible approaches
When considering the adjacent frequency blocks, same area scenario, the possible process of frequency licensing should
guarantee, as far as possible, a “controlled interference” deployment. Emissions from one operator’s frequency block
into a neighbour block will need to be controlled. This can be done by different methodologies.
A first one, already recommended in other frequency bands, imposes, between the assignments, fixed guard bands
evaluated around the most likely equipment to be deployed.
Alternatively, as recommended in the 42 GHz MWS band, a frequency block edge EIRP density emission mask is used.
The block edge mask limits the emissions into a neighbouring operator's frequency block and it enables operators to
place the outermost radio channels with suitable guardbands, inside their assigned block, in order to avoid co
ordination with the neighbour's frequency blocks.
Having considered the pros and cons of these two approaches, in particular taking into account a general shortage of
frequencies in this band, the current study agreed to follow the second approach, i.e. the contiguous assignment of
frequency blocks with the associated Block Edge Mask (BEM) requirement. For further enhancing the spectrum
efficiency, administrations might wish, after the block assignment procedure has been done, to decide later not to
enforce the blockedge mask for neighbouring operators who will apply mutual coordination at the blocks edge in view
to optimise the utilisation of outermost parts of the assigned blocks. In that case, the enforcing rules will apply only in a
“mutually agreed” way or it would be flexibly changed according the actual interference scenario shared by both
operators with theirs planning tools.
1.2.1 The Worst Case deployment scenario (derived from ERC Report 99)
In principle, the most efficient way of evaluating the guard bands would be through a “case by case” evaluation. This
would imply that the administrations should, in the application phase, analyse the actual behaviour, the planned
coverage range, the hubs location, the cellular structure and the cell planning aspects of the system operated by the
operators in any particular area.
However, when Terminal Stations (TS) are taken into account, the number of variables become so huge that the method
is practical only for deterministic evaluation of Central Station (CS) to CS interference, based on I/N objective (then
independent from system characteristics apart the noise figure).
For the above reasons, a more realistic approach is necessary, and hereby only the two examples described in next
sections 1.2.2 and 1.2.3 are mentioned; however, only the last one in section 1.2.3 would be extensively used in this
report because looks more attractive from the point of view of spectrum usage and certain results in technology neutral
environment.
1.2.2 The “predefined guard band deployment”
In the first approach, here called “predefined guard band deployment”, the administration would aim to provide, to both
operators and end users, a reasonably interference free environment.
In addition, for maintaining good spectrum efficiency, this method asks for a quite good knowledge of the typical FWA
system technologies used. The guardbands are likely to be determined by the wider band systems therefore the method
is most suited in case the differences among deployed technologies (e.g. channel spacing, NFD and modulation formats)
are expected to be small or in bands already deployed where fixed channel arrangements are recommended. This
approach tends to prevent spectrum efficiency improvement with the technology evolution and thus is not
recommended as a preferred method.
ECC REPORT 33
Page 9
1.2.3 The "guided unplanned deployment" (Block edge emission density mask)
The second approach, here called ”guided unplanned deployment”, implies that additional EIRP density limits are set in
order to allow an “average” interference free scenario to the operators. In this case, the guard band is to be included in
the blocks assigned to the operators; the blocks are to be made consequentially larger. In this case the “interference free
environment” is ensured by the EIRP density limits set by the administration, evaluated for “average worstcase”
interference scenarios.
With this approach, the operator is permitted to use the assigned block as much as the equipment filtering and actual
EIRP allow operation close to the blockedge, leaving to him and the manufacturers the possibility to improve the
spectral efficiency as far as possible.
This method is most suited when very different technologies are used. The EIRP density mask is designed on the basis
of acceptable noise floor increase due to interference from adjacent block; therefore only the knowledge of typical
victim receiver noise figure and antenna gain are necessary. The method is therefore quite independent from ETSI
standards, and is effective for bands that do not have fixed channel arrangements as a deployment constraint.
For a sensible and costeffective regulation, a block edge mask is generally designed on the basis of a small degradation
in an assumed scenario with a low occurrence probability of a worst case (e.g. two directional antennas pointing exactly
each other at close distance).
As for the guardband method described in section 1.2.2 above it is not therefore excluded that in a limited number of
cases specific mitigation techniques might be necessary; operators would still be asked to solve, with conventional site
engineering methods, the "worst cases" that may happen in few cases. In particular when CSs are colocated on the
same building or very close to each other, the statistical approach is not applicable and it is assumed that common
practice of site engineering (e.g. vertical decoupling) is implemented for improving antenna decoupling as much as
possible.
Moreover, for further enhancing the efficiency, administrations are not expected, after the block assignment procedure,
to enforce the blockedge requirements to neighbour operators who will apply mutual coordination at the block edge in
view to optimise the guard bands. In that case, only the maximum "inblock" EIRP/power density applies while the
"outofblock" noise floor will apply only from a "mutually agreed" starting point within the adjacent block.
It is up to operators to possibly further coordinate with other operators using adjacent blocks.
Also adjacent block receiver rejection concurs to a reduced interference scenario, however this is not in the scope of this
Report to set limits for it; nevertheless it is expected that ETSI standards will adequately cover the issue.
2 “SAME AREA  ADJACENT FREQUENCY BLOCKS” INTERFERENCE SCENARIO
2.1 Analysis of the coexistence of two FWA cells in the 3.4  3.8 GHz band
2.1.1 Typical System Parameters
Considering the scenario of a wide suburban area with relatively high traffic demand and a small amount of obstructing
buildings, medium bandwidth systems (3.5/7 MHz) are analysed in LoS conditions as well as similar system in more
restricted and dense urban areas analysed in mostly NLoS conditions.
The examples shown refer to the ETSI EN 301 021 and BRAN HIPERMAN system (including foreseen applications
for indoor terminals with omnidirectional antennas for easy user defined deployment, only for defining a typical
receiver BER thresholds, power output and C/I sensitivity. However, the considerations made are not too sensitive to
the multiple access method, provided that all have similar spectral and linkbudget characteristics. These data are then
“technology independent”, nevertheless for defining typical cell coverage sizes also real modulation formats should be
used; in Table 1 data for two systems types only are referred. Different modulation are obviously possible (e.g. 64
states) but, they would not, in principle, lead to different conclusions on the regulatory framework objective of this
report.
ECC REPORT 33
Page 10
System Type (according typical ETSI definitions)
Type A (typical 4 state) Type B (typical 16 state)
System Channel bandwidth MHz 3.5/7
2
3.5/7
2
Transmitted Power at section D’ (dBm) (CS)
3
30 (rural LOS cases)
35 (NLOS cases)
30 (rural LOS cases)
35 (NLOS cases)
Transmitted Power at section D’ (dBm)
4
(conventional
outdoor TS)
30 30
Transmitted Power at section D’ (dBm)
5
(indoor TS) 23 23
Receiver Noise Figure at section D (dB) 8
6
8
6
Receiver Threshold for BER= 10
6
(dBm)
7
84 76
Critical C/I for BER= 10
6
(dB)
8
~14 ~22
Hub (CS) antenna  90° sector boresight gain (dB) 16 16
CS antenna azimuth and elevation radiation patterns ETSI EN 302 085 [5]
ITUR F.1336
ETSI EN 302 085 [5]
ITUR F.1336
Terminal (TS) outdoor directional antenna boresight
gain (dBi) and RPE
9
16
ETSI EN 302 085 [5]
ITUR F.1336
16
ETSI EN 302 085 [5]
ITUR F.1336
Terminal (TS) indoor omnidirectional antenna gain
(dBi)
10
4 4
Table 1: Typical system data for typical cell size evaluation
The same system parameters will be initially used for both victim and interferer. The 3.5 GHz will be used as radio
frequency throughout the calculations.
2 This channel spacing is considered the most representative for being carried over in the calculation. It is considered that the larger
channel systems would lead the coexistence rules. Nevertheless lover spacing channels (e.g. from 1.5 MHz up), also widely popular,
should more easily fit in that possible framework
3 The lower power level is considered typical for “conventional” PMP systems, while the higher level might be justified for
applications including NLOS propagation and TS indoor with lowgain omnidirectional antennas. This value includes feeder losses
for full indoor applications.
4 This outdoor TS power is assumed as typical for symmetric traffic in conventional PMP systems, mostly with LOS operation. This
value includes feeder losses for full indoor applications. The 35 dBm Maximum Power presently allowed in ETSI ENs (e.g. EN 301
021 and 301 080) is considered not realistic for conventional outdoor TS.
5 Power for indoor, user operated, TS is kept typically lower, according ETSI BRAN contributions, due to the need of containing
size and power consumption. The corresponding uplink margin loss is recovered by other methods (e.g. modulation formats and/or
subchannelling..
6 The Noise Figure estimated from EN 301 021 BER values and typical modulation formats would result in ~12 dB; however this is
justified considering the significant margins used in ETSI EN that have had regulatory valence for old “national type approval” as
well as for R&TTED regimes; therefore for network planning purpose it seems too pessimistic and a value of 8 dB has been
assumed, it should already give enough margin for the possible necessity of a selective RF channel filter of reduced size for TS.
7 This value includes feeder losses for full indoor applications. These values are used only for evaluating reference order of
magnitude of cell sizes in various environments. The actual interference study is based either on I/N in a noise limited scenario or on
C/I between blocks (without specific carriers positioning) in an interference limited scenario; therefore, the results are parametric on
“interference power density”, then independent from the system channel bandwidth. Modulation order is taken care by two different
critical C/I values provided in this table, sophisticated coding is conservatively neglected.
8 These are typical conservative values, independent from the system channel bandwidth. Modulation order is taken care by two
different critical C/I, sophisticated coding is conservatively neglected.
9 An antenna with relatively low gain is frequently used for transmitting and receiving signals at the outstations or in sectors of
central stations of PMP radiorelay systems. These antennas may exhibit a gain of the order of 20 dBi or less. It has been found that
using the reference radiation pattern given in Recommendation ITUR F.699 for these relatively lowgain antennas will result in an
overestimate of the gain for relatively large offaxis angles. As a consequence, the amount of interference caused to other systems
and the amount of interference received from other systems at relatively large offaxis angles will likely be substantially
overestimated if the pattern of Recommendation ITUR F.699 is used. On the other hand ITUR F.1336 gives low gain TS antenna
patterns only for bands below 3 GHz, nevertheless it is considered more appropriate and will be used in this study.”
10
Standing that in these indoor, userdefined deployment the antenna would be of very simple technology without any attempt to
define and/or align the antenna according conventional azimuth and elevation planes, these omnidirectional antennas, unlike the
ETSI standardised ones for CS, would be considered to have constant gain over the whole spherical 360° angle.
ECC REPORT 33
Page 11
Due to the importance of Terminal Station (TS) directional antennas RPEs (and in particular of their main lobe) on the
results shown in this Report, suggest that the use of ETSI RPE for TS antennas might give worstcase results that are
not experienced in practice. ETSI RPEs are generally defined only for “type approval” purpose (i.e. 100% of RPE
values shall be within the mask). Annex2 of ITUR F.1336 gives typical “average” RPE that are more representative of
the field situation; F.1336 recommends RPE for the bands below 3 GHz that here are considered appropriate also in the
3.5 and 3.7 GHz bands; Figure 1 show the difference between those RPE.
The antenna gain is the parameter used in the formulas of Annex 2 of ITUR F.1336 for identifying different RPEs,
therefore it has been used in Figure 1 to reference the different antenna RPEs; the gain range 16 to 20 dB is considered
representative, from the ITUR recommendation F.1336 point of view, of classes of antennas similar to ETSI TS 2 and
TS 3. However the objective of this report would be mainly to consider the impact of different ETSI antenna RPEs for
coexistence studies, not necessarily for studying the increase of cell size. Therefore, while the typical ITUR F.1336
RPEs with gain 16 and 20 dBi will be generally used in all numerical evaluations, the Report will maintain a fixed gain
of 16 dBi, reported in Table 1 as representative of the average value on the market.
In addition, the report also considered omnidirectional TS antennas, assuming the uniform figure of antenna gain in all
directions, i.e. no azimuth or elevation specifics were considered due to assumption of uncontrolled deployment by
users, without any specific alignment towards the CS antenna.
Figure 1: Antenna RPE Comparison
2.1.2 Cell coverage.
For evaluating the interference potential, it is initially needed a rough evaluation of the cells radius in the various
environments using appropriate propagation models.
We will examine two typical scenarios based on LOS and NLOS propagation, they will be conventionally identified as
“rural” or “urban”; however, the results should not be considered strictly applicable for those areas only, but might be
applied whenever the propagation specific LOS or NLOS conditions are applicable (at least for the fraction of the user
terminals concerned). The two scenarios are specific for:
ECC REPORT 33
Page 12
 LOS “Rural” situations, when for wider cells coverage (or high quality of service) it is necessary to rely on
conventional PMP deployment with directional antennas and mostly LOS situation (certainly applicable for the
users at cell border). The cell dimension is noiselimited freespace attenuation and Rayleigh fading, with
correction derived by spherical diffraction phenomenon.
 NLOS “urban” situation, when presence of high building or foreseen service to indoor terminals is likely to
shade most of the links to users terminals, considerably reducing the available coverage. Propagation phenomena
is modelled with SUI model, assuming fast varying multiple reflections.
2.1.2.1 Rural LOS scenario
The scenario examined is a LoS, relatively flat environment without significant obstructions, located in central Europe.
The main propagation modes are assumed to be free space and, possibly, spherical diffraction. The link availability will
be affected by clearair multipath.
The maximum cell radius R
will be calculated from the link budget:
SG + G
CS
+ G
TS
= FSPL + A
sph
+ FM (1)
where:
SG is the “system gain” (i.e. difference in dB of TX output power and RX threshold at given BER 10
6
)
G
CS
and G
TS
are CS and TS antenna gains in dB. For this evaluation we will consider G
TS
=16 dB as worst case
(resulting in smaller cell size).
FSPL is the free space attenuation loss for f=3.5 GHz given by:
FSPL = 92.4+ 20 log(f D) = 103.28 + 20 log(D) (2)
A
sph
is the spherical diffraction attenuation described in ITUR Recommendation P.562 that depends on the height of CS
and TS antennas, relative to the ground.
FM is the fade margin (excess attenuation) required to meet the yearly availability objective.
FM can be evaluated according to ITUR P.530, which covers both the deep fade and shallow fade regions. For the
purpose of the present analysis, it seems adequate to use the deep fade approximation or 10 dB, whichever is greater.
The 10 dB has been chosen as a safe value to ensure proper operation in "normal" clear air propagation.
From ITUR P.5308:
FM=10 log[P
wm
/P
0
], (3)
P
wm
is the probability of exceeding the critical BER during the worst month. Scaling it to a yearly average, for the
assumed geographical area and for 3.5 GHz radio frequency, with the conservative approach that the yearly
unavailability (un
year
%) is spread over four “worst” months only, FM can be written as:
P
wm
%= 3 * un
year
% (4)
P
0
%=5*10
7
* 10
[0.1*(CoClatClon)]
* pl
(1.5)
* (1+c)
(1.4)
* f
(0.89)
* D
3.6
(5)
Assuming C
0
=3.5 (hilly terrain), C
lon
=3 dB (Europe), C
lat
=0 dB (medium latitude); pl=10%; c=0;
P
0
% = 5*10
7
* 10
[0.1*(3.53)]
* 10
(1.5)
* (1+0)
(1.4)
* 3.5
(0.89)
* D
(3.6)
P
0
% = 4.2972*10
5
* D
(3.6)
P
0
/P
wm
= [4.2972*10
5
* D
(3.6)
]
/ (3 * un
year
%)
Substituting (4) and (5) into (3) we obtain:
FM=  48.44+36*log(D)  10*log(un
year
%) (6)
ECC REPORT 33
Page 13
Spherical diffraction attenuation A
sph
can be calculated by subtracting the free space attenuation from the output of the
program GRWAVE (available from ITU). A sample output for two significant cases is shown in Figure 2. Neglecting
the ripple at short distances, which comes from reflections in the plane earth model, A
sph
is approximated as:
A
sph
= 0 (for D < D
0
) (7a)
A
sph
= K
2
(DD
0
) (for D > D
0
) (7b)
D
0
is taken as the point where the total attenuation equals the free space value (i.e. A
sph
= 0 in Figure 2) where spherical
diffraction attenuation starts to be significant. D
0
depends on the heights of hub and terminal antennas above ground
(hc, ht). Values for a few significant cases are shown in the following Figure 2 that shows that K
2
~ 1.3 dB/km is nearly
invariant and that when different CS and TS antenna heights are considered, the mean height value could be used.
0 10 20 30 40 50 60 70 80 90 100
20
0
20
40
60
80
100
120
Spherical Diffraction attenuation  f=3.5GHz
D (km)
d
B
1010
2020
3030
4040
5050
1030
3050
2040
Antenna
heights (m)
(*)
Figure 2: Additional attenuation due to spherical diffraction
Substituting (2), (6) and (7) into (1), the link budget at the cell edge (D=R) can be rewritten as:
SG+ G
CS
+ G
TS
= 54.84 10 log(un
year
%) +56 log(R) + K
2
(RD
0
) (8)
With the assumed equipment parameters, antenna heights, and yearly availability objectives 99.99 %, and 99.999%, the
maximum cell radius values are shown in Table 2.
ECC REPORT 33
Page 14
Availability
System
Antenna
heights
99.99% 99.999%
type G
TS
=16 dB G
TS
=20 dB G
TS
=16 dB G
TS
=20 dB
R [km]
FM
[dB]
R [km]
FM
[dB]
R
[km]
FM
[dB]
R [km]
FM
[dB]
hc =40m
ht = 20m
18.7 17.3 22 19.9 12.4 20.9 14.6 23.4
hc =30m
ht = 30m
18.7 17.3 22 19.9 12.4 20.9 14.6 23.4
A
hc = 30m
ht = 10m
18.7 17.3 (21.2) (19.3) 12.4 20.9 14.6 23.4
hc =20m
ht = 20m
18.7 17.3 22 19.9 12.4 20.9 14.6 23.4
(hc =10m
ht =10m)
(14.7) (13.6) (16.2) (15.1) (11.4) (19.6) (12.9) (21.5)
hc =40m
ht = 20m
13.4 12.2 15.8 14.7 8.9 15.7 10.5 18.3
hc =30m
ht = 30m
13.4 12.2 15.8 14.7 8.9 15.7 10.5 18.3
B
hc = 30m
ht = 10m
13.4 12.2 15.8 14.7 8.9 15.7 10.5 18.3
hc =20m
ht = 20m
13.4 12.2 15.8 14.7 8.9 15.7 10.5 18.3
(hc =10m
ht =10m)
(12) (10.4) (13.5) (12.2) (8.7) (15.4) (10.5) (18.3)
Note: Values in parenthesis denote the impact of spherical diffraction attenuation
Table 2: Cell radius and FM vs. Availability (BER 10
6
)
The conclusions of Table 2 show that, for most cases of practical antenna height, the cell radius is limited by the system
gains considered and by the free space loss only. Hence spherical diffraction is not yet affecting the propagation;
moreover, antenna heights are not affecting the area coverage. The cases with CS and TS antenna heights = 10 m (see
Figure 2) are the only ones where spherical diffraction attenuation has some impact by reducing the cell size. The latter
cases are shown only as explanatory example of the phenomenon, however, hc = 10m is not considered realistic and
therefore will no longer be taken into account in further evaluations.
2.1.2.2 Urban NLOS scenario
For urban propagation models there are not consolidated ITU models. However, the one recently adopted by IEEE
802.16 for similar coexistence studies [4] might be profitably used; it describes he associated path attenuation, in dB,
with a Gaussian probability distribution function (p.d.f.), with mean value (here called A
50
) and a shadowing variance
with standard deviation “o” (between ~8 to 10 dB). Overimposed to this attenuation probabilistic spread, another “fast
fading” occurrence probability described with SUI propagation model, derived from multiple facets attenuation due to
quick variable situations (e.g. windows openings, leaves, cars etc.)
In IEEE 802.16 an “excess attenuation” for all TSs is introduced (mostly based on tests made over wooden/hilly areas
among low rise buildings, typical for most US cities outside their relatively small downtown) increasing with distance
from CS and subdivided into three different categories (Erceg categories A, B and C) according the terrain
characteristics and 6 SUI models. However, subsequent tests made in European urban environment has shown that
some of the variants (notably Erceg B with intermediate SUI 24) are suitably applicable as well as Erceg C for rural
NLoS (or wide spread suburban areas).
The following basic principles describe the IEEE model:
 The path loss (PL) can be seen as the summation of basic free space loss (FSL) and the excess loss (Lex) due
to the local blockage conditions or reduction of antenna gains: PL(dB) = FSL (dB) + Lex (dB)
 The path loss can be modelled as follows: PL(dB) = A0(dB) + 10 n log(d/d0) + S(dB) , where the exponent n
represents the decay of path loss and depends on the operating frequency, antenna heights and propagation
environments. The reference path loss A0 (also sometimes referred as A
50
in the document) at a distance d0
ECC REPORT 33
Page 15
(generally d0 is assumed =100m considering that propagation closer to 100 m might be dominated by other
unpredictable factors such as reflections) from the transmitter is typically found through field measurements.
The shadowing loss S denotes a zero mean Gaussian random variable (Lognormal variable) with a standard
deviation (also in decibels).
The detailed evaluation of cell size as a function of desired availability is reported in Annex 1.
Results might be summarised in cell sizes of 2/3 km radius (depending on the predominant modulation format) in
suburban areas for conventional Outdoor TS coverage ~95% and availability objective of 99.99% (QoS) services.
In urban scenarios cell sizes drop down to 1/2 km (depending on the predominant modulation format) for conventional
Outdoor TS coverage ~95% and availability objective of 99.99% (QoS) services and is maintained around the same size
for indoor applications provided that the objectives would be relaxed to ~70% coverage and 99% availability (best
effort service). Alternatively the cell size should be further reduced to 0.5/1 km.
2.1.3 Interference protection factor (IPF)
The potential coexistence of different cells in adjacent frequency blocks is guaranteed when there is sufficient isolation
between interfering transmitters of one cell and victim receivers of the other cell.
This required isolation, generally referred as Interference protection factor (IPF), might be obtained enforcing a block
edge spectrum density mask (in form of either outofblock EIRP density or outofblock absolute power density
emission at antenna port as discussed later in the document.
As a consequence a “technology neutral” approach is hereby used in the form of the above IPF, out of which a specific
example is the transmitter output power density Blockedge mask (BEM) concept, described in Figure 3.
The BEM concept, strictly related to the Net Filter Discrimination (NFD) concept, actually summarised all the
equipment/antenna related IPF contributions and might be best fit in environment where equipment characteristics are
not known beforehand.
This does not imply that the BEM is always the best method, when system characteristics are known (e.g. when one or
few specific technologies are privileged) so that fixed coordination rules might be uniquely set a more detailed
approach might be more appropriate.
Also polarisation decoupling is a factor that might not be prejudged (unless different polarisations are imposed in
licensing two adjacent blocks operators, limiting their free usage of the block) and in the following evaluation is not
taken into account.
The relationship between NFD (in the central portion of adjacent block) and BEM, expressed in terms of EIRP, is
equipment/antenna dependent only and is described as:
P
outdensity
(dBW/MHz) + G
TX
− NFD = X
3
(dBW/MHz) (9)
where X
3
represents, in EIRP density terms, the CS BEM outof block lowest requirement within the central portion of
the adjacent block as described in Figure 3.
ECC REPORT 33
Page 16
TX Output Power Density
(dBW/MHz)
F, from the block edge
X
1
X
3
0
P
MAX
Assigned
block
Adjacent block
Transition zone
(not considered in this
report)
Figure 3: Scheme of the BEM concept
In Figure 3 the “0” frequency reference is intended as the central frequency between contiguous blocks, either having a
fixed external guardband assigned in between or, in case of immediately adjacent blocks, having a corresponding wider
block bandwidth assigned for operators to manage the best an internal guard band between them.
The transition zone, shown between X
1
and X
3
power density levels, is not considered in this report, neither in terms of
IPF or NFD, but should be determined through a tradeoff between the need of a limited amount of guardband, the
possible slight decrement of NFD available for the blockoutermost system channels and the practical feasibility of
equipment filtering for reaching the required X
3
level from inblock P
max
density, which is generally defined for suitable
service and sharing conditions in the licensing process. Annex 3 give some more consideration on the definition of the
transition zone.
It should also be noted that equation 9 assumes that the receiver filters of equipment are matched to the transmit
requirements (i.e. are able to reduce the adjacent block interfering carriers power down to a level not impacting the
receiver linearity or the total amount of inband interfering power). On the other hand, ETSI ENs do not give any
specific requirement or guidance for receive filtering but only minimal C/I rejection capability, leaving to manufacturer
freedom for its best practice, also considering that in the R&TTE Directive spirit receiver performances should be
dominated by market forces.
For convenience, in the following sections, where specific numerical examples are made on the base of representative
system characteristics defined in Table 1 the parameter X
3
only is used with the understanding that NFD is easily
derived from equation 9.
2.1.3.1 Channel arrangements
Prior to evaluating BEM requirements, possible channel arrangement should be analysed, as offered by CEPT/ERC
Recommendations 1403 (3.43.6 GHz) and 1208 (3.63.8 GHz).
Both recommend assignments based on "number of slots" 0.25 MHz wide; apparently only symmetric assignments are
foreseen and no specific mention is made of internal (goreturn) guard band except for the fact that in 3.413.6 GHz the
"odd" 10 MHz automatically create a ~ 10 MHz guard band. However such guard band disappears for the 3.53.6 MHz
(50 MHz duplex) and for all 3.6  3.8 GHz.
That means that unless specific number of slots are reserved in both go and return subbands (wasting at least half of
them), adjacent TX/RX interference is expected also for FDD systems.
ECC REPORT 33
Page 17
Moreover, the recommendations mention that:
"where a duplex frequency allocation is required, the spacing between the lower edges of each paired subband shall be
100 MHz"
and also:
"PMP equipment may be used having a duplex spacing other than exactly 50(100) MHz. However, such equipment
must conform to the limits of the block allocation as defined above."
These sentences and the fact that no recommendation on subband for CS and TS operation is made, clearly show the
intention to admit (on a non discriminatory way) TDD and FDD, symmetric and asymmetric systems.
It has been recently demonstrated by CEPT studies for 40 GHz band the best compromise method for allowing
flexibility and efficient use of the spectrum with the recommended symmetrical assignment, the deployment of
asymmetrical systems being made with mixed uplink/downlink subbands within the symmetrical assignment.
The above considerations and the small duplex spacing, lead to the conclusion that, unless the band should be assigned
for predefined technology (e.g. FDD only) and spectrum waste is envisaged for creating goreturn guard band, a mixed
TX/RX in nearly adjacent assignments should be in any case considered.
2.2 CStoCS interference
A "same area, adjacent frequency blocks" scenario will be assumed (Figure 4). CStoCS interference is particularly
dangerous, since it can cause unavailability of a whole sector
11
. Therefore a worst case analysis will be presented for it.
Both CSs are supposed to face each other in line of sight (worst case situation). The fading events (mainly due to clear
air multipath) are considered as completely uncorrelated. Rain attenuation is negligible at this frequency band.
R
d
Block 1 Block 2
Figure 4: CStoCS interference scenario
As far as propagation model is concerned, the allowed interfering IPF (or EIRP) is calculated for freespace propagation
only, since:
 The distance between CSs is in practice short enough to exclude spherical diffraction.
 Both antennas are in relatively high locations (30m in the example) even in the urban environment. In this case
the mean path loss predicted by the IEEE propagation model (see Annex 1) is near or lower than the freespace
attenuation.
It is further assumed that the allowed degradation of the victim receiver threshold due to interference is
11
In principle, it happens for the part of the time when the two CS are in opposite Tx /Rx modes. This will be 100% of the time in
the case of two FDD systems, at least for the innermost assigned blocks where the mitigation of predefined uplink/downlink duplex
blocks becomes ineffective. When at least one system operates in TDD mode it will be less than 100%. The actual interference
intervals will vary because the two CS T/R periods will not be synchronised. In any case the contribution to availability of
unsynchronised T/R period tends to be negligible when the multipath activity is large and propagation events last far longer than T/R
periods. Therefore this aspect of T/R period impact will not be taken into account.
ECC REPORT 33
Page 18
A
Threshold
= 1 dB, hence the allowed interference spectral density is:
I
S
= N
0
− 6 = −144 +NF −6 (dBW/MHz).
As in the 40 GHz case, it may be assumed that the victim receiver has selectivity that matches the IPF or the block edge
mask. Hence the main carriers of the adjacent block EIRP are always reduced below the interfering outofblock noise
floor so that their residual contribution is negligible; therefore, the required EIRP level in the central portion of the
adjacent block is:
X
3
− 92.4 − 20 log(RF) − 20 log(d) + G
RX
= −144 +NF −6
where: X
3
= P
outdensity
+ G
TX
−NFD represents, in EIRP density terms, the CS BEM outof block lowest requirement in
dBW/MHz within the central portion of the adjacent block (see Figure 3 and paragraph 2.1.3), RF is the frequency in
GHz, “d” the CS distance in km,
with the assumed system reference values shown in Table 1, a plot of the required X
3
value vs. “d” is shown in Figure 5,
giving obviously the same result for both A and B systems, having the same CS antenna gains (16 dBi).
Figure 5: Required CStoCS spacing for un coordinated deployment
If, as initial assumption, one would consider an outofblock emission limit of 50 dBm/MHz at the antenna connector,
one would get for X
3
:
X
3
= ÷ 50 ÷ 30 + G
antCS
= ÷ 50 ÷ 30 +16 = ÷ 64 dBW/MHz
This would allow a minimum uncoordinated distance of about 350m, which seems quite reasonable in a rural
environment, given the typical cell radius values shown in Table 2.
On the other hand a value of X
3
= ÷73 dBW/MHz would lead, for system types B in LOS CS to CS scenarios, to a
minimum uncoordinated distance of less than 100m, which seems more appropriate in a urban scenario where more
dense network and higher number of user terminals in NLOS, lead to smaller cells size.
It is assumed that when lower distances are concerned, there is a likely mutual shading between two elevated buildings
so close by; therefore the two CSs would be likely colocated on the same building. It is then assumed that common
practice of site engineering (e.g. vertical decoupling of few meters, for avoiding near field propagation effects
12
) is
implemented, recovering with sidetoside antenna decoupling the reduction in free space loss.
12
At 3.5 GHz, for a dish antenna of ~same gain (i.e. diameter D ~ 30 cm), the nearfield area is ~ 2m (i.e. ~ 2D
2
/ì); sectorial
antennas should not be far from this figure. It is assumed that, even if colocated, the antennas may have at least 35 m distance (in
height on the same pole). Other arrangement topologist are also possible with similar results.
0 0.5 1 1.5 2
75
70
65
60
55
50
45
d (km)
km
ECC REPORT 33
Page 19
Therefore, these values will be carried over for being validated through further analysis related to CS to TS interference,
in the next section 2.3 for having a suitably low ISOP or unacceptable C/I occurrence probability.
It should be noted that for urban scenarios, the above LoS evaluation is an absolute worstcase. The additional shading
attenuation probability is not a negligible factor and using specific propagation models helps prove this
(e.g. the IEEE 802.16 adopted one depicted in Annex 1.2 and in IEEE document available at
http://grouper.ieee.org/groups/802/16/tg3/contrib/802163c01_29r4.pdf ).
2.3 CStoTS interference
As for the cells coverage area, the CS to TS interference scenario should be analysed for the two cases of LOS and
NLOS propagation, see description of “rural” and “urban” scenarios in section 2.2.1.
2.3.1 Rural LOS scenario
The ISOP approach will be used, due to the random nature of this kind of interference. Also in this case complete
uncorrelation will be assumed between fading events affecting the "wanted" and the interference path.
There will be an area in the victim sector where the receiver threshold degradation will exceed the assumed 1 dB limit.
Its size and shape depends on the distance between CS's and the additional protection from the terminal antenna RPE.
Referring to Figure 5, we will label V the victim TS, W the "wanted" CS and I the interfering CS.
Figure 6: CS to TS interfering scenario
The area where the threshold degradation exceeds the 1 dB objective will be defined by:
X
3
FSPL
(IV)
+ G
antTS(V)
 A
uTS(IW)
> 144 + NF – 6
This formula is commonly used for rural scenario (where availability is dominated by Rayleigh deepfade phenomena)
considering the uncorrelation of deep fading events in the different paths..
A
uTS(IW)
is the additional attenuation given by the TS victim antenna RPE at an angle equal to the difference in azimuth
between the victimtowantedCS and the victimtointerfererCS path (assuming that the victim antenna is aligned at
boresight with the wanted CS).
CS (W)
CS (I)
TS (V)
dWI
Interf. Area (IA)
Antenna: TS 2
Interf. Area (IA)
Antenna: TS 3
ECC REPORT 33
Page 20
Using the TS 2 and TS 3 antenna classes (represented by typical ITUR F.1336 antenna RPE derived with G=16 and 20
dB, still maintaining fixed boresight gain of 16 dB), the "forbidden" interference area IA can be derived and are
represented in Figure 7.
Figure 7: Interference Areas (IA) for victim TS as function of outofblock EIRP density X3
and ITUR TS antenna RPE
Figure 7 is representative of the worst case AI experienced when the two W and I CSs are boresight aligned and at their
minimum distance (derived in section 2.1.3.2 CS to CS scenario).
ISOP is calculated following the approach in ECC Report 99, as ISOP = (1(1P
1
)
Nt
)*P
2
*P
3
where:
 P1 is the probability that one TS falls within the interference area where the margin degradation exceed a
predefined value (here assumed 1 dB) evaluated as P
1
= IA/A
sector
.
 N
t
is the total number of terminals deployed in a sector.
 P
2
is the probability that the attenuation from wanted TX to victim RX and from interfering TX to victim RX
are uncorrelated. In a propagation environment dominated by multipath we assume P
2
=1. This is valid for
“rural” scenario where the cell size is limited only by LoS propagation following Rayleigh statistics and
described in ITUR Recommendation P.530. For urban nearNLoS scenario this is not generally true; however,
due to its shortness, one of the two paths (the interference one) is still here considered LoS for actually
affecting the victim TS region. In addition the assumption P
2
= 1 is conservative.
 P
3
is the probability that operators use adjacent frequency blocks and equal coverage on the same area. It
depends on the number of available blocks, the number of operators, the relative area coverage and the number
of sectors per cell. Assuming 2 or 3 blocks (one per operator) and 4 sectors per cell, P3=1/6 for 2 operators,
P3=~1/2 for 3 operators.
Assuming N
t
= 64 or N
t
= 32 (considered representative of relatively wideband systems adopted in this frequency
bands). An average P
3
= 1/4, a few values of ISOP have been calculated as examples in Table 3a) and Table 3b) with
0,3
0,2
0,1
0
0,1
0,2
0,3
0 0,1 0,2 0,3 0,4 0,5
km
km
X3=64 dBW/MHz  TS 2 antenna class X3=73 dBW/MHz TS 2 antenna class
X3=64 dBW/MHz  TS 3 antenna class X3=73 dBW/MHz  TS 3 antenna class
ECC REPORT 33
Page 21
G
CS
=16dBi, G
TS
= 16 dBi and TS2 and TS3 antenna typical RPE (using typical ITUR F.1336 RPE derived with gain of
16 and 20 dBi, respectively).
Scenario
X3
(dBW/MHz)
TS
Antenna
d
UI
min
(m)
A
Sector
(km
2
)
IA
(km
2
)
P1 ISOP %
N
T
=64
ISOP %
N
T
=32
Rural 64 TS2 350 274.65 0.0445 0.0162 0.258 0.129
Rural 64 TS3 350 274.65 0.02924 0.01065 0.1698 0.085
Rural 73 TS2 100 274.65 0.006 0.0022 0.0352 0.0176
Rural 73 TS3 100 274.65 0.00397 0.00145 0.0231 0.0156
Table 3a): ISOP % as function of outof block EIRP density in some rural scenarios
÷System type A (4 states)÷
Scenario
X3
(dBW/MHz)
TS
Antenna
d
UI
min
[m]
A
Sector
(km
2
)
IA
(km
2
)
P1 ISOP %
N
T
=64
ISOP %
N
T
=32
Rural 64 TS2 350 141.03 0.0445 0.0315 0.499 0.251
Rural 64 TS3 350 141.03 0.02924 0.02073 0.3296 0.1653
Rural 73 TS2 100 141.03 0.00605 0.00429 0.0686 0.0343
Rural 73 TS3 100 141.03 0.00397 0.00282 0.045 0.0225
Table 3b): ISOP % as function of outof block EIRP density in some rural scenarios
÷System type B (16 states)÷
The above data are obtained for the worst case of W and I CS placement (boresight aligned and in closest position);
however the ISOP drops rapidly as the distance increases. Figure 8 shows two examples taken from those in Table 3.
Figure 8: ISOP as function of W and I CS distance
From the above it appears that:
 The use of Class 3 TS antenna reduce the ISOP by ~25% to 35% in comparison to Class 2 one.
dUI [km]
ISOP %
0,00
0,25
0,50
0,75
1,00
1,25
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
64 QAM (NT=64)
X3=64 dBW/MHz
Rmax=13.4 km
4 QAM (NT=64)
X3=64 dBW/MHz
Rmax=18.7 km
ECC REPORT 33
Page 22
 In rural LOS scenario an ISOP < 1% is already obtained with the less demanding NFD (or X
3
) limit of ÷64
dBW/MHz. Therefore the more critical parameter might still be the CS to CS interference.
2.3.2 Urban NLOS scenario
For the urban scenario, on quite lower cell size, availability will not be dominated by deep Raleigh fading but by NLoS
statistics (i.e. SUI channel models of the IEEE 802.16 adopted model) which are fully uncorrelated between wanted
(victim) and unwanted (interferer) paths.
Therefore, the system performance degradation from interference is not considered “noise limited”, but “interference
limited” whenever the C/I falls down a critical value for the assumed modulation format.
In the following evaluations we will conservatively consider CS with the following data:
 TX output power = 35 dBm
 Signal bandwidth = 3.5 MHz
 TX output power density = 29.5 dBm/MHz
 Antenna TX Gain (sectorial 90°) = 16 dB
 Antenna height = 30 / 40 m
 Modulation formats = QPSK / 16QAM
and TS relevant data as:
 NF = 8 dB (including cable connection to antennas)
 Modulation formats = QPSK / 16QAM
 Typical S/N @ BER=10
6
for coded modulation:
 QPSK = 11.2 dB
 16QAM = 18.2 dB
 Typical C/I for 1dB threshold degradation at BER=10
6
:
 QPSK = 18 dB
 16QAM = 25 dB
 Typical C/I for BER=10
6
in unfaded conditions:
 QPSK = 12 dB
 16QAM = 19 dB
 Antenna Gain = G
TS
=18 dB (ETSI TS3 typical RPE, directional rooftop/high balcony outdoor deployment,
typical for high coverage of business high quality connections). For the calculations the typical ITU.R F.1336
radiation pattern will be used.
 Antenna Gain = G
TS
=4 dB (omnidirectional desk indoor deployment, typical for generic purpose residential
IP connections with less coverage/quality requirements). For simulating the actual random building/indoor
propagation losses, which are also different for the C and I path due to different average arrival directions, a
90° sector RP with random orientation will be used, dropping the antenna gain by the 5 dB minimum losses
(i.e. G=÷1 dB).
 Antenna Gain = G
TS
=10 dB (ETSI TS2 typical RPE, directional indoor window deployment, intermediate
situation deployment). For the calculations the typical ITU.R F.1336 radiation pattern will be used.
 Antenna height = h
TS
=4 / 15 m (see also considerations in the Propagation model section)
Propagation models:
As in the objective for the revision we will use the IEEE 802.16 NLOS model described in Annex 1 section A1.2.
Having to manage rooftop outdoor and indoor (ranging from ground floors to higher floors) situations, we need to
differentiate also the propagation model parameters to be used.
ECC REPORT 33
Page 23
It should also be noted that, even if antennas heights are a free parameters in IEEE formulas, formally the models are
valid for TS heights from 2 to 10 m; therefore, conservatively we will not use TS locations higher than 15m.
We would rather use two different Erceg Categories and SUI models, for representing more or less favourable
propagation conditions.
With the above assumptions we have selected two specific models with two variant each:
Business case (directional TS outdoor):
ErcegCategory B+SUI4 as worst case (lower floors or more obstructed situations)
Erceg Category C+SUI2 as best case (higher floors or less obstructed situations)
Shadowing variance: σ = 8 dB
Additional Attenuation = 0 dB (no building entrance loss)
Residential case (omni/low directivity TS indoor):
ErcegCategory B+SUI4 as worst case (lower floors or more obstructed situations)
Erceg Category C+SUI2 as best case (higher floors or less obstructed situations)
Shadowing variance: σ = 10 dB
Additional Attenuation = 5 dB (building entrance loss with window or desk antenna, in any case likely close to a
window) or 15 dB (less favourable indoor location conditions)
TS antenna height:
Three cases are only considered for the C/I evaluation, 4m, 10m and 15m
For higher floors situations, that would in any case fall outside the Erceg model applicability (limited to TS height
~ 10m), they are considered conservative; moreover, being the terminal location, and then the propagation model,
common to wanted and unwanted signals, it will be shown that C/I statistic is practically independent from TS height
and propagation models applied.
The following
Figure 9and Figure 10 show examples (for Erceg Category C at two different h
TS
height) of the attenuation distribution
of the received power due to shadowing at various occurrence probability versus distance, while Figure 11show the
additional distribution of “fastfaded Rice attenuation” for the two selected typical in SUI2 and SUI4 channels,
respectively.
The latter curves are obtained as average of about 10
6
channels trials; in figure the average curve of the two models is
also reported.
In NLOS scenarios there are many possible combinations of parameters to obtain such channel descriptions; in
particular a SUI2 result into a model suitable for a light multipath scenario instead a SUI4 results into a model suitable
for a scenario characterized by strong multipath. Practical experience in the mobile services shows that SUI4 model is
quite severe.
10
1
10
0
10
1
40
60
80
100
120
140
160
180
200
D [km]
A
t
t
[
d
B
]
Path Loss @ 3.5 GHz
NLoS model
Erceg C
hcs = 30 m, hts = 2 m
Shadowing std dev = 10 dB
Free Space
Mean NLos
85% shadowing
95% shadowing
99% shadowing
Figure 9: Example of shadowing attenuation variance
(h
TS
=2m).
10
1
10
0
10
1
40
60
80
100
120
140
160
180
200
D [km]
A
t
t
[
d
B
]
Path Loss @ 3.5 GHz
NLoS model
Erceg C
hcs = 30 m, hts = 15 m
Shadowing std dev = 10 dB
Free Space
Mean NLos
85% shadowing
95% shadowing
99% shadowing
Figure 10:Example of shadowing attenuation variance
(h
TS
=15m)
ECC REPORT 33
Page 24
Figure 11: Examples of Rice fading statistics
Additional antennas and path loss simulation methodology for the indoor TS evaluation
 The antenna radiation patterns are taken into account only in azimuth; this might be marginal for the evaluation of the
“wanted” C, while for the Interfering signal it might be too conservative, even in NLOS propagation, provided that the
most critical cases are TS far from the wanted CS (lower average elevation of C arrival angles) and close to the I CS
(higher average elevation of I arrival angles).
 In case of omnidirectional TS antennas, here used only for indoor applications, there was sensible contribution to the
results from TS that appears to be within the side lobes of the Interfering CS.
At first glance it might look correct; however, it certainly would have been if the antenna would be outdoor, while for
indoor use additional considerations should be made:
The IEEE 802.16 does not cover indoor applications; other available models treat only indoortoindoor propagation.
The widows themselves would act as lateral shielding for the omni antenna so that it might be better described as low
gain “sectorial”;
Nevertheless, the orientation of such antenna should be rendered random on a 360° azimuth (due to the random
position of the window closest to the TS, with respect to the principal direction of the main signal. With this
modelling, some of the cases considered in the C/I statistic would have been discarded for lack of wanted signal
(i.e. the area coverage would become lower for lack of signal and not for excess of C/I);
In this section the same propagation models and system parameters will be used for evaluating the occurrence
probability of C/I ratio between a wanted and an unwanted CS stations.
The interference scenario is described by the following Figure 12.
ECC REPORT 33
Page 25
Figure 12A: Generic model for CSTS interference (Case A)
Figure 12B: Generic model for CSTS interference (Case B)
In principle, wanted and unwanted sectors CS will be also randomly located and the sectors boresight also randomly
oriented.
For simplification, we will study only the case with sectors aligned (when sectors are differently aligned, the interfered
sector will be in any case wholly interfered by one or another sector of the interfering cell).
Case in Figure 12A has proven to be worst case when directional TS antennas are concerned, while case of Figure 12B
would be more severe for applications with TS omnidirectional antennas.
ECC REPORT 33
Page 26
The interfering CS is moved along the axis.
In NLOS propagation; even with CS collocated, their distance, in elevation or in roofspace, will be at least of several
wavelengths so that the statistic components of Shielding and Rician attenuation of wanted RSL and unwanted
interference (I) could be considered uncorrelated.
2.3.2.1 Methodology
The objective is to define the EIRP discrimination between adjacent blocks (e.g. in term of block mask or NFD) in
order to offer a suitable protection between blocks.
The RSL of the possible active Terminal Stations of the wanted cell would be randomly distributed according the
selected propagation model, therefore only few will have RSL close to the threshold, while most will have higher RSL;
the C/I by acceptable those stations would then range from the C/I producing a small degradation of the threshold (e.g.
1 dB was assumed in present Report 33) up to the critical C/I producing BER >10
6
in unfaded condition.
When indoor omnidirectional antennas are concerned, the simulation of the random building entry and indoor losses
will be made artificially assuming the antenna RP as a 90° sectorial with a randomly assigned orientation; this will
result in privileging the C or the I paths according their different average arrival angle (i.e. resulting in more trials
discarded when the C path is unfavourable, according the procedure described in step 1 below). For introducing the
minimal building entry effect (i.e. 5 dB), the gain of the antenna will also be artificially reduced by 5 dB (i.e. G =
÷1 dB).
Conservatively we would consider, as target for all cases, only the C/I for 1dB degradation of 10
6
BER threshold.
The two CS are placed at distance D=0 m (CS colocated), R/4, R/2, ¾R and 0.99*R (cell border) and 1.5*R (only for
validate omnidirectional worst case).
As in the previous cell coverage evaluation, the wanted signals sector is populated by a pattern of TS location.
Using the same Monte Carlo method and propagation models the probability distribution of the RSL is derived.
However, unlike the previous case, here each trial resulting in RSL lower that the BER 10
6
threshold is excluded from
the final probability distribution because, if affected by an interference, it will be irrelevant.
Also in this case, the evaluation of the interblock NFD/EIRP protection requires a number of steps:
Step 1:
Evaluation of the statistic distribution of the wanted RSL (C) related to the shadowing:
For each TS the mean wanted path attenuation (Aw
50
) is evaluated.
Other ~300 independent Monte Carlo trials for the shadowing attenuation are generated.
Trials with RSL < BER 10
6
threshold are discarded.
The overall sum constitutes the distribution function of the wanted “shadowed” component of RSL (C
SH
).
Step 2:
Evaluation of the statistic distribution of the unwanted interference (I) related to the shadowing:
For each TS the mean interference path attenuation (Au
50
) is evaluated.
Other ~300 independent Monte Carlo trials for the shadowing attenuation are generated.
The overall sum constitutes the distribution function of the unwanted “shadowed” component of interference (I
SH
).
Step 3:
Inclusion of the Rician fading component in the overall statistic distribution of the wanted RSL (C):
For each TS the distribution function of wanted path attenuation, derived from the step 1, is convoluted with the
relevant SUI channel attenuation distribution.
The result constitutes the spatial and time distribution function of the wanted RSL (C).
Step 4:
Inclusion of the Rician fading component in the overall statistic distribution of the unwanted interference (I):
For each TS the distribution function of unwanted interference path attenuation, derived from the step 2, is
convoluted with the relevant SUI channel attenuation distribution.
ECC REPORT 33
Page 27
The result constitutes the spatial and time distribution function of the unwanted interference (I).
Step 5:
Evaluation of the overall probability of the C/I:
The overall C and I distribution for each TS are convoluted for defining the probability distribution of C/I.
The mean value of the probability distributions for all TSs is calculated.
The cumulative probability function of C/I is calculated
The result constitutes the mean spatial and time cumulative probability function of C/I < x dB.
Step 6:
Evaluation of the required interblock EIRP/NFD protection:
The two CS are placed at distance D=0 m (CS colocated), R/4, R/2, ¾R and 0.99*R (cell border) and 1.5*R (for
omnidirectional case only in order to validate the worst case situation).
2.3.2.2 Results:
2.3.2.2.1 C/I statistics
The following simplifications have been used for the simulations.
 The antenna radiation patterns are taken into account only in azimuth; this might be marginal for the evaluation of the
“wanted” C, while for the Interfering signal it might be too conservative, even in NLOS propagation, provided that the
most critical cases are TS far from the wanted CS (lower average elevation of C arrival angles) and close to the I CS
(higher average elevation of I arrival angles).
 In case of omnidirectional TS antennas, here used only for indoor applications, the random building/indoor
attenuation, likely different for the C and I paths, are taken into account with the methodology described above.
The following Figure 13 shows examples of the cumulative distributions of the C/I derived with the above
methodology.
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
D = 0
D = R/4
D = R/2
D = 3/4*R
D = 0.99*R
60 55 50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
D = 0
D = R/4
D = R/2
D = 3/4*R
D = 0.99*R
(GTS=18 dB) (GTS=4dB + random building loss 5÷40 dB)
Figure 13: C/I distribution (Erceg Cat B, SUI4, h
CS
=30m, h
TS
=4m) versus D on a 2 km cell (Case A)
In Annex 4 of this document the value of C/I exceeded for more than 99% and, for omnidirectional indoor case, also
for 96 to 99% of the trials are reported.
In addition, provided that the calculations depend on a significant number of input data, and actual situations might
contains various mixture of those data, analysis of the variations produced by each data are reported in next Figures;
(Note that basic data, whenever not subject of the variation are those of Figure 13, Erceg Cat B, SUI4, h
CS
=30, h
TS
=4,
G
TS
=18, R=2km cell, CS distance R/4).
Of particular importance are the last two Figure 19and Figure 20that show the impact of having different W and I CS
height, which variation is not elsewhere considered.
ECC REPORT 33
Page 28
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
Erceg B  SUI4
Erceg C  SUI2
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
h = 4
h = 10
h = 15
Figure 14: Variations with overall propagation model Figure 15: Variations with h
TS
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
G = 1
G = 10
G = 18
Figure 16: Variations with G
TS
( G= ÷1dB represent the omnidirectional indoor case)
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
R = 0.5
R = 1
R = 2
R = 2.7
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
h = 25 m
h = 30 m
h = 35 m
Figure 17: Variations with cell radius
Figure 18: Variations with h
CS
(W=I)
ECC REPORT 33
Page 29
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
h CS
U
= 25 m
h CS
U
= 30 m
h CS
U
= 35 m
Figure 19: Variations of Wanted h
CS
50 45 40 35 30 25 20 15 10 5 0 5 10 15 20
10
3
10
2
10
1
10
0
C/I [dB]
P
r
o
b
a
b
i
l
i
t
y
h CS
I
= 25 m
h CS
I
= 30 m
h CS
I
= 35 m
Figure 20: Variations of Interfering h
CS
2.3.2.2.2 Interblock protection evaluation
2.3.2.2.2.1 Directional TS antennas
Provided that the average number of TS is constant over the sector (being physically limited by the handling capability
of the system and not by the sector area), from Annex 4 it might be seen that the result depends in practice from the W
and I CS positioning relative to the cell radius (D/R ratio) and only marginally from G
TS
; the propagation model and TS
height have nearly no impact.
The worst C/I cases reported in Annex 2 are concentrated, with interfering CS situation of Figure 4A, where the
interfering CS is at 3/4 of the wanted cell; here a worst C/I ~ ÷49 dB with 99% probability is evaluated (for antenna
gain 10 dB, Erceg Cat. B/SUI4) and are practically independent from any parameter a part the antenna gain; 18 dB gain
antennas resulted in better results with worst C/I~ ÷44 dB. With interfering CS situation of Figure 4B, the results are
sensibly better by several dB.
For defining the Required interblock protection, we should compare it with the minimum C/I for 1 dB degradation
(16QAM): ~+25 dB
We should now consider that the 99% occurrence probability is evaluated on the basis that ALL TS locations and
antennas are in the same conditions; in reality, for a given cell size the distribution of antenna types and TS location
would be random among the various cases considered (~30 for each antenna types for a total of ~60 cases with
directional antennas).
From Table 35to Table 38of Annex 4 it may be seen that for a given cell size only 6 cases have actually a C/I ~ ÷49 dB,
next 9 cases have C/I ~ ÷46 dB other ~45 cases have ÷44 < C/I < ÷10 dB.
It seems therefore correct to say that real 99% of the cases would happen with an average C/I among those cases. We
would conservatively assume a factor of 60/15= 4 reduction, corresponding to 6 dB of C/I improvement.
In addition we should consider safeguard margins for some differences in CS EIRP and heights between wanted and
interfering systems:
 Max ΔEIRP W/I = ÷5 dB
 Additional allowance for Δh
CS
W/I: ÷5 dB (from Figure 19 and Figure 20)
With these assumptions the required protection could be evaluated as:
Interblock protection = 25 ÷ (÷49 + 6) + 5 + 5 = 78 dB
Considering that the CS EIRP density considered in the study (see CS data) is +45.5 dBm/MHz, the Out of block EIRP
for protection of adjacent block TSs is:
Out of block EIRP = +45.5 ÷ 78 = ÷33.5 dBm/MHz
It should be considered that this value is still conservative for a number of factors:
 The comparison is made using the C/I for 1dB threshold degradation, while a number of the cases with higher
average RSL might be affected only if C/I exceed the “critical C/I” that for any modulation format is assumed to
be ~ 7 dB lower.
ECC REPORT 33
Page 30
 Being the worst cases concentrated around an area at ¾R of the cell it is likely that a number of those cases already
use a 4QAM format that would require less protection.
 We have not specifically considered the 64 QAM modulation format that would require ~7 dB more stringent
requirements; however, it is understood that systems offering 64QAM will be of “mixedmode” nature with
dynamic change of formats on a per TS basis. Therefore, only the TSs closer to the CS
W
or those with very
favourable propagation will be addressed with this format. In these cases the expected C/I distribution would not
be the worst one (e.g. related to lower W to I distances or better propagation situation in Annex 2).
 It contains 10 dB margin for W/I ΔEIRP and Δh
CS
that, in average, might not be present.
2.3.2.2.2.2 Omnidirectional indoor TS antennas
From the results in Table 39Annex 4, the use of omnidirectional antennas results in interference situation, at the 99%
of occurrence probability of worse C/I cases, is more severe of ~23dB .
This is physically expected consequent to the lesser gain/directivity; however, standing that we wish to consider only
indoor applications, a number of additional considerations, with respect to those made above for terminals with
directional antennas should be taken into account:
1. Systems offering indoor applications are assumed to be necessarily of “mixed mode” type, with dynamic
modulation formats management (actually any new generation BWA system will have this characteristic).
The worst case appears to be when interfering CS is at the wanted cell border; the affected TS are likely the
ones close by, which, being also at cell border will already in average require QPSK format (most applications
foreseen even BPSK modulations offering additional safeguard margin).
Therefore, the ~7dB more robustness to interference (kept as additional safeguard in directional antennas
applications) may be taken into account.
2. The indoor applications, for their own nature, are considered “best effort” service; therefore lower performance
objectives (from the point of view of coverage and availability) are justified. If we look at a less demanding
objective of 96% of worse C/I cases occurrence probability, from Figure 66in annex 4, ~ 17dB (from a C/I ~
÷72 at 1% to C/I ~ ÷55 at 4%) are immediately recovered.
This should not be seen as an absolute degradation of the performance of the interfered system, the 4%
degradation is only relative to actual percentage of coverage of indoor terminals, which is already assumed to
be significantly lower than that of outdoor applications with directional antennas (e.g. if the expected cell
coverage, at 99% availability, for indoor/omnidirectional terminals is 80% in no interference, the coverage
will further drop for adjacent block interference of 4*0.8= 3.2 % resulting in a total 76.8% expected cell
coverage with 99% availability). In addition, being any system generally deployed with a mixture of indoor
and outdoor TS, the overall coverage degradation should be further mediated over all cases.
The above two points justified that it is reasonable to assume that the 23 dB worse C/I situation for indoor omni
directional antennas with respect to conventional outdoor directional cases, may be duly compensated by the
~7 + 17 ~ 21dB, as shown.
Therefore, an overall discussion of the differentiate applications and objectives and of cost/benefits is reported in next
section.
2.3.3 Conclusions for CSTS protection in urban NLOS environment
This new, more complex evaluation, based on real C/I occurrence probability following a certified NLOS propagation
model, gives results that, for directional antennas, are ~ 10 dB less demanding than the value of EIRP previously
evaluated with a more empirical propagation model and based on deterministic areas where RSL thresholds might be
exceeded due to interference levels only.
This result was expected since the actual fade margin due to dynamic propagation (i.e. the Rician fade) is sensibly lower
than Raleigh multipath expected in freespace. Most of the overall system fade margin is dedicated to provide for the
“shadowing” attenuation that is somehow static, once the CS and TS locations are fixed.
Therefore, comparing the interference level with the BER thresholds in NLOS propagation environment will result in
very conservative scenario; actually most of the TS “in deep shadow from their CS” will have RSL significantly higher
than the threshold, without need of very high additional FM.
During preparation of the first published version of this report, the expected scenario for the deployment of FWA in 3.4
3.8 GHz has changed and a wide interest on this kind of applications are now reshaping the market expectations:
 The market, for rural as well as for suburban still highly populated areas, is addressing broadband services
only. This would, in practice, greatly reduce the system characteristics spread (in term of significant power
ECC REPORT 33
Page 31
density and bandwidth options) among various implementations. The recent draft revision of draft
ECC/REC 0405 after public consultation have highlighted the wide preference for typical applications with
channel bandwidth of 3.5 MHz or higher.
 “Conventional” PMP fully outdoor systems, originally designed for mostly LOS (or moderately obstructed
paths requiring professional intervention also for TS deployment), are migrating towards more sophisticated
performance in strong NLOS environment (allowing simpler TS deployment, ideally by the user alone)
 Standards organisations (ETSI/BRAN, IEEE 802.16, WiMAX) have published standards and conformance
tests including multivendor interoperability options based on the above market expectation.
 From administrations point of view, a wide number of consultations on the use/regulations of bands around
3.5 GHz have been made or are in progress in CEPT and worldwide. Most of them are considering the
redefinition of rules assuming only adhoc provisions for “grandfathering” of existing old licensee.
 previous responses from industries show great difficulty in matching the proposed “urban” blockedge mask;
the new focus on potential wide consumer market suggests the adoption of less demanding as possible
regulations for stimulating development and deployment of initial networks.
 Originally the two separate outofblock eirp levels have been considered, according a rural or urban
environment, respectively; however, previous responses from administrations have highlighted the difficulty
in managing two values from regulatory and manufacturers point of view. Therefore it would be advisable to
have just one definition of limits for all application scenarios.
The introduction in the study of indoor TS with omnidirectional applications have shown that their higher sensitivity,
with respect to more conventional applications with directional antennas, might be somehow balanced by their lower
quality objectives on an overall mixture of cases, containing indoor and outdoor applications, with more or less severe
NLOS.
Therefore this report used an assumption that a single mask, without distinction between urban and rural scenarios,
should be suitable, and the outofband EIRP limit of 73 dBW/MHz would be appropriate to ensure coexistence of
both CS>TS and CS>CS, the latter being the defining critical case.
Then, if the BEM were expressed in terms of transmitter output power, the outofband limit would become 73G
a
=
7316=89 dBW/MHz.
2.4 TS to CS interference
2.4.1 Interference evaluation
This evaluation would lead to setting the required NFD (or the X
3
value of the blockedge mask) for TS.
However, the evaluation might be based only on a statistical IPF, common to the evaluation made in the section devoted
to the IPF and guardband methodology (see Annex 2) and its details are there reported.
From the detailed evaluation made in Annex 2, the required NFD or “outofblock” EIRP density for suitably low
(<1%) probability of TS interfering a victim CS, may be summarised, for the worst cases presented, in Table 4
depending on the assumed TS antenna ITUR RPE:
TS antennas class Typical gain (dBi) Required TS “X
3
” value
(outofblock EIRP density)
TS 3 18 ÷ 35 dBW/MHz
TS 2 16 ÷ 37 dBW/MHz
TS indoor Omnidirectional
(no ETSI standard
available)
4 ÷49 dBW/MHz
Table 4: Outofblock EIRP density requirement
for ~ 1% of TS to CS interference probability in urban scenarios
ECC REPORT 33
Page 32
2.4.2 Conclusions for the TS to CS interference study
The expected more stringent requirements for less directivity antennas suggests that the possible BEM limit would be
more suitably defined in term of absolute outofblock power density rather than outofblock EIRP.
In such way a single power density requirement will automatically adapt to the necessary EIRP according the used
antenna, which gain is universally intended to be inversely proportional to antenna directivity.
This would lead to a single value of outofblock power density, derived from the more stringent EIRP requirement
detailed in Annex 2 (A2.2.6) (reduced by the assumed omnidirectional gain of 4 dBi), of ÷53 dBW/MHz at the antenna
connector.
However, the results in the above Table 4are only incidentally linear with down to omni antenna gain; actually in this
case also the additional indoor to outdoor propagation attenuation has been included (evaluated to be at ~ 15 dB in
average).
Therefore the limit of ÷53 dBW/MHz should be assumed valid for omnidirectional use in prevalent indoor
situations; whenever the prevalence would be expected for outdoor situations then some 15 dB more stringent
value should be required.
2.5 Terminal Station to Terminal Station
This kind of interference, unlike the cases where a CS is involved, would only impair the operation of one terminal;
therefore a worst case approach is considered as too stringent. Due to the random nature of this kind of interference, a
statistical Monte Carlo approach seems more adequate. A possible scenario could be two CSs located at both sides of
the boundary, each pointing into its own service area. Each TS will have its EIRP level set to deliver a signal to the CS
6 dB above the receiver threshold, except one (different in each trial) which would have maximum EIRP, to simulate
the occurrence of a fading event.
However this effect, at least for PMP architectures with directive TS antennas, is not generally considered a limiting
factor for coexistence (when compared to the other interference cases). This evaluation may require further work in case
of mixed PMP and MESH architectures or TS with mobile (outdoor) characteristics are possibly considered.
3 ”ADJACENT AREA  SAME FREQUENCY BLOCK” INTERFERENCE SCENARIO
3.1 Power Flux Density Limits for adjacent FWS service areas
This section focuses on initial interoperator coordination guidelines that would support assignments to FWA operators
in the same frequency block adjacent geographic areas. These guidelines consist of service boundary PFD limits to
assist frequency reuse between neighbouring service areas. The PFD limits are linked with a coordination distance,
which is the distance from the service area boundary within which transmitter stations should be coordinated with
adjacent area operators.
The methodology used in this report follows the same approach as for the 40 GHz band, which was base for the relevant
Annex 4 of draft ERC Recommendation (01)04.
The specific propagation behaviour in 3.5 GHz band was taken into account; in particular the spherical diffraction
attenuation has been introduced as function of the antenna height. Due to the relatively large radius of first Fresnel zone
(~ 50 m) and the typical horizontal pointing of FWA antennas, the spherical diffraction attenuation will play significant
role in defining the respective area and the PFD level for triggering coordination.
The study here presented assume PMP FWS services deployed over flat land, mostly LoS conditions, so that only well
known free space and spherical diffraction propagation models apply. Therefore, the results represent only an example
for this territory and propagation assumption.
However, it might be considered an upper bound for most situation, whenever over the service borderline are not
present quite different geographical situations such as:
 high mountain chains acting as physical barriers, then increasing the path blockage situation and the
NLOS attenuation.
 broad depressions (more than ~10 km wide), e.g. river valley, somehow counteracting the spherical
diffraction effects, then reducing the path attenuations
In these cases, the methodology here offered might still be applied provided that suitable modification of propagation
model are adopted.
ECC REPORT 33
Page 33
The findings of this section may be as follows:
FWA Central Stations (CS) transmitters should be coordinated when the PFD generated at the network’s service area
boundary exceeds the value of PFD (dBW/MHz/m
2
) shown in Figure 24.
Due to the determinant contribution of spherical diffraction, the coordination distance and PFD at the boundary
strongly depend on the antenna (interfering TX and victim RX) heights.
The values derived from Figure 24 can be used to determine coordination distances over flat land. For typical values of
EIRP expected in the 3.5 GHz band, coordination distances are evaluated as ~ 60 to 80 km for PMP CS (see
Figure 23).
For terminal stations’ (TS) EIRP, being similar to that of CS, there is no practical difference apart from the typically
lower height of their antenna.
A reference EIRP = 20 dBW/MHz and antenna heights 20 to 50 m for CS and 10 to 40 for TS are used for this
evaluation.
The range of distance and relevant PFD may be reduced or fixed in case administrations may wish to limit, as close as
the radio site is to the service border, upperbounds for both EIRP and antenna heights above the ground or to define CS
downtilt angles in case that height is exceeded.
This section reviews the methodology behind these figures and proposes the principle of boundary PFD limits as an
appropriate means of controlling the interference environment between operators assigned same frequency block(s) in
neighbouring geographical areas.
The proposed methodology might be also suitable for FSS coordination.
3.1.1 Assumptions
In order to cater for the variety of technologies possible for FWA no assumptions were made regarding duplex method
or multiple access method. To generate the broadest of guidelines the assumption was merely that an interfering
transmitter is deployed in one service area and a victim receiver is operating on the same frequency, but located in an
adjacent service area.
In Table 5, equipment characteristics are reported for interference analysis and for a consequent tentative technology
independent regulatory framework. Those values are not regarded as “typical” for most current system available on the
market, but cater for due allowance for some special cases and possible further technology developments.
ECC REPORT 33
Page 34
Nominal channel bandwidth: 7 MHz
13
Central station EIRP: 20 dBW/MHz
14
Central station antenna gain: 18 dBi
15
Central station antenna radiation pattern (90º): EN 302 085 class C2
Central station antenna height 20 to 50 m
16
Terminal station EIRP
TX
: 20 dBW / MHz
17
Terminal station antenna gain 18 dBi
Terminal station antenna 3dB beam width ~ ±10°
Terminal station antenna radiation pattern: EN 302 085 class TS2
Terminal station antenna height 10 to 40 m
18
Typical Central Station and Terminal station receiver
threshold (10
6
BER)
84 dBm (4QAM)
76 dBm (16QAM)
Nominal ATPC regulated uplink receiver level 6 dB above 10
6
BER threshold
Receiver noise figure 8 dB
19
Interference limit (kTBF – 10 dB)
20
146 dBW / MHz
Table 5: Summary of system characteristics assumed for defining the proposed regulatory framework
In addition the following propagation characteristics have been assumed:
 Line of sight path unless otherwise stated.
 No atmospheric attenuation at 3.5 GHz.
 Spherical diffraction attenuation (1
st
Fresnel zone partially obstructed due to limited antenna height) calculated
following ITUR Rec. P.562.
 ATPC effect at 3.5 GHz should also be taken into account; however, it is assumed that ATPC, in these lower
bands, will be operated by multipath and not by rain, therefore correlation between interfering and victim paths
attenuation is negligible.
13
This channel spacing is considered the most representative for being used in the calculation. It is considered that the larger channel
systems would determine the coexistence rules. Nevertheless lover spacing channels (e.g. from 1.5 MHz up), also widely popular,
should more easily fit in that possible framework.
14
This value includes allowance for feeder losses for full indoor applications. The assumed EIRP is intended to make room for the
highest values allowed by present technology, used in particular applications (e.g. very large coverage in remote areas or when NLoS
area should be covered at best), nevertheless network considerations would generally lead to lower EIRP. In this case it is also
intended that the latter systems would more easily meet any regulatory limit.
15
Even if antenna gain might be slightly lower in typical applications, antenna technology is in fast evolution; therefore 18 dB has
been used for taking into account a not infrequent worst case, while 16 dB has been assumed as typical value in previous section of
this report dealing with “same area – adjacent block”.
16
Antenna height would impact the cell coverage but also the pfd at area boundaries. It is current practice for limiting the latter,
when high antenna location is used, to downtilt the antenna itself for remaining in the boundary pfd limits set by the Administration.
17
This is the worst case, assuming symmetrical uplink/downlink capacity.
18
In principle there should be no limitation to TS antenna height, it being dependent on the customer location. However, the same
consideration made for CS antenna regarding the higher value still applies. For the lower limit, we should consider that second
generation FWA systems might employ techniques which enable them to operate without a clear LoS path. The desire for low cost,
simple (self) installs has resulted in system performance being improved to allow the TS to be deployed within buildings. Hence, TS
heights may be less than 7 meters, and are rarely higher than 2 meters above the subscribers’ building height.
19
Typical front ends noise figure in this band are lower (e.g. ~5 dB). The 8 dB value included allowance for feeder losses and
possible narrowband filters for enhanced selectivity required by dense environment as assumed in this report.
20
For the “adjacent area same frequency block” scenario a more stringent requirement is used (i.e. frequency reuse by another
operator should be more protected than when operators use adjacent blocks of frequency).
ECC REPORT 33
Page 35
3.1.2 Methodology
The PFD threshold has been determined assuming a single interferer and unobstructed LOS, directly aligned path
between interferer and victim essentially a “minimum coupling loss” approach. The PFD limit is then used to derive an
appropriate maximum coordination distance.
The threshold can then be tested using Monte Carlo statistical analysis to check its validity in a typical multiple
interferer environment.
3.1.3 Central Station to Central Station
The generic scenario for the definition of PFD limits at service boundary between two operators reusing the same
frequency block is shown in Figure 21. Operator “B” is assumed to be deploying service, while operator “A” represents
the potential “victim” receivers.
Figure 21: Defining PFD limit at geographical block assignment boundary
3.1.3.1 Worst case single interferer scenario: 3.5 GHz calculations
Assuming a 18 dBi victim antenna gain, the minimum separation between the two CSs (R
min
) vs. the interfering station
EIRP
int
, can be derived from the link budget equation, i.e.,
P
RX
= EIRP
int
– FSPL – A
sph
+ G
RX
where P
RX
is the interference power at the receiver input
FSPL is the free space path loss =20 log(4πR
min
/λ)
A
sph
is the spherical diffraction attenuation depending on the heights (ha and hb) of the two
CS antennas relative to the ground. This has been calculated following ITUR Rec. P.562 and
approximated as:
A
sph
= 0 dB D < D
0
(km)
A
sph
= 1.3 (DD
0
) db D > D
0
(km)
D
0
is the maximum distance where the total calculated attenuation equals the free space attenuation
G
RX
is the receiver antenna gain in the direction of the interferer
P
RX
Max (dBW/MHz) =  146 = EIRP
int
– 92.5 –20log(3.5) –20log(D) – A
sph
+ 18
D
D/2
Boundary between
licensed areas
Operator “B” Interferer radiating
towards the victim in
neighboring area. Representative
EIRP assumed.
PFD “A”
Boundary PFD “B”
Operator “A” Victim receiver looking
towards the interferer.
Max acceptable interference = x dB
below thermal noise floor.
ECC REPORT 33
Page 36
Figure 22 shows the received interference power as a function of the separation distance from the interfering transmitter
for EIRP
int
= 20dBW/MHz and some different cases of antenna heights (ranging from 20m to 50 m). The curves for
different EIRP values can be obtained by simple shift of the same amount.
In Figure 22 flat terrain has been assumed and it shows that in case of different interferer/interfered antenna height, the
mean value of the two can be taken into account (e.g. ha=20m and hb=40m correspond to the case ha=hb=30m).
Flat terrain is assumed to be close to the worst case; it is not likely that operator boundaries lie along a relatively narrow
valley and, even in that case, antennas would be “groundgrazing” aligned.
0 10 20 30 40 50 60 70 80 90 100
220
200
180
160
140
120
100
80
60
Received interference power vs. separation distance
RF=3.5GHz, EIRP=20dBW/MHz
D (km)
d
B
W
/
M
H
z
1010
2020
3030
4040
5050
1030
2040
3050
Antenna
heights (m)
(3)
P
rec
MAX (1)
146
136
P MAX (2)
rec
(1) P
REC
MAX proposed for CSs and TSs at nominal operating EIRP (6 dB above threshold)
(2) P
REC
MAX proposed for TSs (with ATPC enabled) at maximum EIRP
(3) Each curve is valid also for any mixed antenna heights with the same sum value (e.g. 3030 is valid
also for 2040, 2030 is valid for 2525 and so on)
Figure 22: Received interference power vs. separation distance for the CS to CS interference scenario
(3.5 GHz, line of sight)
In Figure 22 two limits are shown. The first (146 dBW/MHz) is valid for little or no degradation of the victim CS
receiver.
The second (136 dBW/MHz) is proposed for TSs at the maximum EIRP (during the small percentage of time when
ATPC is required to operate to counteract multipath attenuation) as discussed later.
The minimum separation, required to meet the –146 dBW / MHz interference criterion defined above, between directly
aligned CSs under clear LOS air conditions is shown in Figure 23 as a function of EIRP, with the antennas height as
parameter.
Within practical antenna heights range (20 to 50 m) the minimum separation distance ranges from 58 to 80 km.
ECC REPORT 33
Page 37
0 5 10 15 20 25 30
30
40
50
60
70
80
90
EIRP(dBW/MHz)
D
m
i
n
(
k
m
)
Minimum separation between CSs vs EIRP
Parameter: antenna heights
1010
2020
4040
3030
5050
1030
2040
3050
Antenna
heights (m)
(*)
Figure 23: Minimum separation between CSs vs. EIRP
for the CS to CS interference scenario (3.5 GHz, line of sight)
If the required separation distance is apportioned equally between the two regions, this will require each operator to
ensure any CS directly aligned with an adjacent operator’s service area boundary is located at least (D
min
/2) km away
from the adjacent service area boundary.
The interference power produced by a CS D
min
/2 km away is calculated again as:
P
rec
(D
min
/2) = EIRP
tx
 FSPL(D
min
/2) A
sph
(D
min
/2) + G
rec
The PFD at this distance can be determined using the formula:
PFD = P
rec
 A
e
,
where:
A
e
= G
rec
+ 10 log(λ
2
/4π) is the receiving antenna effective aperture
A
e
=  14.3 dB m
2
evaluated at 3.5 GHz with G
rec
=18 dB
The PFD at D
min
/2 is shown in Figure 24 as a function of EIRP
tx
for different antenna heights.
Therefore the PFD at the service area boundary should not exceed the values derived from the above relationships, and
summarised in Figure 24.
Data in Figure 24 are obtained with P
rec
(D
min
/2) evaluated assuming a virtual receive antenna height at D
min
/2 to be the
same than that at D
min
(not taking into account any earth surface curving as shown in Figure 25).
The graphs in Figure 24 show crossovers that are due to the different slopes of lineofsight (linear with the distance)
and spherical attenuation (nonlinear with the distance). This impacts only the formal PFD evaluation (and its possible
measurement) at D
min
/2, while at D
min
distance the maximum received interference is still satisfied.
The impact of antenna height and EIRP for defining the boundary PFD coordination triggers is evaluated separately in
following section 3.1.3.2 and 3.1.3.3.
ECC REPORT 33
Page 38
0 5 10 15 20 25 30
105
100
95
90
85
80
75
70
Power Flux Density @ D
min
/2
EIRP(dBW/MHz)
P
F
D
(
d
B
W
/
M
H
z
/
m
2
)
1010
3030
2020
4040
5050
1030
2040
3050
Antenna
heights (m)
(*)
(*) The first value refers to transmit antenna
Figure 24: PFD (Ф) at D
min
/2 (half the minimum distance
derived from Figure 23 between CSs vs. EIRP
tx
Figure 25: Principle for evaluating PFD at D
min
/2 in Figure 24
ECC REPORT 33
Page 39
3.1.3.2 Antenna height
Provided that the separation distance is dominated by spherical diffraction attenuation, Figure 24 shows that different
mix of antenna heights produce little difference when the sum of the TX and victim RX antenna heights is the same
(e.g. 3030 is close to 2040, 2030 is close to 2525 and so on). Therefore values not shown could be extrapolated with
small potential error.
For this reason, the distance from the boundary (D
min
/2) may be managed with the assumption that neighbour
operator have equal antenna height, the suitable value for the boundary PFD (dBW/MHz/m
2
)
could be derived from
Figure 24 without knowing the actual CS antenna height of neighbour operator.
Rationale of this approach is shown in Figure 26, where the PFD at the boundary (D
min
/2) is plotted in function of the
TX antenna height, assuming equal height on potential RX victim side.
Operator “B”, once knowing its own tower height and EIRP, would keep its own minimum distance (D
min
/2)
B
evaluated
assuming the same antenna height for neighbour victim of Operator “A”. In fact, if the victim Operator “A” has higher
or lower antenna height, the difference in D
min
length will be balanced by its own (D
min
/2)
A
portion; the example in
Figure 26 shows that the distances adopted by operators “B” and “A” with antenna heights = 20 and 40 m, respectively,
are D
min
= 30m (“B”) and 38m (“A”) will result in a separation distance of 30+38 = 68 m ~ D
min
for 20/40 m in
Figure 23, therefore respecting the required separation distance.
Nevertheless the lower is the antenna height, the higher are the diffraction attenuation and all other attenuation due to
obstacles such as building, trees etc. generally reducing the probability of worst case occurrence; therefore, as
recommended by ECC Report 97, the use of low antenna heights should be encouraged at the boundary.
5
10
15
20
25
30
35
40
45
50
55
50 40 30 20 10 0 10 20 30 40 50
Dmin/2 (km)
A
n
t
e
n
n
a
s
h
e
i
g
h
t
(
m
)
Area Boundary
"Operator B" area "Operator A" area
EI RP = +2 0 dBW/ M Hz
Figure 26: Example of balanced D
min
obtained by different D
min
/2 for different antenna heights
(EIRP = +20 dBW/MHz)
3.1.3.3 Transmit EIRP
While the antenna height impact on D
min
is self balancing without knowing the victim receiver side (see example of
Figure 4.5), transmit EIRP, which also has impact on D
min
does not have any intrinsic balancing mechanism.
In principle, as far as Operator “B” reduces EIRP, Dmin becomes lower (see Figure 23) and the station, respecting the
PFD at (D
min
)
B
, could be placed closer to the boundary without coordination not risking interference to Operator “A” as
far as the latter uses an equal or higher EIRP for evaluating (D
min
)
A
; however the opposite is no longer true and Operator
“B” receiver would be interfered because being too close to the boundary.
Therefore, a “minimum” EIRP, for evaluating PFD trigger should be defined, actual lower EIRP should be
disregarded in this evaluation and replaced by the agreed value.
ECC REPORT 33
Page 40
3.1.3.4 Conclusions and possible selfregulation method for CSs coordination distance
Unlike what commonly happens in HDFS frequency bands, where line of sight applications give enough clearance from
1
st
Fresnel zone for not considering spherical diffraction attenuation, the above discussion has shown that, in the 3.5
GHz band, the coordination distance, besides EIRP
tx,
, depends on antenna heights of both interfering and victim CS.
In such a way an operator, according to its own actual deployed maximum EIRP
tx
and antenna height, and assuming
victim receiver antenna height at the maximum foreseen (e.g. 50 m), should:
# evaluate the minimum coordination distance (D
min
from Figure 23)
# verify that the PFD at D
min
/2 (service boundary) does not exceed the limits given in Figure 24.
This does not mean that CSs cannot be located closer than D
min
/2 to the boundary. However, the PFD at the boundary
should be no greater than that produced via an unobstructed path by a directly aligned transmitter radiating the same
EIRP. With an antenna height at a distance D
min
/2 from the adjacent service area boundary, in order to allow a similar
transmitter (with the same EIRP and with the tallest antenna mast) at the same D
min
/2 from service boundary.
In this case administrations may wish to limit CSs transmitters in both EIRP
tx
and maximum antenna height
(automatically limiting the maximum coordination distance) or to define downtilt angles in cases when height is
exceeded.
At closer distances to the boundary, additional protection in the form of reduced EIRP in the direction of the boundary
or shielding from terrain or other obstacles will be required. The extent of additional protection required would be
subject of further studies.
3.1.4 Terminal Station to Central Station
3.1.4.1 ATPC impact
The TS is assumed to have ATPC. Under normal conditions each TS is assumed to have its EIRP level set to deliver a
signal to the CS 6 dB above the receiver threshold.
Fadings from clearair multipath on interfering and victim paths are assumed to be uncorrelated. Actually, slight
correlation may be expected for directly aligned line of sight interference scenario but a very rough estimate of the
percentage of time, where both the useful and the interference path might be contemporarily faded, gives negligible
values (based on ITUR P.5308 paragraph 2.3.6).
3.1.4.2 Worst case single interferer scenario, 3.5 GHz calculations
For the worstcase interference scenario, it is assumed that the interfering TS is directed towards a CS located at the
network service area boundary, pointing into its own service area. The worstcase interference arises when the TS is at
the maximum distance from its CS.
This maximum cell size can be determined by considering the downlink power budget, assuming a CS EIRP of
20 dBW / MHz.
This evaluation, assuming multipath environment in rural (flat terrain) scenario, may be found in the previous section
“Same area – Adjacent Block scenario” of this report and is summarised, with the fade margin (FM) in Table 6 as
function of required availability.
Rural Scenario
Availability
System Type 99.99% 99.999%
R
max
(km) FM
0
(dB) R
max
(km) FM
0
(dB)
A 18.7 km 17.3 12.4 km 20.9
B 13.4 km 12.2 8.9 km 15.7
Table 6: Typical cell size in rural scenarios
Therefore, the worstcase interference scenario occurs when the interfering TS is at a distance D
int
= D
min
/2 + R
max
from
the directly aligned victim CS, where D
min
is derived from Figure 23 and R
max
can be taken from Table 6.
With the assumption made of fading uncorrelation, two requirements need to be considered:
ECC REPORT 33
Page 41
a) Interfering TS operating at the "normal" EIRP set by ATPC (unfaded percentage of time ~99.X %)
EIRP
ATPC
= EIRP
max
 FM
0
+ FM
ATPC
FM
0
is the fade margin corresponding to maximum transmitted power (from Table 6). In this case (most of
the time) the received interference power, into the victim CS, should not exceed the required limit (kTBF 
10dB) for not impairing the victim performance and availability.
b) Interfering TS operating at maximum EIRP (faded percentage of time ~ (10099.X) %)
Due to uncorrelation, the victim CS would receive normal level, depending on the availability objective and
the ATPC range, from the useful link (for a percentage of time usually less than 1%) In this case a higher
interference level can be tolerated without impairments.
Assuming that also victim system will work at 6 dB above threshold, we may tolerate up to 3 dB of noise
floor degradation (i.e. up to kTBF= 136 dBW/MHz).
Assuming the TS delivering the assumed maximum EIRP of 20 dBW / MHz, the received signal level at the
victim CS at this distance is derived from Figure 22, for the rural scenario.
3.1.4.3 Examples:
Example 1
Type B interfering system, height of interfering TS h
t
= 10 m, height of victim CS h
c
= 30 m, availability 99.99%.
R
max
= 13.4 km FM
0
= 12.2 dB D
min
= 68 km (for CS to CS interference assuming hc = 30 m on both sides)
D
int
= 13.4 + 68/2 = 47.4 km
EIRP
ATPC
= 20 – 12.2 + 6 = 13.8 dBW/MHz
From Figure 22 (at D
int
and scaled to the actual EIRP level) it is possible to derive:
For case a) an interfering power I =~ 132 – (20  EIRP
ATPC
) =  138.2 dBW/MHz
For case b) an interfering power I =~ 132 dBW/MHz.
Both these levels are higher than requirement.
In case a) in order to receive an interfering power of 146 dBW/MHz, the CS of the interfering system should be placed
at a distance Dx from the border, so that:
D(a) + R
max
+ D
min
/2 =~ 54 km (from Figure 23, at EIRP = 13.8 dBW/MHz)
D(a) = 54 – 13.4  34 = ~ 6.6 km
for case b), in order to receive an interfering power of 136 dBW/MHz (=kTBF), the CS of the interfering system
should be placed at a distance Dy from the border, so that:
D(b) + R
max
+ D
min
/2 =~ 52 km (from Figure 22 with P
rec
Max set to –136 dBW/MHz)
D(b) = 52 – 13.4  34 = ~ 4.6 km
Therefore the minimum distance where a CS (supporting far system type B TSs with height lower than 10 m and
victim CS height lower than 30 m) could be placed is 6.6 km.
Example 2
Type B interfering system, height of interfering TS h
t
= 20 m, height of victim CS h
c
= 40 m, availability 99.99%.
R
max
= 13.4 km FM
0
= 12.2 dB D
min
= 75 km (for CS to CS interference assuming hc = 40 m on both sides)
D
int
= 13.4 + 75/2 = 50.9 km
EIRP
ATPC
= 20 – 12.2 + 6 = 13.8 dBW/MHz
From Figure 22 (at D
int
and scaled to the actual EIRP level) we would derive:
For case a) an interfering power I =~ 121 – (20  EIRP
ATPC
) =  127.2 dBW/MHz
For case b) an interfering power I =~ 121 dBW/MHz.
Also in this example, both these levels are higher than requirement.
In case a) in order to receive an interfering power of 146 dBW/MHz, the CS of the interfering system should be placed
at a distance Dx from the border, so that:
D(a) + R
max
+ D
min
/2 =~ 64 km (from Figure 23, at EIRP = 13.8 dBW/MHz)
D(a) = 64 – 13.4  37.5 = ~ 14.8 km
ECC REPORT 33
Page 42
for case b), in order to receive an interfering power of 136 dBW/MHz (=kTBF), the CS of the interfering system
should be placed at a distance Dy from the border, so that:
D(b) + R
max
+ D
min
/2 =~ 62 km (from Figure 22 with P
rec
Max set to –136 dBW/MHz)
D(b) = 62 – 13.4  37.5 = ~ 11.1 km.
Therefore the minimum distance where a CS (supporting system type B TSs with height lower than 20 m and victim
CS height lower than 40 m) could be placed is 14.8 km.
Example 3
Type A interfering system, height of interfering TS h
t
= 20 m, height of victim CS h
c
= 40 m, availability 99.99%.
R
max
= 18.7 km FM
0
= 17.3 dB D
min
= 75 km (for CS to CS interference assuming hc = 40 m on both sides)
D
int
= 18.7 + 37.5 = 56.2 km
EIRP
ATPC
= 20 – 17.3 + 6 = 8.7 dBW/MHz.
From Figure 22 (at D
int
and scaled to the actual EIRP level) derive:
For case a) an interfering power I =~ 129 – (20  EIRP
ATPC
) =  140.3 dBW/MHz
For case b) an interfering power I =~ 129 dBW/MHz
Also in this example, both these levels are higher than requirement.
In case a) in order to receive an interfering power of 146 dBW/MHz, the CS of the interfering system should be placed
at a distance Dx from the border, so that:
D(a) + R
max
+ D
min
/2 =~ 61 km (from Figure 23, at EIRP = 8.7 dBW/MHz)
D(a) = 61  18.7  37.5 = ~ 4.8 km.
For case b), in order to receive an interfering power of 136 dBW/MHz (=kTBF), the CS of the interfering system
should be placed at a distance Dy from the border, so that:
D(b) + R
max
+ D
min
/2 =~ 62 km (from Figure 22 with P
rec
Max set to –136 dBW/MHz)
D(b) = 62  18.7  37.5 = ~ 5.8 km.
Therefore the minimum distance where a CS (supporting system type A TSs with height lower than 20 m and victim
CS height lower than 40 m) could be placed is 5.8 km.
3.1.4.4 TS to CS Conclusions
From the above examples, a CS, even if pointing away from the border, could not be indifferently placed nearer than the
coordination distance evaluated in Figure 22 and Figure 23. The terminals PFD will become determinant and
engineering of the cell (reduced EIRP and sector beams pointing) should be used to ensure that also TSs PFD (in the
direction of the boundary) does not exceed the values derived from Figure 24.
Figure 27 shows an example of such methodology based on previous examples 2 and 3.
ECC REPORT 33
Page 43
Figure 27: Example of cell sector engineering at service boundary
(assuming h
c
= 40 m on both service areas; for D
min
/2 see Figure 23)
3.1.5 Terminal Station to Terminal Station
This kind of interference, unlike the cases where a CS is involved, would only impair the operation of one terminal;
therefore a worst case approach is considered as too stringent. Due to the random nature of this kind of interference, a
statistical Monte Carlo approach seems more adequate. A possible scenario could be two CSs located at both sides of
the boundary, each pointing into its own service area. Each TS will have its EIRP level set to deliver a signal to the CS
6 dB above the receiver threshold, except one (different in each trial) which would have maximum EIRP, to simulate
the occurrence of a fading event.
However this effect, at least for PMP architectures with directive TS antennas, is not generally considered a limiting
factor for coexistence (when compared to the other interference cases). This evaluation may require further work in case
of mixed PMP and MESH architectures.
As far as TS with omnidirectional antennas are concerned, it is further assumed that:
 the vast majority is deployed indoor; large building/walls blocking attenuation will be always present.
3.2 Conclusions on adjacent service areas boundary block reuse coordination
It is therefore recommended that coordination between operators using same frequency block(s) in the 3.5 GHz band in
adjacent geographic areas should take place for any transmitter (CS and TS assumed to supply very similar EIRP) that
produces a PFD derived from Figure 24 or greater at the service area boundary. The distance from the service area
boundary that will be subject to coordination, as a function of transmitter EIRP, is indicated in
Figure 27.
The proposed PFD guidelines can be tested in Monte Carlo simulations to assess their validity in multiple interferer
scenarios.
X
Rmax
(TypeB)
Rmax
(
TypeA)
X
X
Dmin/2
X
Closest sectorblanked
cells (Type A and B)
Boundary
Standard uncoordinated cells
Blanked sectors of boundary engineered cells (example)
D min/2
D(a)
D(b)
Active sectors of boundary engineered cells (example)
ECC REPORT 33
Page 44
4 CONCLUSIONS OF THE REPORT
This Report has considered a number of facts as initial considerations for deriving the coexistence study:
1. Presently ERC Recommendations 1403 and 1208 for the bands 3.6 GHz and 3.8 GHz do not give
harmonised and detailed suggestion to administration for implementing FWA (such as those produced for
26, 28 and 40 GHz). Those ERC Recommendations offer only channel arrangements.
2. The band is limited and wasted guardbands might drastically reduce the number of licensed operators,
limiting the potential competition for new services.
3. Legacy FWA (WLL) systems have been already licensed in these bands in some countries, and their co
existence with new FWA systems, assigned in accordance with provisions recommended in this study, may
need additional check on a casebycase basis, using the same methodology as described in this report, but
with the parameters corresponding to those of legacy systems.
4. Sharing issues with PP FS systems, FSS, radiolocation (in adjacent band), ENG/OB were not considered in
this report, but are subject of a different CEPT study.
5. At least for CSs, ETSI ENs in these bands are not presently designed for a technology neutral deployment
(this is done only in the 40 GHz MWS EN 301 997) therefore do not contain system controlling parameters,
in terms of EIRP, which would be useful for the desired “technology neutral” and “uncoordinated”
deployment
6. The study suggested that the block edge mask together with the contiguous assignment of frequency blocks
should be the main means for avoiding interference between neighbouring frequency blocks. Since block
edge mask parameters are linked to the size of the block, it is also recommended that the neighbouring
blocks should be of a similar size.
7. It is also shown that, for PMP TSs, the antenna RPE plays a fundamental role in the coexistence; the more
directive is the antenna of TSs, the less demanding might be their NFD (or the EIRP density BEM) required
(offering a flexible tradeoff to the market). That is why the resulting BEM limits outside the block are
described in terms of transmitter output power, allowing operators to make practical use of this phenomena.
8. In addition, basic rules has been set for the coordination distance and PFD boundary levels between
operators reusing the same block in adjacent geographical areas. In this field, the importance of limiting CS
antenna height (or downtilt angle) as possible licensing parameter is highlighted in order of having sensible
coordination distances (i.e. limited by spherical diffraction attenuation).
9. MPMP (MESH) architectures have not been considered in this Report. It is recognised that whilst some of
the results in this report might also be applicable to mixed PMP and MESH deployment, others may clearly
need additional work. These studies might be carried on out in due time if needed and when manufacturers
will be in a position to offer the necessary information.
ECC REPORT 33
Annex 1, Page 45
ANNEX 1: AVERAGE CELL SIZE EVALUATION
A1.a) Simulation Model
The simulation model is illustrated on Figure 28. The cell is subdivided into segments whose angular width isu .
Within each segment, angular arcs are positioned at R
j
multiples of 0.1 R
max
where R
max
is the radius of the cell. There
are thus 10 arcs within each segment. Hence, there are 10 bounded subarea limits within each segment. The area limits
of each may be readily computed.
TS are assumed to be centrally positioned within each subarea. For each TS, the transmission distance is computed.
The impairments relative to LOS are then added. These include Mean Excess Loss (MEL), the random variations to
MEL due to lognormal shadowing and the impact of Rician fading.
For a given simulation, MEL and Rician K are set to the values specified for the SUI channel models. A standard
deviation of o = 9 dB is set for lognormal shadowing. This is a midrange value of the range set for the SUI models. A
random deviate procedure is employed to create shadow loss and Rician K signal variations. For MEL estimates, the CS
antenna elevation was set to 30 m and the TS antenna elevation and gain set to the indicated values.
Setting u = 1 degree results in 360 estimates of signal level. When the signal level of an estimate is found to be less
than the specified performance threshold, the subarea associated with the estimate is accumulated in an "excluded area"
running total. At the completion of a simulation, the ratio of the running total to the cell area defines the % of the cell
that cannot meet coverage requirements for 99.999% availability.
The TS antenna RPEs are those derived from ITUR F.1336 (see Figure 1), while antenna gain is kept fixed at 16 dBi as
for general system assumptions in Table 1.
Rj Rj+1 Theta
Rmax
0.1 Rmax
TS Location
Figure 28: Simulation Model
A1.b) Mean Excess Loss (MEL) only
When only MEL was considered, there were no exposures found that exceeded the performance threshold of Type A
Systems. Table 7 and Table 8 illustrate the simulation results for covered areas of Type B systems. Lognormal
shadowing and Rician fading are excluded.
ECC REPORT 33
Page 46, Annex 1
Excluded Area (%) SUI Terrain
Category TS antenna class TS 2
(ITUR RPE G=+16 dBi)
TS2/TS3 intermediate RPE
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G= +20
dBi)
A 18.8 0 0
B 0 0 0
C 0 0 0
Table 7: Type B System MEL Excluded Area for R
max
= 2.0 km, TS Antenna Elevation = 10 m
Excluded Area (%) TS Antenna
Elevation
(m)
SUI
Terrain
Category
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3 intermediate
(ITUR RPE G= +18
dBi)
TS antenna class TS 3
(ITUR RPE with G=20
dBi)
A 51 37.3 35.5
B 19 0 0
10
C 0 0 0
A 39.6 36 19
B 0 0 0
15
C 0 0 0
A 36 19 0
B 0 0 0
20
C 0 0 0
Table 8: Type B System MEL Excluded Area for R
max
= 2.7 km
A1.c) Mean Excess Loss and LogNormal Shadowing
Table 9 through Table 12 show the results of the simulations when both MEL and lognormal shadowing are
considered. As previously noted, the standard deviation for the lognormal fading was set to o = 9 dB.
With the inclusion of lognormal shadowing, it is apparent that even a Type A system will begin to experience coverage
problems. This is constrained to Terrain Category A and R
max
= 2.7 km.
Due to reduced threshold, coverage issues for Type B systems are significantly increased. Referenced to Table 11 and
Table 12, both 2.0 km and 2.7 km cell radius designs exceed coverage objectives in Terrain Categories A and B.
Excluded Area (%) TS Antenna
Elevation
(m)
SUI
Terrain
Category
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3 intermediate
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
A 8.9 5.7 2.8
B 2.8 1.9 1.2
10
C 0.2 0 0
A 5.9 3.9 1.5
B 1.9 1.1 0.45
15
C 0 0 0
A 4.1 2.6 1.5
B 1.6 0.6 0
20
C 0 0 0
Table 9: Type A System Excluded Area Due to MEL and LogNormal Shadowing (R
max
= 2.0 km)
ECC REPORT 33
Annex 1, Page 47
Excluded Area (%) TS Antenna
Elevation
(m)
SUI
Terrain
Category
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3 intermediate
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
A 22.0 16.8 13.4
B 8.7 6.3 4.3
10
C 0.7 0.5 0
A 17.2 11.85 9.4
B 6.9 4.0 3.5
15
C 0 0.36 0
A 12.8 8.1 6.2
B 5.5 2.9 2.3
20
C 0 0 0
Table 10: Type A System Excluded Area Due to MEL and LogNormal Shadowing
(R
max
= 2.7 km)
Excluded Area (%) TS Antenna
Elevation – m
SUI
Terrain
Category
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3 intermediate
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
A 26.8 20.1 16.8
B 12.8 9.7 6.2
10
C 1.9 1.4 1.0
A 20.3 16.0 11.9
B 9.0 6.9 4.2
15
C 0.8 0.46 0
A 17.3 13.1 9.3
B 7.6 5.0 4.2
20
C 0.5 0 0
Table 11: Type B System Excluded Area Due to MEL and LogNormal Shadowing
(R
max
= 2.0 km)
Excluded Area (%) TS Antenna
Elevation
(m)
SUI
Terrain
Category
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3
intermediate RPE
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
A 44.6 39.9 33.4
B 29 23.0 19.0
10
C 5.9 4.0 3.2
A 39.2 33.6 26.4
B 21.7 17.5 13.3
15
C 3.1 1.8 1.2
A 35.4 29.2 23.3
B 18.8 13.0 9.4
20
C 1.8 0.6 0.6
Table 12: Type B System Excluded Area Due to MEL and LogNormal Shadowing
(R
max
= 2.7 km)
A1.d) Mean Excess, LogNormal Shadowing and Rician Fading
Generally speaking, it is not appropriate to interrelate space and time availability. However, in the NLOS transmission
environment, Rician fading is constantly present. In the event that there is no motion associated with the reflective
facets, it simply means that we are in a fixed up or down fade, subject to the vector addition of all of the signal
components.
In order to examine the significance of Rician fading, its impact was examined for each of the SUI channel models by
running 10 simulations for each channel model. The range of variation in the excluded area was noted and these are
presented in Table 13 and Table 14 for R
max
= 2 km. Table 15 and Table 16 examine the same scenarios for R
max
= 2.7
km.
ECC REPORT 33
Page 48, Annex 1
SUI Channel Model Terrain Category Rice K (dB) Excluded Area Spread (%)
SUI1 C 12 0.0  0.15
SUI2 C 8 0.0  0.1
SUI3 B 5 1.7  2.8
SUI4 B 0 3.1 4.2
SUI5/6 A 0 6.9  8.5
Table 13: Impact of Rician Fading on Type A Systems (R
max
= 2 km)
SUI Channel Model Terrain Category Rice K (dB) Excluded Area Spread (%)
SUI1 C 12 0.32  0.78
SUI2 C 8 0.39  0.79
SUI3 B 5 8.5 9.8
SUI4 B 0 11.0  12.4
SUI5/6 A 0 20.6  22.9
Table 14: Impact of Rician Fading on Type B Systems (R
max
= 2 km)
SUI Channel Model Terrain Category Rice K – (dB) Excluded Area Spread (%)
SUI1 C 12 0.04  0.36
SUI2 C 8 0.16  0.46
SUI3 B 5 5.4  6.7
SUI4 B 0 7.2  9.2
SUI5/6 A 0 16.6  20.1
Table 15: Impact of Rician Fading on Type A Systems (R
max
= 2.7 km)
SUI Channel Model Terrain Category Rice K  dB Excluded Area Spread (%)
SUI1 C 12 1.9  2.8
SUI2 C 8 2.0  3.2
SUI3 B 5 20 22.1
SUI4 B 0 22.8 25.2
SUI5/6 A 0 38.0 40.9
Table 16: Impact of Rician Fading on Type B Systems (R
max
= 2.7 km)
A1.e) Simulation Caveat
Section A1.d) does not cover all of the combinations as Sections A1.b) and A1.c): they are just “midrange” values
illustrative sensitivity analysis examples, with TS antenna height = 15 m and TS antenna gain = +18 dBi).
A close examination of the Tables might imply that there are some inconsistencies in the Table entries. However, this is
not the case. For example, examine Table 8 and Table 12 for Terrain Type A and R
max
=2.7 km. It may be noted that the
Excluded Area is less when the MEL plus lognormal shadowing loss is considered compared to that just resulting from
MEL. However, this is a quite possible result of simulation. The shadowing loss exhibits a random +/ variation about
the mean value that can enhance the signal level of some links. The same may be said about Rician fading for which
both up and down fades can occur.
ECC REPORT 33
Annex 1, Page 49
A1.f) Urban scenarios, directional outdoor versus omnidirectional indoor TS
A1.f.1 Coverage and availability objectives:
Being focused on urban scenarios, addressing different kinds of customer need and deployments, ranging from business
to residential internet connections we should also distinguish the objectives for the evaluation of the covered area and
availability objectives.
Defining different objectives for business (mostly outdoor rooftop) and residential (mostly indoor), besides being
commercially applicable, might equalise the coverage area for the two cases that would exhibit different Fade Margin.
For this study we have considered the following cases:
Business deployment:
Coverage = 95% / 80% / 65%
Availability = 99.99%
Residential deployment:
Coverage = 95% / 80% / 65%
Availability = 99.0%
A1.f.2 Methodology
As initial task we could evaluate the impact of the model on the cell radius with the system assumptions made above.
The evaluation is done establishing a pattern (~3000/sector) of possible TS location; then running MonteCarlo
simulation for each location (~300 trials) for finding the RSL distribution and the percentile of “good” trials (RSL
exceeding the equipment threshold).
The calculation is made in four steps covering the separate objectives of geographical coverage and link availability:
Step 1:
Evaluation of the System Gain (SG) defined as:
SG = P
TX
+G
TX
+G
RX
AATTP
TH
where:
P
TX
= TX power [dBm]
G
TX
, G
RX
= Antenna gains [dB]
AATT = Additional Attenuation [dB] (e.g. indoor penetration losses)
P
TH
= RSL threshold @ BER=10
6
[dBm] (function of modulation format, noise figure (NF) e Rx bandwidth)
Step 2:
Evaluation of Shadowing Margin (SHM):
SHM = SHM(Cov%, σ, γ)
where:
Cov% = coverage objective [%]
σ = Shadowing variance [dB]
γ = Erceg’s category (A, B and C) parameter
The SHM is a statistical function, therefore it is not appropriate and exceedingly conservative to impose the coverage
objective at the cell border only (closer areas would have higher coverage); therefore the coverage is evaluate as the
integral of the probability over the whole cell normalised area.
It could be demonstrated that the SHM (additional to A
50
) does not depend on the cell radius itself; the dependence of
required SG, for a certain coverage, on the cell radius is then limited to the A
50
(R) only.
ECC REPORT 33
Page 50, Annex 1
Step 3:
Evaluation of Rician Fast Fading Margin (FFM) derived from cumulative probability of the selected SUI model (see
Figure 3) at the required availability.
FFM = FFM(SUIProb, Avail)
where:
SUIProb is the cumulative attenuation probability curve
Avail = required availability
Step 4:
Evaluation of R
MAX
with the relationship:
SG = A
50,0
+γ*10log(R
MAX
/R
0
)+SHM+FFM
where:
R
0
= a reference distance (0.1 km)
A
50,0
= mean attenuation (dB) at reference distance R
0
, evaluated for the selected Erceg category model.
R
MAX
is therefore finally given by:
R
MAX
=R
0
*10^((SG  A
50,0
–SHMFFM)/(10*γ))
A1.f.3 Cell radius evaluation
With the above methodology, in the following tables different evaluations are made using different CS and TS location
heights, antenna typology and relevant Erceg Categories and SUI models
Business case (availability 99.99%, TS outdoor directional antennas G
TS
=18 dB)
Lower elevation/more obstructed antennas (ErcegCategory B+SUI4; σ = 8 dB)
Table 17: R
MAX
[km] for QPSK formats
Table 18: R
MAX
[km] for 16QAM formats
Higher elevation/less obstructed antennas (ErcegCategory C+SUI2; σ = 8 dB)
Table 19: R
MAX
[km] for QPSK formats
CS = 30 [m] CS = 40 [m]
TS = 4 [m] TS = 10 [m] TS = 4 [m] TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
1.26 1.87 2.38 1.58 2.35 2.98 1.42 2.14 2.75 1.80 2.72 3.48
CS = 30 [m] CS = 40 [m]
TS = 4 [m] TS = 10 [m] TS = 4 [m] TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.87 1.30 1.65 1.09 1.62 2.06 0.96 1.46 1.86 1.22 1.84 2.36
CS = 30 [m] CS = 40 [m]
TS = 4 [m] TS = 10 [m] TS = 4 [m] TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
2.35 3.57 4.59 3.67 5.57 7.16 2.78 4.30 5.58 4.45 6.88 8.93
ECC REPORT 33
Annex 1, Page 51
Table 20: R
MAX
[km] for 16QAM formats
Residential case (availability 99 %, TS indoor window antennas G
TS
=10 dB)
Lower elevation/more obstructed antennas (ErcegCategory B+SUI4; σ = 10 dB)
Table 21: R
MAX
[km] QPSK
Table 22: R
MAX
[km] 16QAM
Higher elevation/less obstructed antennas (ErcegCategory C+SUI2; σ = 10 dB)
Table 23: R
MAX
[km] QPSK
Table 24: R
MAX
[km] 16QAM
Residential case (availability 99 %, TS indoor omnidirectional antennas G
TS
=4 dB)
Lower elevation/more obstructed antennas (ErcegCategory B+SUI4; σ = 10 dB)
CS = 30 [m] CS = 40 [m]
TS = 4 [m] TS = 10 [m] TS = 4 [m] TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
1.59 2.41 3.10 2.48 3.77 4.84 1.84 2.84 3.69 2.94 4.55 5.90
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.84 1.36 1.81 1.06 1.71 2.27 0.93 1.53 2.06 1.18 1.94 2.61
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.58 0.94 1.25 0.73 1.18 1.57 0.63 1.04 1.40 0.8 1.32 1.77
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
1.51 2.49 3.36 2.35 3.89 5.25 1.73 2.94 4.01 2.77 4.70 6.42
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
1.02 1.69 2.27 1.59 2.63 3.55 1.15 1.94 2.65 1.83 3.11 4.25
ECC REPORT 33
Page 52, Annex 1
Table 25: R
MAX
[km] QPSK; Building/indoor losses = 5 dB
Table 26: R
MAX
[km] QPSK; Building/indoor losses = 15 dB
Table 27: R
MAX
[km] 16QAM; Building/indoor losses = 5 dB
Table 28: R
MAX
[km] 16QAM; Building/indoor losses = 15 dB
Higher elevation/less obstructed antennas (ErcegCategory C+SUI2; σ = 10 dB)
Table 29: R
MAX
[km] QPSK; Building/indoor losses = 5 dB
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.71 1.15 1.53 0.90 1.44 1.92 0.78 1.28 1.73 0.99 1.63 2.19
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.42 0.68 0.90 0.53 0.85 1.13 0.45 0.74 0.99 0.57 0.94 1.26
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.49 0.80 1.06 0.62 1.00 1.33 0.53 0.87 1.17 0.67 1.11 1.49
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.29 0.47 0.63 0.37 0.59 0.78 0.30 0.50 0.67 0.39 0.64 0.86
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.86 1.42 1.92 1.34 2.22 2.99 0.96 1.62 2.22 1.53 2.60 3.55
ECC REPORT 33
Annex 1, Page 53
Table 30: R
MAX
[km] QPSK; Building/indoor losses = 15 dB
Table 31: R
MAX
[km] 16QAM; Building/indoor losses = 5 dB
Table 32: R
MAX
[km] 16QAM; Building/indoor losses = 15 dB
A1.g) Conclusions of the annex
The Tables in this annex summarise all cases generated from the input parameters for the business and residential
applications.
The results for the more and the less favourable conditions might also be averaged for having the actual overall
coverage %.
We should consider that those evaluations do not take into account a number of features that last generation systems,
specifically designed for these NLOS applications, offer for improving the situation. They could be for example:
 Adaptive modulation/coding formats (when RSL of a TS is marginal, the lower modulation format and the
strongest error coding is automatically selected)
 Adaptive antenna arrays
 Diversity
 subchannelling
All these techniques enhance the coverage with respect to the conventional PMP characteristics used in this study.
In conclusion, we could qualitatively say that the coverage and availability objectives could be fulfilled with cell radius
ranging from ~0.5 km to ~ 3 km depending on the decision to privilege or not residential traffic with respect to business
only. It should also be taken into account that, addressing in real offered service a mixture of different situations
(residential or business, with more or less favourable expected propagation), the actual expected coverage/cellsize
would be tradeoff considering which market is addressed and the local environment situations.
Therefore, in following study for evaluating the CS to TS interference based on statistical occurrence of C/I, we would
consider four radii cases: 0.5 km, 1 km, 2 km, 2.7 km.
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.49 0.81 1.10 0.77 1.27 1.71 0.53 0.90 1.23 0.85 1.44 1.97
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.58 0.96 1.30 0.91 1.50 2.02 0.63 1.07 1.47 1.01 1.72 2.35
Alt.CS = 30 [m] Alt.CS = 40 [m]
Alt.TS = 4 [m] Alt.TS = 10 [m] Alt.TS = 4 [m] Alt.TS = 10 [m]
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
Cov.
95%
Cov.
80%
Cov.
65%
0.33 0.55 0.74 0.52 0.86 1.16 0.35 0.60 0.81 0.56 0.95 1.30
ECC REPORT 33
Annex 2, Page 54
ANNEX 2: TS TO CS INTERFERENCE EVALUATION
A2.1 Directional outdoor TS
A2.1.1 Rural scenario
A2.1.1.1 System Model and Simulation Methodology
In subsequent sections, estimates of interference susceptibility are based on MonteCarlo simulations that identify the
spatial probability of a victim link experiencing an excessive level of interference. Graphically, the simulation results
are presented as an interference grade of service (GOS) probability shown as a Cumulative Distribution Function (CDF)
vs. C/I.
The TS to CS system model is illustrated on Figure 29. It is computationally convenient to consider the overlaid
sector/cell as being the victim. This is parameterised at some separation distance S between the two CS sites. Within the
victim sector, all TS locations are assumed to employ distance proportional ATPC. Thus, all received signal levels from
victim TS links are assumed to arrive at the victim CS at the same level of signal strength. Thus, it is only necessary to
set a victim TS to CS signal level based on that of a single celledge victim link located at distance R
max
.
Even for the rural environment, the link margin is modest; thus no cell edge ATPC is assumed. It is important to note
that there is no valid technical reason to apply cell edge ATPC except for interference exposures associated with TS to
TS couplings. These are considered to be quite rare. Maximising cell edge signal level reduces the sensitivity of CS to
CS couplings (not examined in this report).
The number of randomly located interference TS locations within a sector is set to N
t
. = 64. There is no statistical
measurement data available to identify how these TS locations should be located. Two extreme possibilities can be
considered. The first assumption might be to assume that the TS locations are randomly distanceproportionally located
referenced to the maximum cell radius R
max
. The second assumption is to assume that the TS locations are randomly
areaproportionally located. In this latter case, approximately 50 % of TS locations would be expected to be located at
greater than 0.75R
max
. Only the area proportional assumption will be subsequently examined.
To account for the assumption that there is no operator coordination, the relative boresight alignment of the two CS
antennas is considered to be unknown. A simulation run is configured to spin the relative sector alignment in 5 degree
increments. A complete simulation run thus consists of 360/5*N
t
interference estimates (N
t
= 64, resulting in 4608
interference estimates).
In the rural environment, only LOS transmission is considered. Thus, the only fading mechanism considered to be
applicable is that of atmospheric Rayleigh multipath. Generally speaking, it is not statistically appropriate to mix
spatial link availability with time varying availability. However, we will examine this situation, with the caveat that it
only applies during the time intervals when Rayleigh fading occurs.
Due to the distance differentials associated with the victim and interference paths, uncorrelated Rayleigh fading is
ensured. To account for Rayleigh fading, it is necessary to generate random Rayleigh deviates that are created from the
uniform random deviates available with computational machine programs. The procedure is based on the Acceptance
Rejection method as detailed in [3] and is summarised in Appendix 2 to Annex 2.
ECC REPORT 33
Annex 2, Page 55
Cell Radius R
Random
Interference
TS Location
S
Random
Location
Victim CS Arc
1
Spin Angle
Cell Edge
Victim Link
TS Location
Victim CS
Figure 29: Simulation Model
The inclusion of Rayleigh fading adds considerable complexity to the simulation model. It is best described by
reference to Figure 30. It is no longer valid to assign the victim TS to be at distance R
max
. Only one victim TS is active
during a given TDMA time block. As the magnitude of Rayleigh fading is distance related, we should now place the
victim TS at some random distance R
v
from it's serving CS. To establish link loss, we must now do the following:

Compute FSL based on the distance R
v.
 Adjust the FSL signal level PT
v
so that it is reduced to be ATPC distance proportional, i.e., by 20log(R
v
/R
max
).
 Determine the Rayleigh fading adjustment as discussed above. Modify the value of PT
v
accordingly.
 Adjust the victim TX signal level via ATPC so that it adjusts to the FSL margin level set at R
max
. If the
Rayleigh fade impairment exceeds this adjustment, then set the victim TX power to be at its maximum level.
This sets the TX power level of the victim link transmitter. However, we now have to examine the TX power of the
interference link. Given that we have a local meteorological environment that induces Rayleigh fading on the victim
link, it is quite valid to assume that the same conditions apply to the interference link. But we now have two
transmission paths to consider. Referenced to Figure 30, the first of these is the link between the interference TS and it's
serving CS. For any one of the N
t
interference TS's, located at some random distance R0, the uncorrelated Rayleigh
fading signal level adjustment is described as above. This fixes the TX power level PT
i
of any single interference link.
However, the interference coupling path is a different uncorrelated Rayleigh path at a distance R
i
. Employing the same
methodology as previously described, a new Rayleigh fading adjustment is determined for this path. Given that both the
uncorrelated Rayleigh faded victim signal level (C) and the interference signal (I) can now be computed, the C/I of each
interference estimate can be determined.
ECC REPORT 33
Annex 2, Page 56
Victim CS
Interference CS
Victim TS
Interference TS(Nt)
Victim Link @ Distance Rv
Interference Link @ Distance R0
Interference Coupling Path @ Distance Ri
Figure 30: Rayleigh Faded Interference Model
A2.1.1.2 Unfaded Simulation Results
For a cell size of Rmax = 8.9 km, Figure 31 trough Figure 34 illustrate the Monte Carlo simulation results (unfaded)
with antenna classes TS2 to TS3 (i.e. with typical RPE derived from ITUR F.1336 TS antenna using Gain = 16, 18 and
20 dBi).
In each figure a value of NFD has been selected, identifying the minimum NFD requirement to “hit” the ~ 1% CDF at
the C/I critical threshold for the system types reported in Table 1.
Figure 31 applies to S being between 3 to 6 km while Figure 32 applies to a CS separation distance S from 0.1 to 2 km.
A comparative examination of Figure 31 and Figure 32 indicates that the poorest CDF results occur when S is large.
These differences can be explained as follows:
 When S is small, and both the interference and victim CS antennas are partially aligned, a high percentage of the
interference TS links are illuminated by the victim CS antenna. Also, when S is small, FSL is comparable on both
links and TS Antenna RPE is modest. With ATPC, both interference and victim link signals would be expected to
arrive at the victim CS at comparable levels. Hence, the major difference in signal level is that of NFD and, as
shown on Figure 32, there is a resultant sharp "knee" in the C/I in the vicinity of the NFD value.
 As S increases, conflicting geometrical results occur. Some interference TS locations are essentially eliminated as
they are behind the victim CS antenna. As well, as interference TS distance from victim CS decreases, angles
increase, and the RPE rejection of the interference TS increases, thus further reducing the number of serious
interference exposures. Countering this, is relative distance proportional ATPC. It now becomes modest on the
interference links, thus setting up an increase in signal level differentials. The C/I "knee" is thus diminished while
the percentage of C/I exposures above the knee is reduced. However the level, and percentage of worst case C/I
exposures, will increase, as shown on Figure 31.
Hence in the subsequent simulations (Figure 33 and Figure 34), we will only present the S > 3 km case, which controls
the NFD requirement.
ECC REPORT 33
Annex 2, Page 57
Figure 31 . Unfaded CDF (S > 3 km, NFD = 32 dB)
(TS 2 antenna class (ITUR RPE G=16 dBi))
Figure 32: Unfaded CDF S < 2 km, NFD = 32 dB)
(TS 2 antenna class (ITUR RPE G=16 dBi))
ECC REPORT 33
Annex 2, Page 58
Figure 33: Unfaded CDF (S > 3 km, NFD = 30 dB)
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
Figure 34: Unfaded CDF (S > 3 km, NFD = 28 dB)
(TS 3 antenna class (ITUR RPE G=20 dBi))
ECC REPORT 33
Annex 2, Page 59
A2.1.1.3 Rayleigh Faded Simulation Results
Figure 35through Figure 38 illustrate the results for the case where the Rayleigh fading distance probability coefficient
d
coeff
is set to 3.. As compared to the unfaded case, the CDF impairments resulting from uncorrelated Rayleigh fading
are quite modest.
To explain this somewhat surprising result, we first note that the median level p.d.f. crossover for Rayleigh occurs at
63%. But this also means that 37% of the links will be in excess of the median level. For the interference links, these TS
transmitters are ATPC adjusted to be lower in power. They thus transmit at a lower power than in the unfaded
coexistence scenario.
For the victim links, a statistical examination of the ATPC adjusted signal level was performed. Here, it was found that
24% of the victim links were required to operate at maximum power. For the remainder, the distance proportional
ATPC range was sufficient to restore the signal level to its unfaded margin level. However, 50% of these maximum
power links were within 3 dB of the unfaded signal margin. Thus, a high percentage of victim links arrive at close to the
same signal level as that for the unfaded scenario.
Figure 35: Rayleigh Faded CDF (S > 3 km, NFD = 32 dB)
(TS 2 antenna class (ITUR RPE G=16 dBi))
ECC REPORT 33
Annex 2, Page 60
Figure 36: Rayleigh Faded CDF (S < 2 km, NFD = 32 dB)
(TS 2 antenna class (ITUR RPE G=16 dBi))
Figure 37: Rayleigh Faded CDF (S > 3 km, NFD = 30 dB)
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
ECC REPORT 33
Annex 2, Page 61
Figure 38: Rayleigh Faded CDF (S > 3 km, NFD = 28 dB)
(TS 3 antenna class (ITUR RPE G=20 dBi))
A2.11..4 Conclusions
It is concluded that the NFD values summarised in the following Table 33 are acceptable values for the TS emissions
associated with TS to CS interference couplings in the rural scenario.
TS antenna class TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3
intermediate RPE
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
Minimum NFD required
for Type B System (dB)
32 30 28
Table 33: Minimum NFD required for Type B Systems Rural scenario
ECC REPORT 33
Annex 2, Page 62
A2.1.2 Urban Scenario
A21..2.1 Simulation Methodology
The simulation model is comparable to that described in section A2.1 and Figure 30 for the rural scenario. Again, there
are three transmission paths to consider. These are the victim link at distance R
v
, the interference link at distance R
0
and
the interferencecoupling path at distance R
i
. For each interference computation it is necessary to set the TX power of
the TS for the first two links. The procedure is as follows:
 Compute FSL based on the distance R
x
equal to R
v
or R
0
.
 Adjust the FSL signal level so that it is reduced to be distance proportional, i.e., 20log(R
x
/R
max
).
 Compute the mean excess path loss based on the distance R
x
.
 Compute the mean value of Rician K based on distance R
x
and relative to the SUI value for K, as specified for
the cell edge at R
max
.
 Determine the Rician fading adjustment by the random deviate method.
 Adjust the RX signal level to account for both mean excess loss and Rician fading.
 Readjust the TX signal levels via ATPC so that some signal margin above the threshold level is restored. This
would typically be somewhere between 6 dB and 15 dB. As subsequently discussed, the simulations found
some degree of C/I performance sensitivity referenced to the margin value selected.
 Set the TS  TX Power level accordingly. If the ATPC range is insufficient to achieve the specified margin,
then set the TX power to P
max
.
The TX power of both the interference and victim links is now set. The signal level of the interferencecoupling path at
distance R
i
is now determined based on the procedure for computation of excess loss and Rician fading described. The
C/I for each interference estimate can now be determined.
A2.1.2.2 Simulation Results
A2.1.2.2.1 Unfaded
Figure 39 through Figure 50 in this section illustrate the CDF vs. C/I results for:
 Rice factor K = 30 dB. For this K value, the probability of a deep fade is extremely low. Hence, this is
essentially the case without fading.
 R
max
= 2.7 km. and R
max
= 2.0 km
 Different TS antenna heights (15 and 20 m)
 Different TS antenna classes TS2 to TS3 (i.e. with typical RPE derived from ITUR F.1336 TS antenna using
Gain = 16, 18 and 20 dBi, still with fixed 16 dBi boresight gain).
Each time, in the presented simulations the NFD used correspond to the minimum required to “hit” the ~ 1% of cases
with C/I over the critical C/I threshold for the system type as reported in Table 1.
Performance degrades noticeably as CS separation distance S increases. This is a result of the excess loss differential
and can be explained as follows:
 When S is small, a large number of interference and victim links are at a comparable distance from their
serving CS sites. Consequently, excess loss is comparable on both interference and victim links.
 When S is large, there are fewer interference links that can illuminate the victim CS. But for those that do so,
the interference distance is small; thus setting up an excess loss differential that strongly favours the
interference link.
ECC REPORT 33
Annex 2, Page 63
Figure 39: Mean Excess Lossbased CDF R
max
= 2.7 km, TS Ant Elev = 15 m, NFD = 45
(TS 2 antenna class (ITUR RPE G=16 dBi))
Figure 40: Mean Excess Lossbased CDF R
max
= 2.7 km, TS Ant Elev = 15 m, NFD = 45 dB
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
ECC REPORT 33
Annex 2, Page 64
Figure 41: Mean Excess Lossbased CDF R
max
= 2.7 km, TS Ant Elev = 15 m, NFD = 43 dB
(TS 3 antenna class (ITUR RPE G=20 dBi))
Figure 42: Mean Excess Loss based CDF R
max
= 2.7 km, TS Ant Elev = 20 m, NFD = 43 dB
(TS 2 antenna class (ITUR RPE G=16 dBi))
ECC REPORT 33
Annex 2, Page 65
Figure 43: Mean Excess Loss based CDF R
max
= 2.7 km, TS Ant Elev = 20 m, NFD = 43 dB
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
Figure 44: Mean Excess Loss based CDF R
max
= 2.7 km, TS Ant Elev = 20 m, NFD = 41 dB
(TS 3 antenna class (ITUR RPE G=20 dBi))
ECC REPORT 33
Annex 2, Page 66
Figure 45: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 15 m, NFD = 43 dB
(TS 2 antenna class (ITUR RPE G=16 dBi))
Figure 46: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 15 m, NFD = 42 dB
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
ECC REPORT 33
Annex 2, Page 67
Figure 47: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 15 m, NFD = 40 dB
(TS 3 antenna class (ITUR RPE G=20 dBi))
Figure 48: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 20 m, NFD = 40 dB
(TS 2 antenna class (ITUR RPE G=16 dBi))
ECC REPORT 33
Annex 2, Page 68
Figure 49: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 20 m, NFD = 40 dB
(TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi))
Figure 50: Mean Excess Loss based CDF R
max
= 2.0 km, TS Ant Elev = 20 m, NFD = 39 dB
(TS 3 antenna class (ITUR RPE G=20 dBi))
ECC REPORT 33
Annex 2, Page 69
The following Table 34 summarizes, for the most critical system type B, the main findings, in terms of minimum
required NFD value, for the various configurations in the urban scenario:
TS Antenna class
TS Antenna
Height (m)
TS antenna class TS 2
(ITUR RPE G=16 dBi)
TS2/TS3
intermediate RPE
(ITUR RPE G= +18 dBi)
TS antenna class TS 3
(ITUR RPE with G=20 dBi)
Minimum NFD value required (dB)
System Type B
(Cell size 2.7 km)
15 45 45 43
20 43 43 41
System Type B
(Cell size 2.0 km)
15 43 42 40
20 40 40 39
Table 34: Minimum NFD required for Type B Systems Urban scenario
In case of 4QAM system (system type A), there is an 8 dB increase in system gain. Thus, the critical receiver levels
drop to 14.2 dB and 20.2 dB and the CDFs values improve.
A2.1.2.2.2 Rician Faded
If we run simulations for the SUI1 channel model with cell edge Rice K = 12 dB, with cell radius R
max
= 2.7 km,
except for differences in detail, there will be very little difference between the previous unfaded results and the Rician
faded case. This result is expected, Rician fading is modest for K = 12 dB. As well, the uncorrelated fading relationship
results in an "averaging out" of fading differentials between the interference and victim paths. Coexistence performance
criteria are thus dominated by the excess loss differential associated with nearNLoS transmission.
For simulation for a cell radius of R
max
= 2 km, a SUI2 channel is assumed with a mean value of Rice K equal to 9 dB
at cell edge. Note that the maximum value for S has to be reduced to 1.5 km, reflecting the smaller value of R
max
.
In spite of the reduced value of K, the results will be little changed from those of the previous SUI1 case. Due to the
smaller cell size, excess path loss at cell edge is reduced, resulting in a larger fade margin. As the excess loss
differential was previously concluded to control CDF vs. C/I performance, this loss differential reduction is sufficient to
offset the increased probability of deep fades.
A2.1.2.3 Conclusions, directional outdoor antennas
From the preceding analysis and simulations, the following may be concluded:
1. The system gain set for the assumed transmission model constrains nearNLoS operation to be within the SUI
1 and SUI2 transmission environment. To operate in more severe nearNLoS environments, additional system
gain is required. While means exist to provide some increase in system gain, they are outside the scope of this
Report.
2. With the use of ITUR F.1336 TS antenna RPE, representative of reasonably designed ETSI antennas, a NFD
between 40 dB and 45 dB looks adequate for acceptable percentages of interference impairment, depending
both on antenna gain and TS antenna heights.
A.2.2 Omnidirectional indoor TS
A.2.2.1 Input parameters and models
A.2.2.1.1. Objectives
In present version of Report 33 only directional outdoor TS antennas are considered. The study for rural and urban
scenarios, based on the fact that TS antenna directivity will eliminate most of the possible source of interference, show
that urban worst case could be studied in unfaded conditions looking for a suitably low probability of a NLOS
interfering TS to reach a victim CS with a C/I exceeding a fixed threshold level for the most stringent modulation
format.
ECC REPORT 33
Annex 2, Page 70
For the omnidirectional case a different approach should be used; in this case each TS will contribute to an aggregate
interference on the victim CS.
Therefore, the statistical study will be based on:
 Evaluation of a maximum terminal density
 Evaluation of an activity factor averaged over that population
 Distribution of the TS population among various households and offices
 Evaluation of a typical propagation model according the selected location of each TS
 Evaluation of the probability and cumulative distribution functions of the aggregate interference through an high
number of Monte Carlo trials (snapshots).
 Comparison with the noise floor of the CS receiver.
A.2.2.1.2 CS data (same used for CSTS evaluation):
TX output power = not relevant
Signal bandwidth = 3.5 MHz
TX output power density = not relevant
Antenna Gain (sectorial 90°) = 16 dB
Antenna pattern ITUR F.1336
Antenna height = 30 / 40 m
NF = 8 dB (including cable connection to antennas) or 5 dB (ETSI BRAN working assumptions)
Noise floor = ÷109 to ÷106 dBm/MHz
Modulation formats = not relevant
A.2.2.1.3 TS omnidirectional indoor out of block eirp:
The characteristics used for the CSTS evaluation should be integrated with TX characteristics not specified for that
purpose; however, for this TSCS evaluation, only the expected outofblock emissions would be necessary.
Reference eirp limit
As reference value we will use the present eirp limit (s÷37 dBW/MHz) reported in draft REC0405, defined for
directional TS with antenna gain of ~ 16dB, reduced by the expected gain difference.
This is reasonable considering that TSs are generally designed for being coupled with a number of different antennas;
therefore it should not concern the manufacturers.
Therefore, with TS omnidirectional antenna gain ~ 4dB, the out of block emissions would become:
Omni OOB eirp s ÷ 37 ÷ (16÷4) ~ ÷ 49 dBW/MHz
Besides the lesser antenna gain, this value is also justified by lesser expected output power for these TS applications.
Note: ETSI BRAN assumptions for this kind of applications is limited to +20 dBm typical; however, unless specific
limitations might be necessary for sharing purposes, in practice certain percentage of higher power terminals might also
be expected.
ATPC range
Even if ETSI ENs assumed ATPC as an option and do not specify any range, modern PMP systems implement quite
large range of ATPC on TSs in order to balance all TS arrival powers to the CS at minimum operating level improving
linearity of CS operation and keeping intercell interference as low as possible. ATPC of ~ 40/50 dB are quite common.
Out of block eirp emission will possibly be further reduced, even if not of the same amount due to noise limitation.
This would add sensible additional margin that we would evaluate in at least 10 dB in average.
ECC REPORT 33
Annex 2, Page 71
A.2.2.1.4 TS activity factor and density:
The aggregate interference of a number of terminals on the territory will be proportional to their density (D
TS
) and to
their average activity factor (A
f
).
Modern PMP systems might manage quite large number of Terminal stations per BS sector; however, all TSs should
share the overall BS capacity/sector. Therefore, this results in an maximum average activity factor.
PMP systems may manage a number (N
T
) of TS per BS sector; these TS are sharing a constant capacity per sector.
Whichever the capacity the average maximum activity factor (A
f
) is defined as:
A
f
= 1/N
T
Modern PMP systems may manage a quite large number (N
T
) of TS per BS sector; typically 256 or even more (e.g. up
to 1024 addresses are foreseen in BRAN HM and IEEE 802.16 standards protocols).
However, we should consider some physical and market limitations, derived from the actual number of households over
a certain area.
The following considerations are taken from the draft ETSI BRAN System Reference Document (SRD) for
fixed/nomadic convergence presented to the 2
nd
JPTBWA meeting:
Reprinted from Doc. JPTBWA(05)17 Annex 1 (may be found at ERO document server at www.ero.dk/download )
Considering that the maximum penetration rate is 30% for the target broadband services, in suburban and rural
areas, results the maximum subscriber number. The urban penetration was considered also 30% due to the special
attractiveness of the nomadic usage.
Households L = 2km L=1km L = 500 m L = 250m
Area (sq km) 4 1 0.25 0.0625
Urban 12000 3000 750 187.5
Suburban 4000 1000 250 62.5
Rural 1000 250 62.5 15.625
Table 7:Total number of households / cell
L = 2km L=1km L = 500 m L = 250m
Urban
3600 900 225 56
Suburban 1200 300 75 19
Rural 300 75 19 5
Table 8: Total number of subscribers / cell
The above considerations, over a simplified square cells scenario, tell that, in average, the households’ physical
constraints maintain the maximum terminal density (urban scenario) constant at:
D
TSi
~ 900 / km
2
They would be subdivided on sectors containing a number of terminals that, in case of reuse 4 deployment, ranges from:
Urban: ~14 < N
T
< ~ 256 or more, depending on maximum terminal handling and data rate offered
Suburban ~5 < N
T
< ~ 256 or more, depending on maximum terminal handling and data rate offered
ECC REPORT 33
Annex 2, Page 72
The lower N
T
value is, in principle, the most unfavourable for our aggregate interference evaluation because it will
result in the possibly higher activity factor (e.g. A
f
~ 1/14). However, also in this case, some practical considerations on
the actual uplink data rate should be taken into account.
The same ETSI BRAN draft SRD contains evaluation of the maximum data traffic/cell; its aim to evaluating the
spectrum need, therefore it made an average uplink + downlink traffic estimation.
Reprinted from Doc. JPTBWA(05)17 Annex 1 (may be found at ERO document server at www.ero.dk/download )
The traffic estimation will be done for the assumption of VDSLlike services, allowing broadband data and VoD, using
the shared traffic assumption.
Supplementary, will be calculated the data traffic generated by broadcast services, needed for tripleplay service
concept.
The 5 subscribers/cell, in rural like deployment, cannot provide any positive business case; due to this, the variant of
250m cell in Rural deployment has been omitted in the following calculations.
A.3.2.1 Shared traffic, VDSL like
The data rate calculation is done for the following assumptions:
 Peak data rate: 7Mb/s, UL+DL, shared between 20 users
 VoD using MPEG2, regular video, at 2Mb/s: 20% of users
 VoD using MPEG4, for HDTV, at 6Mb/s: 10% of users
 2 frequencies / cell
L = 2km L=1km L = 500 m L = 250m
Urban (Mb/s)
4860 1222 320 88
Suburban (Mb/s) 1620 412 108 34
Rural (Mb/s) 412 108 34
Table 9: Shared traffic / cell for VDSLlike services, Mb/s
The above data show that, with the same assumption of reuse 4 deployment, the capacity per sector (DL + UL) ranges
from 22 Mbit/s to maximum handled by the system.
Assuming a 1/3 UL/DL ratio as reasonable average, we could estimate:
Urban ~7 Mbit/s < ULdata/sector < maximum handled by the system
subUrban ~2.3 Mbit/s < ULdata/sector < maximum handled by the system.
In modern PMP systems designed for fixed/nomadic purpose the UL maximum data capacity is in the range of few tens
of Mbit/s (e.g. the BRAN HM objectives are for minimum 25 Mbit/s per sector, which is a reasonable value standing
the limited spectrum availability in 3.5 GHz bands).
Assuming this value as typical we could estimate that the full UL capacity in urban scenario might not be reached in the
lower cell size deployment of radius L=250 m, where a minimum N
T
~ 14 has been estimated. The activity factor, in
this worst assumed case, will be further limited by a factor 1/4 due to physical deployment reasons.
With same reasoning in suburban case, while the maximum D
TS
would be ~ 1/3 the A
f
will have the same upper bound
of ~1/56.
A.2.2.2 Summary of TS density and OOB eirp reference values:
Following the above evaluation of possible worst cases an average mitigations we would derive the parameters to be
used for the overall aggregation of interfering TSs.
ECC REPORT 33
Annex 2, Page 73
The device densities evaluated above are the overall expected TS population; however, we should consider that:
 A fraction of population belongs to the victim operator and then are not to be counted; assuming the market
shared by 3 operators only 2/3 of the TSs will add as interference.
 A fraction only of the interfering TS will be of indoor/omnidirectional nature here evaluated
 The interfering TS population will be spread over various adjacent channels; those operating on channels closer
to the wanted CS channels would predominate over those on farter channels.
Therefore it is reasonable to assume that only 1/3 of the potential interferers would actually concur to the aggregation.
Regarding the OOBlock eirp the reference value it would be derived as:
OOBlock (RECOMMENDATION) ÷ ATPC ÷ (Delta antenna gain) ÷ 10 log(A
f
)
÷ 37 ÷ 10 ÷ (164) 10log56 ~ ÷76.5 dBW/MHz or ÷ 46.5 dBm/MHz
Sector radius 0.25 km:
TS density urban: D
TS
= 900/3 = 300/km
2
TS density suburban: D
TS
= 300/3 = 100/km
2
Activity factor urban A
f
= (1/14)*1/4 = 1/56
Activity factor suburban A
f
= (1/5)*1/12 = 1/56
Reference OOBlock eirp ÷46.6 dBm/MHz
Sector radius 0.5 km:
TS density urban: D
TS
= 900/3 = 300/km
2
TS density suburban: D
TS
= 300/3 = 100/km
2
Activity factor urban A
f
= 1/56
Activity factor suburban A
f
~ (1/19)*1/3 ~ 1/56
Reference OOBlock eirp ÷46.6 dBm/MHz
These are the worst considered cases; in all other cell radius it can be seen that the factor (D
TS
* A
f
), which affects the
aggregate interference, have equal or lower value due to the constant density and a possible lower activity factor.
A.2.2.3 Scenarios
The analysis is made using a simulation program developed, within ECCTG3, for evaluating the aggregate interference
from a population density of indoor TS communication applications to a Fixed Service station (i.e. a FWA CS, for our
purpose). Therefore it is well suited for any other randomly distributed source of interference (i.e. a population of TSs
belonging to other FWA systems).
Annex 1 shows the basics of the model used.
A.2.2.4 Propagation models:
As for the CSTS interference study, the IEEE 802.16 adopted Erceg propagation model is used for NLOS paths,
added to an indoortooutdoor attenuation statistic as explained in Annex 1.
Erceg B variant has shown to be quite tailored for typical NLOS paths in Milan urban areas.
Cases of TS distributed on higher buildings will use freespace for the outdoor portion of the path, while maintaining
the indoor statistical distribution of attenuations.
Annex 1 gives all details.
A.2.2.5 Simulation results
In the following Pmed is the mean value of the aggregate interference evaluated as sum in power of each the aggregate
result snapshot divided by the number of total snapshot and is used as quick reference of the result, corresponding to
~ 90% probability of not exceeding the value.
ECC REPORT 33
Annex 2, Page 74
A.2.2.5.1 h
CS
= 30 m
Pmed Sector1 = 116.73 dB
Pmed Sector2 = 133.79 dB
Pmed Totale = 116.64 dB
Details per sector and propagation models split:
Sector 1:
Erceg B situations (73.13 %): Pmed = 119.23 dB
FSpace situations (26.87 %): Pmed = 120.32 dB
Sector 2:
Erceg B situations (72.95 %): Pmed = 168.01 dB
FSpace situations (27.05 %): Pmed = 133.80 dB
Figure 51: Total PdF for h
CS
= 30 m
A.2.2.5.2 h
CS
= 40 m
Pmed Sector1 = 118.27 dB
Pmed Sector2 = 133.79 dB
Pmed Totale = 118.16 dB
ECC REPORT 33
Annex 2, Page 75
Details per sector and propagation models split:
Sector 1:
Erceg B situations (73.13 %): Pmed = 121.28 dB
FSpace situations (26.87 %): Pmed = 121.29 dB
Sector 2:
Erceg B situations (72.95 %): Pmed = 163.84 dB
FSpace situations (27.05 %): Pmed = 133.80 dB
Figure 52: Total PdF for h
CS
= 40 m
ECC REPORT 33
Annex 2, Page 76
Figure 53: CDF comparison
A.2.2.6 Conclusions
The above MonteCarlo simulations show that:
 The contribution of the surrounding suburban area (sector 2) is negligible.
 With high cumulative probability (e.g. 95%) of the 40000 snapshots the aggregate interference to a CS antenna
30m height is lower than ÷115 dBm/MHz. This mean that, assuming the actual floors average height of ~3.4m,
on top of a 6 storey + ground floor building with 6m mast, which could be considered a quite worse situation in
urban areas
 This worst case is already 6 to °9 dB lower than the CS noise floor therefore contributing by less than 0.5 to °1
dB to threshold degradation.
 Higher CS antennas are already ~ 5dB more protected or alternatively the cumulative probability of similar
protection will become ~99%.
Therefore we could conclude that the proposed EIRP limit of 49 dBW/MHz is considered suitable also for omni
directional indoor TS, provided that it is reduced linearly with the antenna gain difference.
The study reported here shows that an outofblock EIRP value of 49 dBW/MHz, further linearly reduced by the
difference in antenna gain, coupled to the indoor use and its additional average attenuation might be considered safe
enough for the deployment of a considerable number of omnidirectional indoor terminals.
The linear scaling with antenna gain, for omnidirectional antennas, is however incidental; actually, omnidirectional
case here studied benefits also of the indoortooutdoor attenuation, which, in average, contributes by ~15 dB.
Therefore, omnidirectional antennas should be restricted, with the eirp estimated herein, to indoor use. Otherwise an
additional reduction of the limit is required.
ECC REPORT 33
Appendix 1 to Annex 2, Page 77
APPENDIX 1 TO ANNEX 2: MODEL BASIC DESCRIPTION
The model is based on data taken from aerial pictures of the city of Milan in northern Italy; it may represent a typical
European city.
Data on buildings height on a great 21 x 21 km area and on a more restricted 8 x 8 km urban area (see Figure 54)
encompassing more or less urban and suburban areas) were evaluated over ~ 184000 building entries, ~80000 of which
in the narrower urban area; by difference, the distribution on the suburban frame area is derived.
Building density is ~ 1200/km
2
in the urban area and ~ 280/km
2
in the suburban frame area.
The cumulative probability distribution of building height is shown in
Figure 55.
The probability distribution of floor # and its cumulative probability (assuming them ~ 3m height each) has been
derived from the total number of buildings and the related ~ 790000 floors (see Figure 56).
In such a way we could define through simple Monte Carlo trials an elevation for each interfering TS entry, assigning to
it the appropriate propagation model.
Figure 54: Milan example areas:
urban (square 8x8 km) and suburban (external cornice 21x21 km)
21 x 21 km
8 x 8 km
ECC REPORT 33
Appendix 1 to Annex 2, Page 78
Building height probability distribution
(urban area)
1,E06
1,E05
1,E04
1,E03
1,E02
1,E01
1,E+00
3 9
1
5
2
1
2
7
3
3
3
9
4
5
5
1
5
7
6
3
6
9
7
5
8
1
8
7
9
3
9
9
1
0
5
1
1
1
1
1
7
1
2
3
Building height (m)
H
i
e
g
h
t
p
r
o
b
a
b
i
l
i
t
y
/
t
o
t
a
l
N
u
m
b
e
r
o
f
b
u
i
l
d
i
n
g
s
Building height probability distribution
(suburban frame area)
1,E06
1,E05
1,E04
1,E03
1,E02
1,E01
1,E+00
3 9
1
5
2
1
2
7
3
3
3
9
4
5
5
1
5
7
6
3
6
9
7
5
8
1
8
7
9
3
9
9
1
0
5
1
1
1
1
1
7
1
2
3
Building height (m)
H
i
e
g
h
t
p
r
o
b
a
b
i
l
i
t
y
/
t
o
t
a
l
N
u
m
b
e
r
o
f
b
u
i
l
d
i
n
g
s
Figure 55: Probability distribution of building heights within Milan,
urban and suburban cornice areas
Floors # distribution
(Urban)
1,E06
1,E05
1,E04
1,E03
1,E02
1,E01
1,E+00
1 2 3 4 5 6 7 8 9
1
0
1
1
1
2
1
3
1
4
1
5
1
6
1
7
1
8
1
9
2
0
2
1
2
2
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
0
3
1
3
2
3
3
3
4
3
5
3
6
3
7
3
8
3
9
4
0
4
1
4
2
Floor #
F
l
o
o
r
#
p
r
o
b
a
b
i
l
i
t
y
/
t
o
t
a
l
N
u
m
b
e
r
o
f
f
l
o
o
r
s
Floors # distribution
(suburban frame area)
1,E06
1,E05
1,E04
1,E03
1,E02
1,E01
1,E+00
1 2 3 4 5 6 7 8 9
1
0
1
1
1
2
1
3
1
4
1
5
1
6
1
7
1
8
1
9
2
0
2
1
2
2
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
0
3
1
3
2
3
3
3
4
3
5
3
6
3
7
3
8
3
9
4
0
4
1
4
2
Floor #
F
l
o
o
r
#
p
r
o
b
a
b
i
l
i
t
y
/
t
o
t
a
l
N
u
m
b
e
r
o
f
f
l
o
o
r
s
Note: Floor #1 means ground floor
Figure 56: Probability distribution of floor # within the areas of Figure 54
ECC REPORT 33
Appendix 1 to Annex 2, Page 79
Geometrical description of the model
The scenario has been described as semicircles areas around FWA base station shown in Figure 57.
It simulate the worst case of a CS sector at urban area border facing the urban area, it is also quite conservative because
the same 8 km radius of urban density is maintained up to ± 180°, while it should have been reduced to ±4 km.
Figure 57: Description of different TS distribution areas in the model
Each scenario is represented by a 180° semicircular area; each one is subdivided in sectors where TS deployment is
different i.e.:
Sector 1 portion with urban terminal density
Sector 2 portion with Suburban terminal density
In each sector a pattern of TS devices is established with the required density over the horizontal plane.
Description of MonteCarlo snapshots
Step 1: Assigning a TS height
Each TS device is given with an elevation randomly derived from the relevant building/floor distribution.
Figure 58 pictorially show the random height assignment.
12 10 8 6 4 2 0 2 4 6 8 10 12
0
2
4
6
8
10
12
14
16
18
20
km
1
2
ECC REPORT 33
Appendix 1 to Annex 2, Page 80
Figure 58: Visual example of TS height distribution over model areas
Step 2: Defining the suitable outdoor propagation model
Each TS now allocated in the space, will be given a propagation model depending its height (i.e. Erceg B or freespace);
arrival angle and distance is calculated and path loss, including relevant antenna RPE contribution is calculated.
 TS distributed below the mean building height (i.e. ~15m in urban building distribution and ~10m for sub
urban):
Outdoor Propagation Erceg B; Gaussian shadowing variance o = 9 dB (truncated at ± 2 o)
 TS distributed on height above: will be considered LOS freespace.
Step 3: Defining the additional indoor to outdoor attenuation
For each TS, an indoor + building exit additional attenuation is picked up from the random distribution described
below.
For defining a suitable distribution we consider areas of a “typical floor” and assume that each building will face the FS
station with one side only (if the building orientation is tilted, only portions of two sides will count, maintaining the
same average exposition).
On the side facing the victim antenna the lower attenuation is expected and on the opposite side higher attenuation
would be added.
Assuming typical building area to be ~400 m
2
will give an average 20 x 20m square building footprint. A continuous
distribution derived through the convolution (representing the joint probability) of building through wall attenuation
Gaussian distribution given in ITUR P.1411 and the indoor attenuation, additional to free space, given by the two slope
Siwiak
21
model (d
t
=3m) extended over the floors area. An additional step of 10db, representing the additional
attenuation of internal elevators and stairs areas, is introduced in the middle of the building as shown in Figure 59 and
Figure 60.
This seems also conservative when considering that no floor trespassing attenuation is considered; this would add high
contribution to TSs closer to the victim CS, which might significantly contribute standing the limited elevation RPE
discrimination of CS antennas.
21
The path loss PL(d) derived from the Siwiak study, where f
m
is the geometrical mean of the signal frequency, and c is the velocity
of propagation and d
t
is the break point where the 20dB freespace attenuation slope, gives way to a 30dB attenuation slope:
PL(d )= ÷10log{[c/4tdfm]
2
[
1÷
exp(
÷
(dt /d)
¸÷2
)]} PL(d )= ÷10log{[c/4tdfm]
2
[
1÷
exp(
÷
(dt /d)
¸÷2
)]}
ECC REPORT 33
Appendix 1 to Annex 2, Page 81
0 2 4 6 8 10 12 14 16 18 20 22 24
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
dB
P
r
o
b
a
b
i
l
i
t
y
[
%
]
Gaussian Distribution
Standard deviation: o = 5 dB
Mean: µ = 12 dB
Figure 59: Probability density and cumulative probability of
the P.1411 penetration distribution
0
2
4
6
8
10
12
14
16
18
20
0 2 4 6 8 10 12 14 16 18 20
Distance from the exit window (m)
D
e
l
t
a
a
t
t
e
n
u
a
t
i
o
n
o
v
e
r
f
r
e
e
s
p
a
c
e
(
d
B
)
dt=3m
dt=3m+10dB step
Figure 60: Attenuation of the indoor path additional to freespace
(each value assumed to have equal probability)
The joint probability obtained from convolution of the two distributions is shown in Figure 61 as probability density
and cumulative probability
ECC REPORT 33
Appendix 1 to Annex 2, Page 82
10 0 10 20 30 40 50 60 70
0
1
2
3
4
5
6
7
8
9
10
Atten. [dB]
P
r
o
b
a
b
i
l
i
t
y
[
%
]
PdF
10 0 10 20 30 40 50 60 70
0
10
20
30
40
50
60
70
80
90
100
Atten. [dB]
P
r
o
b
a
b
i
l
i
t
y
[
%
]
CdF
Figure 61: Probability density and cumulative probability of
the joint attenuation functions.
Step 4
Using the attenuation of each TS device, evaluated in previous steps 1 to 3, the received power density generated by
each TS is evaluated and cumulated for all TS population within the same snapshot subdivided into the different sectors
and into the overall aggregation.
Step 5
After concluding the first snapshot, described in the previous steps, other ~39.999 similar snapshots are conducted and
a probability density and cumulative probability function of the aggregate PSD is calculated.
Separate evaluations are made for the mean power density value (Note) in each sector and in the whole area.
Note: the mean power density values are obtained as the power sum of aggregation results of all snapshots
divided by 40000 (number of snapshots). It roughly corresponds to the value not exceeded for 90% of the
snapshots.
ECC REPORT 33
Appendix 2 to Annex 2, Page 83
APPENDIX 2 TO ANNEX 2: ACCEPTANCEREJECTION METHOD
i. Generate three uniform random deviates U1, U2, U3. U3 is a spare deviate to be subsequently described.
ii. Let F
max
be the maximum value of a normalized Rayleigh distribution.
iii. Compute a probability point P
r
(3U2) based on the Rayleigh probability equation and within a finite truncated
range for U2. Setting the range for U2 to be within (0, 3) allows Rayleigh fades to span the range from ÷·
to +10 dB.
iv. Examine the ratio u = P
r
(3U2)/ F
max
. If the ratio u is less than U1, then accept U2 as the random deviate. If
not, then reject the triplet and start again.
v. Once accepted as a valid Rayleigh deviate, the adjustment to the FSL signal level is 20log(U2).
Random deviate U3 was not required in the preceding. However, once U1 and U2 are accepted, the associated U3 value
is employed to identify the probability of Rayleigh fading at some transmission distance R
x
. For Rayleigh fading, the
probability of it's occurrence is known to vary as the 3
rd
power of the distance [4]. The simulation assumptions are as
follows:
a) Under Rayleigh fading conditions, set the probability of a Rayleigh fade at maximum distance R
max
to be
) (
max
R µ = 1. For some lesser distance, say R
x
, set the probability to be µ( ) ( / )
max
R R R
x x
d
coeff
= , where
d
coeff
= 3.
b) Compare the value of µ( ) R
x
with that of U3. If U3 > µ( ) R
x
, then conclude that there is no Rayleigh
fading on the link. If U3 < µ( ) R
x
, then set the Rayleigh fading adjustment to be that given by step v.
above.
ECC REPORT 33
Annex 3, Page 84
ANNEX 3: EXAMPLES FOR MANAGING A CS BLOCKEDGE MASK
When it is considered appropriate a complete “technology independent/uncoordinated use of the bands, the BEM
methodology seems the easiest way to contain mutual interference while minimising the need for detailed coordination.
Using a blockedge mask concept implies that operators should meet the requirements having freedom on three
elements only:
1. The EIRP level
2. The minimum frequency separation from edge of outermost channels
3. The transmit spectrum mask attenuation enhancement.
The first parameter is intended for maximising coverage, while the other two are strictly related to the actual equipment
implementation. Manufacturers might improve the transmitter spectrum mask (and then the possibility of going closer
to the block edge) by actually offering guaranteed masks that, at least for the CSs, are tighter than the minimum ETSI
requirement.
Managing these three elements, equipment manufacturer and an operator can define systems parameters that better fit
the network requirements addressed (e.g. for rural or for urban applications).
The provisions in this section have been based on coexistence studies described in this report; it should here be noted
that those studies are mostly made with statistical tools and typical radio systems, deployment and service performance
objectives assumptions.
Consequently, a block edge mask is generally designed on the basis of a small level of degradation in an assumed
interference scenario with a low occurrence probability of a worst case (e.g. low probability of two directional antennas
pointing exactly at each other). It is not therefore excluded that in a limited number of cases specific mitigation
techniques might be necessary.
In particular when CSs are colocated on the same building, the statistical approach is not applicable and it is assumed
that common practice of site engineering (e.g. vertical decoupling) is implemented for improving antenna decoupling as
much as possible.
Also adjacent block receiver rejection concurs to a reduced interference scenario, however this is not in the scope of this
recommendation to set limits for it; nevertheless it is expected that ETSI standards will adequately cover the issue.
It should be also noted that when TDD or mixed FDD/TDD systems are placed in immediately adjacent bands, the
probability of occurrence of worse cases of interference between CSs is quite higher than in situations where only FDD
are deployed. Therefore, even if the mask here proposed offer a suitably low occurrence probability of interference for
such cases, when TDD systems are concerned additional mitigation techniques (geographic separation of stations,
natural/physical shielding, etc) or additional coordination (including networks synchronisation) between operators
should be implemented as far as possible.
The current studies has shown that less directional antennas (either CS or TS) generally produce more probability of
interference; therefore outofblock emissions in terms of EIRP should be more stringent for lower directivity (and
consequently with lower gain) antennas. That is why the recommended block edge mask limits outside the block are
here described in terms of transmitter output power, allowing operators to make practical use of this phenomenon by
obtaining higher EIRP when using highly directional hence less interfering antennas, while EIRP would be
automatically lowered when low gain (e.g. omnidirectional antennas) are employed.
The transition zone, shown between X
1
and X
3
power density levels (see Figure 3), is not here considered in terms of
IPF or NFD, but should be determined through a tradeoff between the need of a limited amount of guardband, the
possible slight decrement of NFD available for the blockoutermost system channels and the practical feasibility of
equipment filtering for reaching the required X
3
level from inblock P
max
density, which is generally defined for suitable
service and sharing conditions in the licensing process.
Implementing a smooth transition zone implies that the amount of protection within a block is not constant but increases
as far as the operating channel is removed from the edge.
Therefore, the example mask here considered assumes adjacent blocks to be sized from 4 typical system channels plus
an internal guard band as recommended in Annex 1 of ECC/REC(04)05, with increasingly protected frequency areas:
 Internal guard bands areas where protection is not offered unless the interested operators would practice active
coordination
ECC REPORT 33
Annex 3, Page 85
 Outermost system channels areas where protection is given with high probability, but in few worse cases
coordination between CS might be needed. However, in most cases, administrations require CS notification of
emissions parameters (i.e. transmitter EIRP and antenna data with geographical locations) and make the data
base available; therefore, the operator, with the knowledge of other CS already deployed, may operate self
coordination avoiding worse cases and eventually use innermost systems channels in the block that are more
protected.
 Innermost system channels areas where protection is given with very high probability.
The increasing dropdown attenuation at the block edge has the scope of easing TX filtering. Its size has been chosen
taking into account the block size typically needed for broadband applications that will likely dominate current and
future FWA licenses in these bands. The dropdown portion, with higher permitted outofblock emission, mostly fall
within a guardband (internal or external to the block), thus not affecting expected mutual interference.
In the following examples the transition zone of blockedge mask is analyses against the 16 QAM TDMA type B and
OFDMA/TDMA type F from ETSI 302 3262 equipment masks for 3.5 MHz channels.
In particular the following considerations are made:
 Analysis of typical block size: 4*3.5 + 1*3.5(guard) = 17.5 MHz block.
 outermost system channel placed at the reference distance of 3.5 MHz from edge (i.e. not wasting spectrum but also
not gaining any)
 Analysis over 16 QAM (considered a sensible reference). In modern adaptive equipment QPSK is considered to fit
the higher order masks; 64 QAM is considered an option that not all systems might offer.
 5/6 resonators filter as medium high complexity realization or 7 resonators filters as feasible limit for stable and
effective hardware design.
 wideband filters (i.e. 14 MHz bandwidth) for maintaining frequency agility within the block.
 System CS output power +35 dBm (i.e. ~+30.5 dBm/MHz of power density), considered a typical target for indoor
TS coverage.
The result in terms of the maximum power density allowed by the mask is shown in Figure 62. Without extra filtering
the system could not meet the mask unless with very low power, with 5 cavity filter the power should be reduced by
~10 dB, while using 6 and 7 cavity the target is feasible with the desired power.
As a second example, in Figure 63 systems at proposed power density are allowed to be placed nearer to blockedge as
far as the filter complexity increases. Also in this case unfiltered equipment can not meet the edge mask unless with
very large power reduction.
ECC REPORT 33
Annex 3, Page 86
TDMA
OFDMA/TDMA
Figure 62: Examples of increasing power density with RF filtering at same edge distance
ECC REPORT 33
Annex 3, Page 87
TDMA
OFDMA/TDMA
Figure 63: Example of decreasing edge distance with RF filtering at same max EIRP
ECC REPORT 33
Annex 4, Page 88
ANNEX 4: EVALUATION OF THE C/I RATIO EXCEEDED FOR MORE THAN 99% (AND 97% FOR
OMNIDIRECTIONAL CASE) OF PROBABILITY.
The following Figures summarise the worst cases of the large number of different conditions evaluated in Table 35to
Table 39. In addition, for evaluating the possible degradation gradient, Figure 65compares similar evaluation of
directional antennas with 1% and 3% occurrence probability.
Figure 66summarise the worst omnidirectional situation at various occurrence probabilities.
60
50
40
30
20
10
0
0,00 0,20 0,40 0,60 0,80 1,00 1,20
D/R
C
/
I
(
d
B
)
Ag=10dB / C / SUI2 / 15m
Ag=10dB / B / SUI4 / 4m
Ag=18dB / C / SUI2 / 15m
Ag=18dB / B / SUI4 / 4m
Figure 64: Summary results for worse C/I cases (1% occurrence probability) with directional antennas
(Derived from Table 35through Table 38)
Figure 65: Comparisons of worse C/I cases (Erceg BSUI4)
(1% and 3% occurrence probability) with 10 dB directional antennas
Fig 11A
1% prob.
Fig 11B
1% prob.
Fig 11B
3% prob.
Fig 11A
3% prob.
0 0.25 0.5 0.75 1 1.25 1.5
50
40
30
20
10
0
1
0
D / R
C
/
I
(
d
B
)
ECC REPORT 33
Annex 4, Page 89
Figure 66: Summary results for worse C/I cases (Erceg BSUI4)
(1% to 5% occurrence probability) with omnidirectional antennas
Table 35: C/I exceeded for more than 99%
Directional antennas Cell radius 0.5 and 1 km
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
0,5 ErcegB SUI4 4 10 0 32,82
0,5 ErcegB SUI4 10 10 0 32,82
0,5 ErcegB SUI4 15 10 0 32,82
0,5 ErcegC SUI2 4 10 0 31
0,5 ErcegC SUI2 10 10 0 31
0,5 ErcegC SUI2 15 10 0 31
0,5 ErcegB SUI4 10 18 0 32,82
0,5 ErcegB SUI4 15 18 0 32,82
0,5 ErcegB SUI4 4 18 0 32,81
0,5 ErcegC SUI2 4 18 0 31
0,5 ErcegC SUI2 10 18 0 31
0,5 ErcegC SUI2 15 18 0 31
0,5 ErcegB SUI4 4 10 1/4 40,44
0,5 ErcegB SUI4 10 10 1/4 40,44
0,5 ErcegB SUI4 15 10 1/4 40,44
0,5 ErcegC SUI2 15 10 1/4 38,03
0,5 ErcegC SUI2 4 10 1/4 38,02
0,5 ErcegC SUI2 10 10 1/4 38,02
0,5 ErcegB SUI4 10 18 1/4 37,63
0,5 ErcegB SUI4 15 18 1/4 37,63
0,5 ErcegB SUI4 4 18 1/4 37,62
0,5 ErcegC SUI2 4 18 1/4 35,33
0,5 ErcegC SUI2 10 18 1/4 35,33
0,5 ErcegC SUI2 15 18 1/4 35,33
0,5 ErcegB SUI4 4 10 1/2 46,24
0,5 ErcegB SUI4 10 10 1/2 46,24
0,5 ErcegB SUI4 15 10 1/2 46,24
0,5 ErcegC SUI2 15 10 1/2 43,39
Fig 11A
5% prob.
Fig 11A
1% prob.
Fig 11B
5% prob.
Fig 11B
1% prob.
Fig 11B
4% prob.
Fig 11B
3% prob.
Fig 11A
4% prob.
Fig 11A
3% prob.
ECC REPORT 33
Annex 4, Page 90
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
0,5 ErcegC SUI2 4 10 1/2 43,38
0,5 ErcegC SUI2 10 10 1/2 43,38
0,5 ErcegB SUI4 4 18 1/2 41,74
0,5 ErcegB SUI4 10 18 1/2 41,74
0,5 ErcegB SUI4 15 18 1/2 41,74
0,5 ErcegC SUI2 4 18 1/2 39,01
0,5 ErcegC SUI2 10 18 1/2 39,01
0,5 ErcegC SUI2 15 18 1/2 39,01
0,5 ErcegB SUI4 4 10 3/4 49,77
0,5 ErcegB SUI4 10 10 3/4 49,77
0,5 ErcegB SUI4 15 10 3/4 49,77
0,5 ErcegC SUI2 4 10 3/4 46,57
0,5 ErcegC SUI2 10 10 3/4 46,57
0,5 ErcegC SUI2 15 10 3/4 46,57
0,5 ErcegB SUI4 4 18 3/4 44,11
0,5 ErcegB SUI4 10 18 3/4 44,11
0,5 ErcegB SUI4 15 18 3/4 44,11
0,5 ErcegC SUI2 4 18 3/4 41
0,5 ErcegC SUI2 10 18 3/4 41
0,5 ErcegC SUI2 15 18 3/4 41
0,5 ErcegB SUI4 10 10 0.99 23,52
0,5 ErcegB SUI4 15 10 0.99 23,52
0,5 ErcegB SUI4 4 10 0.99 23,51
0,5 ErcegC SUI2 4 10 0.99 20,22
0,5 ErcegC SUI2 10 10 0.99 20,22
0,5 ErcegC SUI2 15 10 0.99 20,22
0,5 ErcegB SUI4 4 18 0.99 13,14
0,5 ErcegB SUI4 10 18 0.99 13,14
0,5 ErcegB SUI4 15 18 0.99 13,14
0,5 ErcegC SUI2 4 18 0.99 9,81
0,5 ErcegC SUI2 10 18 0.99 9,81
0,5 ErcegC SUI2 15 18 0.99 9,81
Table 36: C/I exceeded for more than 99%
Directional antennas Cell radius 1 km
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
1 ErcegB SUI4 4 10 0 32,82
1 ErcegB SUI4 10 10 0 32,82
1 ErcegB SUI4 15 10 0 32,82
1 ErcegC SUI2 15 10 0 31,01
1 ErcegC SUI2 4 10 0 31,00
1 ErcegC SUI2 10 10 0 31,00
1 ErcegB SUI4 4 18 0 32,82
1 ErcegB SUI4 10 18 0 32,82
1 ErcegB SUI4 15 18 0 32,82
1 ErcegC SUI2 4 18 0 31,00
1 ErcegC SUI2 10 18 0 31,00
1 ErcegC SUI2 15 18 0 31,00
1 ErcegB SUI4 4 10 1/4 40,44
1 ErcegB SUI4 10 10 1/4 40,44
1 ErcegB SUI4 15 10 1/4 40,44
1 ErcegC SUI2 4 10 1/4 38,03
1 ErcegC SUI2 10 10 1/4 38,03
1 ErcegC SUI2 15 10 1/4 38,02
1 ErcegB SUI4 4 18 1/4 37,63
ECC REPORT 33
Annex 4, Page 91
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
1 ErcegB SUI4 10 18 1/4 37,63
1 ErcegB SUI4 15 18 1/4 37,62
1 ErcegC SUI2 4 18 1/4 35,33
1 ErcegC SUI2 10 18 1/4 35,33
1 ErcegC SUI2 15 18 1/4 35,33
1 ErcegB SUI4 4 10 1/2 46,24
1 ErcegB SUI4 10 10 1/2 46,24
1 ErcegB SUI4 15 10 1/2 46,24
1 ErcegC SUI2 4 10 1/2 43,38
1 ErcegC SUI2 10 10 1/2 43,38
1 ErcegC SUI2 15 10 1/2 43,38
1 ErcegB SUI4 4 18 1/2 41,74
1 ErcegB SUI4 10 18 1/2 41,74
1 ErcegB SUI4 15 18 1/2 41,74
1 ErcegC SUI2 4 18 1/2 39,01
1 ErcegC SUI2 10 18 1/2 39,01
1 ErcegC SUI2 15 18 1/2 39,01
1 ErcegB SUI4 10 10 3/4 49,77
1 ErcegB SUI4 15 10 3/4 49,77
1 ErcegB SUI4 4 10 3/4 49,76
1 ErcegC SUI2 4 10 3/4 46,57
1 ErcegC SUI2 10 10 3/4 46,57
1 ErcegC SUI2 15 10 3/4 46,57
1 ErcegB SUI4 4 18 3/4 44,11
1 ErcegB SUI4 10 18 3/4 44,11
1 ErcegB SUI4 15 18 3/4 44,11
1 ErcegC SUI2 4 18 3/4 41,00
1 ErcegC SUI2 10 18 3/4 41,00
1 ErcegC SUI2 15 18 3/4 41,00
1 ErcegB SUI4 10 10 0.99 23,52
1 ErcegB SUI4 4 10 0.99 23,51
1 ErcegB SUI4 15 10 0.99 23,51
1 ErcegC SUI2 4 10 0.99 20,22
1 ErcegC SUI2 10 10 0.99 20,22
1 ErcegC SUI2 15 10 0.99 20,22
1 ErcegB SUI4 4 18 0.99 13,14
1 ErcegB SUI4 10 18 0.99 13,14
1 ErcegB SUI4 15 18 0.99 13,14
1 ErcegC SUI2 15 18 0.99 9,82
1 ErcegC SUI2 4 18 0.99 9,81
1 ErcegC SUI2 10 18 0.99 9,81
Table 37: C/I exceeded for more than 99%
Directional antennas Cell radius 2 km
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
2 ErcegB SUI4 4 10 0 32,82
2 ErcegB SUI4 10 10 0 32,82
2 ErcegB SUI4 15 10 0 32,82
2 ErcegC SUI2 4 10 0 31,00
2 ErcegC SUI2 10 10 0 31,00
2 ErcegC SUI2 15 10 0 31,00
2 ErcegB SUI4 4 18 0 32,82
2 ErcegB SUI4 15 18 0 32,82
2 ErcegB SUI4 10 18 0 32,81
2 ErcegC SUI2 4 18 0 31,00
ECC REPORT 33
Annex 4, Page 92
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
2 ErcegC SUI2 10 18 0 31,00
2 ErcegC SUI2 15 18 0 31,00
2 ErcegB SUI4 4 10 1/4 40,44
2 ErcegB SUI4 15 10 1/4 40,44
2 ErcegB SUI4 10 10 1/4 40,43
2 ErcegC SUI2 10 10 1/4 38,03
2 ErcegC SUI2 15 10 1/4 38,03
2 ErcegC SUI2 4 10 1/4 38,02
2 ErcegB SUI4 4 18 1/4 37,62
2 ErcegB SUI4 10 18 1/4 37,62
2 ErcegB SUI4 15 18 1/4 37,62
2 ErcegC SUI2 4 18 1/4 35,33
2 ErcegC SUI2 10 18 1/4 35,33
2 ErcegC SUI2 15 18 1/4 35,33
2 ErcegB SUI4 10 10 1/2 46,24
2 ErcegB SUI4 15 10 1/2 46,24
2 ErcegB SUI4 4 10 1/2 46,23
2 ErcegC SUI2 4 10 1/2 43,38
2 ErcegC SUI2 10 10 1/2 43,38
2 ErcegC SUI2 15 10 1/2 43,38
2 ErcegB SUI4 4 18 1/2 41,74
2 ErcegB SUI4 10 18 1/2 41,74
2 ErcegB SUI4 15 18 1/2 41,74
2 ErcegC SUI2 4 18 1/2 39,01
2 ErcegC SUI2 10 18 1/2 39,01
2 ErcegC SUI2 15 18 1/2 39,01
2 ErcegB SUI4 4 10 3/4 49,77
2 ErcegB SUI4 10 10 3/4 49,77
2 ErcegB SUI4 15 10 3/4 49,76
2 ErcegC SUI2 4 10 3/4 46,57
2 ErcegC SUI2 10 10 3/4 46,57
2 ErcegC SUI2 15 10 3/4 46,57
2 ErcegB SUI4 4 18 3/4 44,11
2 ErcegB SUI4 10 18 3/4 44,11
2 ErcegB SUI4 15 18 3/4 44,11
2 ErcegC SUI2 4 18 3/4 41,00
2 ErcegC SUI2 10 18 3/4 41,00
2 ErcegC SUI2 15 18 3/4 41,00
2 ErcegB SUI4 4 10 0.99 23,52
2 ErcegB SUI4 10 10 0.99 23,51
2 ErcegB SUI4 15 10 0.99 23,51
2 ErcegC SUI2 4 10 0.99 20,22
2 ErcegC SUI2 10 10 0.99 20,22
2 ErcegC SUI2 15 10 0.99 20,22
2 ErcegB SUI4 4 18 0.99 13,15
2 ErcegB SUI4 10 18 0.99 13,14
2 ErcegB SUI4 15 18 0.99 13,14
2 ErcegC SUI2 10 18 0.99 9,82
2 ErcegC SUI2 4 18 0.99 9,81
2 ErcegC SUI2 15 18 0.99 9,81
Table 38: C/I exceeded for more than 99%
Directional antennas Cell radius 2.7 km
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
2,7 ErcegB SUI4 4 10 0 32,82
ECC REPORT 33
Annex 4, Page 93
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
2,7 ErcegB SUI4 10 10 0 32,82
2,7 ErcegB SUI4 15 10 0 32,82
2,7 ErcegC SUI2 4 10 0 31,00
2,7 ErcegC SUI2 10 10 0 31,00
2,7 ErcegC SUI2 15 10 0 31,00
2,7 ErcegB SUI4 10 18 0 32,82
2,7 ErcegB SUI4 15 18 0 32,82
2,7 ErcegB SUI4 4 18 0 32,81
2,7 ErcegC SUI2 4 18 0 31,00
2,7 ErcegC SUI2 10 18 0 31,00
2,7 ErcegC SUI2 15 18 0 31,00
2,7 ErcegB SUI4 10 10 1/4 40,44
2,7 ErcegB SUI4 15 10 1/4 40,44
2,7 ErcegB SUI4 4 10 1/4 40,43
2,7 ErcegC SUI2 4 10 1/4 38,03
2,7 ErcegC SUI2 15 10 1/4 38,03
2,7 ErcegC SUI2 10 10 1/4 38,02
2,7 ErcegB SUI4 10 18 1/4 37,63
2,7 ErcegB SUI4 15 18 1/4 37,63
2,7 ErcegB SUI4 4 18 1/4 37,62
2,7 ErcegC SUI2 4 18 1/4 35,33
2,7 ErcegC SUI2 10 18 1/4 35,33
2,7 ErcegC SUI2 15 18 1/4 35,33
2,7 ErcegB SUI4 4 10 1/2 46,24
2,7 ErcegB SUI4 10 10 1/2 46,24
2,7 ErcegB SUI4 15 10 1/2 46,24
2,7 ErcegC SUI2 15 10 1/2 43,39
2,7 ErcegC SUI2 4 10 1/2 43,38
2,7 ErcegC SUI2 10 10 1/2 43,38
2,7 ErcegB SUI4 4 18 1/2 41,74
2,7 ErcegB SUI4 10 18 1/2 41,74
2,7 ErcegB SUI4 15 18 1/2 41,74
2,7 ErcegC SUI2 4 18 1/2 39,01
2,7 ErcegC SUI2 10 18 1/2 39,01
2,7 ErcegC SUI2 15 18 1/2 39,01
2,7 ErcegB SUI4 4 10 3/4 49,77
2,7 ErcegB SUI4 10 10 3/4 49,77
2,7 ErcegB SUI4 15 10 3/4 49,77
2,7 ErcegC SUI2 4 10 3/4 46,57
2,7 ErcegC SUI2 10 10 3/4 46,57
2,7 ErcegC SUI2 15 10 3/4 46,57
2,7 ErcegB SUI4 4 18 3/4 44,11
2,7 ErcegB SUI4 10 18 3/4 44,11
2,7 ErcegB SUI4 15 18 3/4 44,11
2,7 ErcegC SUI2 4 18 3/4 41,00
2,7 ErcegC SUI2 10 18 3/4 41,00
2,7 ErcegC SUI2 15 18 3/4 41,00
2,7 ErcegB SUI4 10 10 0.99 23,52
2,7 ErcegB SUI4 4 10 0.99 23,51
2,7 ErcegB SUI4 15 10 0.99 23,51
2,7 ErcegC SUI2 4 10 0.99 20,22
2,7 ErcegC SUI2 10 10 0.99 20,22
2,7 ErcegC SUI2 15 10 0.99 20,22
2,7 ErcegB SUI4 4 18 0.99 13,14
2,7 ErcegB SUI4 10 18 0.99 13,14
2,7 ErcegB SUI4 15 18 0.99 13,14
ECC REPORT 33
Annex 4, Page 94
R [km] Erceg Category SUI h TS [m] Ant.Gain [dB] D
CS
/R
C/I [dB]
at 99%
2,7 ErcegC SUI2 4 18 0.99 9,82
2,7 ErcegC SUI2 10 18 0.99 9,81
2,7 ErcegC SUI2 15 18 0.99 9,81
Table 39: C/I exceeded for more than 99% or 97%
OmniDirectional antennas  Cell radius 0.5, 1 and 2 km
R [km] Erceg Category SUI h TS [m] D CS [m]
C/I [dB]
at 99%
C/I [dB]
at 97%
0,5 ErcegC SUI2 4 0 32,25 26,16
0,5 ErcegC SUI2 10 0 32,25 26,16
0,5 ErcegC SUI2 15 0 32,25 26,16
0,5 ErcegB SUI4 4 0 31,91 25,9
0,5 ErcegB SUI4 10 0 31,91 25,9
0,5 ErcegB SUI4 15 0 31,91 25,9
0,5 ErcegB SUI4 10 1/4 43,82 35,52
0,5 ErcegB SUI4 15 1/4 43,62 35,48
0,5 ErcegB SUI4 4 1/4 43,47 35,4
0,5 ErcegC SUI2 10 1/4 42,76 34,85
0,5 ErcegC SUI2 15 1/4 42,75 34,89
0,5 ErcegC SUI2 4 1/4 42,67 34,94
0,5 ErcegB SUI4 10 1/2 52,98 43,19
0,5 ErcegB SUI4 4 1/2 52,73 43,16
0,5 ErcegB SUI4 15 1/2 51,94 42,76
0,5 ErcegC SUI2 10 1/2 51,59 42,03
0,5 ErcegC SUI2 15 1/2 51,57 42,09
0,5 ErcegC SUI2 4 1/2 51,33 41,93
0,5 ErcegB SUI4 15 3/4 60,32 48,51
0,5 ErcegB SUI4 4 3/4 60,12 48,67
0,5 ErcegB SUI4 10 3/4 59,38 48,07
0,5 ErcegC SUI2 15 3/4 58,17 46,77
0,5 ErcegC SUI2 4 3/4 57,8 46,71
0,5 ErcegC SUI2 10 3/4 57,07 46,11
0,5 ErcegB SUI4 4 0.99 52,52 37,67
0,5 ErcegB SUI4 15 0.99 51,57 37,75
0,5 ErcegB SUI4 10 0.99 51,14 37,47
0,5 ErcegC SUI2 15 0.99 50,98 37,65
0,5 ErcegC SUI2 10 0.99 50,51 36,44
0,5 ErcegC SUI2 4 0.99 49,63 36,36
1 ErcegC SUI2 4 0 32,25 26,16
1 ErcegC SUI2 10 0 32,25 26,16
1 ErcegC SUI2 15 0 32,25 26,16
1 ErcegB SUI4 4 0 31,91 25,9
1 ErcegB SUI4 10 0 31,91 25,9
1 ErcegB SUI4 15 0 31,91 25,9
1 ErcegB SUI4 4 1/4 43,84 35,48
1 ErcegB SUI4 10 1/4 43,63 35,33
1 ErcegB SUI4 15 1/4 43,17 35,17
1 ErcegC SUI2 10 1/4 42,88 34,94
1 ErcegC SUI2 4 1/4 42,83 34,9
1 ErcegC SUI2 15 1/4 42,69 34,89
1 ErcegB SUI4 10 1/2 53,55 43,46
1 ErcegB SUI4 15 1/2 52,93 43,25
ECC REPORT 33
Annex 4, Page 95
R [km] Erceg Category SUI h TS [m] D CS [m]
C/I [dB]
at 99%
C/I [dB]
at 97%
1 ErcegC SUI2 4 1/2 52,68 42,8
1 ErcegB SUI4 4 1/2 52,62 42,99
1 ErcegC SUI2 10 1/2 52,06 42,47
1 ErcegC SUI2 15 1/2 51,78 42,18
1 ErcegB SUI4 4 3/4 59,02 47,83
1 ErcegB SUI4 15 3/4 58,87 47,82
1 ErcegB SUI4 10 3/4 58,67 47,52
1 ErcegC SUI2 15 3/4 57,19 46,33
1 ErcegC SUI2 4 3/4 57,11 46,04
1 ErcegC SUI2 10 3/4 56,13 45,51
1 ErcegB SUI4 15 0.99 53,53 38,79
1 ErcegB SUI4 10 0.99 52,44 37,62
1 ErcegB SUI4 4 0.99 52,39 38,6
1 ErcegC SUI2 4 0.99 50,91 36,02
1 ErcegC SUI2 10 0.99 50,70 36,83
1 ErcegC SUI2 15 0.99 49,86 35,86
2 ErcegC SUI2 4 0 32,25 26,16
2 ErcegC SUI2 10 0 32,25 26,16
2 ErcegC SUI2 15 0 32,25 26,16
2 ErcegB SUI4 4 0 31,91 25,9
2 ErcegB SUI4 10 0 31,91 25,9
2 ErcegB SUI4 15 0 31,91 25,9
2 ErcegB SUI4 10 1/4 43,61 35,48
2 ErcegB SUI4 4 1/4 43,30 35,3
2 ErcegC SUI2 15 1/4 43,10 35,04
2 ErcegB SUI4 15 1/4 43,05 35,22
2 ErcegC SUI2 10 1/4 42,81 34,84
2 ErcegC SUI2 4 1/4 42,67 34,72
2 ErcegB SUI4 15 1/2 54,24 43,81
2 ErcegB SUI4 4 1/2 53,94 43,92
2 ErcegB SUI4 10 1/2 53,50 43,34
2 ErcegC SUI2 10 1/2 51,95 42,02
2 ErcegC SUI2 15 1/2 51,52 41,99
2 ErcegC SUI2 4 1/2 51,05 41,85
2 ErcegB SUI4 10 3/4 60,22 48,81
2 ErcegB SUI4 4 3/4 59,94 48,62
2 ErcegB SUI4 15 3/4 59,24 47,98
2 ErcegC SUI2 10 3/4 58,12 46,84
2 ErcegC SUI2 4 3/4 57,31 45,85
2 ErcegC SUI2 15 3/4 57,31 46,36
2 ErcegB SUI4 4 0.99 53,23 38,66
2 ErcegB SUI4 15 0.99 50,51 36,18
2 ErcegB SUI4 10 0.99 50,32 36,6
2 ErcegC SUI2 10 0.99 50,25 36,01
2 ErcegC SUI2 4 0.99 49,79 36,45
2 ErcegC SUI2 15 0.99 49,20 35,33
ECC REPORT 33
Annex 4, Page 96
REFERENCES:
[1] Y. Okumura, E. Ohmori, T. Kawano, K. Fukuda "Field Strength and its Variability in VHF and UHF
landMobile Radio Service" (in "Communication Channels: Characterization and Behaviour", Editor B.
Goldberg, IEEE Press 1976).
[2] J.D. Parsons "The Mobile Radio Propagation Channel" Pentech Press, London.
[3] S. R. Saunders "Antennas and Propagation for Wireless Communication Systems" J. Wiley & Sons.
[4] IEEE 802.16.3c01/29r4, Channel Models for Fixed Wireless Applications, 20010717.
[5] ETSI EN 302 085 v1.1.2,(200102) Fixed Radio Systems; PointtoMultipoint Antennas; Antennas for
pointtomultipoint fixed radio systems in the 3 GHz to 11 GHz band.
[6] Coexistence Same Area Simulations at 3.5 GHz (Inbound), C802.16.2a02/08,02/03/16.
_____________________________
ECC REPORT 33
Page 2
EXECUTIVE SUMMARY AND CONCLUSIONS Summary The scope of this ECC Report is to provide uptodate guidelines for efficient, technology independent deployment of 3.5 GHz (or 3.7 GHz) PointtoMultipoint (PMP) Fixed Wireless Systems (FWS), traditionally referred to as Fixed Wireless Access (FWA) systems. The Report recognises that the current technology for FWA in bands around 3.5 GHz is in continuous extensive evolution since first ERC Recommendations 1403 and 1208 were developed. A detailed study on the coexistence of various technologies was needed in order to provide guidance to Administrations that wish to adopt an efficient and technology neutral approach to the deployment rules in these bands. It is also noted that ETSI ENs in these bands are not presently designed for a technology neutral deployment (this is done only in the 40 GHz Multimedia Wireless Systems in EN 301 997) therefore do not contain system controlling parameters, in terms of EIRP, useful for the desired “technology neutral” and “uncoordinated” deployment. Not having any ECC harmonised guidance for such deployment, the ENs are still bound to a cellbycell “coordinated deployment” concept actually not used in most of the licensing regimes. This report might generate future feedback actions in revising also ETSI ENs accordingly. Aspects that relate to sharing issues with PointtoPoint FS, FSS, radiolocation (in adjacent band) and ENG/OB are not considered in this Report. However they should be taken into account when applying any method of deployment suggested in this document. The applicability limits of the current Report are as follows: Application is mostly devoted to “block assignment” licensing methods, rather than “channel assignment” method. The guidelines presented have been maintained, as far as possible, independent from the access methods described in the ETSI ENs (e.g. EN 302 3261). MPMP (MESH) architectures are not yet considered. In order to include MESH architectures, a number of assumptions on “typical” application (e.g. on the omnidirectional/directional antenna use) need to be defined in order to devise the typical intraoperators, mixed MPMP/PMP interference scenarios for which simulations would have to be carried. Channel sizes and modulation schemes are also not specifically considered unless for defining “typical” system parameters. It should be noted that high state modulations (e.g. 64/128 QAM) have not been specifically addressed in the typical system parameters; nevertheless they would not change the general framework of this report. This may be considered during future update. FDD/TDD, symmetric/asymmetric deployments are considered. Additionally, system independent, EIRP density limits and/or guardbands at the edge of deployed region (pfd boundary conditions) as well as at the edge of assigned spectrum (block edge boundary conditions) are considered as licensing conditions for neighbouring operators’ coexistence (similarly to the latest principles in ERC Recommendation 0104 in the 40 GHz band). Presently, the spectrum blocks assigned to an operator vary widely from country to country  from 10 MHz up to 28MHz (single or duplex) blocks have been typically assigned. The block allocation size and the frequency reuse plan employed by the operator to achieve a multicell and multisector deployment drives the channel bandwidth of the systems presently on the market to be typically no greater than 7MHz. Conversely, the requirement for higher data throughputs is driving the need for wider channel widths (e.g. up to ~28 MHz) and therefore correspondingly wider spectrum blocks assignment in the future. Therefore, system channel bandwidths and block sizes are not fixed, even if typical data for current technologies are used for feasibility analysis of the “blockedge” constraints. The report considers two different aspects of deployment scenarios for two operators: 1. Operating in the same or partly overlapping area with adjacent bands assignment
The approved ETSI EN 302 326 will supersede equipment EN 301 021, EN 301 124, EN 301 744, EN 301 080, EN 301 253, antenna EN 302 085 and related HEN 301 753; technical characteristics in EN 302 326 are identical, or very similar, to those present in the ENs listed.
1
ECC REPORT 33
Page 3
2.
Operating in adjacent or nearby areas and reusing the same band assignment.
A number of different methods have been used to assess the severity of interference. These are: Worst Case (WC) (generally used for Central Station (CS) to CS interference) and for PFD limits at geographical boundary for frequency (block) reuse Interference Scenario Occurrence Probability (ISOP) (for CS to Terminal Stations (TS) interference between adjacent blocks in rural LoS situations) Monte Carlo simulations for CDF (cumulative distribution function) vs. C/I (for CS to TS interference between adjacent blocks in urban mostly NLoS situations and for TS to CS interference between adjacent blocks). For the above methods it has been possible to estimate the probability of interference between FWA systems. From these results, estimates have been made of the frequency and/or geographical spacing needed between these systems in order to reduce the level of interference to an acceptably low level. Absolute recommendations cannot be made because some system parameters are not defined by the available standards and because the effects of buildings and terrain are very difficult to model. The report therefore gives guidelines that will lead to acceptably low levels or low probability of interference in most cases. For the above methods that might be described as: The Worst Case (WC) method derives system deployment parameters to ensure that interference is always below a set threshold for all cases. The Interference Scenario Occurrence Probability (ISOP) is defined as the probability that an operator places at least one terminal in the IA. ISOP is related to the number of terminals deployed by the operator, and possibly to the cell planning methodology. The ISOP method evaluates the NFD or the outofblock rejection required in order to meet an interference probability lower than a certain value. The Cumulative Distribution Function (CDF), derived from Monte Carlo simulation of large number of “trial” TSs with a certain equipment/antenna/propagation model, is defined as the probability that a certain percentage of those trials would result in a C/I of the victim CS exceeding a predefined target limit. The Report derives the following alternative parameters, useful for defining an “uncoordinated technology independent” deployment: The Interference protection factor (IPF) and associated guardband method used to define the amount of isolation required from the interfering station to victim receivers in adjacent frequency block in terms of Net Filter Discrimination (NFD), obtained also by frequency separation (guard bands) and EIRP limitation. The Block Edge EIRP Density Mask (BEM) method is used for directly limiting the EIRP density in the adjacent block, and for assessing the CS to CS worst case interference, the CS to TS interference through acceptable ISOP value and the TS to CS through acceptable probability of exceeding a limit C/I to the victim CS. An important finding of this Report is that stringent protection requirement (e.g. in terms of BEM or NFD) is required only for CS emissions; the protection factor for TS is far less stringent and reduces as the antenna directivity is improved. Another important conclusion is a significant impact of CS antenna height on coordination distance for the frequency block reuse; due to the low LoS attenuation with distance, sensible size of coordination distance and associated PFD value are obtained only considering spherical diffraction attenuation. If the CS antenna height is not limited (or a downtilt angle is required) as a licensing parameter, it is nearly impossible to tell how far away the block may be reused. The example presented, made with typical system values, led to examples of BEM coherent with a “technology neutral” deployment of different systems in adjacent blocks. Receiver filters are assumed to be stringent enough to maintain the potential NFD implicit in the BEM (i.e. have sufficient outofblock selectivity for avoiding non linear distortion in the RX frontend chain). In annexes to the report, some further technical background and studies for related issues are also reported. They include urban obstructed propagation (nearNLoS) models and examples of practical application of RF filtering for easing the CS absolute EIRP BEM fulfilment when using equipmentgeneric relative spectrum masks defined by ETSI.
the more directive is the antenna of TSs. ENG/OB were not considered in this report. Since block edge mask parameters are linked to the size of the block. At least for CSs. FSS. using the same methodology as described in this report. 5. the importance of limiting CS antenna height (or downtilt angle) as possible licensing parameter is highlighted in order of having sensible coordination distances (i. 8. Sharing issues with PP FS systems. which would be useful for the desired “technology neutral” and “uncoordinated” deployment The study suggested that the block edge mask together with the contiguous assignment of frequency blocks should be the main means for avoiding interference between neighbouring frequency blocks.ECC REPORT 33 Page 4 Conclusions This Report has considered a number of facts as initial considerations for deriving the coexistence study: 1. limiting the potential competition for new services. . and their coexistence with new FWA systems. it is also recommended that the neighbouring blocks should be of a similar size. the antenna RPE plays a fundamental role in the coexistence. assigned in accordance with provisions recommended in this study. limited by spherical diffraction attenuation).8 GHz do not give harmonised and detailed suggestion to administration for implementing FWA (such as those produced for 26. The band is limited and wasted guardbands might drastically reduce the number of licensed operators. Those ERC Recommendations offer only channel arrangements. MPMP (MESH) architectures have not been considered in this Report. may need additional check on a casebycase basis. These studies might be carried on out in due time if needed and when manufacturers will be in a position to offer the necessary information. for PMP TSs. the less demanding might be their NFD (or the EIRP density BEM) required (offering a flexible tradeoff to the market). 3. That is why the resulting BEM limits outside the block are described in terms of transmitter output power. 9. allowing operators to make practical use of this phenomenon.e. in terms of EIRP. It is recognised that whilst some of the results in this report might also be applicable to mixed PMP and MESH deployment. Legacy FWA (WLL) systems have been already licensed in these bands in some countries. 2. In addition. others may clearly need additional work.6 GHz and 3. Presently ERC Recommendations 1403 and 1208 for the bands 3. ETSI ENs in these bands are not presently designed for a technology neutral deployment (this is done only in the 40 GHz MWS EN 301 997) therefore do not contain system controlling parameters. but are subject of a different CEPT study. 7. In this field. radiolocation (in adjacent band). but with the parameters corresponding to those of legacy systems. basic rules has been set for the coordination distance and PFD boundary levels between operators reusing the same block in adjacent geographical areas. 6. It is also shown that. 28 and 40 GHz). 4.
.............................................................................................................3 Central Station to Central Station ........... 46 MEAN EXCESS.....................2 The “predefined guard band deployment” ................ 8 1.................................................................................................14 2..................................1.................................... 19 2..1 Methodology...2.....1.......... 9 2...................................1..........................................................................................................................................................5 GHz calculations ...........................................................................2.1................................ 7 1.....................................f.............1 Typical System Parameters...........1 SCOPE ..........2...................................................................................................................................................................................1 The Worst Case deployment scenario (derived from ERC Report 99)............................ 32 3.........................................f................................................................... 33 3.......................................................................................................................................................4................ 3..........2 Conclusions for the TS to CS interference study........1..................1.................................1 3.............................. 49 Cell radius evaluation..........................16 2............1......4...3..................................................................................................................................................................................... 45 MEAN EXCESS LOSS AND LOGNORMAL SHADOWING ..................................... 32 3 ”ADJACENT AREA .. 43 3.................1 Interference protection factor (IPF)........40 3............... 45 A1.................2 3.......2 3..................................1 ANALYSIS OF THE COEXISTENCE OF TWO FWA CELLS IN THE 3...........................................................2...............................................2 Rural LOS scenario .......5 TERMINAL STATION TO TERMINAL STATION ........................................................................................................... 47 SIMULATION CAVEAT ............................................................8 GHZ BAND.............3...................................................................................... 31 2..................................................................................2..1 3............3................. 30 2............................27 2........................................................................2 CSTOCS INTERFERENCE .................. 32 2..........................SAME FREQUENCY BLOCK” INTERFERENCE SCENARIO .....................4................................................27 2. 7 1............................3......................................f.................. 9 2 “SAME AREA ......3 The "guided unplanned deployment" (Block edge emission density mask)..................... DIRECTIONAL OUTDOOR VERSUS OMNIDIRECTIONAL INDOOR TS..........1.................................................. 8 1....1 Rural LOS scenario...............3................................................................................................................................................................................................................................2 Methodology........................ 35 3.....................5 Terminal Station to Terminal Station .............................................................................. 31 2......................5 GHz calculations ...................................................................................................................................................................................................2 A1........... 49 Methodology.......4 3.......2..................3 A1............................................................................2.... 53 .........2 Interblock protection evaluation ................................................................................................4.........................................2.................................................................................... 15 Channel arrangements ................................3................ 44 ANNEX 1: AVERAGE CELL SIZE EVALUATION ..............3 3........................1...........................................................................................................2.4 ..............3 CSTOTS INTERFERENCE................... 45 MEAN EXCESS LOSS (MEL) ONLY ............................................................1 C/I statistics .......................................................1..........................................B) A1....................................1 A1...............................................................2 Results: ..........................3.........................................................3.................2.........3.................1........................ 50 CONCLUSIONS OF THE ANNEX.......................................................1...............................................29 2.....................3......ADJACENT FREQUENCY BLOCKS” INTERFERENCE SCENARIO .4...........................40 Worst case single interferer scenario..................................................................D) A1............................................. 9 2..........................................2 Cell coverage..........1 POWER FLUX DENSITY LIMITS FOR ADJACENT FWS SERVICE AREAS .1....1 2...........39 Conclusions and possible selfregulation method for CSs coordination distance ...............12 Urban NLOS scenario ...........................1........................................................... 19 2......................... 8 1............ 22 2...........G) SIMULATION MODEL ..................................3..................................................... 40 ATPC impact .....................................................................................................4 TS TO CS INTERFERENCE ........ 9 2...............3 2.................................................................................................................. 43 4 CONCLUSIONS OF THE REPORT ............40 Examples: ...2...................................................................................4 Worst case single interferer scenario: 3...............3..................... 48 URBAN SCENARIOS.....................................................3 Conclusions for CSTS protection in urban NLOS environment ....................................ECC REPORT 33 Page 5 INDEX TABLE 1 INTRODUCTION........................................1.......................................... 35 3..................E) A1...........................................3...42 3..............................F) A1..........................................4.1.............................................................. ...........2 CONCLUSIONS ON ADJACENT SERVICE AREAS BOUNDARY BLOCK REUSE COORDINATION .................................2 THE FREQUENCY LICENSING POLICY AND THE POSSIBLE APPROACHES .................................1...............................................4 Terminal Station to Central Station .....................................39 Transmit EIRP .................. 11 2.........................................26 2............................................................................................................................................................................................ 32 3................ 17 2.1 Interference evaluation ........... 49 Coverage and availability objectives: ...........................................................3 3............................................................2 Urban NLOS scenario ........ LOGNORMAL SHADOWING AND RICIAN FADING ..............................................................................1.....41 TS to CS Conclusions.............1 Assumptions.........................C) A1...............................................................................................................................1......A) A1......35 Antenna height..........................
72 Scenarios .............................. 73 Simulation results.............2........ECC REPORT 33 Page 6 ANNEX 2: TS TO CS INTERFERENCE EVALUATION.......2....... A.... directional outdoor antennas.....................................................................................................................2...3 A...2.........................2 A.........................1..2....................................................2.........................5.....................2.....2.................................................................................... ........4 A..........6 APPENDIX 1 TO ANNEX 2: MODEL BASIC DESCRIPTION...............................................................3 A2..........................................2....................................................1 Simulation Methodology ..1.....................................................................................................................................................................................................................................................2............................................................. 62 A21..........................1 Unfaded ....................................1.......2 A......... 77 APPENDIX 2 TO ANNEX 2: ACCEPTANCEREJECTION METHOD........62 A2...............3........................2............................................1 A2.... 88 REFERENCES:...........................................2 OMNIDIRECTIONAL INDOOR TS................3 Conclusions.......1 Input parameters and models ..........................................................................72 A..................................................................... 54 A2................................................................................... 73 Propagation models:..........5....2................. 69 A.......1...74 hCS = 40 m ............................................................2.......................................1..........................2..... 83 ANNEX 3: EXAMPLES FOR MANAGING A CS BLOCKEDGE MASK ................ 84 ANNEX 4: EVALUATION OF THE C/I RATIO EXCEEDED FOR MORE THAN 99% (AND 97% FOR OMNIDIRECTIONAL CASE) OF PROBABILITY.......................................................3 A......................................................................................59 Conclusions.....................................................2 Urban Scenario ........................................................62 A2.............2....................................69 A2.........1 Rural scenario....... VDSL like..............2 Summary of TS density and OOB eirp reference values: .........................................................2...........1.................69 A.................56 Rayleigh Faded Simulation Results................................................................................1...........2.........1.........2............................................................... 69 A..................................................74 A.............................................................2...................2...........................4 A.....................................................................2..................................................11.........................................................................2........................................................2......................................2......................... 54 A2................................... 96 ...1.................................................................................... 73 Conclusions............... 76 hCS = 30 m .................................2....................................................................70 TS activity factor and density:.........5 A.4 System Model and Simulation Methodology..............................................................................................................................................1..................................71 Shared traffic................................................ 54 A2.................................................70 TS omnidirectional indoor out of block eirp:............................................2 Simulation Results ...................2......2 Rician Faded ..............................1...54 Unfaded Simulation Results ...........................................1.................................................1..........................69 CS data (same used for CSTS evaluation): .......1 DIRECTIONAL OUTDOOR TS .......................................................2..................................................................2.........................................2...........................................2...1........................1.....................1 A.............1..........................................................1 Objectives...................................................................................................................................1................................................2.......62 A2.............2.....2 A2.......................................................................61 A2...........................................................2.........
Nowadays more experience has been gained from recent studies for the 26 and 28 GHz bands. Measures are suggested in this report for dealing with the issue of interoperator coexistence both between adjacent frequency blocks and between neighbouring geographic areas. on harmonised radio frequency arrangements for Multipoint systems. and new market realities calling for wide spread use of flexibly deployed user terminals with nondirectional antennas. In order to cater for the mix of technologies and services to be delivered it is most appropriate that a block (or blocks) of spectrum is made available to a potential operator in a manner consistent with the technology and market that the operator may wish to address. This report aims to assist the administrations in the assignment of frequency blocks to the operators who operate FWA systems in the available bands between 3. finalised by ECC Recommendation 0104.1).2. In principle no assumption has been made regarding the architecture of any FWA network. should be able to be deployed without mutual interference when operating in: a) adjacent frequency blocks in the same area or. and most of all for the 42 GHz MWS band. These bands were subject to the previous ERC Recommendation 1403. EN 301 080.16 standardisation activities. During the development of this report. Taking into account the final conclusions of this study.3. CEPT may wish to liaise with ETSI in order to align CEPT and ETSI deliverables on this issue.1 INTRODUCTION Scope The scope of this report is to investigate the coexistence of PointtoMultipoint (PMP) Fixed Wireless Systems (FWS).8 GHz BAND 1 1.4 GHz to 3. . in particular developments with ETSI BRAN (HIPERMAN) and IEEE 802.4 . EN 301 253 and EN 301 744. therefore they were not specifically considered as a starting point in this study. This report incorporates and enriches the information in earlier reports and recommendations. traditionally referred to as Fixed Wireless Access (FWA). Systems. These systems are developed in accordance with the ETSI EN 301 021. A key principle of the assignment guidelines is that even though a technology specific channelisation scheme is expected to operate within an assigned block this channelisation is not the basis for the assignment process. Other ECC work has reported and concluded on MESH systems in higher frequency millimetric bands.1 and B2.ECC REPORT 33 Page 7 THE ANALYSIS OF THE COEXISTENCE OF POINTTOMULTIPOINT FWS CELLS IN THE 3. it is important that the block assignment process supports systems for both symmetric and asymmetric traffic as well as systems that employ FDD and TDD techniques. finalised by ERC Report 99 and Recommendations 0005 and 0103. Operators are free to subdivide the assigned frequency block in the most efficient way for deploying or redeploying the selected technology. others may clearly need additional work. It is recognised that whilst some of the results in this report might also be applicable to mixed PMP and MESH architectures. The basis for these measures is to allow deployment with the minimum of coordination although more detailed coordination is encouraged as an interoperator issue. however MPMP (MESH) architectures have not been considered in detail in this Report. It should be noted that the current ETSI masks for PMP equipment in this band were not designed with the block assignment coordination in mind. b) the same frequency block(s) in adjacent areas.1. In conjunction with the CEPT channel plan defined by the ERC Recommendations 1403 (sections A1 and B1) and 1208 (sections B2.8 GHz. Due to the flexibility required in newly deployed services. the account was taken of the latest developments of FWA technologies. owned by different operators. ERC Report 97 qualitatively summarised requirements for modern licensing process and has also been taken into account in developing this report. These studies might be carried out in due time if needed and when manufacturers will be in a position to offer the necessary information. EN 301 124.
here called “predefined guard band deployment”. Emissions from one operator’s frequency block into a neighbour block will need to be controlled. inside their assigned block. Alternatively. In addition. as recommended in the 42 GHz MWS band. The block edge mask limits the emissions into a neighbouring operator's frequency block and it enables operators to place the outermost radio channels with suitable guardbands. the cellular structure and the cell planning aspects of the system operated by the operators in any particular area. This would imply that the administrations should. . administrations might wish. in order to avoid coordination with the neighbour's frequency blocks. the planned coverage range. for maintaining good spectrum efficiency. fixed guard bands evaluated around the most likely equipment to be deployed. the possible process of frequency licensing should guarantee. as far as possible.2. only the last one in section 1. the administration would aim to provide. channel spacing.3 would be extensively used in this report because looks more attractive from the point of view of spectrum usage and certain results in technology neutral environment. same area scenario.ECC REPORT 33 Page 8 In order to cope with the oftenconflicting requirements of a number of technologies in terms of efficient and appropriate block assignments.2.2 and 1.2 The frequency licensing policy and the possible approaches When considering the adjacent frequency blocks.e. This can be done by different methodologies. in the application phase.g. In that case. however. i. which balance constraints as far as possible on any specific technology. the contiguous assignment of frequency blocks with the associated Block Edge Mask (BEM) requirement. between the assignments. The guardbands are likely to be determined by the wider band systems therefore the method is most suited in case the differences among deployed technologies (e. Having considered the pros and cons of these two approaches.1 The Worst Case deployment scenario (derived from ERC Report 99) In principle. A first one. the hubs location. this method asks for a quite good knowledge of the typical FWA system technologies used. For further enhancing the spectrum efficiency. a more realistic approach is necessary. the most efficient way of evaluating the guard bands would be through a “case by case” evaluation. the current study agreed to follow the second approach.2. to both operators and end users. a frequency block edge EIRP density emission mask is used. NFD and modulation formats) are expected to be small or in bands already deployed where fixed channel arrangements are recommended. However. after the block assignment procedure has been done. For the above reasons. 1. 1.2. imposes. and hereby only the two examples described in next sections 1. some compromise is suggested in order to develop reasonable assignment guidelines. 1. when Terminal Stations (TS) are taken into account. a “controlled interference” deployment. a reasonably interference free environment. the number of variables become so huge that the method is practical only for deterministic evaluation of Central Station (CS) to CS interference. in particular taking into account a general shortage of frequencies in this band. the enforcing rules will apply only in a “mutually agreed” way or it would be flexibly changed according the actual interference scenario shared by both operators with theirs planning tools. already recommended in other frequency bands. This approach tends to prevent spectrum efficiency improvement with the technology evolution and thus is not recommended as a preferred method.3 are mentioned.2. based on I/N objective (then independent from system characteristics apart the noise figure). to decide later not to enforce the blockedge mask for neighbouring operators who will apply mutual coordination at the blocks edge in view to optimise the utilisation of outermost parts of the assigned blocks. analyse the actual behaviour.2 The “predefined guard band deployment” In the first approach.
lead to different conclusions on the regulatory framework objective of this report. therefore only the knowledge of typical victim receiver noise figure and antenna gain are necessary. evaluated for “average worstcase” interference scenarios. provided that all have similar spectral and linkbudget characteristics. In that case. The EIRP density mask is designed on the basis of acceptable noise floor increase due to interference from adjacent block. It is up to operators to possibly further coordinate with other operators using adjacent blocks. 64 states) but. they would not.g. for further enhancing the efficiency. nevertheless for defining typical cell coverage sizes also real modulation formats should be used. vertical decoupling) is implemented for improving antenna decoupling as much as possible.1 “SAME AREA . nevertheless it is expected that ETSI standards will adequately cover the issue. The examples shown refer to the ETSI EN 301 021 and BRAN HIPERMAN system (including foreseen applications for indoor terminals with omnidirectional antennas for easy user defined deployment. however this is not in the scope of this Report to set limits for it. implies that additional EIRP density limits are set in order to allow an “average” interference free scenario to the operators.ADJACENT FREQUENCY BLOCKS” INTERFERENCE SCENARIO Analysis of the coexistence of two FWA cells in the 3. a block edge mask is generally designed on the basis of a small degradation in an assumed scenario with a low occurrence probability of a worst case (e. power output and C/I sensitivity. operators would still be asked to solve. medium bandwidth systems (3. . For a sensible and costeffective regulation.2 above it is not therefore excluded that in a limited number of cases specific mitigation techniques might be necessary. after the block assignment procedure. As for the guardband method described in section 1. the considerations made are not too sensitive to the multiple access method. Also adjacent block receiver rejection concurs to a reduced interference scenario. the guard band is to be included in the blocks assigned to the operators.1 Typical System Parameters Considering the scenario of a wide suburban area with relatively high traffic demand and a small amount of obstructing buildings. two directional antennas pointing exactly each other at close distance). With this approach. However. the blocks are to be made consequentially larger. here called ”guided unplanned deployment”. Moreover. with conventional site engineering methods. These data are then “technology independent”. only the maximum "inblock" EIRP/power density applies while the "outofblock" noise floor will apply only from a "mutually agreed" starting point within the adjacent block.g.2. in Table 1 data for two systems types only are referred. to enforce the blockedge requirements to neighbour operators who will apply mutual coordination at the block edge in view to optimise the guard bands. the "worst cases" that may happen in few cases.ECC REPORT 33 Page 9 1.1. In this case.g. Different modulation are obviously possible (e. in principle. and is effective for bands that do not have fixed channel arrangements as a deployment constraint.5/7 MHz) are analysed in LoS conditions as well as similar system in more restricted and dense urban areas analysed in mostly NLoS conditions. the operator is permitted to use the assigned block as much as the equipment filtering and actual EIRP allow operation close to the blockedge. only for defining a typical receiver BER thresholds.4 . This method is most suited when very different technologies are used.8 GHz band 2. leaving to him and the manufacturers the possibility to improve the spectral efficiency as far as possible. In this case the “interference free environment” is ensured by the EIRP density limits set by the administration.2. In particular when CSs are colocated on the same building or very close to each other.3 The "guided unplanned deployment" (Block edge emission density mask) The second approach.3. The method is therefore quite independent from ETSI standards. administrations are not expected. 2 2. the statistical approach is not applicable and it is assumed that common practice of site engineering (e.
2 This channel spacing is considered the most representative for being carried over in the calculation. due to the need of containing size and power consumption. sophisticated coding is conservatively neglected. The 3.1336 16 ETSI EN 302 085 [5] ITUR F.1336 4 23 86 76 ~22 16 ETSI EN 302 085 [5] ITUR F.g. also widely popular. These antennas may exhibit a gain of the order of 20 dBi or less.5 MHz up). EN 301 021 and 301 080) is considered not realistic for conventional outdoor TS. independent from the system channel bandwidth. . from 1. these omnidirectional antennas. Modulation order is taken care by two different critical C/I values provided in this table.1336 16 ETSI EN 302 085 [5] ITUR F. according ETSI BRAN contributions. This value includes feeder losses for full indoor applications. These values are used only for evaluating reference order of magnitude of cell sizes in various environments. Modulation order is taken care by two different critical C/I.699 for these relatively lowgain antennas will result in an overestimate of the gain for relatively large offaxis angles. On the other hand ITUR F. unlike the ETSI standardised ones for CS. nevertheless it is considered more appropriate and will be used in this study. the results are parametric on “interference power density”.1336 gives low gain TS antenna patterns only for bands below 3 GHz. however this is justified considering the significant margins used in ETSI EN that have had regulatory valence for old “national type approval” as well as for R&TTED regimes. 7 This value includes feeder losses for full indoor applications.5 GHz will be used as radio frequency throughout the calculations. This value includes feeder losses for full indoor applications.. 4 This outdoor TS power is assumed as typical for symmetric traffic in conventional PMP systems. sophisticated coding is conservatively neglected. 6 The Noise Figure estimated from EN 301 021 BER values and typical modulation formats would result in ~12 dB. should more easily fit in that possible framework 3 The lower power level is considered typical for “conventional” PMP systems. therefore for network planning purpose it seems too pessimistic and a value of 8 dB has been assumed.90° sector boresight gain (dB) CS antenna azimuth and elevation radiation patterns Terminal (TS) outdoor directional antenna boresight gain (dBi) and RPE 9 System Type (according typical ETSI definitions) Type A (typical 4 state) Type B (typical 16 state) 3.g.5/7 2 30 (rural LOS cases) 30 (rural LOS cases) 35 (NLOS cases) 35 (NLOS cases) 30 30 23 86 84 ~14 16 ETSI EN 302 085 [5] ITUR F.699 is used. then independent from the system channel bandwidth. It has been found that using the reference radiation pattern given in Recommendation ITUR F. The actual interference study is based either on I/N in a noise limited scenario or on C/I between blocks (without specific carriers positioning) in an interference limited scenario. modulation formats and/or subchannelling. userdefined deployment the antenna would be of very simple technology without any attempt to define and/or align the antenna according conventional azimuth and elevation planes. 9 An antenna with relatively low gain is frequently used for transmitting and receiving signals at the outstations or in sectors of central stations of PMP radiorelay systems. The corresponding uplink margin loss is recovered by other methods (e. The 35 dBm Maximum Power presently allowed in ETSI ENs (e.g.” 10 Standing that in these indoor. mostly with LOS operation. 5 Power for indoor. It is considered that the larger channel systems would lead the coexistence rules. would be considered to have constant gain over the whole spherical 360 angle. the amount of interference caused to other systems and the amount of interference received from other systems at relatively large offaxis angles will likely be substantially overestimated if the pattern of Recommendation ITUR F.5/72 3. TS is kept typically lower. Nevertheless lover spacing channels (e.ECC REPORT 33 Page 10 System Channel bandwidth MHz Transmitted Power at section D’ (dBm) (CS)3 Transmitted Power at section D’ (dBm)4 (conventional outdoor TS) Transmitted Power at section D’ (dBm)5 (indoor TS) Receiver Noise Figure at section D (dB) Receiver Threshold for BER= 106 (dBm)7 Critical C/I for BER= 106 (dB)8 Hub (CS) antenna .1336 4 Terminal (TS) indoor omnidirectional antenna gain (dBi)10 Table 1: Typical system data for typical cell size evaluation The same system parameters will be initially used for both victim and interferer. therefore. user operated. it should already give enough margin for the possible necessity of a selective RF channel filter of reduced size for TS. As a consequence. while the higher level might be justified for applications including NLOS propagation and TS indoor with lowgain omnidirectional antennas. 8 These are typical conservative values.
of classes of antennas similar to ETSI TS 2 and TS 3. In addition. not necessarily for studying the increase of cell size.ECC REPORT 33 Page 11 Due to the importance of Terminal Station (TS) directional antennas RPEs (and in particular of their main lobe) on the results shown in this Report.7 GHz bands. the report also considered omnidirectional TS antennas. ETSI RPEs are generally defined only for “type approval” purpose (i. from the ITUR recommendation F. therefore it has been used in Figure 1 to reference the different antenna RPEs.1336 for identifying different RPEs. suggest that the use of ETSI RPE for TS antennas might give worstcase results that are not experienced in practice.2 Cell coverage. Figure 1: Antenna RPE Comparison 2.5 and 3. while the typical ITUR F.1336 recommends RPE for the bands below 3 GHz that here are considered appropriate also in the 3. The two scenarios are specific for: . For evaluating the interference potential.e. the gain range 16 to 20 dB is considered representative.1.1336 gives typical “average” RPE that are more representative of the field situation. they will be conventionally identified as “rural” or “urban”. no azimuth or elevation specifics were considered due to assumption of uncontrolled deployment by users. the results should not be considered strictly applicable for those areas only. however. Annex2 of ITUR F. reported in Table 1 as representative of the average value on the market. F. but might be applied whenever the propagation specific LOS or NLOS conditions are applicable (at least for the fraction of the user terminals concerned).e. 100% of RPE values shall be within the mask). Therefore. the Report will maintain a fixed gain of 16 dBi. it is initially needed a rough evaluation of the cells radius in the various environments using appropriate propagation models. assuming the uniform figure of antenna gain in all directions.1336 point of view. However the objective of this report would be mainly to consider the impact of different ETSI antenna RPEs for coexistence studies. We will examine two typical scenarios based on LOS and NLOS propagation. The antenna gain is the parameter used in the formulas of Annex 2 of ITUR F. Figure 1 show the difference between those RPE. without any specific alignment towards the CS antenna. i.1336 RPEs with gain 16 and 20 dBi will be generally used in all numerical evaluations.
whichever is greater.ECC REPORT 33 Page 12 LOS “Rural” situations.2. it seems adequate to use the deep fade approximation or 10 dB. relative to the ground.44+36*log(D) . Propagation phenomena is modelled with SUI model.48.1 The scenario examined is a LoS.2972*105* D(3.89) * D(3.e.28 + 20 log(D) (2) (1) Asph is the spherical diffraction attenuation described in ITUR Recommendation P.5) * (1+0)(1.1.2972*105 * D(3.5 (hilly terrain). which covers both the deep fade and shallow fade regions. Rural LOS scenario 2.4+ 20 log(f D) = 103. FM can be written as: Pwm %= 3 * unyear % (4) P0%=5*107 * 10[0.89) * D3. located in central Europe. For this evaluation we will consider GTS=16 dB as worst case (resulting in smaller cell size).6) P0% = 4. FM is the fade margin (excess attenuation) required to meet the yearly availability objective.1*(CoClatClon)] * pl(1.10*log(unyear%) (6) (5) . when presence of high building or foreseen service to indoor terminals is likely to shade most of the links to users terminals.4) * 3. considerably reducing the available coverage. relatively flat environment without significant obstructions. Clon=3 dB (Europe).6)] / (3 * unyear%) Substituting (4) and (5) into (3) we obtain: FM= .5 GHz given by: FSPL = 92. when for wider cells coverage (or high quality of service) it is necessary to rely on conventional PMP deployment with directional antennas and mostly LOS situation (certainly applicable for the users at cell border).6) P0/Pwm = [4. NLOS “urban” situation. difference in dB of TX output power and RX threshold at given BER 106) GCS and GTS are CS and TS antenna gains in dB. The cell dimension is noiselimited freespace attenuation and Rayleigh fading. The main propagation modes are assumed to be free space and.530. with the conservative approach that the yearly unavailability (unyear%) is spread over four “worst” months only. For the purpose of the present analysis. The link availability will be affected by clearair multipath. The maximum cell radius R will be calculated from the link budget: SG + GCS + GTS = FSPL + Asph + FM where: SG is the “system gain” (i.1*(3. assuming fast varying multiple reflections. spherical diffraction.4) * f(0. pl=10%. Clat=0 dB (medium latitude). FSPL is the free space attenuation loss for f=3.5 GHz radio frequency.5(0. P0% = 5*107 * 10[0.5) * (1+)(1.6 Assuming C0=3.53)] * 10(1.562 that depends on the height of CS and TS antennas. =0. Scaling it to a yearly average. The 10 dB has been chosen as a safe value to ensure proper operation in "normal" clear air propagation. for the assumed geographical area and for 3.5308: FM=10 log[Pwm /P0]. with correction derived by spherical diffraction phenomenon. possibly. From ITUR P. FM can be evaluated according to ITUR P. (3) Pwm is the probability of exceeding the critical BER during the worst month.
and yearly availability objectives 99. the maximum cell radius values are shown in Table 2. the mean height value could be used. the link budget at the cell edge (D=R) can be rewritten as: SG+ GCS + GTS = 54. Neglecting the ripple at short distances.99 %. and 99. Spherical Diffraction attenuation . ht). Values for a few significant cases are shown in the following Figure 2 that shows that K2 1.84 10 log(unyear%) +56 log(R) + K2 (RD0) (8) With the assumed equipment parameters. antenna heights.ECC REPORT 33 Page 13 Spherical diffraction attenuation Asph can be calculated by subtracting the free space attenuation from the output of the program GRWAVE (available from ITU).f=3. D0 depends on the heights of hub and terminal antennas above ground (hc.5GHz 120 Antenna heights (m) (*) 1010 100 1030 2020 2040 3030 80 3050 4040 5050 60 dB 40 20 0 20 0 10 20 30 40 50 D (km) 60 70 80 90 100 Figure 2: Additional attenuation due to spherical diffraction Substituting (2).999%. which comes from reflections in the plane earth model. Asph = 0 in Figure 2) where spherical diffraction attenuation starts to be significant. .3 dB/km is nearly invariant and that when different CS and TS antenna heights are considered.e. (6) and (7) into (1). A sample output for two significant cases is shown in Figure 2. Asph is approximated as: Asph= 0 (for D < D0) (7a) Asph = K2 (DD0) (for D D0) (7b) D0 is taken as the point where the total attenuation equals the free space value (i.
7 8. The cases with CS and TS antenna heights = 10 m (see Figure 2) are the only ones where spherical diffraction attenuation has some impact by reducing the cell size.7 8.9 20. derived from multiple facets attenuation due to quick variable situations (e.8 14.7 15.7) (13. moreover. However.d.4 12.1. another “fastfading” occurrence probability described with SUI propagation model.3) 12.9 ht = 20m (hc =10m (12) (10. The latter cases are shown only as explanatory example of the phenomenon.6) (16.1) (11.3 22 19.9 20.ECC REPORT 33 Page 14 Availability System type Antenna heights 99.2 15.5) 18.6) 15.7 (15.4 ht = 20m hc =30m 18.4 23.3 (18.6 14.2 15.2.7 15.6 14.3) hc =40m 18. with mean value (here called A50) and a shadowing variance with standard deviation “” (between ~8 to 10 dB).3 22 19. leaves.5) (12.8 14. for most cases of practical antenna height.4 A ht = 10m hc =20m 18.2 15.16 for similar coexistence studies [4] might be profitably used. However. it describes he associated path attenuation.9) 10.99% GTS=16 dB FM R [km] [dB] GTS=20 dB FM R [km] [dB] 99.9 ht = 20m hc =30m 13.3 18. antenna heights and propagation environments.4 23. the one recently adopted by IEEE 802.3 18.8 14. windows openings.3 22 19.4) ht =10m) hc =40m 13. 2. however. Overimposed to this attenuation probabilistic spread.9 12.7 17.5 10.9 (19.4) GTS=20 dB FM R [km] [dB] 14.5 10.2 15. The reference path loss A0 (also sometimes referred as A50 in the document) at a distance d0 .6 (12.9 ht = 30m hc = 30m 13.9 12. cars etc.6 14. hc = 10m is not considered realistic and therefore will no longer be taken into account in further evaluations. typical for most US cities outside their relatively small downtown) increasing with distance from CS and subdivided into three different categories (Erceg categories A.5 (10.9 20.9 12.f.7 8.7 17.2) (15.4) (13.5 10.16 an “excess attenuation” for all TSs is introduced (mostly based on tests made over wooden/hilly areas among low rise buildings.g.5) 23. in dB.2) (8. Availability (BER 106) The conclusions of Table 2 show that.3 (21.7 8.3 18.2 Urban NLOS scenario For urban propagation models there are not consolidated ITU models.4 12.7 17.2) (19. where the exponent n represents the decay of path loss and depends on the operating frequency. B and C) according the terrain characteristics and 6 SUI models. with a Gaussian probability distribution function (p. antenna heights are not affecting the area coverage. The following basic principles describe the IEEE model: The path loss (PL) can be seen as the summation of basic free space loss (FSL) and the excess loss (Lex) due to the local blockage conditions or reduction of antenna gains: PL(dB) = FSL (dB) + Lex (dB) The path loss can be modelled as follows: PL(dB) = A0(dB) + 10 n log(d/d0) + S(dB) . subsequent tests made in European urban environment has shown that some of the variants (notably Erceg B with intermediate SUI 24) are suitably applicable as well as Erceg C for rural NLoS (or wide spread suburban areas).4 23.4 12.9 B ht = 10m hc =20m 13.4 (21.7) ht =10m) Note: Values in parenthesis denote the impact of spherical diffraction attenuation Table 2: Cell radius and FM vs.7 15. the cell radius is limited by the system gains considered and by the free space loss only. Hence spherical diffraction is not yet affecting the propagation.) In IEEE 802.).8 14.4 ht = 20m (hc =10m (14.4 ht = 30m hc = 30m 18.4 12.999% GTS=16 dB R FM [km] [dB] 20.7 17.
described in Figure 3. when one or few specific technologies are privileged) so that fixed coordination rules might be uniquely set a more detailed approach might be more appropriate. limiting their free usage of the block) and in the following evaluation is not taken into account. is equipment/antenna dependent only and is described as: Poutdensity (dBW/MHz) + GTX − NFD = X3 (dBW/MHz) (9) where X3 represents. 2. in EIRP density terms. the CS BEM outof block lowest requirement within the central portion of the adjacent block as described in Figure 3.99% (QoS) services and is maintained around the same size for indoor applications provided that the objectives would be relaxed to ~70% coverage and 99% availability (best effort service). As a consequence a “technology neutral” approach is hereby used in the form of the above IPF.1. The relationship between NFD (in the central portion of adjacent block) and BEM.g. strictly related to the Net Filter Discrimination (NFD) concept. out of which a specific example is the transmitter output power density Blockedge mask (BEM) concept. when system characteristics are known (e. generally referred as Interference protection factor (IPF). might be obtained enforcing a blockedge spectrum density mask (in form of either outofblock EIRP density or outofblock absolute power density emission at antenna port as discussed later in the document.ECC REPORT 33 Page 15 (generally d0 is assumed =100m considering that propagation closer to 100 m might be dominated by other unpredictable factors such as reflections) from the transmitter is typically found through field measurements. Also polarisation decoupling is a factor that might not be prejudged (unless different polarisations are imposed in licensing two adjacent blocks operators.99% (QoS) services.3 Interference protection factor (IPF) The potential coexistence of different cells in adjacent frequency blocks is guaranteed when there is sufficient isolation between interfering transmitters of one cell and victim receivers of the other cell. Results might be summarised in cell sizes of 2/3 km radius (depending on the predominant modulation format) in suburban areas for conventional Outdoor TS coverage ~95% and availability objective of 99. actually summarised all the equipment/antenna related IPF contributions and might be best fit in environment where equipment characteristics are not known beforehand. expressed in terms of EIRP. Alternatively the cell size should be further reduced to 0.5/1 km. . The detailed evaluation of cell size as a function of desired availability is reported in Annex 1. The BEM concept. This does not imply that the BEM is always the best method. This required isolation. In urban scenarios cell sizes drop down to 1/2 km (depending on the predominant modulation format) for conventional Outdoor TS coverage ~95% and availability objective of 99. The shadowing loss S denotes a zero mean Gaussian random variable (Lognormal variable) with a standard deviation (also in decibels).
which is generally defined for suitable service and sharing conditions in the licensing process.63. However such guard band disappears for the 3. The transition zone.8 GHz. It should also be noted that equation 9 assumes that the receiver filters of equipment are matched to the transmit requirements (i. Both recommend assignments based on "number of slots" 0. For convenience.6 . . adjacent TX/RX interference is expected also for FDD systems. 2.25 MHz wide. possible channel arrangement should be analysed. either having a fixed external guardband assigned in between or.ECC REPORT 33 Page 16 TX Output Power Density (dBW/MHz) Assigned block PMAX Adjacent block Transition zone (not considered in this report) X1 X3 0 F. but should be determined through a tradeoff between the need of a limited amount of guardband. are able to reduce the adjacent block interfering carriers power down to a level not impacting the receiver linearity or the total amount of inband interfering power). in the following sections. leaving to manufacturer freedom for its best practice.53. That means that unless specific number of slots are reserved in both go and return subbands (wasting at least half of them). from the block edge Figure 3: Scheme of the BEM concept In Figure 3 the “0” frequency reference is intended as the central frequency between contiguous blocks. where specific numerical examples are made on the base of representative system characteristics defined in Table 1 the parameter X3 only is used with the understanding that NFD is easily derived from equation 9.1 Channel arrangements Prior to evaluating BEM requirements. also considering that in the R&TTE Directive spirit receiver performances should be dominated by market forces. is not considered in this report.43.1.6 GHz the "odd" 10 MHz automatically create a ~ 10 MHz guard band.8 GHz). On the other hand.3. Annex 3 give some more consideration on the definition of the transition zone. shown between X1 and X3 power density levels. the possible slight decrement of NFD available for the blockoutermost system channels and the practical feasibility of equipment filtering for reaching the required X3 level from inblock Pmax density. in case of immediately adjacent blocks. apparently only symmetric assignments are foreseen and no specific mention is made of internal (goreturn) guard band except for the fact that in 3.413.3. having a corresponding wider block bandwidth assigned for operators to manage the best an internal guard band between them. neither in terms of IPF or NFD. as offered by CEPT/ERC Recommendations 1403 (3.6 MHz (50 MHz duplex) and for all 3. ETSI ENs do not give any specific requirement or guidance for receive filtering but only minimal C/I rejection capability.6 GHz) and 1208 (3.e.
2.2 CStoCS interference A "same area. since: The distance between CSs is in practice short enough to exclude spherical diffraction. In any case the contribution to availability of unsynchronised T/R period tends to be negligible when the multipath activity is large and propagation events last far longer than T/R periods." These sentences and the fact that no recommendation on subband for CS and TS operation is made. It is further assumed that the allowed degradation of the victim receiver threshold due to interference is 11 In principle. at least for the innermost assigned blocks where the mitigation of predefined uplink/downlink duplex blocks becomes ineffective. the deployment of asymmetrical systems being made with mixed uplink/downlink subbands within the symmetrical assignment.g. CStoCS interference is particularly dangerous. symmetric and asymmetric systems. the spacing between the lower edges of each paired subband shall be 100 MHz" and also: "PMP equipment may be used having a duplex spacing other than exactly 50(100) MHz. It has been recently demonstrated by CEPT studies for 40 GHz band the best compromise method for allowing flexibility and efficient use of the spectrum with the recommended symmetrical assignment. since it can cause unavailability of a whole sector11. the recommendations mention that: "where a duplex frequency allocation is required. This will be 100% of the time in the case of two FDD systems. the allowed interfering IPF (or EIRP) is calculated for freespace propagation only. a mixed TX/RX in nearly adjacent assignments should be in any case considered. However. unless the band should be assigned for predefined technology (e. The actual interference intervals will vary because the two CS T/R periods will not be synchronised. Rain attenuation is negligible at this frequency band. . FDD only) and spectrum waste is envisaged for creating goreturn guard band. Therefore a worst case analysis will be presented for it. Both antennas are in relatively high locations (30m in the example) even in the urban environment.ECC REPORT 33 Page 17 Moreover. Therefore this aspect of T/R period impact will not be taken into account. The fading events (mainly due to clear air multipath) are considered as completely uncorrelated. The above considerations and the small duplex spacing. it happens for the part of the time when the two CS are in opposite Tx /Rx modes. Both CSs are supposed to face each other in line of sight (worst case situation). lead to the conclusion that. adjacent frequency blocks" scenario will be assumed (Figure 4). When at least one system operates in TDD mode it will be less than 100%. clearly show the intention to admit (on a non discriminatory way) TDD and FDD. d R Block 1 Block 2 Figure 4: CStoCS interference scenario As far as propagation model is concerned. In this case the mean path loss predicted by the IEEE propagation model (see Annex 1) is near or lower than the freespace attenuation. such equipment must conform to the limits of the block allocation as defined above.
e. It is assumed that when lower distances are concerned. giving obviously the same result for both A and B systems. for system types B in LOS CS to CS scenarios.5 2 Figure 5: Required CStoCS spacing for un coordinated deployment If. It is then assumed that common practice of site engineering (e. the antennas may have at least 35 m distance (in height on the same pole). km 45 50 55 60 65 70 75 0 0.3). Hence the main carriers of the adjacent block EIRP are always reduced below the interfering outofblock noise floor so that their residual contribution is negligible. the nearfield area is ~ 2m (i. it may be assumed that the victim receiver has selectivity that matches the IPF or the block edge mask. vertical decoupling of few meters. therefore. the CS BEM outof block lowest requirement in dBW/MHz within the central portion of the adjacent block (see Figure 3 and paragraph 2. Other arrangement topologist are also possible with similar results.g. recovering with sidetoside antenna decoupling the reduction in free space loss. for a dish antenna of ~same gain (i. As in the 40 GHz case. to a minimum uncoordinated distance of less than 100m. a plot of the required X3 value vs. for avoiding near field propagation effects12) is implemented. as initial assumption. given the typical cell radius values shown in Table 2. which seems quite reasonable in a rural environment. even if colocated. the required EIRP level in the central portion of the adjacent block is: X3− 92. 12 At 3. lead to smaller cells size. diameter D ~ 30 cm). therefore the two CSs would be likely colocated on the same building. sectorial antennas should not be far from this figure. On the other hand a value of X3 = 73 dBW/MHz would lead.5 1 d (km) 1.1. which seems more appropriate in a urban scenario where more dense network and higher number of user terminals in NLOS. “d” the CS distance in km. It is assumed that. there is a likely mutual shading between two elevated buildings so close by.e. with the assumed system reference values shown in Table 1. one would consider an outofblock emission limit of 50 dBm/MHz at the antenna connector.4 − 20 log(RF) − 20 log(d) + GRX = −144 +NF −6 where: X3 = Poutdensity + GTX −NFD represents. in EIRP density terms. ~ 2D2/). one would get for X3: X3 = 50 30 + GantCS = 50 30 +16 = 64 dBW/MHz This would allow a minimum uncoordinated distance of about 350m.ECC REPORT 33 Page 18 Threshold = 1 dB. RF is the frequency in GHz. “d” is shown in Figure 5.5 GHz. . having the same CS antenna gains (16 dBi). hence the allowed interference spectral density is: IS = N0 − 6 = −144 +NF −6 (dBW/MHz).
in the next section 2.pdf ).ieee.g.org/groups/802/16/tg3/contrib/802163c01_29r4. Area (IA) Antenna: TS 3 Figure 6: CS to TS interfering scenario The area where the threshold degradation exceeds the 1 dB objective will be defined by: X3 FSPL(IV) + GantTS(V) . the above LoS evaluation is an absolute worstcase.3.16 adopted one depicted in Annex 1.1 Rural LOS scenario The ISOP approach will be used. Area (IA) Antenna: TS 2 Interf.2 and in IEEE document available at http://grouper.3 for having a suitably low ISOP or unacceptable C/I occurrence probability. the CS to TS interference scenario should be analysed for the two cases of LOS and NLOS propagation.ATS(IW) 144 + NF – 6 This formula is commonly used for rural scenario (where availability is dominated by Rayleigh deepfade phenomena) considering the uncorrelation of deep fading events in the different paths. Referring to Figure 5.1. The additional shading attenuation probability is not a negligible factor and using specific propagation models helps prove this (e.ECC REPORT 33 Page 19 Therefore. due to the random nature of this kind of interference. Also in this case complete uncorrelation will be assumed between fading events affecting the "wanted" and the interference path. It should be noted that for urban scenarios. Its size and shape depends on the distance between CS's and the additional protection from the terminal antenna RPE.3 CStoTS interference As for the cells coverage area. these values will be carried over for being validated through further analysis related to CS to TS interference. 2. TS (V) CS (W) dWI CS (I) Interf. see description of “rural” and “urban” scenarios in section 2. 2.. W the "wanted" CS and I the interfering CS.2. . we will label V the victim TS. the IEEE 802. There will be an area in the victim sector where the receiver threshold degradation will exceed the assumed 1 dB limit. ATS(IW) is the additional attenuation given by the TS victim antenna RPE at an angle equal to the difference in azimuth between the victimtowantedCS and the victimtointerfererCS path (assuming that the victim antenna is aligned at boresight with the wanted CS).
however.ECC REPORT 33 Page 20 Using the TS 2 and TS 3 antenna classes (represented by typical ITUR F. This is valid for “rural” scenario where the cell size is limited only by LoS propagation following Rayleigh statistics and described in ITUR Recommendation P. P3=1/6 for 2 operators.530.TS 3 antenna class Figure 7: Interference Areas (IA) for victim TS as function of outofblock EIRP density X3 and ITUR TS antenna RPE Figure 7 is representative of the worst case AI experienced when the two W and I CSs are boresight aligned and at their minimum distance (derived in section 2.3 0. An average P3 = 1/4. the relative area coverage and the number of sectors per cell.3 0. the "forbidden" interference area IA can be derived and are represented in Figure 7. Assuming 2 or 3 blocks (one per operator) and 4 sectors per cell.4 0. the number of operators. P2 is the probability that the attenuation from wanted TX to victim RX and from interfering TX to victim RX are uncorrelated.3 0 0.5 km X3=64 dBW/MHz .1.TS 2 antenna class X3=64 dBW/MHz . Nt is the total number of terminals deployed in a sector. P3=~1/2 for 3 operators. In a propagation environment dominated by multipath we assume P2=1. For urban nearNLoS scenario this is not generally true.1 0.2 CS to CS scenario). P3 is the probability that operators use adjacent frequency blocks and equal coverage on the same area.1 0.1 km 0 0.TS 3 antenna class X3=73 dBW/MHz TS 2 antenna class X3=73 dBW/MHz .3.2 0. as ISOP = (1(1P1)Nt)*P2*P3 where: P1 is the probability that one TS falls within the interference area where the margin degradation exceed a predefined value (here assumed 1 dB) evaluated as P1 = IA/Asector. In addition the assumption P2 = 1 is conservative. a few values of ISOP have been calculated as examples in Table 3a) and Table 3b) with . Assuming Nt = 64 or Nt = 32 (considered representative of relatively wideband systems adopted in this frequency bands). due to its shortness. still maintaining fixed boresight gain of 16 dB). It depends on the number of available blocks. one of the two paths (the interference one) is still here considered LoS for actually affecting the victim TS region.2 0. 0.1336 antenna RPE derived with G=16 and 20 dB.2 0. ISOP is calculated following the approach in ECC Report 99.
65 274. Scenario Rural Rural Rural Rural X3 (dBW/MHz) TS Antenna TS2 TS3 TS2 TS3 dUI min (m) ASector (km ) 274.00605 0.02924 0.0231 ISOP % NT=32 0.0445 0.251 0. however the ISOP drops rapidly as the distance increases.00397 2 P1 0.0176 0.0022 0.0156 64 64 73 73 350 350 100 100 Table 3a): ISOP % as function of outof block EIRP density in some rural scenarios System type A (4 states) Scenario Rural Rural Rural Rural X3 (dBW/MHz) TS Antenna TS2 TS3 TS2 TS3 dUI min [m] ASector (km ) 141.006 0.129 0.3296 0.1698 0.02924 0.00282 ISOP % NT=64 0.0343 0.02073 0.50 0.00 64 QAM (NT=64) X3=64 dBW/MHz Rmax=13.ECC REPORT 33 Page 21 GCS=16dBi.0162 0. respectively).03 141.00397 2 P1 0.1653 0.03 141.0315 0.085 0.0686 0.258 0. Figure 8 shows two examples taken from those in Table 3.045 ISOP % NT=32 0.03 2 IA (km ) 0.499 0.75 0.25 0.4 km 4 QAM (NT=64) X3=64 dBW/MHz Rmax=18.65 2 IA (km ) 0.03 141. ISOP % 1.0445 0. GTS = 16 dBi and TS2 and TS3 antenna typical RPE (using typical ITUR F.7 km 0.0352 0.65 274.00145 ISOP % NT=64 0.00 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 dUI [km] Figure 8: ISOP as function of W and I CS distance From the above it appears that: The use of Class 3 TS antenna reduce the ISOP by ~25% to 35% in comparison to Class 2 one.65 274.0225 64 64 73 73 350 350 100 100 Table 3b): ISOP % as function of outof block EIRP density in some rural scenarios System type B (16 states) The above data are obtained for the worst case of W and I CS placement (boresight aligned and in closest position). .1336 RPE derived with gain of 16 and 20 dBi.00429 0.01065 0.25 1.
G=1 dB).e.1336 radiation pattern will be used. which are also different for the C and I path due to different average arrival directions.2 dB 16QAM = 18. we need to differentiate also the propagation model parameters to be used.ECC REPORT 33 Page 22 In rural LOS scenario an ISOP < 1% is already obtained with the less demanding NFD (or X3) limit of 64 dBW/MHz.2 Urban NLOS scenario For the urban scenario. intermediate situation deployment).16 NLOS model described in Annex 1 section A1. SUI channel models of the IEEE 802.e. For the calculations the typical ITU. Antenna height = hTS=4 / 15 m (see also considerations in the Propagation model section) Propagation models: As in the objective for the revision we will use the IEEE 802. Antenna Gain = GTS=4 dB (omnidirectional desk indoor deployment. availability will not be dominated by deep Raleigh fading but by NLoS statistics (i.2. dropping the antenna gain by the 5 dB minimum losses (i.5 dBm/MHz Antenna TX Gain (sectorial 90) = 16 dB Antenna height = 30 / 40 m Modulation formats = QPSK / 16QAM NF = 8 dB (including cable connection to antennas) Modulation formats = QPSK / 16QAM Typical S/N @ BER=106 for coded modulation: QPSK = 11. 2. directional rooftop/high balcony outdoor deployment. typical for generic purpose residential IP connections with less coverage/quality requirements). For simulating the actual random building/indoor propagation losses. Antenna Gain = GTS=10 dB (ETSI TS2 typical RPE. Therefore the more critical parameter might still be the CS to CS interference. .2 dB and TS relevant data as: Typical C/I for 1dB threshold degradation at BER=106: QPSK = 18 dB 16QAM = 25 dB Typical C/I for BER=106 in unfaded conditions: QPSK = 12 dB 16QAM = 19 dB Antenna Gain = GTS=18 dB (ETSI TS3 typical RPE.R F.16 adopted model) which are fully uncorrelated between wanted (victim) and unwanted (interferer) paths.1336 radiation pattern will be used. In the following evaluations we will conservatively consider CS with the following data: TX output power = 35 dBm Signal bandwidth = 3. Therefore. For the calculations the typical ITU.R F. on quite lower cell size. directional indoor window deployment. a 90 sector RP with random orientation will be used.5 MHz TX output power density = 29. Having to manage rooftop outdoor and indoor (ranging from ground floors to higher floors) situations.3. the system performance degradation from interference is not considered “noise limited”. typical for high coverage of business high quality connections). but “interference limited” whenever the C/I falls down a critical value for the assumed modulation format.
In NLOS scenarios there are many possible combinations of parameters to obtain such channel descriptions. The latter curves are obtained as average of about 106 channels trials.Category C+SUI2 as best case (higher floors or less obstructed situations) Shadowing variance: σ = 10 dB Additional Attenuation = 5 dB (building entrance loss with window or desk antenna. Figure 10:Example of shadowing attenuation variance (hTS=15m) .ECC REPORT 33 Page 23 It should also be noted that. Path Loss @ 3. that would in any case fall outside the Erceg model applicability (limited to TS height ~ 10m). while Figure 11show the additional distribution of “fastfaded Rice attenuation” for the two selected typical in SUI2 and SUI4 channels. hts = 15 m Shadowing std dev = 10 dB 10 D [km] 0 10 1 10 1 Figure 9: Example of shadowing attenuation variance (hTS=2m). We would rather use two different Erceg Categories and SUI models. common to wanted and unwanted signals. 4m. respectively. in any case likely close to a window) or 15 dB (less favourable indoor location conditions) TS antenna height: Three cases are only considered for the C/I evaluation. being the terminal location. in figure the average curve of the two models is also reported. in particular a SUI2 result into a model suitable for a light multipath scenario instead a SUI4 results into a model suitable for a scenario characterized by strong multipath.5 GHz 200 180 160 140 120 100 80 60 40 1 10 NLoS model Erceg C hcs = 30 m. With the above assumptions we have selected two specific models with two variant each: Business case (directional TS outdoor): ErcegCategory B+SUI4 as worst case (lower floors or more obstructed situations) Erceg. formally the models are valid for TS heights from 2 to 10 m. hts = 2 m Shadowing std dev = 10 dB 10 D [km] 0 Path Loss @ 3. and then the propagation model. they are considered conservative.Category C+SUI2 as best case (higher floors or less obstructed situations) Shadowing variance: σ = 8 dB Additional Attenuation = 0 dB (no building entrance loss) Residential case (omni/low directivity TS indoor): ErcegCategory B+SUI4 as worst case (lower floors or more obstructed situations) Erceg. for representing more or less favourable propagation conditions. therefore. it will be shown that C/I statistic is practically independent from TS height and propagation models applied. moreover. 10m and 15m For higher floors situations. conservatively we will not use TS locations higher than 15m. The following Figure 9and Figure 10 show examples (for Erceg Category C at two different hTS height) of the attenuation distribution of the received power due to shadowing at various occurrence probability versus distance.5 GHz 200 Free Space Mean NLos 85% shadowing 95% shadowing 99% shadowing Free Space Mean NLos 85% shadowing 95% shadowing 99% shadowing 180 160 140 Att [dB] 120 100 80 60 40 1 10 Att [dB] NLoS model Erceg C hcs = 30 m. Practical experience in the mobile services shows that SUI4 model is quite severe. even if antennas heights are a free parameters in IEEE formulas.
The interference scenario is described by the following Figure 12. Nevertheless. some of the cases considered in the C/I statistic would have been discarded for lack of wanted signal (i. even in NLOS propagation. however. it certainly would have been if the antenna would be outdoor. while for the Interfering signal it might be too conservative. . In this section the same propagation models and system parameters will be used for evaluating the occurrence probability of C/I ratio between a wanted and an unwanted CS stations. provided that the most critical cases are TS far from the wanted CS (lower average elevation of C arrival angles) and close to the I CS (higher average elevation of I arrival angles). with respect to the principal direction of the main signal. this might be marginal for the evaluation of the “wanted” C.In case of omnidirectional TS antennas.The antenna radiation patterns are taken into account only in azimuth. while for indoor use additional considerations should be made: The IEEE 802. The widows themselves would act as lateral shielding for the omni antenna so that it might be better described as lowgain “sectorial”. the orientation of such antenna should be rendered random on a 360 azimuth (due to the random position of the window closest to the TS. At first glance it might look correct. other available models treat only indoortoindoor propagation. here used only for indoor applications. .e. there was sensible contribution to the results from TS that appears to be within the side lobes of the Interfering CS. With this modelling.ECC REPORT 33 Page 24 Figure 11: Examples of Rice fading statistics Additional antennas and path loss simulation methodology for the indoor TS evaluation .16 does not cover indoor applications. the area coverage would become lower for lack of signal and not for excess of C/I).
Case in Figure 12A has proven to be worst case when directional TS antennas are concerned. wanted and unwanted sectors CS will be also randomly located and the sectors boresight also randomly oriented. For simplification. we will study only the case with sectors aligned (when sectors are differently aligned. while case of Figure 12B would be more severe for applications with TS omnidirectional antennas.ECC REPORT 33 Page 25 Figure 12A: Generic model for CSTS interference (Case A) Figure 12B: Generic model for CSTS interference (Case B) In principle. . the interfered sector will be in any case wholly interfered by one or another sector of the interfering cell).
g. Step 4: Inclusion of the Rician fading component in the overall statistic distribution of the unwanted interference (I): For each TS the distribution function of unwanted interference path attenuation. Trials with RSL < BER 106 threshold are discarded. 2. as target for all cases. 1 dB was assumed in present Report 33) up to the critical C/I producing BER >106 in unfaded condition. the simulation of the random building entry and indoor losses will be made artificially assuming the antenna RP as a 90 sectorial with a randomly assigned orientation. if affected by an interference.1 Methodology The objective is to define the EIRP discrimination between adjacent blocks (e. even with CS collocated. derived from the step 2. while most will have higher RSL. is convoluted with the relevant SUI channel attenuation distribution. The result constitutes the spatial and time distribution function of the wanted RSL (C).5*R (only for validate omnidirectional worst case). will be at least of several wavelengths so that the statistic components of Shielding and Rician attenuation of wanted RSL and unwanted interference (I) could be considered uncorrelated. in term of block mask or NFD) in order to offer a suitable protection between blocks. The two CS are placed at distance D=0 m (CS colocated). . As in the previous cell coverage evaluation.3. ¾R and 0. therefore only few will have RSL close to the threshold. The RSL of the possible active Terminal Stations of the wanted cell would be randomly distributed according the selected propagation model. G = 1 dB).e. R/2. 5 dB). R/4. Also in this case.ECC REPORT 33 Page 26 The interfering CS is moved along the axis. In NLOS propagation. The overall sum constitutes the distribution function of the wanted “shadowed” component of RSL (CSH). The overall sum constitutes the distribution function of the unwanted “shadowed” component of interference (ISH). Step 3: Inclusion of the Rician fading component in the overall statistic distribution of the wanted RSL (C): For each TS the distribution function of wanted path attenuation. the evaluation of the interblock NFD/EIRP protection requires a number of steps: Step 1: Evaluation of the statistic distribution of the wanted RSL (C) related to the shadowing: For each TS the mean wanted path attenuation (Aw50) is evaluated. according the procedure described in step 1 below). Other ~300 independent Monte Carlo trials for the shadowing attenuation are generated. the wanted signals sector is populated by a pattern of TS location. the C/I by acceptable those stations would then range from the C/I producing a small degradation of the threshold (e. it will be irrelevant. the gain of the antenna will also be artificially reduced by 5 dB (i. Other ~300 independent Monte Carlo trials for the shadowing attenuation are generated. their distance. is convoluted with the relevant SUI channel attenuation distribution.99*R (cell border) and 1. here each trial resulting in RSL lower that the BER 106 threshold is excluded from the final probability distribution because. Conservatively we would consider. However. derived from the step 1. For introducing the minimal building entry effect (i. Step 2: Evaluation of the statistic distribution of the unwanted interference (I) related to the shadowing: For each TS the mean interference path attenuation (Au50) is evaluated. unlike the previous case.e. only the C/I for 1dB degradation of 106 BER threshold. in elevation or in roofspace. resulting in more trials discarded when the C path is unfavourable. Using the same Monte Carlo method and propagation models the probability distribution of the RSL is derived.2. this will result in privileging the C or the I paths according their different average arrival angle (i.e. When indoor omnidirectional antennas are concerned.g.
whenever not subject of the variation are those of Figure 13. which variation is not elsewhere considered. this might be marginal for the evaluation of the “wanted” C. R/4.2.3. R=2km cell. Erceg Cat B.99*R 10 Probability 0 10 Probability 1 1 D= D= D= D= D= 0 R/4 R/2 3/4*R 0. hCS=30m. provided that the calculations depend on a significant number of input data. hCS=30. The cumulative probability function of C/I is calculated The result constitutes the mean spatial and time cumulative probability function of C/I < x dB. SUI4. SUI4. GTS=18. analysis of the variations produced by each data are reported in next Figures. . are taken into account with the methodology described above. here used only for indoor applications.2. likely different for the C and I paths. R/2.2 Results: 2. and actual situations might contains various mixture of those data. .ECC REPORT 33 Page 27 The result constitutes the spatial and time distribution function of the unwanted interference (I). 10 0 10 D= D= D= D= D= 0 R/4 R/2 3/4*R 0. (Note that basic data. .3. 2. CS distance R/4). even in NLOS propagation. hTS=4.5*R (for omnidirectional case only in order to validate the worst case situation). also for 96 to 99% of the trials are reported. Step 5: Evaluation of the overall probability of the C/I: The overall C and I distribution for each TS are convoluted for defining the probability distribution of C/I. hTS=4m) versus D on a 2 km cell (Case A) In Annex 4 of this document the value of C/I exceeded for more than 99% and.In case of omnidirectional TS antennas. for omnidirectional indoor case.1 C/I statistics The following simplifications have been used for the simulations. In addition.The antenna radiation patterns are taken into account only in azimuth. The following Figure 13 shows examples of the cumulative distributions of the C/I derived with the above methodology. the random building/indoor attenuation. The mean value of the probability distributions for all TSs is calculated. Of particular importance are the last two Figure 19and Figure 20that show the impact of having different W and I CS height.99*R (cell border) and 1. provided that the most critical cases are TS far from the wanted CS (lower average elevation of C arrival angles) and close to the I CS (higher average elevation of I arrival angles). while for the Interfering signal it might be too conservative. ¾R and 0. Step 6: Evaluation of the required interblock EIRP/NFD protection: The two CS are placed at distance D=0 m (CS colocated).99*R 10 2 10 2 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 10 3 60 55 50 45 40 35 30 25 20 15 10 5 C/I [dB] 0 5 10 15 20 (GTS=18 dB) (GTS=4dB + random building loss 5÷40 dB) Figure 13: C/I distribution (Erceg Cat B.2.
5 1 2 2.ECC REPORT 33 Page 28 10 0 10 0 Erceg B .7 10 Probability 0 h = 25 m h = 30 m h = 35 m 10 Probability 1 1 10 2 10 2 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 Figure 17: Variations with cell radius Figure 18: Variations with hCS (W=I) .SUI4 Erceg C .SUI2 10 Probability 1 h=4 h = 10 h = 15 10 Probability 1 10 2 10 2 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 Figure 14: Variations with overall propagation model 10 0 Figure 15: Variations with hTS G = 1 G = 10 G = 18 10 Probability 10 1 2 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 Figure 16: Variations with GTS ( G= 1dB represent the omnidirectional indoor case) 10 0 10 R= R= R= R= 0.
Erceg Cat. here a worst C/I 49 dB with 99% probability is evaluated (for antenna gain 10 dB. B/SUI4) and are practically independent from any parameter a part the antenna gain. the results are sensibly better by several dB. We would conservatively assume a factor of 60/15= 4 reduction. The worst C/I cases reported in Annex 2 are concentrated. while a number of the cases with higher average RSL might be affected only if C/I exceed the “critical C/I” that for any modulation format is assumed to be ~ 7 dB lower.3.2. from Annex 4 it might be seen that the result depends in practice from the W and I CS positioning relative to the cell radius (D/R ratio) and only marginally from GTS.1 Directional TS antennas Provided that the average number of TS is constant over the sector (being physically limited by the handling capability of the system and not by the sector area).5 78 = 33. It seems therefore correct to say that real 99% of the cases would happen with an average C/I among those cases. In addition we should consider safeguard margins for some differences in CS EIRP and heights between wanted and interfering systems: Max ΔEIRP W/I = 5 dB Additional allowance for ΔhCS W/I: 5 dB (from Figure 19 and Figure 20) Interblock protection = 25 (49 + 6) + 5 + 5 = 78 dB Considering that the CS EIRP density considered in the study (see CS data) is +45. where the interfering CS is at 3/4 of the wanted cell.2 Interblock protection evaluation Figure 20: Variations of Interfering hCS 2. With interfering CS situation of Figure 4B.2.3. 18 dB gain antennas resulted in better results with worst C/I 44 dB. the Out of block EIRP for protection of adjacent block TSs is: Out of block EIRP = +45. With these assumptions the required protection could be evaluated as: . with interfering CS situation of Figure 4A. corresponding to 6 dB of C/I improvement.2.ECC REPORT 33 Page 29 10 0 10 h CSU = 25 m h CSU= 30 m h CSU = 35 m 0 h CSI = 25 m h CSI = 30 m h CSI = 35 m 10 Probability 1 10 Probability 1 10 2 10 2 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 10 3 50 45 40 35 30 25 20 15 10 C/I [dB] 5 0 5 10 15 20 Figure 19: Variations of Wanted hCS 2.2. next 9 cases have C/I 46 dB other ~45 cases have 44 < C/I < 10 dB.5 dBm/MHz It should be considered that this value is still conservative for a number of factors: The comparison is made using the C/I for 1dB threshold degradation. for a given cell size the distribution of antenna types and TS location would be random among the various cases considered (~30 for each antenna types for a total of ~60 cases with directional antennas). For defining the Required interblock protection. in reality.2.5 dBm/MHz. From Table 35to Table 38of Annex 4 it may be seen that for a given cell size only 6 cases have actually a C/I 49 dB. we should compare it with the minimum C/I for 1 dB degradation (16QAM): +25 dB We should now consider that the 99% occurrence probability is evaluated on the basis that ALL TS locations and antennas are in the same conditions. the propagation model and TS height have nearly no impact.
the coverage will further drop for adjacent block interference of 4*0. It contains 10 dB margin for W/I ΔEIRP and ΔhCS that. are considered “best effort” service. Therefore. in average. for rural as well as for suburban still highly populated areas. The above two points justified that it is reasonable to assume that the 23 dB worse C/I situation for indoor omnidirectional antennas with respect to conventional outdoor directional cases. more complex evaluation. greatly reduce the system characteristics spread (in term of significant power . are ~ 10 dB less demanding than the value of EIRP previously evaluated with a more empirical propagation model and based on deterministic areas where RSL thresholds might be exceeded due to interference levels only. as shown. 2. Most of the overall system fade margin is dedicated to provide for the “shadowing” attenuation that is somehow static. the ~7dB more robustness to interference (kept as additional safeguard in directional antennas applications) may be taken into account. The worst case appears to be when interfering CS is at the wanted cell border. Therefore. the affected TS are likely the ones close by. with dynamic modulation formats management (actually any new generation BWA system will have this characteristic). During preparation of the first published version of this report.8% expected cell coverage with 99% availability).2.e. however. is addressing broadband services only.2. In these cases the expected C/I distribution would not be the worst one (e. gives results that. from Figure 66in annex 4. based on real C/I occurrence probability following a certified NLOS propagation model. therefore lower performance objectives (from the point of view of coverage and availability) are justified. We have not specifically considered the 64 QAM modulation format that would require ~7 dB more stringent requirements. for their own nature. If we look at a less demanding objective of 96% of worse C/I cases occurrence probability. the expected scenario for the deployment of FWA in 3.ECC REPORT 33 Page 30 Being the worst cases concentrated around an area at ¾R of the cell it is likely that a number of those cases already use a 4QAM format that would require less protection. with respect to those made above for terminals with directional antennas should be taken into account: 1. once the CS and TS locations are fixed. a number of additional considerations. which. is more severe of ~23dB . comparing the interference level with the BER thresholds in NLOS propagation environment will result in very conservative scenario.2 % resulting in a total 76. 2. This is physically expected consequent to the lesser gain/directivity.2. may be duly compensated by the ~7 + 17 21dB. Systems offering indoor applications are assumed to be necessarily of “mixed mode” type. This result was expected since the actual fade margin due to dynamic propagation (i. the use of omnidirectional antennas results in interference situation. only the TSs closer to the CSW or those with very favourable propagation will be addressed with this format. Therefore. the overall coverage degradation should be further mediated over all cases. however.g. Therefore. if the expected cell coverage. it is understood that systems offering 64QAM will be of “mixedmode” nature with dynamic change of formats on a per TS basis. for indoor/omnidirectional terminals is 80% in no interference. This should not be seen as an absolute degradation of the performance of the interfered system.2 Omnidirectional indoor TS antennas From the results in Table 39Annex 4. at 99% availability. ~ 17dB (from a C/I 72 at 1% to C/I 55 at 4%) are immediately recovered. 2.8 GHz has changed and a wide interest on this kind of applications are now reshaping the market expectations: The market. actually most of the TS “in deep shadow from their CS” will have RSL significantly higher than the threshold.g.3.3. being any system generally deployed with a mixture of indoor and outdoor TS. the 4% degradation is only relative to actual percentage of coverage of indoor terminals. at the 99% of occurrence probability of worse C/I cases. an overall discussion of the differentiate applications and objectives and of cost/benefits is reported in next section. being also at cell border will already in average require QPSK format (most applications foreseen even BPSK modulations offering additional safeguard margin). without need of very high additional FM. In addition. for directional antennas. in practice. This would. related to lower W to I distances or better propagation situation in Annex 2).8= 3. standing that we wish to consider only indoor applications. might not be present.3 Conclusions for CSTS protection in urban NLOS environment This new. The indoor applications. which is already assumed to be significantly lower than that of outdoor applications with directional antennas (e.43. the Rician fade) is sensibly lower than Raleigh multipath expected in freespace.
the required NFD or “outofblock” EIRP density for suitably low (<1%) probability of TS interfering a victim CS. previous responses from administrations have highlighted the difficulty in managing two values from regulatory and manufacturers point of view. The introduction in the study of indoor TS with omnidirectional applications have shown that their higher sensitivity. with respect to more conventional applications with directional antennas. WiMAX) have published standards and conformance tests including multivendor interoperability options based on the above market expectation.5 GHz have been made or are in progress in CEPT and worldwide. Therefore this report used an assumption that a single mask.4.4 TS to CS interference 2. From administrations point of view. Therefore it would be advisable to have just one definition of limits for all application scenarios. common to the evaluation made in the section devoted to the IPF and guardband methodology (see Annex 2) and its details are there reported. the evaluation might be based only on a statistical IPF. containing indoor and outdoor applications. However. a wide number of consultations on the use/regulations of bands around 3. 2. ideally by the user alone) Standards organisations (ETSI/BRAN. should be suitable. however. in Table 4 depending on the assumed TS antenna ITUR RPE: TS antennas class TS 3 TS 2 Typical gain (dBi) 18 16 Required TS “X3” value (outofblock EIRP density) 35 dBW/MHz 37 dBW/MHz 49 dBW/MHz TS indoor Omnidirectional 4 (no ETSI standard available) Table 4: Outofblock EIRP density requirement for ~ 1% of TS to CS interference probability in urban scenarios . The recent draft revision of draft ECC/REC 0405 after public consultation have highlighted the wide preference for typical applications with channel bandwidth of 3. respectively.5 MHz or higher.1 Interference evaluation This evaluation would lead to setting the required NFD (or the X3 value of the blockedge mask) for TS. might be somehow balanced by their lower quality objectives on an overall mixture of cases. the outofband limit would become 73Ga=7316=89 dBW/MHz. without distinction between urban and rural scenarios. IEEE 802. may be summarised. From the detailed evaluation made in Annex 2. “Conventional” PMP fully outdoor systems. with more or less severe NLOS. previous responses from industries show great difficulty in matching the proposed “urban” blockedge mask. originally designed for mostly LOS (or moderately obstructed paths requiring professional intervention also for TS deployment).16. for the worst cases presented. Most of them are considering the redefinition of rules assuming only adhoc provisions for “grandfathering” of existing old licensee. the latter being the defining critical case. according a rural or urban environment.ECC REPORT 33 Page 31 density and bandwidth options) among various implementations. the new focus on potential wide consumer market suggests the adoption of less demanding as possible regulations for stimulating development and deployment of initial networks. Originally the two separate outofblock eirp levels have been considered. and the outofband EIRP limit of 73 dBW/MHz would be appropriate to ensure coexistence of both CS>TS and CS>CS. if the BEM were expressed in terms of transmitter output power. are migrating towards more sophisticated performance in strong NLOS environment (allowing simpler TS deployment. Then.
However.1 ”ADJACENT AREA . This would lead to a single value of outofblock power density. broad depressions (more than ~10 km wide).SAME FREQUENCY BLOCK” INTERFERENCE SCENARIO Power Flux Density Limits for adjacent FWS service areas This section focuses on initial interoperator coordination guidelines that would support assignments to FWA operators in the same frequency block adjacent geographic areas. at least for PMP architectures with directive TS antennas. the results in the above Table 4are only incidentally linear with down to omni antenna gain. The specific propagation behaviour in 3. mostly LoS conditions. in particular the spherical diffraction attenuation has been introduced as function of the antenna height. e. The methodology used in this report follows the same approach as for the 40 GHz band. which is the distance from the service area boundary within which transmitter stations should be coordinated with adjacent area operators. each pointing into its own service area.2 Conclusions for the TS to CS interference study The expected more stringent requirements for less directivity antennas suggests that the possible BEM limit would be more suitably defined in term of absolute outofblock power density rather than outofblock EIRP. 2. actually in this case also the additional indoor to outdoor propagation attenuation has been included (evaluated to be at ~ 15 dB in average). The PFD limits are linked with a coordination distance. then reducing the path attenuations In these cases. so that only well known free space and spherical diffraction propagation models apply. is not generally considered a limiting factor for coexistence (when compared to the other interference cases). of 53 dBW/MHz at the antenna connector. it might be considered an upper bound for most situation.2.5 Terminal Station to Terminal Station This kind of interference. However this effect. Therefore. whenever over the service borderline are not present quite different geographical situations such as: high mountain chains acting as physical barriers.ECC REPORT 33 Page 32 2. Each TS will have its EIRP level set to deliver a signal to the CS 6 dB above the receiver threshold. However. This evaluation may require further work in case of mixed PMP and MESH architectures or TS with mobile (outdoor) characteristics are possibly considered. a statistical Monte Carlo approach seems more adequate. A possible scenario could be two CSs located at both sides of the boundary. In such way a single power density requirement will automatically adapt to the necessary EIRP according the used antenna. 3 3. except one (different in each trial) which would have maximum EIRP. the methodology here offered might still be applied provided that suitable modification of propagation model are adopted. .6) (reduced by the assumed omnidirectional gain of 4 dBi). which gain is universally intended to be inversely proportional to antenna directivity. Therefore the limit of 53 dBW/MHz should be assumed valid for omnidirectional use in prevalent indoor situations. whenever the prevalence would be expected for outdoor situations then some 15 dB more stringent value should be required. therefore a worst case approach is considered as too stringent. the spherical diffraction attenuation will play significant role in defining the respective area and the PFD level for triggering coordination. derived from the more stringent EIRP requirement detailed in Annex 2 (A2. to simulate the occurrence of a fading event. Due to the relatively large radius of first Fresnel zone ( 50 m) and the typical horizontal pointing of FWA antennas. river valley. somehow counteracting the spherical diffraction effects. would only impair the operation of one terminal. the results represent only an example for this territory and propagation assumption. Due to the random nature of this kind of interference.4. which was base for the relevant Annex 4 of draft ERC Recommendation (01)04.g. then increasing the path blockage situation and the NLOS attenuation.5 GHz band was taken into account. These guidelines consist of service boundary PFD limits to assist frequency reuse between neighbouring service areas. The study here presented assume PMP FWS services deployed over flat land. unlike the cases where a CS is involved.
1 Assumptions In order to cater for the variety of technologies possible for FWA no assumptions were made regarding duplex method or multiple access method. Those values are not regarded as “typical” for most current system available on the market. but cater for due allowance for some special cases and possible further technology developments. To generate the broadest of guidelines the assumption was merely that an interfering transmitter is deployed in one service area and a victim receiver is operating on the same frequency. 3. as close as the radio site is to the service border. the coordination distance and PFD at the boundary strongly depend on the antenna (interfering TX and victim RX) heights. In Table 5. A reference EIRP = 20 dBW/MHz and antenna heights 20 to 50 m for CS and 10 to 40 for TS are used for this evaluation. equipment characteristics are reported for interference analysis and for a consequent tentative technology independent regulatory framework. being similar to that of CS. For typical values of EIRP expected in the 3. Due to the determinant contribution of spherical diffraction. The values derived from Figure 24 can be used to determine coordination distances over flat land. but located in an adjacent service area. For terminal stations’ (TS) EIRP. This section reviews the methodology behind these figures and proposes the principle of boundary PFD limits as an appropriate means of controlling the interference environment between operators assigned same frequency block(s) in neighbouring geographical areas. The proposed methodology might be also suitable for FSS coordination.5 GHz band. The range of distance and relevant PFD may be reduced or fixed in case administrations may wish to limit. coordination distances are evaluated as 60 to 80 km for PMP CS (see Figure 23). upperbounds for both EIRP and antenna heights above the ground or to define CS downtilt angles in case that height is exceeded. .1.ECC REPORT 33 Page 33 The findings of this section may be as follows: FWA Central Stations (CS) transmitters should be coordinated when the PFD generated at the network’s service area boundary exceeds the value of PFD (dBW/MHz/m2) shown in Figure 24. there is no practical difference apart from the typically lower height of their antenna.
it is assumed that ATPC. Even if antenna gain might be slightly lower in typical applications. In principle there should be no limitation to TS antenna height. For the lower limit. 13 This channel spacing is considered the most representative for being used in the calculation. when high antenna location is used. to downtilt the antenna itself for remaining in the boundary pfd limits set by the Administration. used in particular applications (e.g.5 GHz. we should consider that second generation FWA systems might employ techniques which enable them to operate without a clear LoS path. Spherical diffraction attenuation (1st Fresnel zone partially obstructed due to limited antenna height) calculated following ITUR Rec. 17 18 16 This is the worst case. It is considered that the larger channel systems would determine the coexistence rules. The assumed EIRP is intended to make room for the highest values allowed by present technology. simple (self) installs has resulted in system performance being improved to allow the TS to be deployed within buildings. Typical front ends noise figure in this band are lower (e. also widely popular. No atmospheric attenuation at 3. nevertheless network considerations would generally lead to lower EIRP.ECC REPORT 33 Page 34 Nominal channel bandwidth: Central station EIRP: Central station antenna gain: Central station antenna radiation pattern (90º): Central station antenna height Terminal station EIRPTX: Terminal station antenna gain Terminal station antenna 3dB beam width Terminal station antenna radiation pattern: Terminal station antenna height Typical Central Station and Terminal station receiver threshold (106 BER) Nominal ATPC regulated uplink receiver level Receiver noise figure 7 MHz13 20 dBW/MHz14 18 dBi15 EN 302 085 class C2 20 to 50 m 16 20 dBW / MHz 17 18 dBi ~ ±10 EN 302 085 class TS2 10 to 40 m 18 84 dBm (4QAM) 76 dBm (16QAM) 6 dB above 106 BER threshold 8 dB 19 Interference limit (kTBF – 10 dB) 20 146 dBW / MHz Table 5: Summary of system characteristics assumed for defining the proposed regulatory framework In addition the following propagation characteristics have been assumed: Line of sight path unless otherwise stated. and are rarely higher than 2 meters above the subscribers’ building height. frequency reuse by another operator should be more protected than when operators use adjacent blocks of frequency). however. antenna technology is in fast evolution. In this case it is also intended that the latter systems would more easily meet any regulatory limit. ~5 dB).e. should more easily fit in that possible framework. The 8 dB value included allowance for feeder losses and possible narrowband filters for enhanced selectivity required by dense environment as assumed in this report.5 MHz up). 20 19 .same frequency block” scenario a more stringent requirement is used (i. it being dependent on the customer location.5 GHz should also be taken into account. TS heights may be less than 7 meters. in these lower bands. It is current practice for limiting the latter. therefore 18 dB has been used for taking into account a not infrequent worst case. However. ATPC effect at 3. therefore correlation between interfering and victim paths attenuation is negligible. P. 14 This value includes allowance for feeder losses for full indoor applications. the same consideration made for CS antenna regarding the higher value still applies. For the “adjacent area. The desire for low cost. assuming symmetrical uplink/downlink capacity. Nevertheless lover spacing channels (e.g. very large coverage in remote areas or when NLoS area should be covered at best).562. Hence.g. will be operated by multipath and not by rain. while 16 dB has been assumed as typical value in previous section of this report dealing with “same area – adjacent block”. from 1. 15 Antenna height would impact the cell coverage but also the pfd at area boundaries.
directly aligned path between interferer and victim essentially a “minimum coupling loss” approach.1. can be derived from the link budget equation.1. Operator “A” Victim receiver looking towards the interferer. The threshold can then be tested using Monte Carlo statistical analysis to check its validity in a typical multiple interferer environment. the interfering station EIRPint. Operator “B” Interferer radiating towards the victim in neighboring area.5) –20log(D) – Asph + 18 . i.2 Methodology The PFD threshold has been determined assuming a single interferer and unobstructed LOS. D/2 D PFD “A” Boundary PFD “B” Boundary between licensed areas Figure 21: Defining PFD limit at geographical block assignment boundary 3. Max acceptable interference = x dB below thermal noise floor. Representative EIRP assumed.146 = EIRPint – 92. 3.1.e. This has been calculated following ITUR Rec.5 –20log(3.ECC REPORT 33 Page 35 3. PRX = EIRPint – FSPL – Asph + GRX where PRX is the interference power at the receiver input FSPL is the free space path loss =20 log(4πRmin/λ) Asph is the spherical diffraction attenuation depending on the heights (ha and hb) of the two CS antennas relative to the ground.3.1 Worst case single interferer scenario: 3.. while operator “A” represents the potential “victim” receivers.3 Central Station to Central Station The generic scenario for the definition of PFD limits at service boundary between two operators reusing the same frequency block is shown in Figure 21.5 GHz calculations Assuming a 18 dBi victim antenna gain.562 and approximated as: Asph = 0 dB D < D0 (km) Asph = 1. P. Operator “B” is assumed to be deploying service. The PFD limit is then used to derive an appropriate maximum coordination distance. the minimum separation between the two CSs (Rmin) vs.3 (DD0) db D D0 (km) D0 is the maximum distance where the total calculated attenuation equals the free space attenuation GRX is the receiver antenna gain in the direction of the interferer PRX Max (dBW/MHz) = .
even in that case. 2030 is valid for 2525 and so on) Figure 22: Received interference power vs. between directly aligned CSs under clear LOS air conditions is shown in Figure 23 as a function of EIRP. separation distance RF=3.5GHz. Within practical antenna heights range (20 to 50 m) the minimum separation distance ranges from 58 to 80 km. EIRP=20dBW/MHz 60 80 100 120 PrecMAX (2) dBW/MHz 136 140 146 P MAX (1) rec Antenna heights (m) (3) 5050 160 180 4040 3050 3030 2040 200 2020 1030 1010 220 0 10 20 30 40 50 D (km) 60 70 80 90 100 (1) PRECMAX proposed for CSs and TSs at nominal operating EIRP (6 dB above threshold) (2) PRECMAX proposed for TSs (with ATPC enabled) at maximum EIRP (3) Each curve is valid also for any mixed antenna heights with the same sum value (e. . separation distance for the CS to CS interference scenario (3. it is not likely that operator boundaries lie along a relatively narrow valley and.ECC REPORT 33 Page 36 Figure 22 shows the received interference power as a function of the separation distance from the interfering transmitter for EIRPint= 20dBW/MHz and some different cases of antenna heights (ranging from 20m to 50 m). In Figure 22 flat terrain has been assumed and it shows that in case of different interferer/interfered antenna height. The minimum separation. with the antennas height as parameter. antennas would be “groundgrazing” aligned. Received interference power vs. The second (136 dBW/MHz) is proposed for TSs at the maximum EIRP (during the small percentage of time when ATPC is required to operate to counteract multipath attenuation) as discussed later. Flat terrain is assumed to be close to the worst case.g. The first (146 dBW/MHz) is valid for little or no degradation of the victim CS receiver.5 GHz. the mean value of the two can be taken into account (e. The curves for different EIRP values can be obtained by simple shift of the same amount. ha=20m and hb=40m correspond to the case ha=hb=30m). 3030 is valid also for 2040. line of sight) In Figure 22 two limits are shown. required to meet the –146 dBW / MHz interference criterion defined above.g.
3. .3.ECC REPORT 33 Page 37 Minimum separation between CSs vs EIRP Parameter: antenna heights 90 5050 4040 3050 80 3030 2040 70 2020 1030 Dmin (km) 60 1010 Antenna heights (m) (*) 50 40 30 0 5 10 15 EIRP(dBW/MHz) 20 25 30 Figure 23: Minimum separation between CSs vs. while at Dmin distance the maximum received interference is still satisfied. line of sight) If the required separation distance is apportioned equally between the two regions.14.1. Data in Figure 24 are obtained with Prec(Dmin/2) evaluated assuming a virtual receive antenna height at Dmin/2 to be the same than that at Dmin (not taking into account any earth surface curving as shown in Figure 25).5 GHz. This impacts only the formal PFD evaluation (and its possible measurement) at Dmin/2.1. The interference power produced by a CS Dmin/2 km away is calculated again as: Prec(Dmin/2) = EIRPtx .Ae. and summarised in Figure 24. Therefore the PFD at the service area boundary should not exceed the values derived from the above relationships. The impact of antenna height and EIRP for defining the boundary PFD coordination triggers is evaluated separately in following section 3.FSPL(Dmin/2) Asph (Dmin/2) + Grec The PFD at this distance can be determined using the formula: PFD = Prec .5 GHz with Grec=18 dB The PFD at Dmin/2 is shown in Figure 24 as a function of EIRPtx for different antenna heights. EIRP for the CS to CS interference scenario (3. The graphs in Figure 24 show crossovers that are due to the different slopes of lineofsight (linear with the distance) and spherical attenuation (nonlinear with the distance).3. this will require each operator to ensure any CS directly aligned with an adjacent operator’s service area boundary is located at least (Dmin/2) km away from the adjacent service area boundary.2 and 3.3 dB m2 evaluated at 3. where: Ae = Grec + 10 log(λ2/4π) is the receiving antenna effective aperture Ae = .
ECC REPORT 33 Page 38 Power Flux Density @ Dmin/2 70 Antenna heights (m) (*) 5050 4040 3050 3030 2040 75 80 2020 PFD (dBW/MHz/m ) 2 85 1030 90 1010 95 100 105 0 5 10 15 EIRP(dBW/MHz) 20 25 30 (*) The first value refers to transmit antenna Figure 24: PFD (Ф) at Dmin/2 (half the minimum distance derived from Figure 23 between CSs vs. EIRPtx Figure 25: Principle for evaluating PFD at Dmin/2 in Figure 24 .
a “minimum” EIRP. are Dmin= 30m (“B”) and 38m (“A”) will result in a separation distance of 30+38 = 68 m Dmin for 20/40 m in Figure 23. trees etc. Nevertheless the lower is the antenna height.2 Antenna height Provided that the separation distance is dominated by spherical diffraction attenuation. the suitable value for the boundary PFD (dBW/MHz/m2) could be derived from Figure 24 without knowing the actual CS antenna height of neighbour operator. 3030 is close to 2040. Therefore. For this reason. where the PFD at the boundary (Dmin/2) is plotted in function of the TX antenna height. as recommended by ECC Report 97. the distance from the boundary (Dmin/2) may be managed with the assumption that neighbour operator have equal antenna height. therefore.g. Operator “B”. Therefore values not shown could be extrapolated with small potential error. In principle. Area Boundary 55 50 45 40 35 30 25 20 15 10 5 50 40 30 20 10 0 EI R P = +2 0 d B W / M Hz Antennas height (m) Dm in /2 (km) 10 20 30 40 50 "Operator B" area "Operator A" area Figure 26: Example of balanced Dmin obtained by different Dmin/2 for different antenna heights (EIRP = +20 dBW/MHz) 3. actual lower EIRP should be disregarded in this evaluation and replaced by the agreed value. respecting the PFD at (Dmin)B. assuming equal height on potential RX victim side. the difference in Dmin length will be balanced by its own (Dmin/2)A portion. if the victim Operator “A” has higher or lower antenna height. the use of low antenna heights should be encouraged at the boundary. respectively.ECC REPORT 33 Page 39 3.3. the higher are the diffraction attenuation and all other attenuation due to obstacles such as building.1.3 Transmit EIRP While the antenna height impact on Dmin is self balancing without knowing the victim receiver side (see example of Figure 4. for evaluating PFD trigger should be defined. would keep its own minimum distance (Dmin/2)B evaluated assuming the same antenna height for neighbour victim of Operator “A”. generally reducing the probability of worst case occurrence. which also has impact on Dmin does not have any intrinsic balancing mechanism. 2030 is close to 2525 and so on). Figure 24 shows that different mix of antenna heights produce little difference when the sum of the TX and victim RX antenna heights is the same (e. however the opposite is no longer true and Operator “B” receiver would be interfered because being too close to the boundary. the example in Figure 26 shows that the distances adopted by operators “B” and “A” with antenna heights = 20 and 40 m. as far as Operator “B” reduces EIRP.1. In fact. once knowing its own tower height and EIRP. .5). transmit EIRP.3. Dmin becomes lower (see Figure 23) and the station. could be placed closer to the boundary without coordination not risking interference to Operator “A” as far as the latter uses an equal or higher EIRP for evaluating (Dmin)A. Rationale of this approach is shown in Figure 26. therefore respecting the required separation distance.
1. may be found in the previous section “Same area – Adjacent Block scenario” of this report and is summarised. the PFD at the boundary should be no greater than that produced via an unobstructed path by a directly aligned transmitter radiating the same EIRP.2 Worst case single interferer scenario. according to its own actual deployed maximum EIRPtx and antenna height. This maximum cell size can be determined by considering the downlink power budget. 3. However. assuming a CS EIRP of 20 dBW / MHz. pointing into its own service area.4 Conclusions and possible selfregulation method for CSs coordination distance Unlike what commonly happens in HDFS frequency bands. gives negligible values (based on ITUR P. The worstcase interference arises when the TS is at the maximum distance from its CS. with the fade margin (FM) in Table 6 as function of required availability.999% Rmax (km) FM0 (dB) Rmax (km) FM0 (dB) 18. the above discussion has shown that.ECC REPORT 33 Page 40 3.4 km 20. Fadings from clearair multipath on interfering and victim paths are assumed to be uncorrelated.1..9 13.6). where both the useful and the interference path might be contemporarily faded.3 12. In such a way an operator. The extent of additional protection required would be subject of further studies. where line of sight applications give enough clearance from 1st Fresnel zone for not considering spherical diffraction attenuation. 3.5 GHz calculations For the worstcase interference scenario. At closer distances to the boundary. the worstcase interference scenario occurs when the interfering TS is at a distance Dint = Dmin/2 + Rmax from the directly aligned victim CS. slight correlation may be expected for directly aligned line of sight interference scenario but a very rough estimate of the percentage of time. in the 3. besides EIRPtx.2 8.1 Terminal Station to Central Station ATPC impact The TS is assumed to have ATPC.5308 paragraph 2. should: evaluate the minimum coordination distance (Dmin from Figure 23) verify that the PFD at Dmin/2 (service boundary) does not exceed the limits given in Figure 24. where Dmin is derived from Figure 23 and Rmax can be taken from Table 6. assuming multipath environment in rural (flat terrain) scenario. two requirements need to be considered: . Under normal conditions each TS is assumed to have its EIRP level set to deliver a signal to the CS 6 dB above the receiver threshold. and assuming victim receiver antenna height at the maximum foreseen (e. it is assumed that the interfering TS is directed towards a CS located at the network service area boundary.4. This does not mean that CSs cannot be located closer than Dmin/2 to the boundary. the coordination distance.4 km 12.4. With an antenna height at a distance Dmin/2 from the adjacent service area boundary. in order to allow a similar transmitter (with the same EIRP and with the tallest antenna mast) at the same Dmin/2 from service boundary. Rural Scenario Availability 99. In this case administrations may wish to limit CSs transmitters in both EIRPtx and maximum antenna height (automatically limiting the maximum coordination distance) or to define downtilt angles in cases when height is exceeded.7 Table 6: Typical cell size in rural scenarios System Type A B Therefore. depends on antenna heights of both interfering and victim CS.7 km 17.99% 99. Actually. 3.g.5 GHz band.3. With the assumption made of fading uncorrelation.9 km 15.3.1.4 3. additional protection in the form of reduced EIRP in the direction of the boundary or shielding from terrain or other obstacles will be required.1. This evaluation. 50 m).
4 km FM0 = 12. into the victim CS.6 km Therefore the minimum distance where a CS (supporting far system type B TSs with height lower than 10 m and victim CS height lower than 30 m) could be placed is 6.EIRPATPC) = . Assuming the TS delivering the assumed maximum EIRP of 20 dBW / MHz.99%. we may tolerate up to 3 dB of noise floor degradation (i. so that: D(a) + Rmax + Dmin/2 =~ 54 km (from Figure 23. Rmax = 13.8 dBW/MHz From Figure 22 (at Dint and scaled to the actual EIRP level) we would derive: For case a) an interfering power I =~ 121 – (20 . up to kTBF= 136 dBW/MHz). Rmax = 13. at EIRP = 13. from the useful link (for a percentage of time usually less than 1%) In this case a higher interference level can be tolerated without impairments.99%. should not exceed the required limit (kTBF 10dB) for not impairing the victim performance and availability. the CS of the interfering system should be placed at a distance Dy from the border.8 dBW/MHz From Figure 22 (at Dint and scaled to the actual EIRP level) it is possible to derive: For case a) an interfering power I =~ 132 – (20 . the CS of the interfering system should be placed at a distance Dx from the border. height of interfering TS ht = 10 m.EIRPATPC) = . Both these levels are higher than requirement.2 dBW/MHz For case b) an interfering power I =~ 132 dBW/MHz.2 + 6 = 13.e.ECC REPORT 33 Page 41 a) b) Interfering TS operating at the "normal" EIRP set by ATPC (unfaded percentage of time ~99. so that: D(b) + Rmax + Dmin/2 =~ 52 km (from Figure 22 with Prec Max set to –136 dBW/MHz) D(b) = 52 – 13. the victim CS would receive normal level.127.1.X %) EIRPATPC = EIRPmax .8 dBW/MHz) D(a) = 54 – 13. Example 2 Type B interfering system. depending on the availability objective and the ATPC range. availability 99. at EIRP = 13. Examples: 3.9 km EIRPATPC= 20 – 12.8 dBW/MHz) D(a) = 64 – 13.6 km.X) %) Due to uncorrelation. the CS of the interfering system should be placed at a distance Dx from the border.34 = ~ 4. Also in this example.37.4 . the received signal level at the victim CS at this distance is derived from Figure 22.4 .4 .2 dB Dmin = 75 km (for CS to CS interference assuming hc = 40 m on both sides) Dint = 13. In case a) in order to receive an interfering power of 146 dBW/MHz.5 = ~ 14.34 = ~ 6.138. In this case (most of the time) the received interference power. Assuming that also victim system will work at 6 dB above threshold.2 dB Dmin = 68 km (for CS to CS interference assuming hc = 30 m on both sides) Dint = 13.2 dBW/MHz For case b) an interfering power I =~ 121 dBW/MHz. for the rural scenario. so that: D(a) + Rmax + Dmin/2 =~ 64 km (from Figure 23. Interfering TS operating at maximum EIRP (faded percentage of time ~ (10099. height of interfering TS ht = 20 m.4 km FM0 = 12.4 + 68/2 = 47.6 km for case b).FM0 + FMATPC FM0 is the fade margin corresponding to maximum transmitted power (from Table 6). both these levels are higher than requirement. height of victim CS hc = 30 m.4.4 + 75/2 = 50.3 Example 1 Type B interfering system. availability 99.8 km .4 km EIRPATPC= 20 – 12.2 + 6 = 13. height of victim CS hc = 40 m. in order to receive an interfering power of 136 dBW/MHz (=kTBF). In case a) in order to receive an interfering power of 146 dBW/MHz.
ECC REPORT 33
Page 42
for case b), in order to receive an interfering power of 136 dBW/MHz (=kTBF), the CS of the interfering system should be placed at a distance Dy from the border, so that: D(b) + Rmax + Dmin/2 =~ 62 km (from Figure 22 with Prec Max set to –136 dBW/MHz) D(b) = 62 – 13.4  37.5 = ~ 11.1 km. Therefore the minimum distance where a CS (supporting system type B TSs with height lower than 20 m and victim CS height lower than 40 m) could be placed is 14.8 km. Example 3 Type A interfering system, height of interfering TS ht = 20 m, height of victim CS hc = 40 m, availability 99.99%. Rmax = 18.7 km FM0 = 17.3 dB Dmin = 75 km (for CS to CS interference assuming hc = 40 m on both sides) Dint = 18.7 + 37.5 = 56.2 km EIRPATPC= 20 – 17.3 + 6 = 8.7 dBW/MHz. From Figure 22 (at Dint and scaled to the actual EIRP level) derive: For case a) an interfering power I =~ 129 – (20  EIRPATPC) =  140.3 dBW/MHz For case b) an interfering power I =~ 129 dBW/MHz Also in this example, both these levels are higher than requirement. In case a) in order to receive an interfering power of 146 dBW/MHz, the CS of the interfering system should be placed at a distance Dx from the border, so that: D(a) + Rmax + Dmin/2 =~ 61 km (from Figure 23, at EIRP = 8.7 dBW/MHz) D(a) = 61  18.7  37.5 = ~ 4.8 km. For case b), in order to receive an interfering power of 136 dBW/MHz (=kTBF), the CS of the interfering system should be placed at a distance Dy from the border, so that: D(b) + Rmax + Dmin/2 =~ 62 km (from Figure 22 with Prec Max set to –136 dBW/MHz) D(b) = 62  18.7  37.5 = ~ 5.8 km. Therefore the minimum distance where a CS (supporting system type A TSs with height lower than 20 m and victim CS height lower than 40 m) could be placed is 5.8 km. 3.1.4.4 TS to CS Conclusions
From the above examples, a CS, even if pointing away from the border, could not be indifferently placed nearer than the coordination distance evaluated in Figure 22 and Figure 23. The terminals PFD will become determinant and engineering of the cell (reduced EIRP and sector beams pointing) should be used to ensure that also TSs PFD (in the direction of the boundary) does not exceed the values derived from Figure 24. Figure 27 shows an example of such methodology based on previous examples 2 and 3.
ECC REPORT 33
Page 43
Rmax (TypeA)
X
X
X
X
Rmax (TypeB)
Closest sectorblanked cells (Type A and B)
D(a) D(b)
Dmin/2 Boundary Standard uncoordinated cells
Dmin/2
Blanked sectors of boundary engineered cells (example) Active sectors of boundary engineered cells (example)
Figure 27: Example of cell sector engineering at service boundary (assuming hc = 40 m on both service areas; for Dmin/2 see Figure 23)
3.1.5
Terminal Station to Terminal Station
This kind of interference, unlike the cases where a CS is involved, would only impair the operation of one terminal; therefore a worst case approach is considered as too stringent. Due to the random nature of this kind of interference, a statistical Monte Carlo approach seems more adequate. A possible scenario could be two CSs located at both sides of the boundary, each pointing into its own service area. Each TS will have its EIRP level set to deliver a signal to the CS 6 dB above the receiver threshold, except one (different in each trial) which would have maximum EIRP, to simulate the occurrence of a fading event. However this effect, at least for PMP architectures with directive TS antennas, is not generally considered a limiting factor for coexistence (when compared to the other interference cases). This evaluation may require further work in case of mixed PMP and MESH architectures. As far as TS with omnidirectional antennas are concerned, it is further assumed that: the vast majority is deployed indoor; large building/walls blocking attenuation will be always present. 3.2 Conclusions on adjacent service areas boundary block reuse coordination
It is therefore recommended that coordination between operators using same frequency block(s) in the 3.5 GHz band in adjacent geographic areas should take place for any transmitter (CS and TS assumed to supply very similar EIRP) that produces a PFD derived from Figure 24 or greater at the service area boundary. The distance from the service area boundary that will be subject to coordination, as a function of transmitter EIRP, is indicated in Figure 27. The proposed PFD guidelines can be tested in Monte Carlo simulations to assess their validity in multiple interferer scenarios.
ECC REPORT 33
Page 44
4
CONCLUSIONS OF THE REPORT
This Report has considered a number of facts as initial considerations for deriving the coexistence study: 1. Presently ERC Recommendations 1403 and 1208 for the bands 3.6 GHz and 3.8 GHz do not give harmonised and detailed suggestion to administration for implementing FWA (such as those produced for 26, 28 and 40 GHz). Those ERC Recommendations offer only channel arrangements. 2. 3. The band is limited and wasted guardbands might drastically reduce the number of licensed operators, limiting the potential competition for new services. Legacy FWA (WLL) systems have been already licensed in these bands in some countries, and their coexistence with new FWA systems, assigned in accordance with provisions recommended in this study, may need additional check on a casebycase basis, using the same methodology as described in this report, but with the parameters corresponding to those of legacy systems. Sharing issues with PP FS systems, FSS, radiolocation (in adjacent band), ENG/OB were not considered in this report, but are subject of a different CEPT study. At least for CSs, ETSI ENs in these bands are not presently designed for a technology neutral deployment (this is done only in the 40 GHz MWS EN 301 997) therefore do not contain system controlling parameters, in terms of EIRP, which would be useful for the desired “technology neutral” and “uncoordinated” deployment The study suggested that the block edge mask together with the contiguous assignment of frequency blocks should be the main means for avoiding interference between neighbouring frequency blocks. Since block edge mask parameters are linked to the size of the block, it is also recommended that the neighbouring blocks should be of a similar size. It is also shown that, for PMP TSs, the antenna RPE plays a fundamental role in the coexistence; the more directive is the antenna of TSs, the less demanding might be their NFD (or the EIRP density BEM) required (offering a flexible tradeoff to the market). That is why the resulting BEM limits outside the block are described in terms of transmitter output power, allowing operators to make practical use of this phenomena. In addition, basic rules has been set for the coordination distance and PFD boundary levels between operators reusing the same block in adjacent geographical areas. In this field, the importance of limiting CS antenna height (or downtilt angle) as possible licensing parameter is highlighted in order of having sensible coordination distances (i.e. limited by spherical diffraction attenuation). MPMP (MESH) architectures have not been considered in this Report. It is recognised that whilst some of the results in this report might also be applicable to mixed PMP and MESH deployment, others may clearly need additional work. These studies might be carried on out in due time if needed and when manufacturers will be in a position to offer the necessary information.
4. 5.
6.
7.
8.
9.
0. TS are assumed to be centrally positioned within each subarea. MEL and Rician K are set to the values specified for the SUI channel models. Within each segment. the transmission distance is computed. When the signal level of an estimate is found to be less than the specified performance threshold. For each TS.a) Simulation Model The simulation model is illustrated on Figure 28. there were no exposures found that exceeded the performance threshold of Type A Systems. The area limits of each may be readily computed. These include Mean Excess Loss (MEL). Lognormal shadowing and Rician fading are excluded. Table 7 and Table 8 illustrate the simulation results for covered areas of Type B systems.ECC REPORT 33 Annex 1. . the CS antenna elevation was set to 30 m and the TS antenna elevation and gain set to the indicated values. A random deviate procedure is employed to create shadow loss and Rician K signal variations. For a given simulation. the subarea associated with the estimate is accumulated in an "excluded area" running total. the random variations to MEL due to lognormal shadowing and the impact of Rician fading.1336 (see Figure 1). The cell is subdivided into segments whose angular width is . The impairments relative to LOS are then added.999% availability. The TS antenna RPEs are those derived from ITUR F. while antenna gain is kept fixed at 16 dBi as for general system assumptions in Table 1. This is a midrange value of the range set for the SUI models. Page 45 ANNEX 1: AVERAGE CELL SIZE EVALUATION A1. Setting = 1 degree results in 360 estimates of signal level.1 Rmax where Rmax is the radius of the cell. For MEL estimates. A standard deviation of = 9 dB is set for lognormal shadowing.b) Mean Excess Loss (MEL) only When only MEL was considered. Hence. there are 10 bounded subarea limits within each segment. the ratio of the running total to the cell area defines the % of the cell that cannot meet coverage requirements for 99.1 Rmax Theta Rj Rj+1 TS Location Rmax Figure 28: Simulation Model A1. At the completion of a simulation. angular arcs are positioned at Rj multiples of 0. There are thus 10 arcs within each segment.
the standard deviation for the lognormal fading was set to = 9 dB.2 C 0. As previously noted.9 1.3 35.5 B 1. Due to reduced threshold.9 5.8 1.0 km and 2.8 B 2.7 km TS antenna class TS 2 (ITUR RPE G=16 dBi) TS antenna class TS 2 (ITUR RPE G=+16 dBi) Excluded Area (%) TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi) TS Antenna Elevation (m) 10 15 20 A1.2 0 0 15 A 5. With the inclusion of lognormal shadowing.6 1.6 0 C 0 0 0 Table 9: Type A System Excluded Area Due to MEL and LogNormal Shadowing (Rmax = 2.9 3. coverage issues for Type B systems are significantly increased.45 C 0 0 0 20 A 4.1 0. Annex 1 SUI Terrain Category TS antenna class TS 3 (ITUR RPE with G= +20 dBi) 18.0 km.5 B 1.7 km cell radius designs exceed coverage objectives in Terrain Categories A and B.6 0.5 19 0 0 0 0 0 39.8 0 0 A 0 0 0 B 0 0 0 C Table 7: Type B System MEL Excluded Area for Rmax= 2.9 1. it is apparent that even a Type A system will begin to experience coverage problems.c) Mean Excess Loss and LogNormal Shadowing Table 9 through Table 12 show the results of the simulations when both MEL and lognormal shadowing are considered.7 2.0 km) .ECC REPORT 33 Page 46. both 2.1 2. TS Antenna Elevation = 10 m SUI Terrain Category A B C A B C A B C Excluded Area (%) TS2/TS3 intermediate TS antenna class TS 3 (ITUR RPE G= +18 (ITUR RPE with G=20 dBi) dBi) 51 37. TS Antenna Elevation (m) 10 SUI Excluded Area (%) Terrain TS antenna class TS 2 TS antenna class TS 3 TS2/TS3 intermediate (ITUR RPE G=16 dBi) (ITUR RPE with G=20 dBi) (ITUR RPE G= +18 dBi) Category A 8.7 km. This is constrained to Terrain Category A and Rmax = 2.9 1.6 36 19 0 0 0 0 0 0 36 19 0 0 0 0 0 0 0 Table 8: Type B System MEL Excluded Area for Rmax = 2. Referenced to Table 11 and Table 12.
9 1.8 20. Rician fading is constantly present.9 A 9. LogNormal Shadowing and Rician Fading Generally speaking.3 B 0.4 A 21.3 B 3.4 A 6.85 9.3 A 18.46 0 C 17.8 13.2 B 0.7 0.7 6.5 0 C 17. subject to the vector addition of all of the signal components. it simply means that we are in a fixed up or down fade.9 33.0 16.4 A 29 23.3 B 0 0 0 C Table 10: Type A System Excluded Area Due to MEL and LogNormal Shadowing (Rmax = 2. .0 19.8 A 12.9 2.5 13.7 6. it is not appropriate to interrelate space and time availability.0 B 5. The range of variation in the excluded area was noted and these are presented in Table 13 and Table 14 for Rmax = 2 km. Table 15 and Table 16 examine the same scenarios for Rmax = 2.0 3.6 C Table 12: Type B System Excluded Area Due to MEL and LogNormal Shadowing (Rmax = 2.0 3.5 B 0 0.2 B 0.5 2.9 4. However.36 0 C 12.2 A 5.2 B 1.2 C 39.0 9.7 km.3 13.0 6.6 39.7 17.5 0 0 C Table 11: Type B System Excluded Area Due to MEL and LogNormal Shadowing (Rmax = 2. Page 47 TS Antenna Elevation (m) 10 15 20 SUI Excluded Area (%) Terrain TS antenna class TS 3 TS antenna class TS 2 TS2/TS3 intermediate (ITUR RPE G=16 dBi) (ITUR RPE with G=20 dBi) (ITUR RPE G= +18 dBi) Category 22.8 13. In the event that there is no motion associated with the reflective facets.8 9.8 1.7 km) Excluded Area (%) SUI Terrain TS antenna class TS 2 TS2/TS3 intermediate TS antenna class TS 3 (ITUR RPE G=16 dBi) (ITUR RPE with G=20 dBi) (ITUR RPE G= +18 dBi) Category 26.3 A 7.7 km) A1.4 B 1.1 16.0 C 20.d) Mean Excess.4 29.ECC REPORT 33 Annex 1.8 0.9 4. In order to examine the significance of Rician fading.8 0.0 km) SUI Terrain Category Excluded Area (%) TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi) TS Antenna Elevation – m 10 15 20 TS Antenna Elevation (m) 10 15 20 TS antenna class TS 2 (ITUR RPE G=16 dBi) TS antenna class TS 3 (ITUR RPE with G=20 dBi) 44. in the NLOS transmission environment.0 4.2 23.6 5.4 A 8.8 8.2 33.1 9.9 4.1 1.0 11.3 16.2 C 35.1 6.4 1.6 26.2 11.3 4. its impact was examined for each of the SUI channel models by running 10 simulations for each channel model.6 0.
1 4.0.4 SUI4 B 0 20.4 . It may be noted that the Excluded Area is less when the MEL plus lognormal shadowing loss is considered compared to that just resulting from MEL.6. .7 SUI3 B 0 7. Annex 1 SUI Channel Model Terrain Category Rice K (dB) Excluded Area Spread (%) 12 0.2.ECC REPORT 33 Page 48.9 SUI5/6 A Table 14: Impact of Rician Fading on Type B Systems (Rmax = 2 km) SUI Channel Model Terrain Category Rice K – (dB) Excluded Area Spread (%) 12 0.8 25.32 . examine Table 8 and Table 12 for Terrain Type A and Rmax =2. this is a quite possible result of simulation. For example.2 SUI4 B 0 38.0.7 .0 .04 .2 SUI4 B 0 6. However. with TS antenna height = 15 m and TS antenna gain = +18 dBi).20.36 SUI1 C 8 0.1 SUI3 B 0 22.dB Excluded Area Spread (%) 12 1.1 SUI5/6 A Table 15: Impact of Rician Fading on Type A Systems (Rmax = 2.1 SUI2 C 5 1. The same may be said about Rician fading for which both up and down fades can occur.8 SUI3 B 0 11.2 .e) Simulation Caveat Section A1.2 SUI4 B 0 16.2 SUI2 C 5 20 22.46 SUI2 C 5 5.8 SUI1 C 8 2.9 .0 .7 km.79 SUI2 C 5 8.variation about the mean value that can enhance the signal level of some links. this is not the case.0 40.0.0.22.39 .12.15 SUI1 C 8 0.9.8 SUI3 B 0 3.6 .3.0.c): they are just “midrange” values illustrative sensitivity analysis examples.2.9 .d) does not cover all of the combinations as Sections A1.7 km) A1.78 SUI1 C 8 0.8.0. A close examination of the Tables might imply that there are some inconsistencies in the Table entries.6 .9 SUI5/6 A Table 16: Impact of Rician Fading on Type B Systems (Rmax = 2.16 .b) and A1.0 .0 .5 SUI5/6 A Table 13: Impact of Rician Fading on Type A Systems (Rmax = 2 km) SUI Channel Model Terrain Category Rice K (dB) Excluded Area Spread (%) 12 0.7 km) SUI Channel Model Terrain Category Rice K . However. The shadowing loss exhibits a random +/.5 9.
the dependence of required SG. γ) where: Cov% = coverage objective [%] σ = Shadowing variance [dB] γ = Erceg’s category (A. B and C) parameter The SHM is a statistical function. GRX = Antenna gains [dB] AATT = Additional Attenuation [dB] (e.1 Coverage and availability objectives: Being focused on urban scenarios.f. directional outdoor versus omnidirectional indoor TS A1. therefore it is not appropriate and exceedingly conservative to impose the coverage objective at the cell border only (closer areas would have higher coverage). noise figure (NF) e Rx bandwidth) Step 2: Evaluation of Shadowing Margin (SHM): SHM = SHM(Cov%.99% Residential deployment: Coverage = 95% / 80% / 65% Availability = 99. then running MonteCarlo simulation for each location (~300 trials) for finding the RSL distribution and the percentile of “good” trials (RSL exceeding the equipment threshold).ECC REPORT 33 Annex 1. might equalise the coverage area for the two cases that would exhibit different Fade Margin. Defining different objectives for business (mostly outdoor rooftop) and residential (mostly indoor). The evaluation is done establishing a pattern (~3000/sector) of possible TS location. for a certain coverage. The calculation is made in four steps covering the separate objectives of geographical coverage and link availability: Step 1: Evaluation of the System Gain (SG) defined as: SG = PTX+GTX+GRXAATTPTH where: PTX = TX power [dBm] GTX. ranging from business to residential internet connections we should also distinguish the objectives for the evaluation of the covered area and availability objectives. therefore the coverage is evaluate as the integral of the probability over the whole cell normalised area.f. σ. It could be demonstrated that the SHM (additional to A50) does not depend on the cell radius itself.g. indoor penetration losses) PTH = RSL threshold @ BER=106 [dBm] (function of modulation format.f) Urban scenarios. addressing different kinds of customer need and deployments. Page 49 A1. For this study we have considered the following cases: Business deployment: Coverage = 95% / 80% / 65% Availability = 99.2 Methodology As initial task we could evaluate the impact of the model on the cell radius with the system assumptions made above.0% A1. on the cell radius is then limited to the A50(R) only. . besides being commercially applicable.
95% TS = 4 [m] Cov. 80% Cov. 65% Cov.96 1.67 5. 80% Cov.75 1.58 2.35 3.45 6.98 1. 95% CS = 40 [m] TS = 10 [m] Cov. evaluated for the selected Erceg category model. σ = 8 dB) CS = 30 [m] TS = 4 [m] Cov.42 2.59 3.87 2. 95% Cov. FFM = FFM(SUIProb. σ = 8 dB) CS = 30 [m] TS = 4 [m] Cov.ECC REPORT 33 Page 50. 80% Cov. 65% Cov. in the following tables different evaluations are made using different CS and TS location heights.57 4. 65% Cov. 65% Cov.22 1. 65% 2. Annex 1 Step 3: Evaluation of Rician Fast Fading Margin (FFM) derived from cumulative probability of the selected SUI model (see Figure 3) at the required availability.16 2.78 4. antenna typology and relevant Erceg Categories and SUI models Business case (availability 99. 65% Cov. 65% Cov. TS outdoor directional antennas GTS=18 dB) Lower elevation/more obstructed antennas (ErcegCategory B+SUI4.06 0. 65% 0. 95% TS = 10 [m] Cov.65 1.26 1. 95% Cov. 95% TS = 4 [m] Cov.0 = mean attenuation (dB) at reference distance R0.88 8. 65% Cov.48 Table 17: RMAX [km] for QPSK formats CS = 30 [m] TS = 4 [m] Cov.80 2.72 3.58 4. 95% CS = 40 [m] TS = 10 [m] Cov.0 –SHMFFM)/(10*γ)) A1. 80% Cov.57 7.30 1.f.84 2.86 1. 80% Cov.93 Table 19: RMAX [km] for QPSK formats .14 2.38 1. 95% Cov. 65% Cov.09 1. 95% TS = 4 [m] Cov. 80% Cov. 65% Cov. 95% CS = 40 [m] TS = 10 [m] Cov.46 1. 80% Cov.1 km) A50. RMAX is therefore finally given by: RMAX =R0*10^((SG .87 1. 95% TS = 10 [m] Cov.62 2.36 Table 18: RMAX [km] for 16QAM formats Higher elevation/less obstructed antennas (ErcegCategory C+SUI2.0 +γ*10log(RMAX /R0)+SHM+FFM where: R0 = a reference distance (0.30 5. 65% 1.A50. Avail) where: SUIProb is the cumulative attenuation probability curve Avail = required availability Step 4: Evaluation of RMAX with the relationship: SG = A50. 80% Cov. 95% TS = 10 [m] Cov. 80% Cov. 80% Cov.99%.3 Cell radius evaluation With the above methodology. 80% Cov. 80% Cov.35 2.
80% Cov. 95% CS = 40 [m] TS = 10 [m] Cov.73 1.84 1.TS = 10 [m] Cov. 95% Cov. 80% Alt. 95% Cov.8 1.TS = 4 [m] Cov.63 1.25 Table 24: RMAX [km] 16QAM Residential case (availability 99 %.69 2. 95% Alt.TS = 10 [m] Cov. 95% Alt. 80% Cov.CS = 30 [m] Alt.51 2. 65% Cov. 65% Cov. 65% Cov.CS = 30 [m] Alt. 65% Cov. 65% Cov.TS = 4 [m] Cov.35 3.CS = 40 [m] Alt.TS = 4 [m] Cov. 95% Cov.TS = 10 [m] Cov.77 Table 22: RMAX [km] 16QAM Higher elevation/less obstructed antennas (ErcegCategory C+SUI2. 65% Cov. 65% 0.84 2.94 4.40 0. TS indoor omnidirectional antennas GTS=4 dB) Lower elevation/more obstructed antennas (ErcegCategory B+SUI4. 65% Cov.32 1. 80% Cov.TS = 10 [m] Cov.TS = 4 [m] Cov.27 1. 80% Cov. 95% Alt.94 2.90 Table 20: RMAX [km] for 16QAM formats Residential case (availability 99 %.70 6. 80% Cov.84 1.TS = 4 [m] Cov.18 1.65 1. 80% Cov. TS indoor window antennas GTS=10 dB) Lower elevation/more obstructed antennas (ErcegCategory B+SUI4.04 1. σ = 10 dB) Alt.48 3.55 1.69 2. 95% Alt. 95% Cov.36 2.42 Table 23: RMAX [km] QPSK Alt.ECC REPORT 33 Annex 1. 65% Cov.77 4.25 1.94 1.36 1. 80% Cov. 65% 1.TS = 10 [m] Cov.53 2. 65% 1. 80% Cov.59 2. σ = 10 dB) Alt.89 5. 80% Cov. 80% Alt.49 3. 95% Cov.55 5. 95% Cov. 95% TS = 10 [m] Cov.CS = 40 [m] Alt. 80% Cov. 80% Alt. 95% Cov.77 4.CS = 30 [m] Alt.TS = 4 [m] Cov. σ = 10 dB) .94 2.25 0. 80% Cov. 80% Cov.01 2.61 Table 21: RMAX [km] QPSK Alt. 95% Cov.TS = 10 [m] Cov. 95% Alt.58 0.18 1.CS = 30 [m] Alt.06 1.71 2. 80% Alt.CS = 40 [m] Alt.TS = 10 [m] Cov. 80% Cov.TS = 4 [m] Cov.TS = 10 [m] Cov. 95% Cov.59 2.11 4. 95% Alt. 65% 0. 65% Cov. 95% TS = 4 [m] Cov.02 1. 65% Cov.15 1. 80% Cov.93 1.27 0. 65% Cov. 65% Cov. 95% Alt.81 1.CS = 40 [m] Alt.TS = 4 [m] Cov.73 2. 65% Cov. 80% Cov. Page 51 CS = 30 [m] TS = 4 [m] Cov.06 1.10 2.57 0. 65% Cov.83 3. 65% 1.63 3.41 3.84 3. 80% Cov.94 4. 65% Cov. 95% Alt.
95% Alt.86 Table 28: RMAX [km] 16QAM. 95% Alt.22 2.37 0. 65% Cov.TS = 10 [m] Cov. 95% Cov.TS = 10 [m] Cov. 65% Cov. 65% 0. σ = 10 dB) Alt.90 1.64 0.87 1.CS = 40 [m] Alt.TS = 4 [m] Cov.68 0.29 0.CS = 40 [m] Alt. Building/indoor losses = 5 dB Alt.TS = 4 [m] Cov.ECC REPORT 33 Page 52. 80% Cov.53 0. 80% Cov. 65% Cov.92 0. 65% Cov. 65% 0. 95% Cov.TS = 4 [m] Cov. 65% Cov.34 2.13 0.TS = 10 [m] Cov. 95% Alt.74 0.TS = 4 [m] Cov.63 0. 95% Cov.TS = 10 [m] Cov. 65% Cov.53 2. 65% Cov. Building/indoor losses = 15 dB Higher elevation/less obstructed antennas (ErcegCategory C+SUI2. 95% Alt. 80% Cov. 80% Cov.50 0.19 Table 25: RMAX [km] QPSK. 95% Alt. 65% 0. Building/indoor losses = 5 dB Alt. 95% Cov. 80% Cov.CS = 40 [m] Alt. 80% Cov. 80% Cov.99 0.39 0.67 1.26 Table 26: RMAX [km] QPSK. 80% Alt.85 1.17 0.59 0. 80% Alt.CS = 30 [m] Alt.60 3. Building/indoor losses = 5 dB .28 1. Building/indoor losses = 15 dB Alt.45 0.62 1.63 2. 95% Cov.CS = 30 [m] Alt. 95% Alt.47 0.80 1.49 0. 95% Alt.62 2.57 0. 80% Alt.TS = 4 [m] Cov.53 0.TS = 10 [m] Cov.42 0.67 0.TS = 10 [m] Cov. Annex 1 Alt. 95% Cov.TS = 10 [m] Cov.96 1.42 1. 65% Cov.CS = 40 [m] Alt. 65% Cov.11 1.CS = 30 [m] Alt.15 1.TS = 4 [m] Cov.53 0. 65% Cov.06 0. 80% Cov.TS = 4 [m] Cov. 80% Cov. 95% Cov. 95% Alt.TS = 10 [m] Cov.78 0.94 1.71 1. 80% Cov. 80% Cov.99 0. 65% Cov. 80% Cov.49 Table 27: RMAX [km] 16QAM.99 1.90 0.55 Table 29: RMAX [km] QPSK. 65% Cov. 65% Cov.00 1.33 0.TS = 10 [m] Cov. 95% Cov.73 0. 95% Cov.22 1.92 1. 65% Cov. 65% 0. 65% 0.44 1.86 1. 80% Cov.CS = 40 [m] Alt.30 0. 80% Cov. 95% Alt. 80% Alt.78 1. 80% Cov.TS = 4 [m] Cov. 80% Alt. 95% Alt.TS = 10 [m] Cov.TS = 4 [m] Cov.TS = 4 [m] Cov. 65% Cov. 95% Cov.CS = 30 [m] Alt.CS = 30 [m] Alt.
95% Cov. We should consider that those evaluations do not take into account a number of features that last generation systems. 65% Cov.85 1. 1 km. Building/indoor losses = 15 dB Alt.TS = 4 [m] Cov.01 1.96 1.81 0. 2.77 1.90 1. 95% Cov. 65% Cov.95 1.60 0. The results for the more and the less favourable conditions might also be averaged for having the actual overall coverage %.30 0.CS = 30 [m] Alt.35 0.74 0. addressing in real offered service a mixture of different situations (residential or business.35 Table 31: RMAX [km] 16QAM. 80% Cov. 95% Cov. the actual expected coverage/cellsize would be tradeoff considering which market is addressed and the local environment situations. 95% Alt.44 1. 65% Cov. 95% Alt.72 2. Building/indoor losses = 5 dB Alt.71 0.27 1. .86 1. 80% Alt.TS = 10 [m] Cov. Page 53 Alt. They could be for example: Adaptive modulation/coding formats (when RSL of a TS is marginal. we would consider four radii cases: 0.5 km.63 1. in following study for evaluating the CS to TS interference based on statistical occurrence of C/I.02 0. 95% Alt. Building/indoor losses = 15 dB A1. It should also be taken into account that.ECC REPORT 33 Annex 1.TS = 10 [m] Cov.58 0.07 1. 65% Cov.TS = 10 [m] Cov. 65% Cov. with more or less favourable expected propagation).16 0.10 0. 80% Cov. 80% Alt.91 1.TS = 10 [m] Cov. 80% Cov. 65% 0.50 2.23 0. 2 km. we could qualitatively say that the coverage and availability objectives could be fulfilled with cell radius ranging from ~0.56 0. In conclusion. 80% Cov. 95% Cov.CS = 40 [m] Alt. Therefore.52 0.CS = 30 [m] Alt. offer for improving the situation.TS = 4 [m] Cov. 95% Cov.g) Conclusions of the annex The Tables in this annex summarise all cases generated from the input parameters for the business and residential applications. 80% Cov.55 0. 95% Alt. specifically designed for these NLOS applications. 65% Cov. 65% 0.53 0.CS = 30 [m] Alt. 65% Cov.TS = 4 [m] Cov.TS = 4 [m] Cov. 80% Cov.TS = 10 [m] Cov. 65% Cov.49 0.TS = 10 [m] Cov.CS = 40 [m] Alt. 80% Cov.7 km.30 Table 32: RMAX [km] 16QAM. 80% Alt.CS = 40 [m] Alt. 80% Cov. 95% Cov.33 0.TS = 4 [m] Cov. 65% Cov. 95% Alt.TS = 4 [m] Cov. 80% Cov.47 1. 95% Alt. the lower modulation format and the strongest error coding is automatically selected) Adaptive antenna arrays Diversity subchannelling All these techniques enhance the coverage with respect to the conventional PMP characteristics used in this study.97 Table 30: RMAX [km] QPSK.81 1. 65% 0.5 km to ~ 3 km depending on the decision to privilege or not residential traffic with respect to business only.
1 System Model and Simulation Methodology In subsequent sections. Graphically. Maximising cell edge signal level reduces the sensitivity of CS to CS couplings (not examined in this report). . it is necessary to generate random Rayleigh deviates that are created from the uniform random deviates available with computational machine programs. the simulation results are presented as an interference grade of service (GOS) probability shown as a Cumulative Distribution Function (CDF) vs. it is not statistically appropriate to mix spatial link availability with time varying availability. resulting in 4608 interference estimates). only LOS transmission is considered. Even for the rural environment. Thus. uncorrelated Rayleigh fading is ensured. It is important to note that there is no valid technical reason to apply cell edge ATPC except for interference exposures associated with TS to TS couplings. To account for the assumption that there is no operator coordination. Thus.75Rmax. The second assumption is to assume that the TS locations are randomly areaproportionally located. thus no cell edge ATPC is assumed. Two extreme possibilities can be considered. Page 54 ANNEX 2: TS TO CS INTERFERENCE EVALUATION A2. the only fading mechanism considered to be applicable is that of atmospheric Rayleigh multipath. These are considered to be quite rare.1. It is computationally convenient to consider the overlaid sector/cell as being the victim. A simulation run is configured to spin the relative sector alignment in 5 degree increments. In the rural environment. The procedure is based on the AcceptanceRejection method as detailed in [3] and is summarised in Appendix 2 to Annex 2. Only the area proportional assumption will be subsequently examined. The number of randomly located interference TS locations within a sector is set to Nt. estimates of interference susceptibility are based on MonteCarlo simulations that identify the spatial probability of a victim link experiencing an excessive level of interference. The TS to CS system model is illustrated on Figure 29. Within the victim sector. it is only necessary to set a victim TS to CS signal level based on that of a single celledge victim link located at distance Rmax. In this latter case. = 64. we will examine this situation. the relative boresight alignment of the two CS antennas is considered to be unknown. However. A complete simulation run thus consists of 360/5*Nt interference estimates (Nt = 64.1. Generally speaking. Thus. This is parameterised at some separation distance S between the two CS sites. To account for Rayleigh fading. The first assumption might be to assume that the TS locations are randomly distanceproportionally located referenced to the maximum cell radius Rmax. all TS locations are assumed to employ distance proportional ATPC.1 Directional outdoor TS A2. approximately 50 % of TS locations would be expected to be located at greater than 0. the link margin is modest. Due to the distance differentials associated with the victim and interference paths. There is no statistical measurement data available to identify how these TS locations should be located. with the caveat that it only applies during the time intervals when Rayleigh fading occurs. all received signal levels from victim TS links are assumed to arrive at the victim CS at the same level of signal strength. C/I.ECC REPORT 33 Annex 2.1.1 Rural scenario A2.
It is no longer valid to assign the victim TS to be at distance Rmax. i. If the Rayleigh fade impairment exceeds this adjustment. Only one victim TS is active during a given TDMA time block. This fixes the TX power level PTi of any single interference link. Given that both the uncorrelated Rayleigh faded victim signal level (C) and the interference signal (I) can now be computed. then set the victim TX power to be at its maximum level. To establish link loss. As the magnitude of Rayleigh fading is distance related. we must now do the following: Compute FSL based on the distance Rv. Employing the same methodology as previously described. located at some random distance R0. Modify the value of PTv accordingly. we now have to examine the TX power of the interference link.e.Adjust the FSL signal level PTv so that it is reduced to be ATPC distance proportional. However. the C/I of each interference estimate can be determined. Referenced to Figure 30. It is best described by reference to Figure 30. Given that we have a local meteorological environment that induces Rayleigh fading on the victim link. a new Rayleigh fading adjustment is determined for this path. Page 55 Cell Edge Victim Link TS Location Random Interference TS Location Random Location Victim CS Arc S Cell Radius R Spin Angle 1 Victim CS Figure 29: Simulation Model The inclusion of Rayleigh fading adds considerable complexity to the simulation model. the interference coupling path is a different uncorrelated Rayleigh path at a distance Ri. .Determine the Rayleigh fading adjustment as discussed above. . it is quite valid to assume that the same conditions apply to the interference link.ECC REPORT 33 Annex 2. However. . we should now place the victim TS at some random distance Rv from it's serving CS. But we now have two transmission paths to consider. .Adjust the victim TX signal level via ATPC so that it adjusts to the FSL margin level set at Rmax. This sets the TX power level of the victim link transmitter. the uncorrelated Rayleigh fading signal level adjustment is described as above.. the first of these is the link between the interference TS and it's serving CS. by 20log(Rv/Rmax). For any one of the Nt interference TS's.
A comparative examination of Figure 31 and Figure 32 indicates that the poorest CDF results occur when S is large. In each figure a value of NFD has been selected. angles increase. will increase. Figure 31 applies to S being between 3 to 6 km while Figure 32 applies to a CS separation distance S from 0.9 km. there is a resultant sharp "knee" in the C/I in the vicinity of the NFD value. Some interference TS locations are essentially eliminated as they are behind the victim CS antenna.ECC REPORT 33 Annex 2. It now becomes modest on the interference links. and both the interference and victim CS antennas are partially aligned. As well. 18 and 20 dBi). identifying the minimum NFD requirement to “hit” the ~ 1% CDF at the C/I critical threshold for the system types reported in Table 1. As S increases. conflicting geometrical results occur. With ATPC. as shown on Figure 32.1336 TS antenna using Gain = 16. the major difference in signal level is that of NFD and. Countering this. thus setting up an increase in signal level differentials. both interference and victim link signals would be expected to arrive at the victim CS at comparable levels. These differences can be explained as follows: When S is small. Figure 31 trough Figure 34 illustrate the Monte Carlo simulation results (unfaded) with antenna classes TS2 to TS3 (i. FSL is comparable on both links and TS Antenna RPE is modest.2 Unfaded Simulation Results For a cell size of Rmax = 8. a high percentage of the interference TS links are illuminated by the victim CS antenna. and the RPE rejection of the interference TS increases. we will only present the S > 3 km case.1. Hence in the subsequent simulations (Figure 33 and Figure 34). as shown on Figure 31. thus further reducing the number of serious interference exposures. Hence. which controls the NFD requirement. when S is small. as interference TS distance from victim CS decreases.1 to 2 km. However the level. and percentage of worst case C/I exposures. .1. The C/I "knee" is thus diminished while the percentage of C/I exposures above the knee is reduced.e. is relative distance proportional ATPC. Also. with typical RPE derived from ITUR F. Page 56 Victim Link @ Distance Rv Victim CS Interference Coupling Path @ Distance Ri Victim TS Interference Link @ Distance R0 Interference TS(Nt) Interference CS Figure 30: Rayleigh Faded Interference Model A2.
ECC REPORT 33 Annex 2. NFD = 32 dB) (TS 2 antenna class (ITUR RPE G=16 dBi)) Figure 32: Unfaded CDF S < 2 km. NFD = 32 dB) (TS 2 antenna class (ITUR RPE G=16 dBi)) . Unfaded CDF (S > 3 km. Page 57 Figure 31 .
Page 58 Figure 33: Unfaded CDF (S > 3 km. NFD = 30 dB) (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) Figure 34: Unfaded CDF (S > 3 km.ECC REPORT 33 Annex 2. NFD = 28 dB) (TS 3 antenna class (ITUR RPE G=20 dBi)) .
1. To explain this somewhat surprising result. They thus transmit at a lower power than in the unfaded coexistence scenario. However.. the CDF impairments resulting from uncorrelated Rayleigh fading are quite modest.1.f. For the victim links. it was found that 24% of the victim links were required to operate at maximum power. For the interference links. crossover for Rayleigh occurs at 63%. Thus. 50% of these maximum power links were within 3 dB of the unfaded signal margin. NFD = 32 dB) (TS 2 antenna class (ITUR RPE G=16 dBi)) . a statistical examination of the ATPC adjusted signal level was performed. Page 59 A2.3 Rayleigh Faded Simulation Results Figure 35through Figure 38 illustrate the results for the case where the Rayleigh fading distance probability coefficient dcoeff is set to 3.d. For the remainder. a high percentage of victim links arrive at close to the same signal level as that for the unfaded scenario. the distance proportional ATPC range was sufficient to restore the signal level to its unfaded margin level. Figure 35: Rayleigh Faded CDF (S > 3 km. But this also means that 37% of the links will be in excess of the median level.ECC REPORT 33 Annex 2. Here. As compared to the unfaded case. we first note that the median level p. these TS transmitters are ATPC adjusted to be lower in power.
Page 60 Figure 36: Rayleigh Faded CDF (S < 2 km.ECC REPORT 33 Annex 2. NFD = 32 dB) (TS 2 antenna class (ITUR RPE G=16 dBi)) Figure 37: Rayleigh Faded CDF (S > 3 km. NFD = 30 dB) (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) .
NFD = 28 dB) (TS 3 antenna class (ITUR RPE G=20 dBi)) A2.4 Conclusions It is concluded that the NFD values summarised in the following Table 33 are acceptable values for the TS emissions associated with TS to CS interference couplings in the rural scenario. TS antenna class TS antenna class TS 2 (ITUR RPE G=16 dBi) TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi) TS antenna class TS 3 (ITUR RPE with G=20 dBi) 32 30 Minimum NFD required for Type B System (dB) Table 33: Minimum NFD required for Type B Systems Rural scenario 28 . Page 61 Figure 38: Rayleigh Faded CDF (S > 3 km.ECC REPORT 33 Annex 2..11.
7 km.Set the TS .1 Unfaded Figure 39 through Figure 50 in this section illustrate the CDF vs. as specified for the cell edge at Rmax.2 Urban Scenario A21. . As subsequently discussed. Performance degrades noticeably as CS separation distance S increases. . Hence. still with fixed 16 dBi boresight gain).When S is small. Page 62 A2.When S is large. and Rmax = 2. C/I results for: Rice factor K = 30 dB. The signal level of the interferencecoupling path at distance Ri is now determined based on the procedure for computation of excess loss and Rician fading described. The TX power of both the interference and victim links is now set. If the ATPC range is insufficient to achieve the specified margin. A2. . .Compute the mean value of Rician K based on distance Rx and relative to the SUI value for K. the interference distance is small. thus setting up an excess loss differential that strongly favours the interference link. in the presented simulations the NFD used correspond to the minimum required to “hit” the ~ 1% of cases with C/I over the critical C/I threshold for the system type as reported in Table 1.1336 TS antenna using Gain = 16.Adjust the RX signal level to account for both mean excess loss and Rician fading.1. 18 and 20 dBi. the interference link at distance R0 and the interferencecoupling path at distance Ri. The C/I for each interference estimate can now be determined. Again.e. The procedure is as follows: . the probability of a deep fade is extremely low.2. These are the victim link at distance Rv. Consequently.. . For this K value. This is a result of the excess loss differential and can be explained as follows: . a large number of interference and victim links are at a comparable distance from their serving CS sites.2 Simulation Results A2.TX Power level accordingly. For each interference computation it is necessary to set the TX power of the TS for the first two links.2.Compute FSL based on the distance Rx equal to Rv or R0. 20log(Rx/Rmax). .1 Simulation Methodology The simulation model is comparable to that described in section A2.e. then set the TX power to Pmax.Determine the Rician fading adjustment by the random deviate method. Rmax = 2. with typical RPE derived from ITUR F.Adjust the FSL signal level so that it is reduced to be distance proportional. .2.0 km Different TS antenna heights (15 and 20 m) Different TS antenna classes TS2 to TS3 (i. there are three transmission paths to consider.1.. i.2.Compute the mean excess path loss based on the distance Rx. . this is essentially the case without fading. there are fewer interference links that can illuminate the victim CS.ECC REPORT 33 Annex 2.1. Each time. excess loss is comparable on both interference and victim links. . This would typically be somewhere between 6 dB and 15 dB.1 and Figure 30 for the rural scenario. But for those that do so. the simulations found some degree of C/I performance sensitivity referenced to the margin value selected.Readjust the TX signal levels via ATPC so that some signal margin above the threshold level is restored.
NFD = 45 dB (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) .7 km. TS Ant Elev = 15 m. Page 63 Figure 39: Mean Excess Lossbased CDF Rmax = 2. NFD = 45 (TS 2 antenna class (ITUR RPE G=16 dBi)) Figure 40: Mean Excess Lossbased CDF Rmax = 2. TS Ant Elev = 15 m.ECC REPORT 33 Annex 2.7 km.
ECC REPORT 33 Annex 2. NFD = 43 dB (TS 3 antenna class (ITUR RPE G=20 dBi)) Figure 42: Mean Excess Loss based CDF Rmax = 2. NFD = 43 dB (TS 2 antenna class (ITUR RPE G=16 dBi)) .7 km.7 km. TS Ant Elev = 20 m. TS Ant Elev = 15 m. Page 64 Figure 41: Mean Excess Lossbased CDF Rmax = 2.
TS Ant Elev = 20 m. NFD = 41 dB (TS 3 antenna class (ITUR RPE G=20 dBi)) . TS Ant Elev = 20 m.ECC REPORT 33 Annex 2. Page 65 Figure 43: Mean Excess Loss based CDF Rmax = 2. NFD = 43 dB (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) Figure 44: Mean Excess Loss based CDF Rmax = 2.7 km.7 km.
NFD = 42 dB (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) . NFD = 43 dB (TS 2 antenna class (ITUR RPE G=16 dBi)) Figure 46: Mean Excess Loss based CDF Rmax = 2. TS Ant Elev = 15 m.ECC REPORT 33 Annex 2.0 km.0 km. TS Ant Elev = 15 m. Page 66 Figure 45: Mean Excess Loss based CDF Rmax = 2.
Page 67 Figure 47: Mean Excess Loss based CDF Rmax = 2. TS Ant Elev = 15 m. TS Ant Elev = 20 m. NFD = 40 dB (TS 2 antenna class (ITUR RPE G=16 dBi)) .0 km. NFD = 40 dB (TS 3 antenna class (ITUR RPE G=20 dBi)) Figure 48: Mean Excess Loss based CDF Rmax = 2.0 km.ECC REPORT 33 Annex 2.
Page 68 Figure 49: Mean Excess Loss based CDF Rmax = 2. NFD = 40 dB (TS2/TS3 intermediate RPE (ITUR RPE G= +18 dBi)) Figure 50: Mean Excess Loss based CDF Rmax = 2.0 km. NFD = 39 dB (TS 3 antenna class (ITUR RPE G=20 dBi)) .0 km.ECC REPORT 33 Annex 2. TS Ant Elev = 20 m. TS Ant Elev = 20 m.
a NFD between 40 dB and 45 dB looks adequate for acceptable percentages of interference impairment.2 dB and the CDFs values improve. additional system gain is required.0 km) 15 20 Table 34: Minimum NFD required for Type B Systems Urban scenario In case of 4QAM system (system type A).1. with cell radius Rmax = 2. depending both on antenna gain and TS antenna heights. there is an 8 dB increase in system gain.1336 TS antenna RPE. for the most critical system type B. A. there will be very little difference between the previous unfaded results and the Rician faded case. As well. A2. the results will be little changed from those of the previous SUI1 case.2. For simulation for a cell radius of Rmax = 2 km. Page 69 The following Table 34 summarizes.ECC REPORT 33 Annex 2. resulting in a larger fade margin. the critical receiver levels drop to 14. As the excess loss differential was previously concluded to control CDF vs.2. . excess path loss at cell edge is reduced.2.2. Note that the maximum value for S has to be reduced to 1. To operate in more severe nearNLoS environments. This result is expected. Rician fading is modest for K = 12 dB. In spite of the reduced value of K.2. show that urban worst case could be studied in unfaded conditions looking for a suitably low probability of a NLOS interfering TS to reach a victim CS with a C/I exceeding a fixed threshold level for the most stringent modulation format.3 Conclusions. this loss differential reduction is sufficient to offset the increased probability of deep fades.2. in terms of minimum required NFD value. the following may be concluded: 1.2 Rician Faded If we run simulations for the SUI1 channel model with cell edge Rice K = 12 dB.2 Omnidirectional indoor TS A.7 km) 15 20 45 43 43 40 Minimum NFD value required (dB) 45 43 42 40 43 41 40 39 System Type B (Cell size 2. the main findings.1 Input parameters and models A. 2.2.1. The system gain set for the assumed transmission model constrains nearNLoS operation to be within the SUI1 and SUI2 transmission environment. With the use of ITUR F. Thus.5 km. except for differences in detail. Due to the smaller cell size. a SUI2 channel is assumed with a mean value of Rice K equal to 9 dB at cell edge. Objectives In present version of Report 33 only directional outdoor TS antennas are considered. for the various configurations in the urban scenario: TS Antenna class TS antenna class TS 3 TS Antenna TS antenna class TS 2 TS2/TS3 (ITUR RPE G=16 dBi) (ITUR RPE with G=20 dBi) Height (m) intermediate RPE (ITUR RPE G= +18 dBi) System Type B (Cell size 2.2 dB and 20. The study for rural and urban scenarios. the uncorrelated fading relationship results in an "averaging out" of fading differentials between the interference and victim paths. they are outside the scope of this Report. While means exist to provide some increase in system gain. representative of reasonably designed ETSI antennas. C/I performance. reflecting the smaller value of Rmax. A2. based on the fact that TS antenna directivity will eliminate most of the possible source of interference.1.7 km.1.2. Coexistence performance criteria are thus dominated by the excess loss differential associated with nearNLoS transmission. directional outdoor antennas From the preceding analysis and simulations.
modern PMP systems implement quite large range of ATPC on TSs in order to balance all TS arrival powers to the CS at minimum operating level improving linearity of CS operation and keeping intercell interference as low as possible. in this case each TS will contribute to an aggregate interference on the victim CS. Out of block eirp emission will possibly be further reduced. in practice certain percentage of higher power terminals might also be expected.2. the statistical study will be based on: .1.Evaluation of a maximum terminal density .2. for this TSCS evaluation. Therefore.1336 Antenna height = 30 / 40 m NF = 8 dB (including cable connection to antennas) or 5 dB (ETSI BRAN working assumptions) Noise floor = 109 to 106 dBm/MHz Modulation formats = not relevant A.Distribution of the TS population among various households and offices .2. ATPC of ~ 40/50 dB are quite common. Therefore. however. this value is also justified by lesser expected output power for these TS applications. however. the out of block emissions would become: Omni OOB eirp 37 (164) 49 dBW/MHz Besides the lesser antenna gain. Note: ETSI BRAN assumptions for this kind of applications is limited to +20 dBm typical. A.1. . defined for directional TS with antenna gain of 16dB. unless specific limitations might be necessary for sharing purposes. This would add sensible additional margin that we would evaluate in at least 10 dB in average. even if not of the same amount due to noise limitation.Comparison with the noise floor of the CS receiver. with TS omnidirectional antenna gain 4dB. ATPC range Even if ETSI ENs assumed ATPC as an option and do not specify any range. only the expected outofblock emissions would be necessary. Reference eirp limit As reference value we will use the present eirp limit (37 dBW/MHz) reported in draft REC0405.ECC REPORT 33 Annex 2.2.Evaluation of an activity factor averaged over that population . reduced by the expected gain difference.2 CS data (same used for CSTS evaluation): TX output power = not relevant Signal bandwidth = 3. therefore it should not concern the manufacturers. This is reasonable considering that TSs are generally designed for being coupled with a number of different antennas. .5 MHz TX output power density = not relevant Antenna Gain (sectorial 90) = 16 dB Antenna pattern ITUR F.Evaluation of the probability and cumulative distribution functions of the aggregate interference through an high number of Monte Carlo trials (snapshots). Page 70 For the omnidirectional case a different approach should be used.3 TS omnidirectional indoor out of block eirp: The characteristics used for the CSTS evaluation should be integrated with TX characteristics not specified for that purpose.Evaluation of a typical propagation model according the selected location of each TS .
in case of reuse 4 deployment.5 15. Modern PMP systems might manage quite large number of Terminal stations per BS sector. derived from the actual number of households over a certain area. Whichever the capacity the average maximum activity factor (Af) is defined as: Af = 1/NT Modern PMP systems may manage a quite large number (NT) of TS per BS sector.5 62. The urban penetration was considered also 30% due to the special attractiveness of the nomadic usage. JPTBWA(05)17 Annex 1 (may be found at ERO document server at www.4 TS activity factor and density: The aggregate interference of a number of terminals on the territory will be proportional to their density (DTS) and to their average activity factor (Af).1. typically 256 or even more (e. However. up to 1024 addresses are foreseen in BRAN HM and IEEE 802. ranges from: Urban: ~14 < NT < ~ 256 or more.25 750 250 62. in average.625 Table 7:Total number of households / cell L = 2km 3600 1200 300 L=1km 900 300 75 L = 500 m 225 75 19 L = 250m 56 19 5 Urban Suburban Rural Table 8: Total number of subscribers / cell The above considerations. Therefore.g. depending on maximum terminal handling and data rate offered Suburban ~5 < NT < ~ 256 or more.ECC REPORT 33 Annex 2.5 L = 250m 0.2.ero. over a simplified square cells scenario. in suburban and rural areas. we should consider some physical and market limitations. Page 71 A. these TS are sharing a constant capacity per sector. all TSs should share the overall BS capacity/sector.2. results the maximum subscriber number. Households Area (sq km) Urban Suburban Rural L = 2km 4 12000 4000 1000 L=1km 1 3000 1000 250 L = 500 m 0.16 standards protocols). The following considerations are taken from the draft ETSI BRAN System Reference Document (SRD) for fixed/nomadic convergence presented to the 2nd JPTBWA meeting: Reprinted from Doc. tell that.dk/download ) Considering that the maximum penetration rate is 30% for the target broadband services. this results in an maximum average activity factor. the households’ physical constraints maintain the maximum terminal density (urban scenario) constant at: DTSi 900 / km2 They would be subdivided on sectors containing a number of terminals that. PMP systems may manage a number (NT) of TS per BS sector.0625 187. depending on maximum terminal handling and data rate offered . however.
shared between 20 users VoD using MPEG2.g. Reprinted from Doc. needed for tripleplay service concept. with the same assumption of reuse 4 deployment. regular video.1 Shared traffic. its aim to evaluating the spectrum need. which is a reasonable value standing the limited spectrum availability in 3. However.ero.2. . cannot provide any positive business case. Supplementary. will be further limited by a factor 1/4 due to physical deployment reasons.dk/download ) The traffic estimation will be done for the assumption of VDSLlike services. The activity factor.g. The 5 subscribers/cell. where a minimum NT 14 has been estimated. also in this case. in principle. while the maximum DTS would be ~ 1/3 the Af will have the same upper bound of ~1/56.2. A.2 Summary of TS density and OOB eirp reference values: Following the above evaluation of possible worst cases an average mitigations we would derive the parameters to be used for the overall aggregation of interfering TSs. VDSL like Peak data rate: 7Mb/s. we could estimate: Urban ~7 Mbit/s < ULdata/sector < maximum handled by the system subUrban ~2.3. Assuming this value as typical we could estimate that the full UL capacity in urban scenario might not be reached in the lower cell size deployment of radius L=250 m. at 2Mb/s: 20% of users VoD using MPEG4. in this worst assumed case.ECC REPORT 33 Annex 2. due to this. using the shared traffic assumption. at 6Mb/s: 10% of users 2 frequencies / cell L = 2km 4860 1620 412 L=1km 1222 412 108 L = 500 m 320 108 34 L = 250m 88 34 The data rate calculation is done for the following assumptions: Urban (Mb/s) Suburban (Mb/s) Rural (Mb/s) Table 9: Shared traffic / cell for VDSLlike services. some practical considerations on the actual uplink data rate should be taken into account. the variant of 250m cell in Rural deployment has been omitted in the following calculations. With same reasoning in suburban case. the BRAN HM objectives are for minimum 25 Mbit/s per sector. A. will be calculated the data traffic generated by broadcast services. the capacity per sector (DL + UL) ranges from 22 Mbit/s to maximum handled by the system. The same ETSI BRAN draft SRD contains evaluation of the maximum data traffic/cell. allowing broadband data and VoD. Mb/s The above data show that. Page 72 The lower NT value is.3 Mbit/s < ULdata/sector < maximum handled by the system. In modern PMP systems designed for fixed/nomadic purpose the UL maximum data capacity is in the range of few tens of Mbit/s (e. for HDTV.5 GHz bands). therefore it made an average uplink + downlink traffic estimation. in rural like deployment. Assuming a 1/3 UL/DL ratio as reasonable average. JPTBWA(05)17 Annex 1 (may be found at ERO document server at www. the most unfavourable for our aggregate interference evaluation because it will result in the possibly higher activity factor (e.2. Af 1/14). UL+DL.
we should consider that: A fraction of population belongs to the victim operator and then are not to be counted.25 km: TS density urban: TS density suburban: Activity factor urban Activity factor suburban Reference OOBlock eirp Sector radius 0. A fraction only of the interfering TS will be of indoor/omnidirectional nature here evaluated The interfering TS population will be spread over various adjacent channels.2. Cases of TS distributed on higher buildings will use freespace for the outdoor portion of the path.5 dBm/MHz Sector radius 0. in all other cell radius it can be seen that the factor (DTS * Af). those operating on channels closer to the wanted CS channels would predominate over those on farter channels. the IEEE 802. . added to an indoortooutdoor attenuation statistic as explained in Annex 1.6 dBm/MHz DTS = 900/3 = 300/km2 DTS = 300/3 = 100/km2 Af = (1/14)*1/4 = 1/56 Af = (1/5)*1/12 = 1/56 46.e. A. a population of TSs belonging to other FWA systems).16 adopted Erceg propagation model is used for NLOS paths. have equal or lower value due to the constant density and a possible lower activity factor.e.2.ECC REPORT 33 Annex 2. which affects the aggregate interference.5 km: TS density urban: TS density suburban: Activity factor urban Activity factor suburban Reference OOBlock eirp DTS = 900/3 = 300/km2 DTS = 300/3 = 100/km2 Af = 1/56 Af (1/19)*1/3 1/56 46.2. Erceg B variant has shown to be quite tailored for typical NLOS paths in Milan urban areas. for evaluating the aggregate interference from a population density of indoor TS communication applications to a Fixed Service station (i.2. for our purpose). Therefore it is well suited for any other randomly distributed source of interference (i.2. a FWA CS. A. Annex 1 shows the basics of the model used. corresponding to ~ 90% probability of not exceeding the value. Therefore it is reasonable to assume that only 1/3 of the potential interferers would actually concur to the aggregation. however. Page 73 The device densities evaluated above are the overall expected TS population.3 Scenarios The analysis is made using a simulation program developed. Regarding the OOBlock eirp the reference value it would be derived as: OOBlock (RECOMMENDATION) ATPC (Delta antenna gain) 10 log(Af) 37 10 (164) 10log56 76.5 dBW/MHz or 46. assuming the market shared by 3 operators only 2/3 of the TSs will add as interference. A. Annex 1 gives all details.6 dBm/MHz These are the worst considered cases.4 Propagation models: As for the CSTS interference study.5 Simulation results In the following Pmed is the mean value of the aggregate interference evaluated as sum in power of each the aggregate result snapshot divided by the number of total snapshot and is used as quick reference of the result.2. within ECCTG3. while maintaining the indoor statistical distribution of attenuations.
01 dB Pmed = 133.5.16 dB .79 dB = 118. Page 74 A.32 dB = 116.13 %): FSpace situations (26.23 dB Pmed = 120.73 dB = 133.79 dB = 116.80 dB Pmed = 119.2.2 hCS = 40 m Pmed Sector1 Pmed Sector2 Pmed Totale = 118.1 hCS = 30 m Pmed Sector1 Pmed Sector2 Pmed Totale Sector 1: Erceg B situations (73.05 %): Pmed = 168.87 %): Sector 2: Erceg B situations (72.5.2.64 dB Details per sector and propagation models split: Figure 51: Total PdF for hCS = 30 m A.ECC REPORT 33 Annex 2.2.27 dB = 133.2.95 %): FSpace situations (27.
ECC REPORT 33 Annex 2.95 %): FSpace situations (27.87 %): Sector 2: Erceg B situations (72.13 %): FSpace situations (26. Page 75 Details per sector and propagation models split: Sector 1: Erceg B situations (73.05 %): Pmed = 163.28 dB Pmed = 121.29 dB Figure 52: Total PdF for hCS = 40 m .80 dB Pmed = 121.84 dB Pmed = 133.
Therefore. in average. Page 76 Figure 53: CDF comparison A. . Higher CS antennas are already ~ 5dB more protected or alternatively the cumulative probability of similar protection will become ~99%. omnidirectional case here studied benefits also of the indoortooutdoor attenuation.2. This mean that.5 to 1 dB to threshold degradation. to indoor use. which.4m.ECC REPORT 33 Annex 2. contributes by ~15 dB. with the eirp estimated herein. assuming the actual floors average height of ~3. Therefore we could conclude that the proposed EIRP limit of 49 dBW/MHz is considered suitable also for omnidirectional indoor TS. actually. further linearly reduced by the difference in antenna gain. Otherwise an additional reduction of the limit is required. With high cumulative probability (e. The linear scaling with antenna gain. for omnidirectional antennas. The study reported here shows that an outofblock EIRP value of 49 dBW/MHz. omnidirectional antennas should be restricted. coupled to the indoor use and its additional average attenuation might be considered safe enough for the deployment of a considerable number of omnidirectional indoor terminals.g. which could be considered a quite worse situation in urban areas This worst case is already 6 to 9 dB lower than the CS noise floor therefore contributing by less than 0.6 Conclusions The above MonteCarlo simulations show that: The contribution of the surrounding suburban area (sector 2) is negligible.2. provided that it is reduced linearly with the antenna gain difference. on top of a 6 storey + ground floor building with 6m mast. is however incidental. 95%) of the 40000 snapshots the aggregate interference to a CS antenna 30m height is lower than 115 dBm/MHz.
The probability distribution of floor # and its cumulative probability (assuming them ~ 3m height each) has been derived from the total number of buildings and the related ~ 790000 floors (see Figure 56). Page 77 APPENDIX 1 TO ANNEX 2: MODEL BASIC DESCRIPTION The model is based on data taken from aerial pictures of the city of Milan in northern Italy.ECC REPORT 33 Appendix 1 to Annex 2. In such a way we could define through simple Monte Carlo trials an elevation for each interfering TS entry. the distribution on the suburban frame area is derived. Data on buildings height on a great 21 x 21 km area and on a more restricted 8 x 8 km urban area (see Figure 54) encompassing more or less urban and suburban areas) were evaluated over ~ 184000 building entries. Building density is ~ 1200/km2 in the urban area and ~ 280/km2 in the suburban frame area. ~80000 of which in the narrower urban area. by difference. it may represent a typical European city. assigning to it the appropriate propagation model. The cumulative probability distribution of building height is shown in Figure 55. 21 x 21 km 8 x 8 km Figure 54: Milan example areas: urban (square 8x8 km) and suburban (external cornice 21x21 km) .
E01 Floor # probability/total Number of floors 1.E04 1.ECC REPORT 33 Appendix 1 to Annex 2.E03 1.E04 1.E03 1.E06 99 10 5 11 1 11 7 12 3 3 9 81 87 57 15 21 27 33 39 45 51 63 69 75 93 1.E05 1.E02 1. Page 78 Building height probability distribution (urban area) Building height probability distribution (suburban frame area) 1.E+00 1.E06 Floor # Note: Floor #1 means ground floor Figure 56: Probability distribution of floor # within the areas of Figure 54 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Floor # 123 .E06 3 9 15 21 27 33 39 45 51 57 63 69 75 81 87 93 99 105 111 117 Building height (m) Building height (m) Figure 55: Probability distribution of building heights within Milan.E05 1.E01 1.E04 1.E03 Floor # probability/total Number of floors 1.E+00 1.E02 1.E01 1.E01 Hieght probability/total Number of buildings 1.E+00 Floors # distribution (suburban frame area) 1. urban and suburban cornice areas Floors # distribution (Urban) 1.E+00 1.E02 Hieght probability/total Number of buildings 1.E03 1.E05 1.E05 1.E04 1.E02 1.E06 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 1.
e. Figure 58 pictorially show the random height assignment. Description of MonteCarlo snapshots Step 1: Assigning a TS height Each TS device is given with an elevation randomly derived from the relevant building/floor distribution. each one is subdivided in sectors where TS deployment is different i. it is also quite conservative because the same 8 km radius of urban density is maintained up to ± 180. while it should have been reduced to ±4 km. . Page 79 Geometrical description of the model The scenario has been described as semicircles areas around FWA base station shown in Figure 57.ECC REPORT 33 Appendix 1 to Annex 2. It simulate the worst case of a CS sector at urban area border facing the urban area.: Sector 1 Sector 2 portion with urban terminal density portion with Suburban terminal density In each sector a pattern of TS devices is established with the required density over the horizontal plane. 20 18 16 14 12 10 8 6 4 2 0 12 10 8 6 4 2 2 1 0 km 2 4 6 8 10 12 Figure 57: Description of different TS distribution areas in the model Each scenario is represented by a 180 semicircular area.
~15m in urban building distribution and ~10m for suburban): Outdoor Propagation Erceg B. Erceg B or freespace). A continuous distribution derived through the convolution (representing the joint probability) of building through wall attenuation Gaussian distribution given in ITUR P. gives way to a 30dB attenuation slope: PL(d )= 10log{[c/ dfm] 2 [exp((dt /d) )]} . including relevant antenna RPE contribution is calculated.  21 The path loss PL(d) derived from the Siwiak study. maintaining the same average exposition). For defining a suitable distribution we consider areas of a “typical floor” and assume that each building will face the FS station with one side only (if the building orientation is tilted. On the side facing the victim antenna the lower attenuation is expected and on the opposite side higher attenuation would be added. An additional step of 10db. given by the two slope Siwiak21 model (dt=3m) extended over the floors area. TS distributed below the mean building height (i. is introduced in the middle of the building as shown in Figure 59 and Figure 60.1411 and the indoor attenuation. This seems also conservative when considering that no floor trespassing attenuation is considered. arrival angle and distance is calculated and path loss. Assuming typical building area to be ~400 m2 will give an average 20 x 20m square building footprint.e. additional to free space. where fm is the geometrical mean of the signal frequency. which might significantly contribute standing the limited elevation RPE discrimination of CS antennas.ECC REPORT 33 Appendix 1 to Annex 2. and c is the velocity of propagation and dt is the break point where the 20dB freespace attenuation slope. an indoor + building exit additional attenuation is picked up from the random distribution described below. Page 80 Figure 58: Visual example of TS height distribution over model areas Step 2: Defining the suitable outdoor propagation model Each TS now allocated in the space. representing the additional attenuation of internal elevators and stairs areas.e. only portions of two sides will count. Gaussian shadowing variance = 9 dB (truncated at ± 2 ) . this would add high contribution to TSs closer to the victim CS. will be given a propagation model depending its height (i. Step 3: Defining the additional indoor to outdoor attenuation For each TS.TS distributed on height above: will be considered LOS freespace.
Page 81 100.0 80.1411 penetration distribution 20 18 16 Delta attenuation over free space (dB) 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 Distance from the exit window (m) dt=3m dt=3m+10dB step Figure 60: Attenuation of the indoor path additional to freespace (each value assumed to have equal probability) The joint probability obtained from convolution of the two distributions is shown in Figure 61 as probability density and cumulative probability .0 40.0 10.0 Probability [%] 60.0 30.ECC REPORT 33 Appendix 1 to Annex 2.0 50.0 0.0 Gaussian Distribution Standard deviation: = 5 dB Mean: = 12 dB 0 2 4 6 8 10 12 dB 14 16 18 20 22 24 Figure 59: Probability density and cumulative probability of the P.0 20.0 90.0 70.
999 similar snapshots are conducted and a probability density and cumulative probability function of the aggregate PSD is calculated. Page 82 PdF 10 9 8 7 Probability [%] Probability [%] 6 5 4 3 2 1 0 10 100 90 80 70 60 50 40 30 20 10 0 10 CdF 0 10 20 30 Atten. other ~39. [dB] 40 50 60 70 0 10 20 30 Atten. It roughly corresponds to the value not exceeded for 90% of the snapshots. Step 4 Using the attenuation of each TS device. Separate evaluations are made for the mean power density value (Note) in each sector and in the whole area. Step 5 After concluding the first snapshot. Note: the mean power density values are obtained as the power sum of aggregation results of all snapshots divided by 40000 (number of snapshots). described in the previous steps.ECC REPORT 33 Appendix 1 to Annex 2. . evaluated in previous steps 1 to 3. [dB] 40 50 60 70 Figure 61: Probability density and cumulative probability of the joint attenuation functions. the received power density generated by each TS is evaluated and cumulated for all TS population within the same snapshot subdivided into the different sectors and into the overall aggregation.
ii. once U1 and U2 are accepted. the probability of it's occurrence is known to vary as the 3rd power of the distance [4]. where ( Rx ) with that of U3. Let Fmax be the maximum value of a normalized Rayleigh distribution. Generate three uniform random deviates U1. then accept U2 as the random deviate. then conclude that there is no Rayleigh U3 < ( Rx ) . The simulation assumptions are as follows: a) Under Rayleigh fading conditions. If not. Examine the ratio u = Pr(3U2)/ Fmax. b) Compare the value of fading on the link. Setting the range for U2 to be within (0. iii.ECC REPORT 33 Appendix 2 to Annex 2. Page 83 APPENDIX 2 TO ANNEX 2: ACCEPTANCEREJECTION METHOD i. U3 is a spare deviate to be subsequently described. v. If the ratio u is less than U1. then set the Rayleigh fading adjustment to be that given by step v. set the probability of a Rayleigh fade at maximum distance Rmax to be ( Rmax ) = 1. the adjustment to the FSL signal level is 20log(U2). For some lesser distance. If above. iv. For Rayleigh fading. . set the probability to be ( Rx ) ( Rx / Rmax ) d coeff 3. say Rx. Random deviate U3 was not required in the preceding. then reject the triplet and start again. U3. Once accepted as a valid Rayleigh deviate. If U3 > ( Rx ) . U2. Compute a probability point Pr(3U2) based on the Rayleigh probability equation and within a finite truncated range for U2. d coeff . 3) allows Rayleigh fades to span the range from to +10 dB. the associated U3 value is employed to identify the probability of Rayleigh fading at some transmission distance Rx. However.
the BEM methodology seems the easiest way to contain mutual interference while minimising the need for detailed coordination. Page 84 ANNEX 3: EXAMPLES FOR MANAGING A CS BLOCKEDGE MASK When it is considered appropriate a complete “technology independent/uncoordinated use of the bands. shown between X1 and X3 power density levels (see Figure 3). Therefore. Therefore. allowing operators to make practical use of this phenomenon by obtaining higher EIRP when using highly directional hence less interfering antennas. it should here be noted that those studies are mostly made with statistical tools and typical radio systems. The transition zone. Manufacturers might improve the transmitter spectrum mask (and then the possibility of going closer to the block edge) by actually offering guaranteed masks that. The first parameter is intended for maximising coverage. etc) or additional coordination (including networks synchronisation) between operators should be implemented as far as possible. which is generally defined for suitable service and sharing conditions in the licensing process. Using a blockedge mask concept implies that operators should meet the requirements having freedom on three elements only: 1. The EIRP level 2. for rural or for urban applications). equipment manufacturer and an operator can define systems parameters that better fit the network requirements addressed (e.g.g. The transmit spectrum mask attenuation enhancement. Also adjacent block receiver rejection concurs to a reduced interference scenario. It should be also noted that when TDD or mixed FDD/TDD systems are placed in immediately adjacent bands. vertical decoupling) is implemented for improving antenna decoupling as much as possible. The current studies has shown that less directional antennas (either CS or TS) generally produce more probability of interference. It is not therefore excluded that in a limited number of cases specific mitigation techniques might be necessary.ECC REPORT 33 Annex 3. with increasingly protected frequency areas: Internal guard bands areas where protection is not offered unless the interested operators would practice active coordination . is not here considered in terms of IPF or NFD. however this is not in the scope of this recommendation to set limits for it. at least for the CSs. In particular when CSs are colocated on the same building. therefore outofblock emissions in terms of EIRP should be more stringent for lower directivity (and consequently with lower gain) antennas. the probability of occurrence of worse cases of interference between CSs is quite higher than in situations where only FDD are deployed. Implementing a smooth transition zone implies that the amount of protection within a block is not constant but increases as far as the operating channel is removed from the edge. while the other two are strictly related to the actual equipment implementation. The minimum frequency separation from edge of outermost channels 3. a block edge mask is generally designed on the basis of a small level of degradation in an assumed interference scenario with a low occurrence probability of a worst case (e. nevertheless it is expected that ETSI standards will adequately cover the issue. the possible slight decrement of NFD available for the blockoutermost system channels and the practical feasibility of equipment filtering for reaching the required X3 level from inblock Pmax density.g. Managing these three elements. but should be determined through a tradeoff between the need of a limited amount of guardband. are tighter than the minimum ETSI requirement. The provisions in this section have been based on coexistence studies described in this report. even if the mask here proposed offer a suitably low occurrence probability of interference for such cases. low probability of two directional antennas pointing exactly at each other).g. the example mask here considered assumes adjacent blocks to be sized from 4 typical system channels plus an internal guard band as recommended in Annex 1 of ECC/REC(04)05. natural/physical shielding. Consequently. omnidirectional antennas) are employed. when TDD systems are concerned additional mitigation techniques (geographic separation of stations. deployment and service performance objectives assumptions. while EIRP would be automatically lowered when low gain (e. the statistical approach is not applicable and it is assumed that common practice of site engineering (e. That is why the recommended block edge mask limits outside the block are here described in terms of transmitter output power.
while using 6 and 7 cavity the target is feasible with the desired power.5 MHz block.e. with higher permitted outofblock emission. therefore. but in few worse cases coordination between CS might be needed. The dropdown portion. . in most cases. However. 14 MHz bandwidth) for maintaining frequency agility within the block. Also in this case unfiltered equipment can not meet the edge mask unless with very large power reduction. may operate selfcoordination avoiding worse cases and eventually use innermost systems channels in the block that are more protected. administrations require CS notification of emissions parameters (i. outermost system channel placed at the reference distance of 3. thus not affecting expected mutual interference. transmitter EIRP and antenna data with geographical locations) and make the database available.5 dBm/MHz of power density). As a second example.ECC REPORT 33 Annex 3.5 + 1*3.5 MHz channels.5 MHz from edge (i. wideband filters (i. not wasting spectrum but also not gaining any) Analysis over 16 QAM (considered a sensible reference). in Figure 63 systems at proposed power density are allowed to be placed nearer to blockedge as far as the filter complexity increases. Without extra filtering the system could not meet the mask unless with very low power.e. In modern adaptive equipment QPSK is considered to fit the higher order masks.e.e. with 5 cavity filter the power should be reduced by ~10 dB. ~+30. the operator. Page 85 Outermost system channels areas where protection is given with high probability. In the following examples the transition zone of blockedge mask is analyses against the 16 QAM TDMA type B and OFDMA/TDMA type F from ETSI 302 3262 equipment masks for 3. Its size has been chosen taking into account the block size typically needed for broadband applications that will likely dominate current and future FWA licenses in these bands. The result in terms of the maximum power density allowed by the mask is shown in Figure 62. mostly fall within a guardband (internal or external to the block). 5/6 resonators filter as medium high complexity realization or 7 resonators filters as feasible limit for stable and effective hardware design. with the knowledge of other CS already deployed. In particular the following considerations are made: Analysis of typical block size: 4*3. Innermost system channels areas where protection is given with very high probability.5(guard) = 17. The increasing dropdown attenuation at the block edge has the scope of easing TX filtering. considered a typical target for indoor TS coverage. System CS output power +35 dBm (i. 64 QAM is considered an option that not all systems might offer.
Page 86 TDMA OFDMA/TDMA Figure 62: Examples of increasing power density with RF filtering at same edge distance .ECC REPORT 33 Annex 3.
ECC REPORT 33 Annex 3. Page 87 TDMA OFDMA/TDMA Figure 63: Example of decreasing edge distance with RF filtering at same max EIRP .
ECC REPORT 33 Annex 4, Page 88
ANNEX 4: EVALUATION OF THE C/I RATIO EXCEEDED FOR MORE THAN 99% (AND 97% FOR OMNIDIRECTIONAL CASE) OF PROBABILITY. The following Figures summarise the worst cases of the large number of different conditions evaluated in Table 35to Table 39. In addition, for evaluating the possible degradation gradient, Figure 65compares similar evaluation of directional antennas with 1% and 3% occurrence probability. Figure 66summarise the worst omnidirectional situation at various occurrence probabilities.
0
10
20
30
Ag=10dB / C / SUI2 / 15m Ag=10dB / B / SUI4 / 4m Ag=18dB / C / SUI2 / 15m Ag=18dB / B / SUI4 / 4m
C/I (dB)
40
50
60 0,00
0,20
0,40
0,60 D/R
0,80
1,00
1,20
Figure 64: Summary results for worse C/I cases (1% occurrence probability) with directional antennas (Derived from Table 35through Table 38)
1 0 0 Fig 11A 3% prob. Fig 11A 1% prob.
10
C/I (dB)
20
30 Fig 11B 3% prob. 40 Fig 11B 1% prob.
50
0
0.25
0.5
0.75 D/R
1
1.25
1.5
Figure 65: Comparisons of worse C/I cases (Erceg BSUI4) (1% and 3% occurrence probability) with 10 dB directional antennas
ECC REPORT 33
Annex 4, Page 89
Fig 11A 5% prob. Fig 11A 4% prob. Fig 11A 3% prob. Fig 11A 1% prob. Fig 11B 5% prob. Fig 11B 4% prob. Fig 11B 3% prob. Fig 11B 1% prob.
Figure 66: Summary results for worse C/I cases (Erceg BSUI4) (1% to 5% occurrence probability) with omnidirectional antennas Table 35: C/I exceeded for more than 99% Directional antennas Cell radius 0.5 and 1 km R [km] 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 Erceg Category ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC SUI h TS [m] SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 15 SUI2 4 SUI2 10 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 15 Ant.Gain [dB] 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 DCS/R 0 0 0 0 0 0 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 C/I [dB] at 99% 32,82 32,82 32,82 31 31 31 32,82 32,82 32,81 31 31 31 40,44 40,44 40,44 38,03 38,02 38,02 37,63 37,63 37,62 35,33 35,33 35,33 46,24 46,24 46,24 43,39
ECC REPORT 33 Annex 4, Page 90 C/I [dB] at 99% 43,38 43,38 41,74 41,74 41,74 39,01 39,01 39,01 49,77 49,77 49,77 46,57 46,57 46,57 44,11 44,11 44,11 41 41 41 23,52 23,52 23,51 20,22 20,22 20,22 13,14 13,14 13,14 9,81 9,81 9,81
R [km] 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5 0,5
Erceg Category ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC
SUI h TS [m] SUI2 4 SUI2 10 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15
Ant.Gain [dB] 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18
DCS/R 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
Table 36: C/I exceeded for more than 99% Directional antennas Cell radius 1 km R [km] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Erceg Category ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB SUI h TS [m] SUI4 4 SUI4 10 SUI4 15 SUI2 15 SUI2 4 SUI2 10 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 Ant.Gain [dB] 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 DCS/R 0 0 0 0 0 0 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/4 C/I [dB] at 99% 32,82 32,82 32,82 31,01 31,00 31,00 32,82 32,82 32,82 31,00 31,00 31,00 40,44 40,44 40,44 38,03 38,03 38,02 37,63
00 41.Gain [dB] 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 DCS/R 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 0. Page 91 R [km] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Erceg Category ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC SUI h TS [m] SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 4 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 15 SUI2 4 SUI2 10 Ant.99 0.00 31.82 9.33 35.57 46.51 20.ECC REPORT 33 Annex 4.77 49.99 C/I [dB] at 99% 37.00 .24 46.99 0.77 49.38 41.99 0.57 46.00 31.11 44.38 43.51 23.14 9.01 39.82 32.57 44.00 32.52 23.82 32.99 0.82 31.99 0.11 41.33 35.82 32.74 39.74 41.14 13.81 9.01 39.99 0.Gain [dB] 10 10 10 10 10 10 18 18 18 18 DCS/R 0 0 0 0 0 0 0 0 0 0 C/I [dB] at 99% 32.38 43.81 Table 37: C/I exceeded for more than 99% Directional antennas Cell radius 2 km R [km] 2 2 2 2 2 2 2 2 2 2 Erceg Category ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC SUI h TS [m] SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 15 SUI4 10 SUI2 4 Ant.01 49.74 41.76 46.33 46.99 0.82 32.99 0.11 44.22 20.00 23.99 0.22 13.24 46.81 31.99 0.00 41.62 35.22 20.14 13.24 43.99 0.63 37.
38 43.57 44.62 37.99 0.00 31.01 39.38 43.00 23.99 0.74 41.24 46.11 41.51 20.44 40.22 20.51 23.74 41.ECC REPORT 33 Annex 4.99 0.23 43.22 20.82 9.33 35.81 R [km] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Erceg Category ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC SUI h TS [m] SUI2 10 SUI2 15 SUI4 4 SUI4 15 SUI4 10 SUI2 10 SUI2 15 SUI2 4 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 10 SUI2 4 SUI2 15 Ant.11 44.02 37.81 9.Gain [dB] 10 DCS/R 0 C/I [dB] at 99% 32.24 46.77 49.99 0.14 9.00 41.99 0.22 13.44 40.43 38.Gain [dB] 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 DCS/R 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 0.77 49.33 46.99 0.99 0.33 35.11 44.82 .15 13.99 Table 38: C/I exceeded for more than 99% Directional antennas Cell radius 2.7 km R [km] 2.00 41.62 37.76 46.99 0.38 41.03 38.99 0.7 Erceg Category ErcegB SUI h TS [m] SUI4 4 Ant.99 0.62 35.03 38.00 40. Page 92 C/I [dB] at 99% 31.01 39.99 0.57 46.14 13.52 23.57 46.01 49.74 39.
99 0.33 46.11 41.7 2.7 2.57 44.24 43.7 2.52 23.7 2.44 40.7 2.7 2.7 2.14 13.7 2.Gain [dB] 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 18 18 18 10 10 10 10 10 10 18 18 18 DCS/R 0 0 0 0 0 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 0.7 2.00 31.7 2.01 39.7 2.7 2.7 2.22 20.22 20.00 41.77 49.7 2.7 2.24 46.99 0.7 2.7 2.03 38.00 31.7 2.7 2.24 46.7 2.01 39.00 40.99 0.7 2.7 2.7 2.63 37.99 C/I [dB] at 99% 32.51 20.7 2.00 31.7 2.7 2.74 39.63 37.7 2.99 0.33 35.02 37.82 32.38 41.01 49.99 0.7 2.7 2.00 31.7 2.7 2.7 2.43 38.7 2.7 2.7 2.7 2.7 2.7 2.62 35.57 46.77 46.51 23.7 2.7 2.99 0.7 2.14 13.14 .7 2.7 2.7 2.99 0.7 2.99 0.33 35.7 2.00 41.82 32.03 38.00 23.44 40.7 2.7 2.ECC REPORT 33 Annex 4.7 Erceg Category ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB SUI h TS [m] SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 15 SUI2 10 SUI4 10 SUI4 15 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 15 SUI2 4 SUI2 10 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 10 SUI4 4 SUI4 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 Ant.11 44.7 2.7 2. Page 93 R [km] 2.82 31.77 49.7 2.81 31.00 32.7 2.82 32.74 41.22 13.74 41.57 46.39 43.7 2.7 2.38 43.7 2.11 44.
5 0.94 34.19 43.63 32.99 0.98 50.67 37.91 31.16 26.9 25.81 9.16 26.5 0.55 52.25 31.88 42.25 31.52 51.63 43.5 0.Cell radius 0.81 R [km] 2.99 0.44 36.16 26.89 43.16 42.5 0.32 60.93 48.5 0.9 35.9 25.5 0.5 0.07 52.57 51.5 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Erceg Category ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB SUI h TS [m] SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI4 10 SUI4 15 SUI4 4 SUI2 10 SUI2 15 SUI2 4 SUI4 10 SUI4 4 SUI4 15 SUI2 10 SUI2 15 SUI2 4 SUI4 15 SUI4 4 SUI4 10 SUI2 15 SUI2 4 SUI2 10 SUI4 4 SUI4 15 SUI4 10 SUI2 15 SUI2 10 SUI2 4 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI4 4 SUI4 10 SUI4 15 SUI2 10 SUI2 4 SUI2 15 SUI4 10 SUI4 15 D CS [m] 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 0.91 31.67 48.67 52.51 48.03 42.8 57.ECC REPORT 33 Annex 4.99 0.48 35.59 51.5.9 25.91 31.75 42.16 25.14 50. 1 and 2 km R [km] 0.17 57.5 0.16 25.99 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 C/I [dB] at 99% 32.57 51.99 0.5 0.17 42.98 52.71 46.82 43.7 Erceg Category ErcegC ErcegC ErcegC SUI h TS [m] SUI2 4 SUI2 10 SUI2 15 Ant.91 31.5 0.5 0.12 59.25 32.82 9.5 0.5 0.5 0.4 34.91 43.36 26.99 0.5 0.5 0.9 35.99 Table 39: C/I exceeded for more than 99% or 97% OmniDirectional antennas .5 0.89 34.94 51.09 41.83 42.85 34.5 0.75 37.11 37.91 43.5 0.5 0.25 32.51 49.52 35.5 0.Gain [dB] 18 18 18 DCS/R 0.25 32.46 43.7 2.33 35.33 60.5 0.62 43.99 0.5 0.73 51.76 42. Page 94 C/I [dB] at 99% 9.5 0.5 0.7 2.16 26.07 46.99 0.65 36.5 0.93 C/I [dB] at 97% 26.25 .69 53.5 0.84 43.38 58.94 43.76 42.47 42.17 34.48 35.25 32.9 34.5 0.47 37.77 46.9 25.5 0.
19 57.39 50.9 25.99 0 0 0 0 0 0 1/4 1/4 1/4 1/4 1/4 1/4 1/2 1/2 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 0.78 59.48 35.61 43.99 41.86 32.99 0.22 34.20 C/I [dB] at 97% 42.70 49.66 36.ECC REPORT 33 Annex 4.87 58.91 50.33 .85 48.36 38.13 53.62 47.16 26.81 42. Page 95 R [km] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Erceg Category ErcegC ErcegB ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegB ErcegB ErcegC ErcegB ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC ErcegB ErcegB ErcegB ErcegC ErcegC ErcegC SUI h TS [m] SUI2 4 SUI4 4 SUI2 10 SUI2 15 SUI4 4 SUI4 15 SUI4 10 SUI2 15 SUI2 4 SUI2 10 SUI4 15 SUI4 10 SUI4 4 SUI2 4 SUI2 10 SUI2 15 SUI2 4 SUI2 10 SUI2 15 SUI4 4 SUI4 10 SUI4 15 SUI4 10 SUI4 4 SUI2 15 SUI4 15 SUI2 10 SUI2 4 SUI4 15 SUI4 4 SUI4 10 SUI2 10 SUI2 15 SUI2 4 SUI4 10 SUI4 4 SUI4 15 SUI2 10 SUI2 4 SUI2 15 SUI4 4 SUI4 15 SUI4 10 SUI2 10 SUI2 4 SUI2 15 D CS [m] 1/2 1/2 1/2 1/2 3/4 3/4 3/4 3/4 3/4 3/4 0.84 34.33 46.91 43.24 58.72 43.94 59.99 0.9 25.25 49.50 51.51 38.99 0.86 26.44 52.10 43.22 59.3 35.16 25.81 43.82 47.99 0.30 43.62 52.04 45.31 53.05 60.06 51.99 42.52 51.95 51.83 47.98 46.68 52.02 36.18 36.83 35.18 47.02 58.23 50.25 31.34 42.79 37.79 49.99 C/I [dB] at 99% 52.91 31.52 46.8 42.67 57.53 52.99 0.9 35.47 42.84 45.11 56.16 26.99 0.99 0.6 36.6 36.25 32.67 54.92 43.24 53.32 50.12 57.01 36.62 38.05 42.02 41.31 57.04 35.25 32.51 50.94 53.85 46.81 48.99 0.91 31.45 35.99 0.99 0.
Editor B. Fukuda "Field Strength and its Variability in VHF and UHF landMobile Radio Service" (in "Communication Channels: Characterization and Behaviour". Goldberg. K.5 GHz (Inbound).02/03/16. S.16. Ohmori. J. London. 20010717. IEEE Press 1976). T. Parsons "The Mobile Radio Propagation Channel" Pentech Press. ETSI EN 302 085 v1. C802. R.(200102) Fixed Radio Systems. Wiley & Sons. Kawano. Okumura. _____________________________ [2] [3] [4] [5] [6] . Channel Models for Fixed Wireless Applications. Page 96 REFERENCES: [1] Y. IEEE 802.2a02/08. E. Saunders "Antennas and Propagation for Wireless Communication Systems" J.1. PointtoMultipoint Antennas. Antennas for pointtomultipoint fixed radio systems in the 3 GHz to 11 GHz band.ECC REPORT 33 Annex 4.3c01/29r4. Coexistence Same Area Simulations at 3.D.16.2.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.