Professional Documents
Culture Documents
Change History
Version Status Date Author Owner Reviewed by Reviewed Approver Approval Description of changes
date date
0.1 DRAFT 20-12-2016 T. ASZTALOS T. ASZTALOS GS/NPO_BL/TE/ARFD05-12-2013 P.SABATIER 20-12-2016 First draft
1.0 DRAFT 14-12-16 T. ASZTALOS T. ASZTALOS GS/NPO_BL/TE/ARFD20-12-2016 P. SABATIER 20-12-2016 First release
5.1.2 Antenna.................................................................................................................................. 22
8 REFERENCES .......................................................................................................................... 57
9 Abbreviations ........................................................................................................................ 59
1.1 Object
The network topology of IoT systems is comprised of simple nodes (End Devices) that
collect and transmit a limited amount of data to a central controller or Gateway which provides
connectivity to the Internet and cloud services. Hence, an IoT Network is composed of one (or
several) Gateway(s) and a possibly high number of End Devices (clients), such as wireless sensors,
smart meters, home automation systems and wearables. Gateways (GWs) are usually connected
via a wired Local Area Network (LAN) to the Internet in order to allow communications between
the wireless End Device clients and the network’s application server(s).
Nodes and gateways must be designed to minimize power consumption, provide reliable
and robust network connections and extend wireless connectivity range as far as possible.
Achieving good range depends on the antenna gain, receiver sensitivity and transmitter
power. The antenna gain is usually limited by cost and device form factor. Receiver sensitivity
sets the lower limit on power that can still be received and understood and receiver sensitivity
determines how well it can distinguish the desired transmission from other signals and noise in
the area. Thus, having a receiver with both good sensitivity and selectivity makes it possible to
achieve longer range. A primary factor affecting radio sensitivity is the data rate. The lower the
data rate, the narrower the receiver bandwidth is and the greater the sensitivity of the radio.
On the transmitter side, range is determined by the output power level. A 6 dB increase
in link budget will double the range in an outdoor, line-of-sight environment. However,
regulatory standards limit the allowed output power, and increasing the transmitter power also
increases the current consumption, which can have a negative effect on battery life.
Although there may exist on the market dedicated to IoT technology simulation/planning tools
available in this document it is assumed that radio planner is jointly handling several different
technologies while using NOKIA selected/validated RNP tools. Therefore, 9955 V7.3.1 (Atoll
platform) is the assumed as the used RNP tool for IoT RF design. Support of LPWA design
features is available from release 9955 V7.3.1 b9754 onwards. Nevertheless, the methodology
can be adapted and used with another RNP tool, tool version, provided that LPWA is supported.
Issuing radio network coverage predictions by using a dedicated RNP tool, allows for the
consideration of several key factors that cannot otherwise be taken into account, these being:
The 3GPP Release recently finalized three, new cellular LTE-based IoT standards—Cat-M1, NB-
IoT, and EC-GSM—that offer a cellular alternative to proprietary unlicensed Low-Power Wide-
Area Network (LPWA) technologies.
Figure 1: 3GPP (NB-LTE, LTE-M , EC-GSM) and non-3GPPsolutions (LoRa, Sigfox) IoT
2.1 LoRaWAN
LoRa is a wireless technology that has been developed to enable low data rate communications
to be made over long distances by sensors and actuators for M2M and Internet of Things, IoT
applications.
LoRaWAN or Long Range Wide Area Network, is based on Semtech’s LoRaTM PHY chip, specifically
designed and optimised to provide low-power WANs with features to support low-cost mobile
secure two-way communication in IoT, see [1] & [2]. LoRaWANs use of sub-GHz ISM bands also
means the network can penetrate the core of large structures and subsurface deployments
building penetration and urban range 2-5km, 15km suburban, 45km rural range
ultra low power usage, estimated 10 years end-node operation with two AAA batteries
Channel frequencies and bandwidth (normally 500 kHz, 250 kHz or 125 kHz) defined for
each network
Typically regulator defines how often a node is allowed to transmit (duty cycle, in EU 1%)
LoRaWAN operates in the sub-gigahertz frequency bands and it’s specification varies from
region to region because of regulatory requirement.
LoRa networks operate in unlicensed ISM (Industrial, Scientific and Medical) frequencies:
Europe 863 MHz – 870 MHz (and 433 MHz – 434 MHz)
2.1.1 LoRa EU
LoRaWAN defines ten channels for Europe. Out of which 8 channels are multi data rate from 293
bps to 5.5 kbps.
• One channel can operate at higher data rate with a speed of 11 kbps.
2.1.2 LoRa US
LoRaWAN for North America operates in the 915 MHz ISM Band divided into the following
channel plans:
LoRaWAN does not make use of GFSK modulation in the US902-928 ISM band.
Devices, when transmitting with 125 kHz BW may use a maximum of +30 dBm. The transmission
should never last more than 400 ms. Devices, when transmitting with 500 kHz BW may use a
maximum of +26 dBm
The capacity requirements target in 3GPP TR 45.820, [9], has been set to 40 devices per
household, based on assumptions for London, which correspond to 39413 devices per typical
omni cell with cell site radius of 500m.
Based on 3GPP TR 45.820, IoT Capacity evaluation is performed by running system level
simulations using traffic models defined in TR 45.820 Annex E and the system level simulation
assumptions in Annex D, see [9].
The major IoT assumptions are the 900MHz carrier and LoRA 125kHz working bandwidth, and
cell site radius of 500m per site assuming the omni antennas. The IoT Users dropped uniformly
in entire cell with DL/UL 1Tx-1Rx scheme only. For the further details, please see [9].
Case Household Density Cell site sector Number of devices Number of devices within
per sq km radius, R (m) within a household a cell site sector
Urban 1517 500 m 40 39413
Table 1: the target IoT device density per cell
IoT Mobile Autonomous Reporting (MAR) periodic reports assumption: Periodic uplink
reporting is expected to be common for cellular IoT applications such as smart utility
(gas/water/electric) metering reports, smart agriculture, smart environment etc.
Characteristic
Application payload size distribution Pareto distribution with shape parameter alpha = 2.5
and minimum application payload size = 20 bytes
with a cut off of 200 bytes i.e. payloads higher than
200 bytes are assumed to be 200 bytes.
Periodic inter-arrival time Split of inter-arrival time periodicity for MAR periodic
is: 1 day (40%), 2 hours (40%), 1 hour (15%), and 30
minutes (5%)
Table 2: the target IoT traffic model Assumptions
In the following table, the general IoT capacity estimation may be obtained with the related
inputs: 125kHz BW is used with the total BW 200kHz. The IoT report size at 32byte per packet
and the activity is at 0.00013 packet report per second per device. The max DL/UL data volume
are about 2 million bytes per hour per cell based on the SX1301 specification, [2]. The LoRA
report transmission is assumed with 10% frame erasure rate. With the available network
feedbacks and proposal, about 4.8 reports/second per cell are assumed for Poisson distributed
report rates per cell in TR 45.820, which follows variable LoRA DL/UL data rate of 0.13 ~ 16kbps.
LoRa
Available BW in kHz 200 kHz
Devices density in cell 39413 devices
LoRA report size 32 bytes
Common
LoRA activity 0.00013 pkt/s/device
Inputs
BW utilization (total 125kHz) 63%
max DL/UL data volume
2000000 bytes/hour/cell
(bytes/hour/cell)
Report frame erasure rate 10%
Input Poisson distributed report
4.76 report/s/cell
rates per cell
average arrival rate of reports 5.11 pkt/s/cell
network packet rate R
4.29 pkt/s/cell
(pkt/s/cell)
network packet rate R
Output (pkt/hour/cell) 15429 pkts/hour/cell
DL/UL data volume
(bytes/hour/cell) 493714 bytes/hour/cell
max Capacity report rate in #
33061 devices/cell
devices/cell
Table 3: LoRA Capacity estimation per cell site
Based on the above inputs as one example, the LoRA shows that the max capacity report rate is
about 33k devices per cell, which may generate 15.4k packet per hour per cell, and delivered the
max DL/UL data volume at 494kbytes per hours per cell.
A link budget (LKB) requires information about the desired level of coverage, the hardware to be
used and the propagation characteristics of the area to calculate the maximum path loss that
can be tolerated between the GW and end device nodes/equipment. The calculated maximum
allowed path loss (MAPL) is used to predict the cell radius and derive the coverage area of the
cell.
Different propagation effects and equipment characteristics will impact the performance of a
wireless IoT network, hence the size and shape of the coverage area. We can group these
parameters in three categories:
At GW side:
transmitted power
receive sensitivity (for nodes to GW transmissions)
antenna used (plus cable/connector losses if external antennas are used)
antenna height and
antenna orientation unless Omni
Environment:
The transmitted signal eventually reaches the receiver. There, results depend on the Receive
Sensitivity of that device - i.e., the minimum power required to handle arriving frames at a given
link speed. Receive Sensitivity is a given characteristic of an IoT device and will vary across
products.
The following table shows the sensitivity threshold (minimum) requirement for a typical LoRa EU
network GW device.
Table 4: Receiver sensitivity for a typical LoRa wireless GW device for EU modulations
Equivalent
Sensitivity
Mode bit rate
(dBm)
(kb/s)
FSK 1.2 1.2 -122
LoRa SF = 12 0.293 -137
LoRa SF = 11 0.537 -134.5
LoRa SF = 10 0.976 -132
LoRa SF = 9 1.757 -129
LoRa SF = 8 3.125 -126
LoRa SF = 7 5.468 -123
LoRa SF = 6 9.375 -118
Note that in case of LoRa equipment, receive sensitivity is generally stated either as a function
of spreading factors (SF) or as a function of network speed, as different spreading factor and
encoding rate is requested to fulfil these different data rate requirements (see Table 4).
Performances vary per supported standard and product family.
Device
Spreading Estimated NF in
Mode BW symbol rate Rs Sensitivity
factor (SF) SNR in dB dB
(dBm)
LoRa SF = 12 SF=12 125 kHz 30.5 sym/s -132.0 dBm -17.0 dB 8.0 dB
LoRa SF = 11 SF=11 125 kHz 61.0 sym/s -129.5 dBm -14.5 dB 8.0 dB
LoRa SF = 10 SF=10 125 kHz 122.1 sym/s -127.0 dBm -12.0 dB 8.0 dB
LoRa SF = 9 SF=9 125 kHz 244.1 sym/s -124.0 dBm -9.0 dB 8.0 dB
LoRa SF = 8 SF=8 125 kHz 488.3 sym/s -121.0 dBm -6.0 dB 8.0 dB
LoRa SF = 7 SF=7 125 kHz 976.6 sym/s -118.0 dBm -3.0 dB 8.0 dB
LoRa SF = 6 SF=6 125 kHz 1953.1 sym/s -113.0 dBm 2.0 dB 8.0 dB
Table 5 & Table 6 listed minimum SNR (SNRmin) requirements relate to presented sensitivity
figures via receiver sensitivity definition of:
The following LKB example is with indicating some typical values for the above parameters.
Parameter Unit
Service Bit Rate bps 293 bps 537 bps 977 bps 1758 bps 3125 bps 5469 bps 9375 bps
Symbol Rate bps 31 bps 61 bps 122 bps 244 bps 488 bps 977 bps 1953 bps
Chip Rate Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps
Coding Rate 4/5 4/5 4/5 4/5 4/5 4/5 4/5
Channel Model PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h
GW required SINR dB -17.5 dB -15.0 dB -12.5 dB -9.5 dB -6.5 dB -3.5 dB 1.5 dB
GW Noise Figure dB 6.0 dB 6.0 dB 6.0 dB 6.0 dB 6.0 dB 6.0 dB 6.0 dB
GW Sensitivity dBm -135 dBm -132 dBm -130 dBm -127 dBm -124 dBm -121 dBm -116 dBm
GW Antenna Gain dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi
Cable & Connector Losses dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB
Cell Area Coverage Probability % 95% 95% 95% 95% 95% 95% 95%
Outdoor Shadowing Standard Deviation dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB
Outdoor Shadowing Margin dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB
Signal Selection Diversity (SSD) Gain dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB
Penetration Margin dB 36 dB 36 dB 36 dB 36 dB 36 dB 36 dB 36 dB
Interference Margin dB 3 dB 3 dB 3 dB 3 dB 3 dB 3 dB 3 dB
MAPL is computed for different LoRa modes, different SF values which define the different
offered LoRa service characteristics (bit rates in bps).
Service Bit Rate bps 293 bps 537 bps 977 bps 1758 bps 3125 bps 5469 bps 9375 bps
Symbol Rate bps 31 bps 61 bps 122 bps 244 bps 488 bps 977 bps 1953 bps
Chip Rate Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps 125 Kcps
Coding Rate 4/5 4/5 4/5 4/5 4/5 4/5 4/5
Channel Model PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h PedA 3km/h
Device required SINR dB -14.5 dB -12.0 dB -9.5 dB -6.5 dB -3.5 dB -0.5 dB 4.5 dB
Device Noise Figure dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB
Device Sensitivity dBm -130 dBm -127 dBm -125 dBm -122 dBm -119 dBm -116 dBm -111 dBm
GW Antenna Gain dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi 11.2 dBi
Cable & Connector Losses dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB 0.5 dB
Cell Area Coverage Probability % 95% 95% 95% 95% 95% 95% 95%
Outdoor Shadowing Standard Deviation dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB 8.0 dB
Outdoor Shadowing Margin dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB 8.6 dB
Signal Selection Diversity (SSD) Gain dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB 4.0 dB
Penetration Margin dB 36 dB 36 dB 36 dB 36 dB 36 dB 36 dB 36 dB
Interference Margin dB 3 dB 3 dB 3 dB 3 dB 3 dB 3 dB 3 dB
RNP Design Level -115.9 dBm -113.4 dBm -110.9 dBm -107.9 dBm -104.9 dBm -101.9 dBm -96.9 dBm
The MAPL derived via the LKB reflects the path-loss to be spent over the air! Based on the
minimum amongst UL MAPL and DL MAPL the RNP design threshold of ‘RNP design level’ is
derived. This is an important design parameter as it is indicating the expected level of received
signal (RSSI) in DL at where cell edge is predicted based on minimum MAPL. The areas on the
resulting coverage plot where the calculated value of this minimum RSSI design threshold is
reached are assumed to provide coverage. ‘RSSI design level’ is computed for 95% of reliability
(accounted in ‘Shadow margin’).
The aim of document [17] in part 8 is to present these potential interference cases and provide
antenna isolation recommendation for interference-free coexistence between these LPWAN
networks and existing technologies.
Depending on the region or country, LoRa could be used in different frequency bands, and
therefore different coexistence scenarios may exists, depending on the possible spectrum
configurations. Since LoRa transmitters could be coexisting in the same geographical area with
existing technologies, there is a concern that potential interference from legacy BS transmitters
could interfere with LoRa BS receiver and vice-versa.
Example scenarios:
in EU 433MHz and China 470-510MHz frequency bands, coexistence issues with TETRA or with
LTE450MHz (E-UTRA Band 31) may appear.
For EU 863-870MHz ISM band, the most critical coexistence scenario is the potential
interference with E-UTRA Band 20 (LTE800MHz).
For US 902-928MHz ISM band, the coexistence with the cellular systems operating in E-UTRA
Band 8.
In this section further, the focus is on the EU863-870MHz band and US902-928MHz band. Note
that coexistence of LoRa in other specified bands is very similar with these two main scenarios,
therefore they are not studied in particular.
One of the most common cases of LoRa coexistence is between LoRa operating in EU 863-
870MHz and LTE 800MHz (E-UTRA Band 20).
The LoRa network channels can be freely attributed by the network operator, and they are used
for both UL and DL, there are no dedicated zones for UL and DL, like in US 902-928MHz band.
The LTE E-UTRA Band 20 is defined as 832-862MHz for UL and 791-821MHz for DL.
The reverse scenario, LTE UL interfering the LoRa DL, meaning that the LTE User Equipment
would impact the LoRa end-device, is considered less critical, due to lower powers involved and
localized impact only – that particular LoRa end-device is impacted, not all devices in the area.
In document [17] example isolation calculations are indicated and way of implementing those by
means of antenna separation is recommended.
One other common case of LoRa coexistence is between LoRa operating in US 902-928MHz and
2G/3G/4G operating in 900MHz band (e.g. E-UTRA Band 8).
In US 902-928MHz band, LoRa network channels are attributed in separated parts of the band,
dedicated for uplink or downlink transmission: the LoRa uplink can be between approximately
902-915MHz and the LoRa downlink can be between approximately 923-928MHz.
On the other hand, the LTE E-UTRA Band 8 is defined as 880-915MHz for UL and 925-960MHz
for DL. However, due to allocation for LoRa, the remaining spectrum available for the other
mobile systems is 883-902MHz for uplink and 928-947MHz for downlink.
At a first look, it can be noticed that this scenario is not critical, as the spectrum allocation
ensures uplink-to-uplink and downlink-to-downlink adjacencies, which is common and does not
put a risk of coexistence, especially if the same site locations are being used by both systems.
It can be seen that downlink-to-uplink scenarios, which are the worst cases, are not critical, due
to frequency separation. In fact, both systems protect their own uplink bands and therefore the
uplink of the other system also, consequently.
Like for the other technologies deployed, LPWA site acquisition must be taken into account at
the beginning of the radio planning phase, as the sites analyzed must correspond to those which
are going to be implemented on the field.
Usually the customer should provide candidate sites positions and the existing positions which
could be re-used in order to avoid new sites implementation.
The customer request is generally to re-use a maximum of existing sites and to not have any
new sites implementation. This is particularly feasible in case of IoT due the IoT coverage gain
over these other technologies.
Even when sites positions are given by the customer, they must be analysed and validated by
NOKIA.
Radio site search and validation is an important step in radio network planning and
implementation.
It allows identifying:
If a site is eligible for required coverage, by detecting if there are masks (near
obstruction like higher building or trees) which have not been seen with radio planning
tool
Detection of pattern distortion risk due to the environment and the structure near the
antenna
Analysis of the antenna height compared to the nearest environment. For continuous
coverage type of deployment, sites must not be too high compared to its environment,
as it can increase inter-cell interference.
All the type of equipment available on the rooftop, and their frequency band. Based on
that it can be deduced what type of antenna is required, and where to position it in order
to avoid interferences created by co-sitting.
During a site selection any interference should be detected. The interference detection process
is detailed in the document referenced as [15] in part 8 of this document.
5.1.2 Antenna
Below some default height values which can provide a view of antenna height per environment
type.
These values are only indications as the antenna heights placed on roof top for dense urban and
urban environments and mainly in suburban depend essentially on building heights.
Rural: 25m<= antenna height <45m (for rural case the height above 30m is not to avoid
rooftop obstacle but to maximize coverage)
For antenna placed in middle of large rooftop and to avoid rooftop effect the following rule
must be respected:
Min antenna height = Tangent (0.5*antenna vertical beamwidth + max tilt value supported+ 5°) *
distance between antenna & rooftop edge (the distance concerned is the one in the antenna
azimuth direction)
Antenna selection:
Antenna & OEM basics are detailed in the presentation referenced as [13] in part 8 of this
document.
Frequency band
Gain
Polarization:
o The two main possibilities are dual- polarized antennas in +/-45° polarization or
vertical polarized antennas.
o Antenna decoupling is with dual polarized antenna as 30dB are ensured inside a
same radome
o For outdoor macro-cells network, the antennas essentially used are sectors, as
they allow a better efficiency to manage interferences, with azimuth and tilt.
o Vertical beam width: vertical angular sector in which the attenuation is lower
than 3dB.
o Front to back ratio: essentially for sector antennas, must be at least equal to
25dB
It is not recommended to tune a model based on scanner drive tests done on existing network
positioned with GPS and measurements samples binned during the post-treatment:
• Geo positioning with GPS is not sufficient since GPS latency takes too long, and GPS
accuracy is not good enough in dense urban / urban zones.
• Performing samples binning after the measurement collection generates some geo-
positioning error, since constant vehicular speed is assumed, which is not correct
especially in cities and at traffic lights.
UE drive tests could not be used for model calibration, since UE measurement accuracy depends
on how the UE is positioned, and can fluctuate between 7dB to 22dB error compared to
accurate receiver with rooftop antenna.
CW measurements must be done very precisely, as these measurements are the reference to
tune a propagation model.
The methodology used is very specific on data collection organization, accuracy and explanation.
Position, road choice and information collected must be adapted around the main activity.
Dedicated Rx chain with CW receiver with data collection adapted and wheel trigger to
average samples every 40 wavelength to respect lee criterion (so binning is done
dynamically during the measurements collection)
A specific geo positioning approach must be followed as it can have a great impact on
model accuracy if samples are not correctly positioned
The following recommendations must be followed in order to have some sufficient sites and
enough environments characterized by the measurements done in the area where the model
must be tuned:
Various terrain around each station (dense urban, urban, suburban, rural, flat, hilly…)
For dense urban and urban areas 80% of the sites must have their heights in line with
average buildings one, which is generally around 30m to 40m. 20% of sites could have
heights of around 50m to 60m in order to reach high distance range (3km for1800MHz
to 2100MHz and 5km for 900MHz)
For suburban and rural the sites measured must have their heights in line with the type
of deployment required by the customer. For example if the customer wants to use 60m
towers, and if it is allowed, then measurements must be done with this height range. It is
recommended to keep the measured site height range from 30m to 60m at most if
possible.
For each frequency band, a minimum of about 6 macro sites should be measured for a
model calibrated on a city with a minimum of 20.000 averaged samples (the exact
number is terrain dependent i.e. what route length can be traversed). If several cities
The process and the application method are detailed in [12] in part 8 of this document.
For those designers owing only regular 9955 licenses with LPWA not included a dedicated
section, see section 7, describing some 9955 workarounds of designing LPWA networks by using
other available modules, was included (ex. WiFi,…).
You can use 9955 RNP to predict radio coverage, carry out calculations on fixed locations of
end-devices, and evaluate network capacity.
Figure 7: 9955 LPWA Network Tab
• LoRA
• UNB
• WMBus
• Signal levels
• Number of servers
• Carrier-to-interference-and-noise ratio
• Services areas
• Throughputs per cell
Moreover, the 9955 RNP LPWA ACP can be used for LPWA site selection based on server
redundancy as well as signal level and quality objectives. The ACP can also be used to optimize
operational IoT networks.
This step deals with setting up a relevant RNP model in the RNP Tool of a LPWA network by
counting for the assumed type of environment, used network equipment performances,
assumed terminal capabilities, target traffic model, user service profiles, available type of
services. Defining a network configuration consists in a number of different steps the planner
shall follow:
This is the very first preparation step for a new LPWA project. When setting up a new LPWA RNP
project in the 9955 tool you should pass some of the following preparation steps, before
entering any of network data:
Coordinate system setup. Different coordinate systems are used all over the world. In a new
9955 project coordinate system setup should be done at: 9955 Tools menu Options .
This step should be the very first one in case of any new RNP project set-up.
It is mandatory that coordinate system of all the used maps and sites to be the same and to
match with the one the project was set up. Otherwise, one might be faced with wrong
positioning of network elements on the map which may compromise any planning effort spent
with this project.
For a very good overview on coordinate systems and geographical projection, but as well as for
explanation on how they are implemented in 9955, please check [6].
For more information on parameters to be entered as well as their recommended default values
please check [7] and [8].
This is the second and largest step of preparation of a new LPWA RNP project.
This step concerns the setup of general equipment and technology info in the 9955 RNP Tool.
These issues are linked to:
LPWA Frequency Bands: 9955 LPWA module comes with a set of predefined LPWA frequency
band and frequency channel configurations. This list shall be completed only if the target
frequency band or channel parameters of the going to be implemented network are not found in
the list.
Table 8: LPWA Frequency Bands
9955 supports currently three different LPWA standards: LoRa, UNB and WMBus.
9955 default LPWA Frequency Band definitions are accordingly to these possible options.
Additionally, these accounts of supported different working bands (hence the variety of
definitions in the above table).
Correspondingly, the radio bearer table can be set up with the different bearer types/bearer
characteristics of these different technologies.
Definition of LPWA reception equipment types is another important LPWA project preparation
step. LPWA reception equipment models both the reception characteristics of GW (cell)
Bearer Selection tables: graphs which relate to different bearer indexes, defined in the LPWA
Radio Bearers table, to their respective selection thresholds. Note in the above example, which
stands for LoRa, that only LoRa bearers can be selected by this LoRa equipment. Specific care
shall be taken in order to select only the right working bearers from the full range of available
different LPWA technology bearers.
It is possible to define different selection thresholds, corresponding to the same bearers, for
different reception equipment/mobility type.
Note that by setting those three technology specific cell parameters the target supported LPWA
technology is selected for the project’s GWs model (LoRa in the above example).
Service definition should be project and technology specific as well, i.e. directly linked to the
customer’s requirement/targets. For this reason, no default service definitions to use could be
recommended although 9955 is providing some default values/ranges of some of the
parameters.
Table 12: LoRa Service definition example
Other project defaults, like default propagation model, default resolution or folder configuration
where pathloss matrices are kept can be set at ‘Predictions’ properties:
9955 offers import filters for the most commonly used geographic data formats.
Morpho maps may indicate the following standardized clutter classes, which can easily be
interpreted by the RNP tool. For different classes a mapping will need to be performed.
Table 15: 9955 Standard clutter classes
2 dense urban 5-15 stories, areas within urban perimeters, inner city, very little vegetation,
high density of buildings, most buildings are standing close together, narrow
pedestrian zones and streets
3 mean urban 3-8 stories, areas within urban perimeters, major streets are visible, most
buildings are standing close together, little vegetation, buildings, narrow
pedestrian zones and streets included
4 suburban 2 -4 stories, most buildings are standing close together, little vegetation
5 residential 1-2 stories, middle density of buildings with gardens. Both inside or outside
the city center. e.g. detached houses, villas
6 village 1-2 stories, low density of buildings with gardens outside of a city.
7 rural Wide open land with small isolated buildings, incl. agriculture and low tree
density
11 parks City parks, mixture of trees and low tree density, municipal parks, plantations,
golf courses
12 open area e.g. desert, beach, runways of an airport, big streets etc. huge parking areas
This section deals with LPWA parameters categories of: Mobility Types, User Profiles and
Environment.
The mobility type is characterizing the radio environment where the signal is transmitted. It
should be project specific.
The following mobility definitions are coming with the default LPWA template:
Table 16: 9955 Default LPWA mobility types
These are:
For user profiles, no defaults can be provided for any project, they should be customized based
on the provided end device traffic behaviour information indicated by the customer.
Environment type definitions are linking together ‘User Profiles’ with ‘Mobility types’ (the
environment those users are acting in) by indicating density of users (#of users/square km) with
possible clutter weighting of distribution.
In order to make reliable simulations with an accurate propagation model, the propagation
modelling is of capital importance.
Many propagation models are available for path loss calculation. However, each propagation
model is valid for a given range of parameters. Be careful in the results’ interpretation when
your project parameters are out of this range.
Propagation model selection is not a trivial issue: it should be done according to some
fixed in a contract design targets (there might be some prediction accuracy/network
performance requirements of the customer which can be fulfilled only by using some specific,
adapted to the modelled scenario, propagation model) and be in line with the available and
already imported geodatabases. The table below is summarizing different possible requirements
linked to different pathloss prediction methods and as well is indicating Nokia recommended
type(s) of propagation model(s) to use in those circumstances.
In 9955 new propagation models can be added by the user through the add-ins as module
concept.
The following IRT models are supported by 9955 and recommended by Nokia:
As a rule of thumb:
For small scale LPWA networks Nokia recommends purchasing building maps and using an
IRT model (either Volcano, Crosswave or WinProp). If the prediction speed becomes
prohibitive or building database is not available one can use instead IRT an empirical SPM
model, specifically tuned for macro or small cells, see below.
Building database and IRT model usage is recommended as well in cases of large or
medium size networks in areas where antenna radiations are obstructed by some
surrounding buildings (antenna locations are below the rooftop level of some of the
surrounding buildings) and/or the environment is strongly multipath.
Due to difficulties to have 3D database, ARFD team has created some pseudo outdoor
propagation model based on SPM formula for other technologies like LTE, UMTS or GSM. These
are recommended to be used for outdoor LPWA analysis purpose as this scenario is quite similar
to those of outdoor macro cells one.
In case no tuned model is available, and there is no time to make CW measurements and to
tune one, some default SPM models are available in default 9955 templates which can be
created with 9955V731. These are described in [11] in part 8 of this document.
These models cannot be considered as a tuned ray tracing ones, and presented to a customer
but they can be used with a reasonable reliability on a geographical raster database, and doesn’t
required a 3D database.
These default models are detailed in xls file referenced as [5] in part 8 of this document.
The SPM default models are proposed for 2D and 2.5D geographical databases
These models have been tuned on CW measurements all around the world.
They are not as accurate as a specific tuned model on a given area but ensure a limited risk
I<ERR>I <2dB & standard deviation <8dB
If geographical databases are used with clutter heights, then not these but SPM with
clutter heights models can be used.
Contact ARFD design team (owner of this document) to support and define the most
appropriate propagation model!!
Coverage analysis fundamentally remains the most critical step in the design of any network.
Appropriate selection of propagation models and the fade margins for the area coverage
probability are key factors to eliminate cell design that leads to interference or quality
degradation in the network.
A coverage prediction provided by the RNP tool provide a statistical approach for the analysis,
its main purpose is to get a best guess of what one would measure in some given location if
the analyzed network would be operational. Alternatively, by accounting for a safety margin
(see ‘Cell Edge Coverage probability’ below) one would ensure that taken measurement would
exceed with a certain probability (higher than 50%) predicted (more conservative) levels.
One can see that after the project setup steps (presented in section 6.1) one can proceed with
some predictions which can be:
9955 approach of signal level predictions: at each pixel of the network coverage area, a non-
interfering probe mobile is placed (the probe mobile does not cause any interference in the
system). The pathloss is computed for each probe mobile location; this is the computed path
loss, plus a shadowing margin. The shadowing margin is calculated by the tool based on the cell
edge coverage probability level set by the user and on the model standard deviation associated
to the mobile clutter class. These parameters are commented below:
• Model standard deviation (dB) at the receiver position: The standard deviation
characterizes the propagation environment, and can be set independently for each
clutter type in 9955 V7.3.1.
• Cell edge coverage probability level (%): In 9955 V7.3.1 a cell edge coverage
probability level in % has to be set for each tool prediction. The tool calculates the
shadowing margin to be applied for each clutter type based on the given reliability level
and on the shadowing standard deviation.
The coverage probability level can be expressed as the probability that the received signal
reaches at least the required minimum power level (dBm); or that the predicted path loss is able
to overcome the shadowing.
Important: in case of LKB derived ‘RSSI design level’ is used as design parameter in a prediction
where the scope is cell edge to be predicted, coverage area to be derived, no shadow margin
option has to be used. This is because ‘RSSI design level’ is already computed in the LKB for 95%
of reliability (shadow margin already accounted).
The following facts might impact strongly requested coverage types to be carried out:
there is a contractually agreed coverage KPI (ex. RSSI level of -100 dBm is provided in
98% of the network area with 95% of reliability level)
continuous signal level coverage is expected while throughput KPIs are defined for cell
edge performances
indoor coverage type is targeted, indoor performance KPI was defined
all required project parameters are set as mentioned in the previous sections and target
of the planning is clearly defined
The planning process for this coverage analysis step may consist of:
Issue coverage plot based on Best Server (‘Coverage by Transmitter’) prediction for the
worst limiting scenario (Insert the parameters either from UL or DL link budget for this),
taken with no margin
try to fix a right coverage footprint of the cells by solving suspected overshooting
problems (if any) or not clear dominance area of the cells
Reiterate process for some modified site configuration/network layout
generate coverage statistics and try to provide uniform coverage; pay particular
attention to sectors with too large or too reduced coverage areas
Reiterate process for some modified site configuration/network layout
Identify coverage holes and optimize coverage
Reiterate process for some modified site configuration/network layout
Generate Signal Strength plots (‘Coverage by Signal Level’) in DL for all the major cases
defined by the design targets/design requirements. Ensure that coverage area, i.e. places
where signal level is stronger than RNP RSSI design threshold provided by LKB, is in
planning target.
Assess the Signal Level (RSSI) coverage performance by generating statistics
Change site config/network layout and reiterate process if RSSI level in high priority
areas is not sufficient
Issue ‘Overlapping Zones’ study (‘Overlapping Zones’ type of prediction) for the above
case, taken with some same margin (ex. 4 dB)
configure, at the display options, a higher number of ‘Value Intervals’ for the going to be
displayed ‘Number of Servers’ than available number of channels (ex. fix 4 intervals in
case of default channel number of 3); generate coverage statistics and check the ‘size’
(indicated percentage of focus zone) of areas where the number of covering servers is
higher than available number of channels; those areas, with no regards on the frequency
plan (which can still be missing) will be with high risk of interference as there will exist all
the time at least two servers working on the same channel/carrier;
perform additional local analyses by using the ‘Point Analysis’ feature, in order to check
what are the predicted signal levels coming from the overlapping servers in some critical
places; if result is not satisfactory try to conclude on the main source for this high risk of
interference (ex. check which covering servers are working on the same carrier) and
reiterate process for some modified site configuration (antenna azimuth & tilt, if the
case, reduce Tx Power) by trying to implement one of the two following solutions:
o create cell dominance of a chosen cell in that area (if possible)
o remove from the area one or more of the cells contributing to this high risk
Issue DL/UL raster CINR prediction plot, (‘Coverage by C/(I+N) Level (DL)’, ‘Coverage by
C/(I+N) Level (UL)’), with no margin, by using a representative Terminal type and selected
mobility type.
generate statistics and quantify results; compare against possible existing defined
quality KPI
Perform CINR predictions for a different channel plan if result is not satisfactory
Identify problematic interference areas and iterate back to the first step of the
planning process if necessary after site/cell configuration has been changed accordingly.
This next (and optional) step concerns issuing service coverage plots, which correspond to the
already optimized network signal level and interference conditions.
This can be performed as well based on assumed fixed UL/DL traffic loads of the LPWA cells.
Issue DL/UL raster Service bearer coverage ( ‘Service Area Analysis (DL)’, ‘Service Area
Analysis (UL)’) prediction plots, with no margins, by using the same representative
Terminal and mobility types as for CINR predictions.
Generate statistics (Reports), histograms on the obtained results; compare against
possible defined KPIs
Identify problematic Service Coverage areas and iterate back to the first step of the
planning process if necessary after site/cell configuration has been changed accordingly.
This step has its target of assessing obtained throughput performances. This means that
usually, available capacity provided in the optimized coverage area is checked, unless capacity
related KPIs were defined while network need to be optimized in order to achieve those.
9955 provides the following types of throughput coverage predictions:
options, which are available for either a ‘Coverage by Throughput (DL)’ or a ‘Coverage by
Throughput (UL)’.
Alternatively, options of:
can be selected.
Issue DL raster ‘Coverage by Throughput (DL)’ prediction plot, by using the same
representative Terminal and mobility types as for CINR & Service coverage predictions.
Issue UL raster ‘Coverage by Throughput (UL)’ prediction plot, by using the same
representative Terminal and mobility types as for CINR & Service coverage predictions.
Generate statistics (Reports), histograms on the obtained results.
Although 9955 LPWA does not offer Monte Carlo simulation functionality it supports user
performance analysis based on a subscriber list.
Subscribers, i.e. locations on end devices, can be distributed anywhere in the coverage area.
Additionally, one can inspect and check performances at individual subscriber locations:
To have the 9955 configuration for some LPWA workaround, specific atl files should be used in
which appropriate settings have been done.
To have the 9955 configuration for LoRa workaround, the atl files of ‘LoRa_templateV01.ATL’
should be used, see [3]. This contains predefined settings to model LoRa by using the 9955 WiFi
technology module.
Frequency bands
According to Figure 17 a LoRa frequency band is declared in the European ISM band of 868 MHz.
A channel bandwidth of 125 kHz is specified.
Reception Equipments
Reception equipment definitions are with considering estimated (see Table 5 and Table 6) CINR
(C/I+N) values.
Radio Bearers
Services
Terminals
This workaround allows the same as section 6.2 presented coverage analyses to be carried out.
Although, some definitions will account of the workaround qualities presented in the previous
section 7.1.1.
Typical predictions:
• Overlapping Zones
• Coverage by Throughput
A ‘Coverage by Best Server’ prediction allows visualization of cell footprints, i.e. areas where a
given cell has the quality of being the best serving. It helps managing cell sizes and removing
overshooting problems.
The criterion to be applied for server overlaps is not technology specific, and its efficiency has
already been measured on other technologies like GSM, CDMA, W-CDMA and LTE networks.
Subscribers, i.e. locations on end devices, can be distributed anywhere in the coverage area.
[2] Semtech, ‘SX1272/73 - 860 MHz to 1020 MHz Low Power Long Range Transceiver’,
[Online]. Available: http://www.semtech.com/images/datasheet/sx1272.pdf
[3] IoT 9955 template, ‘LoRa_templateV01.ATL’, atl file for LoRa workaround
[4] LoRa NOKIA Link Budget tool & presentation
[5] 9955 SPM Default propagation models recommended for 400-5500 MHz ed9.xlsx
Sharenet Livelink: https://sharenet-
ims.int.net.nokia.com/livelink/livelink?func=ll&objId=546613535&objAction=browse&viewType=1
[6] 3DF 01955 6080 BGZZA, ‘A9155 Application Note : Coordinate Systems and Geographic
Projections’,
https://mdms-ll.app.alcatel-
lucent.com/livelink/livelink.exe?func=ll&objId=64155525&objAction=browse&viewType=1
https://sharenet-
ims.int.net.nokia.com/livelink/livelink?func=ll&objId=549706310&objAction=browse&vie
wType=1
https://sharenet-
ims.int.net.nokia.com/livelink/livelink?func=ll&objId=549706310&objAction=browse&vie
wType=1
[9] 3GPP TR 45.820, “Cellular system support for ultra-low complexity and low throughput
Internet of Things (CIoT)”, release 13, 2015-11. [Online]. Available:
http://www.3gpp.org/ftp/Specs/archive/45_series/45.820/45820-d10.zip
[10] Qualcomm, Incorporated, “Narrowband IoT (NB-IoT),” RP-151621, 3GPP TSG RAN
Meeting #69, Sept. 2015. [Online]. Available:
http://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_69/Docs/RP-151621.zip
[11] 9955 default propagation models recommended for 400 - 5500 MHz frequency bands
https://wcdma-ll.app.alcatel-
lucent.com/livelink/livelink.exe?func=ll&objId=68078685&objAction=browse&viewType=1