Professional Documents
Culture Documents
009 Interference Management in UMTS Femtocells
009 Interference Management in UMTS Femtocells
www.femtoforum.org
www.femtoforum.org
back to contents
www.femtoforum.org
telephone +44 (0)845 644 5823 fax +44 (0)845 644 5824 email info@femtoforum.org PO Box 23 GL11 5WA UK
Contents
1 2 2.1 2.2 2.3 3 3.1 4 5 6 7 7.1 7.2 7.3 7.4 8 Executive Summary ............................................................................................................................. 6 Femtocells, Femtocell Access Points and the Femto Forum ............................................................... 8 What are Femtocell Access Points?................................................................................................. 8 What Do Femtocells Offer? ............................................................................................................. 9 What is the Femto Forum? ............................................................................................................ 10 Introduction ....................................................................................................................................... 11 Objectives and Methods of this Paper .......................................................................................... 11 Previous Work ................................................................................................................................... 13 Simulation Scenarios and Definitions ................................................................................................ 15 Abbreviations and Defined Terms ..................................................................................................... 19 Scenario A: Macrocell Downlink Interference to the Femtocell UE Receiver 20 Description..................................................................................................................................... 20 Analysis .......................................................................................................................................... 21 Extended scenario: HSDPA coverage............................................................................................. 25 Conclusions .................................................................................................................................... 26 Scenario B: Macrocell UE Uplink Interference to the Femtocell Receiver ........................................ 27
8.1 Description..................................................................................................................................... 27 8.2 Analysis .......................................................................................................................................... 27 8.2.1 HSUPA .................................................................................................................................... 31 8.3 Conclusions .................................................................................................................................... 33 8.3.1 Customer (MUE) impact ........................................................................................................ 34 8.3.2 8.3.3 9 9.1 9.2 9.3 10 Customer (FUE) Impact.......................................................................................................... 34 Mitigation techniques............................................................................................................ 34
Scenario C: Femtocell Downlink Interference to the Macrocell UE Receiver ................................... 35 Description..................................................................................................................................... 35 Analysis .......................................................................................................................................... 37 Scenario analysis and conclusions ................................................................................................. 39 Scenario D: Femtocell Uplink Interference to the Macrocell NodeB Receiver ................................. 40 Introduction ............................................................................................................................... 40 Analysis of Scenario D - 12k2 Voice and HSUPA ........................................................................ 41 Assumptions ...................................................................................................................... 41 Macro Node B Noise Rise .................................................................................................. 43
11
10.3 Conclusions ................................................................................................................................ 46 10.4 Recommendations..................................................................................................................... 46 Scenario E: Femtocell Downlink Interference to nearby Femtocell UE Receiver.............................. 47 11.1 11.2 11.3 Description ............................................................................................................................... 47 Capacity Analysis.................................................................................................................... 48 Conclusions ............................................................................................................................. 51
page
13
12.3 Conclusions ................................................................................................................................ 56 12.4 Recommendations..................................................................................................................... 56 Scenario G: Macrocell Downlink Interference to an adjacent channel Femtocell UE Receiver ........ 57 13.1 13.2 13.2.1 13.2.2 13.2.3 Description................................................................................................................................. 57 Analysis ...................................................................................................................................... 57 Assumptions ...................................................................................................................... 58 Simulation Analysis ............................................................................................................ 58 Theoretical Analysis ........................................................................................................... 59
14
13.3 Conclusions ................................................................................................................................ 59 Scenario H: Macrocell UE Uplink Interference to the adjacent channel Femtocell Receiver ........... 60 14.1 14.2 14.2.1 14.2.2 14.2.3 Description................................................................................................................................. 60 Analysis ...................................................................................................................................... 61 Parameter settings ............................................................................................................ 61 Impact of MUE interference on AMR ................................................................................ 62 Impact of MUE interference on HSUPA............................................................................. 65
15
14.3 Conclusions ................................................................................................................................ 68 14.4 Femto System Impact ................................................................................................................ 68 14.5 Mitigation techniques................................................................................................................ 68 Scenario I: Femtocell Downlink Interference to the adjacent channel macrocell UE Receiver ........ 69 15.1 15.2 15.2.1 15.2.2 15.2.3 Description................................................................................................................................. 69 Analysis ...................................................................................................................................... 70 Parameter settings ............................................................................................................ 70 Impact of Femtocell interference on AMR service ............................................................ 72 Impact of Femtocell interference on HSDPA ..................................................................... 74
16
15.3 Conclusions ................................................................................................................................ 75 15.4 Customer (MUE) Impact ............................................................................................................ 76 15.5 Mitigation techniques................................................................................................................ 76 Scenario J: Femtocell UE Uplink Interference to the adjacent channel Macrocell NodeB Receiver. 76 16.1 16.2 16.2.1 16.2.2 16.3 Introduction ............................................................................................................................... 76 Analysis of Scenario J - 12k2 Voice and HSUPA ......................................................................... 77 Assumptions ...................................................................................................................... 77 Macro Node B Noise Rise .................................................................................................. 80 Conclusions ................................................................................................................................ 81 page 4
Summary of Findings ......................................................................................................................... 94 Overall Conclusions ......................................................................................................................... 105 Further Reading ............................................................................................................................... 106 20.1 Scenario A ................................................................................................................................ 106 20.2 Scenario B ................................................................................................................................ 106 20.3 Scenario C ................................................................................................................................ 106 20.4 Scenario D ................................................................................................................................ 107 20.5 Scenario E ................................................................................................................................ 107 20.6 Scenario F ................................................................................................................................ 107 20.7 Scenario G ................................................................................................................................ 108 20.8 Scenario H ................................................................................................................................ 108 20.9 Scenario I ................................................................................................................................. 108 20.10 Scenario J ................................................................................................................................. 109 20.11 Scenarios Section 17 ............................................................................................................. 109 References ....................................................................................................................................... 110 Simulation Parameters and Path Loss Models ................................................................................ 115 22.1 22.2 22.2.1 22.2.2 22.2.3 Simulation parameters ............................................................................................................ 115 Path Loss Models ..................................................................................................................... 116 Okumura-Hata ................................................................................................................. 116 ITU-R P.1238 .................................................................................................................... 117 System Simulation (Section 17) Path Loss Models .......................................................... 118
21 22
23
page
Femtocells, by virtue of their simultaneous small size, low cost and high performance, are a potentially industry-changing disruptive shift in technology for radio access in cellular networks. Their small size means that the spectrum efficiency they can attain is much greater than that achievable using macrocells alone. Their low cost means they can be deployed as consumer equipment reducing the capital load and operating expenses of the host network. And their high performance means that all this can be gained at no loss of service to the customer, and in many cases, owing to the improved link budgets, improved service. However, for these apparent benefits to translate into real advantage for network operator and consumer alike, we must answer serious questions about the interaction between the femtocell technology and the host macrocellular radio network into which they are deployed. If femtocells can only achieve their potential by disrupting the macro network, then they will be relegated to niche deployments, of little overall relevance to next-generation networks. On the other hand, if the interactions between macro and femto radio layers can be managed to the benefit of all, then their properties (in terms of lowered cost, improved spectrum efficiency and link budget and general performance) can be fully realised, and femtocells will find themselves an essential component of all future radio access network designs. So, what are these interactions? How can they be managed? What does that all mean for the technology, to the operator and to the consumer? These are the questions that this paper is helping to answer. In doing so, it has deliberately maintained a tight focus, according to the priorities of its authors. It is exclusively concerned with W-CDMA as an air interface technology (other teams within Femto Forum are looking at other air interfaces). This paper is concerned primarily with the 850 MHz band in the United States, but is equally applicable to the 900 MHz band in Europe and elsewhere. It should also be broadly applicable to similar bands (eg. 700 MHz). An earlier study has been published with similar results for 2 GHz. It is exclusively a theoretical treatment, using link level and system level simulations to draw its conclusions, although we expect to back these conclusions up in due course with trial campaign data. In view of the residential application that femtocells are addressing, this paper is also concerned with femtocells operating with closed user groups. Perhaps most importantly, this paper stands on the shoulders of giants, drawing on the great mass of study work that has already been undertaken by 3GPP RAN4 participants in analysing these issues, and referencing them for further reading. The interacting components of the femto-enabled network include femtocells themselves, which can be interacting in their downlinks with other nearby femtocells and macrocells; macrocells, which interact with nearby femtocells; and users and user equipment (UEs), which, by virtue of intentional radio links to femtocells and macrocells, may be causing unintentional interactions with both.
Executive Summary
page
page
Femtocell Access Points (FAPs) are low-power radio access points, providing wireless voice and broadband services to customers primarily in the home environment. The FAP provides cellular access in the home and connects this to the operators network through the customers own broadband connection to the Internet. FAPs usually have an output power less than 0.1 Watt, similar to other wireless home network equipment, and they allow a small number (typically less than 10) of simultaneous calls and data sessions at any time. By making the access points small and low-power, they can be deployed far more densely than macrocells (for instance, one per household). The high density of deployment means that the femtocell spectrum is re-used over and over again, far more often than the re-use that the macro network (with its comparatively large cells) can achieve. Trying to reach the same levels of re-use with macrocellular technology would be prohibitively expensive in equipment and site acquisition costs. By using femtocells, the re-use, spectrum efficiency, and therefore the aggregate capacity of the network can be greatly increased at a fraction of the macrocellular cost. A typical deployment scenario is shown in Figure 2-1.
page
Zero-touch installation by end user: Femtocells are installed by the end user without intervention from the operator. The devices will automatically configure themselves to the network, typically using Network Listen capabilities to select settings that minimise interference with the macro network. Moveability: The end user may move their femtocells for example, to another room, or, subject to operator consent, to another location entirely. Backhaul via the end users fixed broadband connection: Femtocells will use the subscribers broadband connection for backhaul, which typically will be shared with other devices in the home. Access control the Closed User Group: The operator and/or end user will be able to control which mobile devices can access the femtocell. For example, subscribers may be able to add guest phone numbers via a web page. Supports a restricted number of simultaneous users: Femtocells will support a limited number (typically, fewerthan ten) of simultaneous calls and data sessions. Femtozone (homezone) tariffs: Mobile services accessed through the femtocell may be offered at a cheaper rate than the same services on the macro network. End users are advised when services are accessed via the femtocell, either by an advisory tone, or a display icon or some other means, so they know when the femto-tariffs apply. Ownership: Various ownership models are possible for example, end users may own their femtocells, just as they own their mobile phones, or the operator may retain ownership, with end users renting the equipment (like a cable modem). Small cell size/millions of cells in the network: The femtocell network can easily extend to millions of devices. Femto as a service platform: Novel mobile services can be made available on the femtocell. For example, a femtocell-aware application on the mobile handset could automatically upload photos to a website when the user enters the home, and download podcasts.
page
The Femto Forum is the only organisation devoted to promoting femtocell technology worldwide. It is a notfor-profit membership organisation, with membership open to providers of femtocell technology and to operators with spectrum licences for providing mobile services. The Forum is international, representing around 100 members from three continents and all parts of the femtocell industry, including: major operators, major infrastructure vendors, specialist femtocell vendors, and vendors of components, subsystems, silicon and software necessary to create femtocells.
The Femto Forum has three main aims: to promote adoption of femtocells by making information available to the industry and the general public, to promote the rapid creation of appropriate open standards and interoperability for femtocells, and to encourage the development of an active ecosystem of femtocell providers, to deliver ongoing innovation of commercially and technically efficient solutions.
The Femto Forum is technology agnostic and independent. It is not a standards-setting body, but works with standards organisations and regulators worldwide to provide an aggregated view of the femtocell market. A full current list of Femto Forum members and further information is available at www.femtoforum.org.
page 10
The benefits of femtocells are not straightforward to realise. While network operators will see significant capacity gains, and end users can expect higher performance, to achieve this the radio layer must be carefully managed. The management of the radio interference between the Macro and Femto Layers is a key industry concern addressed by this paper. Interference adversely affects the capacity of a radio system and the quality of the individual communication links on that system. Adding capacity is always based on a trade-off between interference, quality and capacity. Hence, there is a need for interference management techniques to minimise interference that might otherwise counteract the capacity gains and degrade the quality of the network. 1. The principal objectives of this study are: a) To develop an industry position on the interference risks from femtocell deployments. b) To recommend mitigation techniques and any necessary associated RF parameters and performance requirements, to ensure minimal disruption to the macro network or other femtocells. 2. To achieve these objectives, this paper develops detailed interference scenarios for evaluation and inclusion in the interference management assessment. The scenarios will cover worst-case deployment conditions and assess the respective system impact. 3. An immediate focus is to develop the assessment for W-CDMA, and in doing so devise a process that should be consistent with alternative radio technologies. 4. Two main steps were identified in order to accomplish the above goal: a) First, a baseline set of interference analysis conclusions for UMTS femtocells, based on 3GPP RAN4 interference studies, was required. This would be supplemented with specific analysis of identified micro scenarios, their likelihood, and potential impact. Interference mitigation techniques should also be considered on the understanding that vendor independence be preserved wherever possible. b) Secondly, a recommendation for a common set of behaviours (RF parameters and/or test cases) that can be derived by any UMTS femtocell was required. This is so that the femtocell can configure itself for minimal disruption to either the macrocell layer or other deployed femtocells.
Introduction
page 11
page 12
Analysis in this problem space has already been carried out as part of the 3GPP Home Node B study item. 3GPP RAN4 concluded their study into the radio interface feasibility of Home Node B (aka femtocells) at RAN#39 in March 2008. Their results are presented in [TR25.820]. Part of their study included the analysis of anticipated interference scenarios covering a range of HNB deployments. A summary of their findings is presented in Table 4-1 below. The scenarios for this paper are defined in Section 5. Table 4-1: Scenario (this paper) A B 25.820 scenario id 4 3 Summary of RAN4 conclusions
Previous Work
Macrocell DL interference can generally be overcome, as long as the femtocell has sufficient transmit dynamic range. The femtocell receiver must reach a compromise between protecting itself against uncoordinated interference from the macro UEs, and controlling the interference caused by its own UEs towards the Macro Layer. Adaptive uplink attenuation can improve performance, but consideration must also be given to other system issues like the associated reduction in UE battery life. Downlink interference from a closed-access femtocell will result in coverage holes in the macro network. In co-channel deployments the coverage holes are considerably more significant than when the femtocell is deployed on a dedicated carrier. A number of models are presented for controlling maximum femtocell transmission power, but it is acknowledged that no single mechanism alone provides a definitive solution. Open access deployment should also be considered as a mitigating option. Noise rise on the Macro Layer will significantly reduce macro performance; consequently, the transmit power of the femto UE should be controlled. A number of mechanisms to achieve this are presented, generally providing a compromise between macro and femtocell performance. Again, open access deployment should be seen as a mitigating option in the co-channel case. This scenario has received less coverage than the macro interference cases, but it is noted that the performance of Closed Subscriber Group (CSG) femtocells is significantly degraded unless interference mitigation techniques are used. This is generally a similar problem to macro DL interference in the co-channel scenario. It is difficult to avoid co-channel interference between CSG femtocells, and this limits the interference reductions achieved by deploying the femtocell on a separate carrier from the macro network. Again, interference management techniques are required to manage femto-to-femto interference. Macrocell DL interference can generally be overcome, as long as the femtocell has sufficient transmit dynamic range. page 13
25.820 scenario id 3
The femtocell receiver must reach a compromise between protecting itself against uncoordinated interference from the macro UEs, and controlling the interference caused by its own UEs towards the Macro Layer. This is generally an easier compromise to arrive at with adjacent-channel deployments than it is with co-channel. Downlink interference from a closed-access femtocell will result in coverage holes in the macro network. In adjacent-channel deployments the coverage holes are considerably easier to minimise and control than when the femtocell is deployed on the same carrier as the Macro Layer. A number of models are presented for controlling maximum femtocell transmission power; all except the fixed maximum power approach are generally acceptable. Noise rise on the Macro Layer will significantly reduce macro performance; consequently, the transmit power of the Femto UE should be controlled. A number of mechanisms to achieve this are presented, generally providing a compromise between macro and femtocell performance. Adjacent-channel deployments can generally be accommodated.
In addition to the previous 3GPP analysis work, the Femto Forum conducted an earlier study covering the same scenarios at 2 GHz [FF08]. For this study at 850 MHz, several changes were made to the simulation parameters used in that earlier 2 GHz study: Wall loss was reduced from 20 to 10dB, to reflect greater building penetration at 850 MHz. Macro basestation antenna height was increased from 25 to 30 metres, to reflect the higher antenna heights (larger cell size) typical in North American deployments. The minimum distance from a macro basestation was increased from 30 to 1,000 meters, to again reflect typical North American deployment scenarios where cells are larger and basestations are not typically located in residential areas. This also allowed us to eliminate the use of the ITU P.1411 propagation model, and to use the Okumura-Hata model, simplifying the analysis work.
page 14
The Femto Forum has identified 10 stretch scenarios that explore the limits of operation of femtocells and femtocell subscriber equipment.
The scenarios are summarised in the following tables and figure. Table 5-1: Femtocell Deployments in Shared Spectrum Scenario Description
Macrocell Downlink Interference to A femtocell UE receiver, located on a table next to the apartment window, is in the direct bore sight of a macrocell (1 km distance). the Femtocell UE Receiver (A) The macrocell becomes fully loaded, while a femtocell UE is connected to the femtocell at the edge of its range. Macrocell Uplink Interference to the A femtocell is located on a table within the apartment. Weak Femtocell Receiver (B) coverage of the macro network is obtained throughout the apartment. A user UE1 (that does not have access to the femtocell) is located next to the femtocell and has a call established at full power from the UE1 device. Another device UE2 has an ongoing call at the edge of femtocell coverage. Femtocell Downlink Interference to UE1 is connected to the macro network at the edge of macro the Macrocell UE Receiver (C) coverage. It is also located in the same room as a femtocell (to which it is not allowed to access). The femtocell is fully loaded in the downlink. Femtocell Uplink Interference to the UE1 is located next to the apartment window, in direct bore sight of a macrocell (1 km distance). UE1 is connected to the femtocell Macrocell Node B Receiver (D) at the edge of its range, and is transmitting at full power. Femtocell Downlink Interference to Two apartments are adjacent to each other. Femtocells (AP1 and AP2) are located one within each apartment. The owner of AP2 Nearby Femtocell UE Receivers (E) visits their neighbours apartment, and is on the edge of coverage of their own femtocell (AP2) but very close (<3m) to AP1. The owner of AP1 establishes a call requiring full power from the femtocell. Femtocell Uplink Interference Nearby Femtocell Receivers (F) to Two apartments are adjacent to each other. Femtocells (AP1 and AP2) are located one within each apartment. The owner of AP2 visits their neighbours apartment, and is on the edge of coverage of their own femtocell. The owner of AP2 establishes a call that requires peak UE power to their own femtocell while they are located next to AP1 (< 3m).
page 15
Macrocell Downlink Interference to A femtocell UE is located on a table next to the apartment window, the adjacent-channel Femtocell UE in direct bore sight of a macrocell (1 km distance). The macrocell becomes fully loaded, while a femtocell UE is connected to the Receiver (G) femtocell at the edge of its range. Macrocell Uplink Interference to the A femtocell is located on a table within the apartment. Weak adjacent-channel Femtocell Receiver coverage of the macro network is obtained throughout the (H) apartment. A user (that does not have access to the femtocell) is located next to the femtocell and has a call established at full power from the UE1 device. Another device UE2 has an ongoing call at the edge of femtocell coverage. Femtocell Downlink Interference to Two users (UE1 and UE2) are within an apartment. UE1 is the adjacent-channel Macrocell UE connected to a femtocell at the edge of coverage. UE2 is Receiver (I) connected to the macrocell at the edge of coverage, and located next to the femtocell transmitting at full power. Femtocell Uplink Interference to the A femtocell is located in an apartment, in direct bore sight of a adjacent-channel Macrocell NodeB macrocell (1 km distance). UE1 is connected to the femtocell at the edge of coverage, but next to the widow thus, in the direct bore Receiver (J) sight of the macrocell antenna. In addition to these extreme scenarios, we include shared-spectrum system level simulations specifically modelling the mitigation of downlink interference and uplink noise rise by power control techniques (Section 16). These simulations also model the effect of femtocells on the total throughput and capacity of the network.
page 16
Victim Femto UE DL Rx Macro NodeB DL Tx Macro UE Aggressor UL Tx Femto AP DL Tx Femto UE UL Tx Neighbour Femto UE UL Tx F 5 A, G 4 B, H 3 C, I 2 D, J 1 E 6 Femto AP UL Rx Macro UE DL Rx Macro NodeB UL Rx Neighbour Femto UE DL Rx
AF are the interference scenarios for co-channel deployments GJ are the interference scenarios for adjacent-channel deployments 16 are the equivalent interference scenario IDs used in the 3GPP HNB analyses [TR25.820]
page 17
A,G
F FUE
D,J
Macro
FUE
NodeB
F B,H
MUE
E
F
UE Association Interference
F
C,I
Apartments
FUE MUE
page 18
parameters. The most frequently used are presented here for quick reference. However, a more extensive list has been produced and is available under separate cover. AP BER BS EIRP FAP FUE HUE HNB MNB MUE QoS UE RAN RAT RSCP RTWP LOS P-CPICH Victim Access Point Bit Error Rate (or Bit Error Ratio) the proportion of the total number of bits received that are decoded wrongly Base Station (assumed to be a wide-area BS, as defined in [TS25.104], unless otherwise stated) Equivalent Isotropic Radiated Power a measure of the transmitted power in a particular direction that takes account of the antenna gain in that direction Femto AP, also known as the femtocell Femto UE, also called the Home UE (HUE) Home UE, also called the femto UE (FUE) Home NodeB Macro NodeB Macro UE Quality of Service User Equipment (handset, data terminal or other device) Radio Access Network Radio Access Technology Received Signal Code Power Received Total Wideband Power Line-Of-Sight Primary Common Pilot Channel Is a radio node (macro node-B, or femto access point) whose receiver performance is compromised by interference from one or more other radio nodes (the Aggressor). Alternatively, the Victim may be a radio link, whose quality is degraded by unwanted interference from Aggressor nodes Is a radio node (either macro node-B, femto access point or UE) whose transmissions are compromising the performance of another radio node (the Victim), or which are contributing to the degradation of quality of a (Victim) radio link
Throughout this paper a number of abbreviations are used to identify various system elements and
Aggressor
Deadzone Is an area where the quality of service is so poor as a result of interference that it is not possible to provide the demanded service. Deadzones are also characterised by the fact that in the absence of any interference, a normal service would be possible. Deadzones are often specified in terms of the path loss to the Aggressor transmitter. A 60dB deadzone in the femtocell is, therefore, a region around the femtocell where the path loss to the FAP is less than 60dB.
page 19
A UE is located on a table next to the apartment window that is 1 km distance away from a macrocell. The macrocell is operating at 50% load, while the UE is connected to the femtocell (ie. FUE) at the edge of its range. In this scenario the Victim link is the downlink from the femtocell to the FUE, while the Aggressor transmitter is the downlink from the macrocell. This interpretation of Scenario A is summarised in Figure 7-1.
7.1 Description
page 20
The objective of the analysis of this scenario is to work out the services that can be delivered to a femto UE when it is on the edge of the femtocell the femtocell itself being positioned, as required by the scenario, 1km from the macro. The analysis strategy for this scenario is broken down as follows: The first task is to determine the range of the femtocell as defined by the pilot power. This gives us the maximum range at which the UE can detect and decode the femto beacon, and therefore camp on to it. Secondly, we work out the services that can be offered by the femtocell at the edge of its coverage, given that interference level. The first step is accomplished by the following sequence: Assume a given P-CPICH transmit power for both macro and femto; then find the power due to the macro at the distance given by the scenario (1km); then find the distance from the femto at which the ratio of femto power to macro power is sufficient for the UE to detect the femtocell. This distance is the range of the femtocell as defined by the pilot power the maximum range at which a UE can detect the femtocell and camp on to it.
7.2 Analysis
The second step (to work out the services that can be offered at this range) is accomplished as follows: For voice, work out how much dedicated channel power is required to sustain a voice call, given the interference level calculated in the first step, and reconcile that with the total amount of power available to give the number of voice calls that may be sustained. For data, work out the Ec/Io that can be achieved by allocating all the remaining power to the HSDPA downlink shared channel, and derive a throughput from that, given an industry standard relationship between Ec/Io and throughput. Assumptions for the macrocell are as defined in [FF09] with variant values shown in Table 7-1, which shows the transmit EIRP of the macrocell. The link budget for the macrocell is defined in Table 7-2.
page 21
54 dBm EIRP_m=Ptx_m+Gm-Lc
Table 7-2: Link budget for the received power from macro Node B to UE. Value Distance macro nodeB to UE Height macro nodeB antenna Height UE from ground Path loss UE antenna gain UE connector and body losses Macro nodeB received power at UE Unit Comments d_mu hb hM PL_m is calculated from the Okumura-Hata Model, + 5dB window loss Gue Lc_u
dBm Prx_m=eirp_m-PL_m+Gue-Lc_u
The value Prx_m in Table 7-2 is the power due to the macrocell at the scenario distance (1 km), and takes account of the propagation, plus an allowance for the window loss (5dB). The femtocell assumptions are presented in Table 7-3. Note that three types of femtocell are assumed with the defined femto transmit power classes (10dBm, 15dBm and 21dBm).
page 22
In order to complete the calculation of position of the cell edge according to P-CPICH, we calculate the PCPICH power at the UE and compare it to the power at the UE due to the macrocell. Note that in this scenario we are fixing the UE at the window and moving the femtocell location so the macrocell power is constant at the value calculated in Table 7-2. We use the indoor propagation model ITU-R P.1238, assuming a residential building and same floor operation, the femtocell characteristics from Table 7-3 as well as the same UE characteristics as in Table 7-2. Figure 7-2 shows the femtocell P-CPICH power received at the UE, and the power at the UE from the macrocell as taken from Table 7-2. In order for the FUE to detect the femtocell and camp onto it, the P-CPICH Ec/No must be sufficient. It is assumed that a level of -18 dB will be adequate in this respect. To find the range of the femtocell we need to find the distance below which the P-CPICH power is less than 18 dB below the power from the macrocell. By observing in Figure 7-2 where the P-CPICH power exceeds the bounds on the macro interference power minus 18 dB, it can be seen that even at the 10 dBm transmit power, the FAP has a range of more than 100 m. It is to be noted that this does not necessarily mean that a UE 100m away from the FAP will select the FAP in idle mode. Rather, it means that if the UE is already connected to this FAP, it can still sustain the connection at this distance
2010 Femto Forum Limited | www.femtoforum.org
page 23
Figure 7-2: Received signal strengths at UE, from macrocell and femtocell. Further, it can be seen that, based on Table 7-4, voice services are readily achievable at the edge of coverage, since they require about the same Ec/No as the minimum CPICH Ec/No assumed above. Table 7-4: Required Ec/No for voice connection. Value Chiprate Bitrate of AMR voice call Eb/No requirement for voice connection Ec/No requirement for voice connection Unit W R Eb/No Ec/Io=Eb/No-10*log10(W/R) Comments
Similarly for HSDPA, assuming that 80% of the femtocell power is reserved for HSDPA services (9dB above PCPICH), the HSDPA Ec/No will be at least -1.8 dB (@ 100m from HNB), which corresponds to > 1.5 Mbps, according to the translation equation in [R4-080149].
page 24
P = I SN F R H D S S C 1 6 (1 ) Po + w P o n t + hPn
Equation 7-1
e o
where: SF16 is the spreading factor, PHS-DSCH is the received power of the HS-DSCH, summing over all active HS-PDSCH codes, Pown is the received own-cell interference, is the downlink orthogonality factor (assumed to be 1, fully orthogonal), Pother is the received other-cell interference, Pnoise is the received noise power (here it is assumed that the UE Noise figure is 7dB).
Assuming: The femtocell transmit powers are 10dBm, 15 dBm and 21 dBm, with 80% allocated to HS-DSCH And employing the path loss assumptions of the previous section The UE is still assumed to be 1 km away from the macrocell. The HSDPA throughput for the FUE at different distances from the femtocell is shown in Figure 7-3.
page 25
Figure 7-3: HSDPA throughput vs. UE to femtocell distance for various femtocell Tx powers. It can be seen from Figure 7-3 that the maximum HSDPA throughput can be expected up to 25 m away from the femto, even at the 10 dBm transmit power.
The scenario that has been analysed in this section examines the case of the UE being located in front of a window overlooking a macrocell that is 1 km away. Assuming standard models and parameters, it is shown that, even at 10 dBm transmit power, the femtocell is able to comfortably provide voice to the UE when the femtocell is located as far as 100 m away, and maximum HSDPA throughput can be expected up to 25 m away.
7.4 Conclusions
page 26
A femtocell is located on a table within the apartment. Weak coverage of the macro network is obtained throughout the apartment. A user that does not have access to the femtocell (MUE) is located next to the femtocell. Another user device (FUE) is connected to the femtocell and has an ongoing call at the edge of femtocell coverage. The scenario is depicted in Figure 8-1. In this case the Victim receiver belongs to the femtocell access point (FAP), and the Aggressor transmitter is that of the nearby MUE.
8.1 Description
The general assumptions for the analysis of this scenario are presented in Table 8-1. The link budget for the MUE is shown in Table 8-2; note that three separation distances between the MUE and the femtocell are taken into account (5, 10 and 15m).
8.2 Analysis
page 27
Table 8-2: MUE link budget at the femtocell receiver. Value MUE uplink transmitted power UE antenna gain Connectors/body loss MUE Tx EIRP Unit Comments
18 dBm eirp_mue=Ptx_mue+Gue-Lue
Distance MUE-femtocell MUE-femtocell path loss Femtocell antenna gain Femtocell feeders/connector losses Uplink power received by the femtocell from MUE at different MUE-femtocell separation distances
d_mue PL_mue, Indoor to indoor path loss model , where d=d_mue, f=fc Gf Lf
In Table 8-3, the FUE's minimum transmitted power requirement for holding a voice call is calculated. Note that the power is well within the FUE's capabilities, even at the largest separation distance.
page 28
Units d_fue
Comments
PL_fue Indoor to indoor path loss model (d=d_fue, f=fc) Eb/No_fue [TS25.104] PG_fue PN from [TS25.942] Prx_fue is calculated from equation [Hol06]:
Eb/N0 requirements for a voice call Processing Gain Noise power FUE received power in order to obtain required Eb/N0 for different MUE distances (d_mue)
8.3
dB
dB dBm dBm
(Eb / No ) fue
dBm
Ptx_fue=Prx_fue-Gue+Lue+PL_fueGf+Lf
Table 8-3: FUE transmitter power requirements in order to hold a voice call The values calculated in Table 8-3 for the transmitted power of the FUE required are the same as the one calculated for the 1900Mhz study. The reason for this is that the reduction on frequency affects both FUE and MUE in the same way. Moreover, as the MUE is near to the femtocell, the affect of Noise Power is small in the calculation of Prx_fue.
In Figure 8-2, the results are interpolated for different UE distances and power levels. Note that the plot includes the downlink deadzones created by the femtocell, which affects the MUE. Downlink deadzone assumptions are summarised in Table 8-4.
page 29
Within these zones, the MUE will be re-directed to another WCDMA frequency or Radio Access Technology (RAT) by the macrocells, or the call may be dropped. In both case the interference level in the femtocell reduces, and the uplink power requirements will relax.
page 30
In this section the affects of HSUPA are analysed. The link budget is shown in Table 8-5. Table 8-5: Link budget for HSUPA Value FUE uplink transmitted power UE antenna gain Connectors/body loss FUE Tx EIRP Unit 21 dBm Ptx_fue 0 dBi 3 dB Gue Lue Comments
18 dBm eirp_fue=Ptx_fue+Gue-Lue
Distance FUE-femtocell
5 m
50.16 dB
MUE distance from femtocell MUE-femtocell separation MUE power at femtocell (see Table 8-2 for d_mue=10)
(Ec / No ) fue
The simulation results in Figure 8-3 show the E_DPDCH Ec/No for two cases: FUE is at 5m from the femtocell FUE is at 15m from the femtocell. In both cases, it is expected that the MUE is transmitting at maximum power (21dBm).
page 31
Note that DL deadzones are not taken into account. However, the grey area in the figure represents the maximum extent (11.3m) of the DL deadzone for a femtocell transmitting at +10dBm. This distance would reduce if the FAP was not loaded in the downlink. Note also that the indoor to indoor path loss model, ITU-R P.1238, may underestimate the true path loss outside 15-20m range, as it is likely that other physical features (such as furniture, walls and buildings) will affect radio propagation (this is particularly true in dense urban areas.). A larger path loss reduces MUE interference, which, in turn, allows greater FUE throughput (linked to an increase in FUE-DPDCH Ec/No).
Figure 8-3: HSUPA simulation, Scenario B. E-DPDCH Ec/No compared to throughput for RFC3.
page 32
Figure 8-4: Throughput for HSUPA. 70% max bit rate for all FRCs.
Based on link budget calculations, the affects of uplink interference from one UE on the macrocell and a UE on the femtocell have been analysed; in this work it is assumed that the same frequency is used by the Macro and Femto Layer. In the analysis, it was assumed a femtocell serving an FUE on the physical edge of the cells (assumed to be 15m away) with a 12.2kbps AMR speech call; while a co-channel interference MUE is in the proximity of the femtocell. The analysis results showed that in order to be able to maintain the uplink connection between the FUE and femtocell, the transmitted power requirements are within the capability of the UE. Additionally, the performance of HSUPA on the femto-FUE link has been analysed in the presence of uplink interference from the Macro UE. By simulation, it has been found that in order to obtain HSUPA throughput of at least 2.8Mbps with a category 6 UE, the FUE needs to be near to the femtocell (5m) and transmit at a
2010 Femto Forum Limited | www.femtoforum.org
8.3 Conclusions
page 33
From the point of view of the MUE, the femtocell is a source of interference to the macrocell. However, the macro network can already cope with re-directing UEs to other WCDMA frequencies or RAT if a user is
Those locations with no coverage from alternative WCDMA frequencies or RATs may be adversely affected by poor Eb/No levels, leading to dropped calls. Due to femtocells, the macrocell may also be affected by an increase of uplink interference as femto-UEs increase power levels in order to achieve required quality levels. This may be limited by capping the maximum power level transmitted by FUEs, or limiting uplink throughput.
The minimum separation between MUE and femtocell has a strong affect on the capability to offer the required QoS to the femtocell user. However, the FUE has enough power to sustain a voice call while the MUE is in the coverage range of the femtocell. The downlink deadzone sets a minimum separation between MUE and femtocell meaning that the FUE transmit power is always within its capability. For HSUPA, the user is required to go closer to the femtocell in order to be provided with the best throughput. Simulation has shown that at 5m from the femtocell, good throughput can be achieved for MUEs further away than 12m.
Availability of alternative resources (a second carrier, or underlay RAT) for handing off or reselecting macrousers is the best way to provide good service when macro-users are in the proximity of femtocells.
page 34
In this scenario, MUE is connected to the macro network at the edge of coverage (RSCP<-95dBm). MUE1 is located in the same room as a femtocell (to which it is not allowed to access). The femtocell is fully loaded in the downlink; the femto UE are denoted as FUE. The Victim receiver in this case is the MUE, and the Aggressor is the femtocell downlink transmitter.
9.1 Description
Due to propagation loss and shadow fading effect, the macrocell signal strength varies at different location in the macrocell network coverage area. Femtocells are deployed at different locations in the macrocell network coverage area. Therefore, the down link interference from macrocell to the femtocell users will be location dependent. In order for the Femto to maintain its designed coverage, it should be capable of adjusting its pilot and max transmission power, while not causing undue interference to macrocell users.
Two important parameters need to be calculated or estimated. These are the minimum path loss (PLmin), when the UE is closest to the antenna, and the maximum path loss (PLmax), when the UE is farthest away from the antenna. PLmin will restrict the Femto maximum transmit power to avoid saturating the UE receiver; while PLmax is the maximum acceptable loss where the femto transmit power is sufficient to keep in-house communication with the UE.
For this purpose, we have assumed a certain house layout as an example with defined structure, and we have worked the path loss across the entire area of the house.
2010 Femto Forum Limited | www.femtoforum.org
page 35
Figure 9-2 below shows that path loss is dependent on the area within the house.
-50
Distance in meters
2 0 -2 -4 -6 -8 -10 -10
-60
-70
-80
-90
The maximum indoor path loss is shown to be more than 90 dB in some locations. The minimum outdoor path loss from an indoor Femto can be less than 60 dB. This will be a challenge for operators to balance good indoor coverage while not causing excessive outdoor interference.
Studied in this section is a macrocell user (MUE) at cell edge, located in an apartment where an active femtocell is operating with full capacity. Analysis is given for the following case: For the MUE to detect the macrocell and camp on it, or to maintain a call, the P-CPICH Ec/No must be sufficient. We assume a -20 dB threshold ie. the received P-CPICH RSCP from the macro must be no more than 20dB below the Rx P-CPICH RSCP of the femto. It is assumed that cell-edge PCPICH RSCP for the macro is -103 dBm, and so we can infer that the femto PCPICH RSCP must be lower than -83dBm for the MUE to camp on the macrocell. (Note that techniques for facilitating cell re-selection, such as the use of hysteresis, cell re-selection parameters, HCS, HPLMN, etc, are not discussed here, and are beyond the scope of this paper; the discussion in this paper is on the generic aspect of triggers for cell re-selection only.) We have assumed two scenarios for the location of the femto relative to the macrocell: 100 metres and 1,000 metres away from the macro have been used. We have found that when the femto is deployed in an area in close proximity to the macrocell (ie. 100 metres away), the maximum output power of the femto should be increased beyond 100 mW in order to ensure operation in high coverage. Therefore, when we
page 36
Macrocell configuration: Macrocell site-to-site distance: 100 or 1,000 metres Antenna height: 25 m Antenna gain: 18 dBi Frequency carrier in 850 MHz band Output power of the macro Node B: 20 Watts Town size: 500m radius.
9.2 Analysis
Femto location configuration: House size: 8.3X17.5 (m2) Houses cover 70% of the area Wall penetration loss: 12 dB CPICH power is 10% of max output power. The following figures show the required power (as a proportion of the total macrocell power) needed to support a voice call at 12.2 kbps within the house in the two deployment scenarios.
page 37
MacroPercentage12k 10 8 6 4 55 50 45
Distance in meters
2 0 -2 -4 -6 -8
Distance in meters
40 35 30 25 20 15 10
Figure 9-3: TX power needed for 12.2 kbps for MUE (1000 metres away and 100 metres away respectively).
It is evident that the required power for a well-sustained call at 12.2 kbps is higher in the following two cases: When the MUE is at the edge of the macrocell (ie. 1,000 metres away) and is behind the building where the femto is deployed. In this case the MUE requires the macrocell to transmit the radio link at a higher power to compensate for the high path loss affecting the macro signal and the interference from the femtocell. When the MUE is in close proximity to the femtocell and the MUE is located inside the house. In this case the wall loss is adding additional attenuation to the macro signal.
The following figures show the macro HSDPA throughput within the house in the two deployment scenarios (based on how far the femto is from the macro).
4000 3500
10 8 6
Distance in meters
2 0 -2 -4
2 0 -2 -4
-5
0 Distance in meters
10
-5
0 Distance in meters
10
Figure 9-4: MUE throughput with HSDPA for locations at 1,000 and 100 metres respectively.
page 38
In the scenario presented in this section, the performance of MUE attached to the macrocell is shown to be affected by the femtocell in some locations. This can be mitigated by the use of adaptive power control on femto. Results show that in some cases the MUE might experience deadzone when in close proximity to the femto. One firm conclusion from this analysis is that adaptive power control is necessary for the femtocells. Femtocells will require higher output power when the femtocell is deployed in locations near the centre of the macrocell. Adaptive power control on the femtocell mitigates interference by offering just the required transmit power on the femto, based on the level of interference from macro. However, it is shown that a macrocell UE (MUE) might not receive an adequate signal level from the macro to compensate for the femto interference. This is evident in all places in close proximity to the femto when the macro and femtocells share the same carrier. It is also concluded that there is no apparent and fundamental performance change whether 850 MHz or 2100 MHz is used for the carrier. In general, if a macro network is designed to provide fixed coverage in terms of cells radius, then the macrocell requires lower output power when operating at 850 MHz. Therefore, the interference level seen by a femto is the same, regardless of the carrier frequency.
It is shown that the femto is an effective vehicle for delivering a good carrier re-use. Furthermore, femtocells are an efficient technique for delivering the high-speed data offered by HSPA to femto users. This can be compared with the macrocell case, where cell radius is larger, resulting in the distribution of the potential bandwidth of the HSDPA to a larger number of users. It is also well known that HSPA throughput is affected by the location of the UE; the closer the UE to the centre of the cell, the higher the throughput. This leads us to conclude that small cells like femtocells are an optimum complementary technique for macrocells for addressing high-data usage.
page 39
10 Scenario D: Femtocell Uplink Interference to the Macrocell NodeB Receiver 10.1 Introduction
This document provides an analysis of Femtocell Uplink Interference from femtocell mobiles (FUEs) to a Macrocell NodeB Receiver.
The scenario being investigated is as follows: An FUE is located next to the apartment window that is in sight of a rooftop macrocell (approximately 1,000 m in distance), as shown in Figure 10-1. At the same time, the FUE is connected to the femtocell at the edge of its range, and is transmitting at full power.
In this analysis the impact to the macro Node B is measured by the sensitivity degradation, also referred to as noise rise (or relative increase in uplink Received Total Wide Band Power (RTWP)), experienced by the macro Node B, due to the femto UE. The impact is considered relative to the impact a macro UE will have on a macro Node B from the same location as the femto UE. The rest of this document is structured as follows:
page 40
In Section 10.2, analysis of Scenario D described in [Law08] is presented, including the assumptions used. The analysis shows that the femto UEs impact on the macro Node B is no worse that the impact a macro UE from the same location would cause. In Section10.4, a mitigation technique is suggested that would always ensure there is minimal impact to macro Node Bs due to femtocell UEs.
An analysis of this scenario is presented, based on link budget calculations. The analysis looks at the noise rise at the Macro Node B antenna connector due to the femtocell UE in the described scenario.
A macro Node B with a noise floor based on the assumption that the sensitivity of the Wide macro Node B for 12k2 voice service at the time is equal to -121 dBm (ie. the 3GPP reference sensitivity level for a 12k2 voice service on a Wide Area Node B at the antenna connector [TS25.104]). This sensitivity captures both the loading and noise figure of the macro Node B. The noise floor calculation is shown in Table 10-1.
kbps MHz dB
Required EbNo
8.30
dB
EbNo
(see [FF09])
noise floor
-104.32
dB
nf_ant
Next, the factors that could lead the femto UE to transmit at a power higher than expected are considered. This will occur if the femto UE is at the femtos cell edge, and if the femtocell experiences a noise rise, or its receiver is experiencing a blocking effect, caused by one of the following: A co-channel macro UE. page 41
An adjacent channel macro UE. Another femto UE located very close (~1m Free Space Loss) to the femtocell eg. a laptop with a 3G data card doing a data upload on the same desk as the femtocell.
Subsequently, for the purposes of this scenario, the following assumptions are made: The femto is operating under extreme conditions, experiencing a total noise rise equivalent to 70% loading in the uplink. A 21 dBm class femto 1 is used in the scenario that can provide a coverage path loss of up to 120 dBs (path loss estimate based on minimum RSCP sensitivity of UE of -111 dBm and an 11 dBm CPICH transmit power and assumption of negligible downlink interference from surrounding Node Bs).
Based on these assumptions, the link budget in Table 10-2 estimates the likely femto UE uplink transmission power at the femtocell edge of coverage for a 12K2 voice service and a 2Mbps HSUPA service.
Femto Receiver Noise Floor Femto UE Service Rate Chip rate Femto UE Processing Gain 24.98 dB PG = 10*log(W/R) -94.93 12.2 3.84 -94.93 dBm kbps MHz trnp R W =rnp +IM
Under the same RF conditions a 21 dBm class femto cell will provide larger downlink coverage than a 15dBm class or a 10dBm class femto
2010 Femto Forum Limited | www.femtoforum.org
page 42
DCH performance without rx diversity Required EbNo 8.30 dB EbNo [FF09] EbNo PG for 12K2 Typical EcNo to achieve HSUPA rates Required EcNo Minimum Required Signal Level for Femto UE Femto UE Path loss to Femto Femto UE Tx Power 120 8.39 120 21 dB dBm DLcov Pfue = min(21, max ((Pfmin + DLcov), -50) -111.61 -94.93 dB Pfmin = trnp +EcNo -16.68 0 dB of ~ 2Mbps [Hol06]
The noise rise caused to the macro by a femto UE transmitting at 8.39dBm for a 12K2 voice service and
21dBm for a 2Mbps HSUPA service was calculated, using the link budget in Table 10-3, as 1.44 dB and 9.12 dB respectively. Assuming that a macro UE is at the same location as the femto UE by the window (path loss of 130.77dB from the macro, see Ltot in Table 10-3), Table 10-4 shows that a macro UE operating from the same location as the femto UE will be transmitting at 9.94 dBm, and 21dBm if on a 12k2 voice service and 2Mbps HSUPA data service respectively and, hence, will lead to the same amount of noise rise as the femto UE.
page 43
8.39 0
21 0
dBm dBi
8.39 5
21 5
dBm dB dB
p Lw Ltot
=Pue Gmant +m
Path loss to Macro Node B Femto UE Interference @ macro antenna connector Rise above noise floor Noise rise
130.77
130.77
dB dB dB
Pfue_rec R NR
page 44
page 45
The following conclusions can be drawn: It is unlikely that a femto UE will be transmitting at maximum power, due to the relatively smaller coverage of the femto compared to the macro. When the femto is operating under extreme loading conditions, the analysis for a 12k2 voice service has shown that a femto UE in the described scenario will be transmitting in the region of 8.39 dBm and will cause a noise rise of approximately 1.44 dB. Further, a macro UE on a 12k2 voice service at the same location as the femto UE will transmit at 9.94 dBm and, hence, will lead to a similar amount of noise rise. When the femto is operating under extreme loading conditions, the analysis for a femto UE with 2Mbps HSUPA data service has shown that a femto UE in the described scenario will cause a noise rise amounting to approximately 8.55 dB; however, it should also be noted that a macro UE operating at the same position and on the same service (with the same service requirement) is expected to cause the same amount of noise rise.
10.3 Conclusions
The following recommendations are made. They will help ensure harmonious coexistence of femtocells and macro Node Bs:
10.4 Recommendations
It is desirable to limit the allowed maximum transmission power of a femto UE, to avoid a noise rise to the Macro Layer.
Assuming the femtocell has certain capabilities, then: o The maximum allowed femto UE transmission power can be limited appropriately, such that the noise rise caused by a femto UE when transmitting at its maximum allowed power is limited based on the femtocells proximity to the surrounding Macro Layer Node Bs. This is important, especially when one considers the cumulative effect of multiple femto UEs spread across a network. A similar approach is suggested in [R4-071578]. o The femtocell could also handover a femto UE to a macrocell if an in-service femto UE is at the verge of the femtocell; thereafter, uplink interference to a macrocell from this UE is avoided.
page 46
11.1 Description
AP1
AP2
UE1 UE2
Apartment 1
Apartment 2
Figure 11-1. Scenario E. Adjacent femto with UEs connected to each AP.
page 47
The effect on average throughput for the femto users can be analysed through the use of a Monte-Carlo simulation.
The simulation layout for this scenario is for case 1 and case 2, as shown in Figures 11-2 and 11-3.
15
10
-5
-10
-15 -15
-10
-5
10
15
In the second scenario contained in this section, the effect of neighbouring femtocell interference on the central house (located at coordinates 0,0) is investigated. In cases where a macrocell is present, it is located at coordinates -500m, -500m.
page 48
Village Plan
80
60
40
Y Coordinate in Meter
20
-20
-40
-60
Figure 11-3. Macrocell location relative to the house where the femtos are located. Simulation Configuration for apartment case: Max Femto power = 13dBm (but actual output power is based on auto-configuration) Pilot power = 10% of femto output power External Wall Loss = 15dB Internal Wall Loss = 10dB Door Loss = 5dB Macrocell location = -500, -500 Macrocell antenna height = 25m.
Apartment layout: Two-story building, height = 7m. Femto acess point is located on the ceiling UE height = 1.5m Penetration loss: External wall = 15 dB Window = 1 dB Doors = 3 dB Outer door = 30 dB. page 49
Simulation assumption for case 2 when houses are considered is found in the section describing Scenario C, but is not repeated here. The first simulation result obtained when the femtos use a dedicated carrier shown in Figure 11-4 below. The graph provides the cumulative distribution of HSDPA throughput for the UEs when located in the various locations (ie. flat or house). The results show the CDF for HSDPA throughput for UE1 in two cases: when the AP1 is operating in isolation (ie. AP2 is not there, and nor is UE2) when AP2 is operating in the adjacent location, and AP2 is connected to AP1 in active call. It is evident that the neighbouring femtocells (AP2) and the presence of UE2 do result in throughput degradation to UE1. It is shown that the performance degradation sustained by UE1 is greater in the case of apartment. In the case of users in apartments, the statistics for UE1 getting full throughput drops from more than 90%, to just over 40%.
1 Flat no Neighbour Flat with Neighbour House no Neighbour House with Neighbour
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
500
1000
2500
3000
3500
page 50
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2 Flat at (500 500) Flat at (100 0) House at (500 500) House at (100 0) 0 500 1000 1500 2000 Throughput in kbps 2500 3000 3500
0.1
In Scenario E, the downlink throughput of the UE connected to Femtocell is shown to be affected by the downlink of neighbouring femtocells. The case shows that driving femtocells to provide coverage for adjacent locations deemed to be covered by other femtocells yields performance degradation. The closer the femtocells are, the higher the mutual interference and performance degradation. It is, therefore, strongly recommended that femtocells use effective power control to confine coverage to their premises. Where the UE cannot get service from the femto, this UE should be supported by the macro network. There is a need to make sure that the pilot and transmit power of the femto is carefully adjusted to provide coverage to UEs within the intended area. It can be concluded that the femto coverage should aim to be restricted to a single apartment/house only in order to limit any undue interference between femtos. Adaptive power control is one method to help this. This leaves the issue of supporting visiting UEs being under the control of the macrocell.
11.3 Conclusions
page 51
12.1 Description
Figure 12-1: Illustration of the Interference Scenario F. The analysis on this scenario mainly focuses on how the uplink receiver (UL Rx) of AP1 would be interfered with or impacted by UE2, especially when service is ongoing in UE2. In this contribution the interference or impact is measure by sensitivity degradation, also referred to as noise rise (or relative increase in uplink Received Total Wide Band Power (RTWP)), experienced by AP1 due to UE2.
Analytical analysis is carried out for the above scenario based on link-budget calculations and transceiver performance requirements taken from [FF09].
12.2 Analysis
page 52
12.2.1 Assumptions
For the purposes of analysis the following assumptions are also made: AP1 and AP2 have equal Maximum DL powers, and CPICH channel power ratio is 10%; both AP1 and AP2 have only one 12.2K voice service ongoing; DL load factors are at about 50%; and AP2 has 50% loading in the uplink.
value Femtocell Noise Figure (NF) UE Processing Gain (G) Required Eb/No (EbNo) Sensitivity (S) UL load factor of AP2 (LoadUL) Noise rise due to UL loading (NRload) DL load factor of AP1 8 25 7 -118 50
Unit dB dB dB dBm %
=-108+EbNo-G+NF
3 50 ( ) 50(
dB
=-10*log(1-LoadUL)
)
10.6 104.4
RSCPAP1 RSCPAP 2
The interference at AP1 (Rx) Noise floor at AP1 (PN) Noise rise due to interference (NRinterfer)
-100 1.3
page 53
The sensitivity of a femtocell is based on the assumption that the noise figure is 8dB [FF09]. The sensitivity calculation is shown in Table 12-1. When UE2 get near enough to AP1, UE2 will drop call from AP2. At this point, the interference received at AP1 from UE2 is at the maximum. The assumed Ec/Io (interference margin) required to maintain a voice call is assumed -18dB.
Ec / Io = 10 * log
(1)
(2)
In order to maintain a voice call, the transmit power of UE2 connected to AP2 can be calculated as follows:
(3)
Inter UE 2 _ AP1 = S AP 2 + NRload + PathLossUE 2 _ AP 2 PathLossUE 2 _ AP1 = S AP 2 + NRload + ( Pcpich , AP 2 RSCPUE 2, AP 2 ) ( Pcpich , AP1 RSCPUE 2, AP1 ) = S AP 2 + NRload + ( Pcpich , AP 2 Pcpich , AP1 ) + ( RSCPUE 2, AP1 RSCPUE 2, AP 2 ) dB
The link budget in Table 12-1 estimates the maximum uplink interference to AP1 from UE2 at the cell edge of coverage of AP2 for a 12.2K voice service from formula (4). Both radio paths, from AP1 and AP2 to UE2, with the same model (ITU P.1238), are assumed to undergo the same signal decay loss with the increasing of distance. (4)
page 54
Distance between AP1 and UE2 with the maximum interference 7.5 7 6.5 6 5.5 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 0 5 10 15 20 25 distance between AP cells(m) 30 35
page 55
The following conclusions can be drawn: The closer from UE2 to AP1, the greater interference from UE2 to AP1. The interference reaches its maximum at the point when UE2 is disconnecting from AP2 (call is dropping). However, the analysis is based on the extreme scenarios. Usually, UE2 will handover to a macrocell before call drop, which will avoid the interference to AP1.
12.3 Conclusions
The following recommendations are made; they will help ensure the harmonious coexistence of co-channel femtocells:
12.4 Recommendations
It is desirable to limit the allowed maximum transmission power of UE2 to avoid a noise rise to the nearby AP1 when UE2 is at the verge of AP2.
The AP2 could also handover a UE2 to a macrocell (macrocell on another frequency channel preferred) if in-service UE2 is in the vicinity of the AP1; thereafter, uplink interference to AP1 from this UE2 is avoided.
page 56
13.1 Description
Figure 13-1: Illustration of the Interference Scenario G. The analysis on this scenario mainly focuses on how the downlink receiver (DL Rx) of UE1 would be interfered or impacted by the macro downlink transmission, especially when service is ongoing in UE1. Here, we assume that the distance between the femto UE and macro NodeB is approximately 1,000m. In this contribution, Ec/Io received by the UE1 at a different place within AP1 coverage is used as the metric to evaluate the impact from macro downlink.
Analytical analysis is carried out for the above scenario based on link-budget calculations and transceiver performance requirements taken from [FF09].
13.2 Analysis
page 57
13.2.1 Assumptions
The macrocell is 50% loaded. Okumura-Hata model + window loss and ITU P.1238 are used, respectively, for macrocell path
loss to UE1. ITU P.1238 is used for indoor modelling (for femtocell path loss to UE1). The macrocell is assumed to have a maximum transmit power of 43dBm, running at 50% utilisation; femtocell 10dBm of maximum transmit power and 50% utilisation. AP is1,020m away from macrocell.
Okumura-Hata model + window loss used for macrocell path loss to UE (approximately 1km distance).
The simulation showed that an adjacent macrocell causes little downlink interference to a femtocell.
page 58
Table 13-1 Macrocell Downlink Interference to an adjacent channel Femtocell UE in this worst-case scenario value unit dBm % dBm dBi km dB dB =Okumura-Hata propagation loss +window loss
Maximum Macro Node B Transmit Power Macro Node B Loading Macro NodeB output power (TxPowerMacroNodeB) Macro Node B Antenna Gain (GtMacroNodeB) Distance from UE to Macro NodeB Window loss Path loss from UE to Macro NodeB (PL1) Adjacent channel selectivity of the UE receiver (ACS) UE Antenna Gain (AntG_UE)
43 50 40 17 1 5 131
33 0
dB dBi =TxPowerMacroNodeB +
-110
dBm
GtMacroNodeB - PL - ACS-BLAntG_UE
From the above table, the downlink interference level from an adjacent channel macrocell at the UE receiver is -110dBm, which is less than thermal noise when the UE is located 1km away from the macrocell. Therefore, adjacent channel macrocell causes no downlink interference to Femto UE receiver.
13.3 Conclusions
Both theoretical analysis and simulation results show that Femtocell UE experiences little
page 59
The aim of this interference scenario is to evaluate impact of uplink interference experienced by a femtocell supporting closed access from a UE that is connected to a macro Node B (as it is not in the femto white list), when the UE and femtocell are located in close proximity. A weak signal is received from the macro Node B within the apartment where the femtocell is located. Further, it is assumed that the macro and femto cellular layers are deployed on adjacent frequencies. The impact of interference is evaluated using two services, AMR 12.2 kbps voice, and HSUPA. 3GPP transceiver specifications will be used in the analysis. It will be determined whether any enhancement to specifications is required.
A femtocell is located on a table within the apartment. Weak coverage of the macro network is obtained throughout the apartment. A user (that does not have access to the femtocell) is located next to the femtocell and has a call established at full power from the UE1 device. Another device UE2 has an ongoing call at the edge of femtocell coverage [Law08]. Figure 14-1 illustrates the interference Scenario H.
14.1 Description
page 60
Analytical evaluation is carried out for the interference scenario based on link-budget calculations and transceiver performance requirements, as specified by 3GPP. The uplink frequency is assumed to be 850 MHz (Band V), and the antenna gains of the Femtocell and UEs are equal to unity. The frequency separation between Femtocell UE (FUE) and Macrocell UE (MUE) is 5 MHz. The assumptions used in the analysis are given below.
14.2 Analysis
The parameter settings that are used in the analysis are given below:
MUE parameters
MUE max transmit power, a = 21 dBm (Power Class 4) [TS25.101] Minimum Coupling Loss (MCL) between MUE and Femtocell, b = 45 dB [TS25.141] Antenna gain = 1dBi.
MNB parameters
Receiver sensitivity, RxSens = -121 dBm [TS25.104] Required Eb/N0 for 12.2 kbps voice, Eb_N0 = 8.3 dB (without Rx diversity [TS25.104]) Noise floor = -104.32 dBm (RxSens + 10*log10(3.84e6/12.2e3) - Eb_N0).
FUE parameters
FUE max transmit power, c = 21 dBm (Power Class 4) [TS25.101] HSUPA terminal category = 6 (5.76 Mbps) [TS25.104].
page 61
Indoor-indoor path loss model ITU P.1238, N = 28 (2.8 x 10), n = 1, floor penetration loss factor = 4dB, residential deployment, shadow fading has log-normal distribution with standard deviation of 8 dB [FF09].
AMR voice service is used in the following analysis. Assuming that the MUE is transmitting at maximum power, the minimum allowed path loss between femtocell and MUE is calculated as the difference between the MUE transmit power (21 dBm) and the received signal level of the unwanted signal (-28 dBm). It is equal to 49 dB. This corresponds to a minimum separation of around 3.2m between femtocell and MUE, based on the ITU P.1238 indoor path loss model [FF09]. Clearly, this separation cannot be guaranteed in a residential deployment. Figure 14-2 illustrates the variation in minimum separation between femtocell and MUE for a given MUE transmit power level. One of the mechanisms available to improve robustness against adjacent channel interference is AGC. Under this technique the receiver will dynamically reduce gain of RF front end when it is subject to a blocking signal. The drawback of this technique is that it will result in a receiver sensitivity loss. The next step is to determine whether the reduction in receiver sensitivity makes a significant difference to uplink coverage of a femtocell. The uplink link-budget of AMR 12.2 kbps voice service is given in Table 14-1. It shows that the UE is only required to transmit at -25 dBm to achieve a typical coverage range of 25 m in uplink. Thus, there is
page 62
Table 14-1: Uplink radio link-budget for AMR 12.2 kbps RAB.
Ref.
Value
Units
Formula
0.003 -25.00
mW dBm
As above in dBm
Input, omni-directional antenna b c d e Antenna gain Body loss Cable loss Transmitter EIRP 0.00 -3.00 0.00 -28.00 dBi dB dB dBm pattern. Input Input a+b+c+d
Receiver (Femtocell) Thermal noise f g h i density Receiver noise figure Receiver noise density Receiver noise power -162.00 -96.16 dBm/Hz dBm f+g h + 10*log(3840000) Input, corresponding to 50% load j k l Interference margin Required Eb/N0 Required Ec/I0 -3.00 8.30 -16.68 dB dB dB [FF09]. Input [TS25.104]. Includes the SF gain. i + l - j, minimum requirement is -107 m Receiver sensitivity Receiver antenna n gain 0.00 dBi Input -109.84 dBm dBm [TS25.104]. -174.00 12.00 dBm/Hz dB Input Input
page 63
Under this interference scenario, the femtocell receiver can utilise AGC and reduce the gain of RF front end. As a result, uplink fast power control will command the FUE to increase its transmit power. Thus, the femtocell receiver will be able to tolerate a higher input level of unwanted signal. Figure 14-2 illustrates performance trends with and without AGC, assuming that the front end gain is reduced by 10 dB. Now, the minimum separation between the femtocell and MUE is equal to 1.5 m. A much smaller separation can be supported if the MUE is transmitting at lower power levels. If the FUE transmit power is increased in response to AGC there will also be an increase in interference to neighbouring femtocells, as well as to the macro Node Bs. Next, the impact on noise rise at the Macro Node B is evaluated. The noise floor at the macro Node B is calculated to be -104.32 dBm, as shown in Section 14.2. Assuming that the HUE is transmitting at -15 dBm and the total loss of signal strength up to the macro Node B is 110 dB (cell edge scenario), the received signal level will be -125 dBm. Adding ACS rejection of 63dB the received in-band signal strength will be equal to -188 dBm. Thus, noise rise at the macro Node B due to FUE will be insignificant. However, noise rise at neighbouring femtocells could become important as they will normally operate on the same frequency and may not be separated from each other by large distances. Thus, it is important to ensure that femtocell receiver de-sensitisation occurs only when it is necessary. Further, in order to reduce the risk of a significant noise rise in the Macro Layer due to femtocells, it is recommended to limit the maximum FUE transmit power eg. as suggested in [R4-071578].
page 64
Impact of adjacent channel interference on the Home Node B 3.5 Interfering signal level = -28 dBm Interfering signal level = -18 dBm
2.5
1.5
0.5
0 4
18
20
22
Figure 14-2: Minimum separation between Femtocell and MUE to avoid blocking, for a given MUE transmit power level.
The fixed-reference channel (FRC) no. 3 is used in the following analysis, as it corresponds to the maximum
uplink bit rate that is likely to be supported by femtocells in initial deployments. According to [TS25.104], the femtocell receiver should provide R 30% of max information bit rate at reference value of Ec/No of 2.4 dB and R 70% of max information bit rate at Ec/No of 9.1 dB. R denotes minimum HSUPA throughput. These values are based on the Pedestrian A channel model. The maximum information bit rate with FRC3 is equal to 4059 kbps. Assuming that MUE to FAP separation is fixed at 2 m, and the received MUE signal level at the femto receiver being less than or equal to -28 dBm (from ACS spec.), Figure 14-3 illustrates the variation in EDPDCH Ec/No measured at the femto receiver for a given MUE transmit power level. It is assumed that the FUE to FAP path loss is fixed at 90 dB (coverage edge scenario). Results show that in order to achieve 70% of max information rate, the average transmit power of FUE should be at least -3 dBm. Additionally, MUE transmit power should be kept to below 2.2 dBm. Maximum allowed FUE transmit power level can be signalled by the femtocell (eg. in RRC signalling), while MUE transmit power level cannot be controlled by
page 65
6 4 2 0 -2 -4 -10
-5
10
15
Figure 14-3: E-DPDCH Ec/No variation as a function of MUE transmit power level. Figure 14-4 illustrates the increase in average transmit power level of the FUE required to meet HSUPA throughput requirements, as a function of MUE transmit power level. The curves show that there is sufficient headroom available in uplink under this interference scenario. Figure 14-5 illustrates the variation in E-DPDCH Ec/No as a function of MUE transmit power level, when the FAP to MUE separation is fixed at 5 m. In this case, although the FUE transmit power should be at least -3 dBm, MUE transmit power can increase to 13 dBm to achieve R 30% of max information bit rate.
page 66
-2
-4 -10
-5
10
15
Figure 14-4: Required average FUE transmit power level to meet HSUPA throughput requirements.
10
Figure 14-5: E-DPDCH Ec/No variation as a function of MUE transmit power level. page 67
This section has considered a simple analysis of the interference Scenario H based on link-budget calculations and 3GPP specifications. Analysis considers impact of interference on two services AMR 12.2 kbps voice, and 5 Mbps HSUPA. The relationship between minimum FAP to MUE separation and MUE transmit power level has been derived. It was found that if the MUE is transmitting at the maximum power of 21 dBm it needs to be separated from the femtocell by around 3.2 m. This separation can be reduced further by employing Automatic Gain Control (AGC) at the femtocell receiver. It has been shown that the minimum MUE to FAP separation can be reduced to 1.5 m if a reduction in gain of 10 dB is applied by AGC. The resulting loss in receiver sensitivity will not deteriorate femtocell coverage of voice, as there is sufficient power headroom available at the UE. The performance of HSUPA has been analysed in the presence of uplink interference from the macro UE, which is operating on the adjacent frequency. The femtocell MUE separation is fixed at 2 m and 5 m. The FUE femtocell path loss is fixed at 90 dB, representing the coverage edge scenario. It was seen that in order to obtain 70% of nominal HSUPA bit rate with a category 6 UE, the MUE transmit power should be below 7.5 dBm and 18.5 dBm, respectively. In both cases minimum transmit power required for HSUPA transmission is equal to -3 dBm. As the likelihood of MUE transmitting at high power increases at the macrocell edge, HSUPA throughput at femtocell is expected to deteriorate in this interference scenario.
14.3 Conclusions
If the minimum separation between the MUE and femtocell is not maintained the femtocell receiver may not be able to decode the wanted speech signal at the required QoS level. Similarly, the HSUPA performance will deteriorate gradually as the MUE transmit power is increased for a given separation between the MUE and femtocell receiver.
The ACS specification for the Home Node B has been enhanced recently to accommodate higher levels of blocking signals [TS25.104]. Additional robustness against uplink interference can be provided with AGC. Since reduction in RF front end gain will cause receiver desensitisation, AGC should be activated only when required. It has been shown that there is sufficient power headroom available at the UE to meet typical femtocell coverage requirements for both voice and data services. Further, to maintain overall system stability in uplink, restriction of the maximum FUE transmit power level could be considered [R4-071578]. Some of the factors governing selection of maximum transmit power of FUE are femtocell coverage, service requirements, frequency deployment, distance to nearest macrocell receiver, uplink noise rise margin, etc.
page 68
The aim of this interference scenario is to evaluate the impact of downlink interference experienced by a UE that is connected to the macro Node B from a femtocell, while being located in close proximity to a femtocell. The MUE is not allowed to access the femtocell (ie. closed subscriber group). A weak signal is received from the macro Node B within the apartment where the femtocell is located. Further, it is assumed that the macro- and femto-cellular layers are deployed on adjacent frequencies. Impact of interference is evaluated using two services, AMR 12.2 kbps voice, and HSDPA. 3GPP transceiver specifications will be used in the analysis. It will be determined whether any enhancement to specifications is required.
Two users (UE1 and UE2) are within an apartment. UE1 (FUE) is connected to a femtocell and at the edge of coverage. UE2 (MUE) is connected to the macrocell at the edge of coverage, and located next to the femtocell transmitting at full power [Law08]. Figure 15-1 illustrates the interference Scenario I.
15.1 Description
page 69
Analytical evaluation is carried out for the interference scenario based on link-budget calculations and transceiver performance requirements as specified by 3GPP. The downlink frequency is assumed to be 850 MHz, and the antenna gains of the Femtocell and UEs are equal to unity.
15.2 Analysis
The parameter settings that are used in the analysis are given below [FF09]:
14.4 Mbps HSDPA. Femtocell parametersStatic maximum total transmit power, including control and traffic channels, Pmax = 10, 15, 20 [dBm] Downlink frequency = 850 MHz. Macrocell parametersMax transmit power on DCH = 33 dBm
Total transmit power = 43 dBm HSDPA power allocation = 42 dBm (80% of total power) Antenna gain = 17 dBi Feeder/cable loss = 3 dB. MUE receiver parameters Reference sensitivity level (DPCH_Ec_<REFSENS>) = -115 dBm (Band II), [TS25.101] REFIor = -104.7 dBm (Band II), [TS25.101] Max transmit power = 21 dBm (Power Class 4), [TS25.101] Maximum input power level = -25 dBm, [TS25.101] ACS = 33 dB, [TS25.101] HSDPA terminal category = 10 (14.4 Mbps). The ACS specification is valid as long as the Femtocell Downlink signal is in the range [-25,-52] (dBm) [TS25.101]. Additionally, the DPCH_Ec from the Macro Node B should be in the range [-74, -101] (dBm) [TS25.101]. Figure 15-2 illustrates the region of operation, which meets conditions specified above. page 70
Okomura Hata + Wall/Window loss External wall loss = 10 dB. Indoor-indoor path loss model, [FF09]
-25
-30
Region of operation
-35
-40
-45
-50
-55
-100
-95 -90 -85 -80 Min. Macro NB Downlink signal strength (Ior) [dBm]
-75
Figure 15-2: Macro Node B signal strength relative to the interfering femtocell signal strength measured at the MUE, required for successful decoding of AMR.
page 71
The region of operation, shown in Figure 15-2, gives the maximum strength of the downlink interfering signal versus the minimum strength of wanted signal. Each point in the region of operation translates into distance of separation between femtocell to MUE, versus distance between macro NodeB and MUE. The ITU P.1238 model will be used to calculate path loss between the femtocell and MUE, while the Okumura-Hata model will be used on the link between the macrocell and MUE. Figure 15-3 illustrates impact of downlink interference as a function of femtocell transmit power. The curves are obtained by converting maximum allowed path loss into distance according to specified path loss models. It is assumed that femtocell is transmitting at full power. The general trend is that as the MNB to MUE separation is increased, the distance between femtocell and MUE also needs to be increased, in order to avoid blocking at the MUE. It is clear from Figure 15-3 that downlink interference will not pose any problem to the MUE when it is located close to the macrocell. However, if the MUE is located close to the macrocell edge femtocell, interference could block the downlink signal. Figure 15-3 also illustrates the merits of adaptive control of maximum femto transmit power level, as for a fixed minimum femtocell MUE separation the appropriate femtocell transmit power level depends on the femtocell macrocell path loss. Table 15-1 gives the maximum MNB MUE separation that can be supported for different femtocell transmit power levels, when the femtocell MUE separation is fixed at 5 m. Results are obtained by converting maximum allowed path loss into distance using appropriate path loss model. A recent 3GPP contribution on the same topic suggests that maximum transmit power of a femtocell should be limited to 10 dBm for the adjacent channel deployment scenario [R4-090940].
page 72
2.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0 5 25 20 15 10 Minimum Femtocell - MUE separation [m] 30 35 Pmax = 10 dBm Pmax = 15 dBm Pmax = 20 dBm
Figure 15-3: Maximum MNB - MUE separation as a function of femtocell MUE separation, assuming AMR voice service.
Table 15-1: Maximum Macro NB MUE separation for a given maximum Femtocell transmit power level, when the Femtocell MUE separation is fixed at 5 m. Femtocell transmit power (dBm) 10 15 20 Max. Macro NB - MUE separation (km) 1.0 0.7 0.5
page 73
Next, performance of HSDPA under this interference scenario is analysed using link-budget type calculations. Fixed Reference Channel definition H-Set 6 is selected for analysis purposes [TS25.101]. A Category 10 UE is chosen, as it supports the maximum achievable HSDPA data rate (equal to 14.4 Mbps). The nominal average information bit rate for this FRC is 3219 kbps with QPSK, and 4689 kbps with 16QAM. The UE specification states that the receiver should meet or exceed the information bit throughput R requirements given in Table 15-2. Table 15-2: UE receiver performance requirement (HSDPA), [TS25.101]. Parameter Channel model Ioc [dBm]
Ec / I or
/I I or oc
Based on link budget calculations, the minimum femtocell to MUE separation is found to be 1.7 m, 2.6 m and 3.9 m (to maintain given Ioc), depending on whether Pmax is equal to 10 dBm, 15 dBm or 20 dBm (ITU p.1238 model). Figure 15-4 illustrates the impact of interference in terms of maximum macrocell to MUE separation for a given femtocell to MUE separation. At each point in the curve, femtocell interference is fixed at -60 dBm, while the macrocell G-factor ( Ior / I oc ) is maintained at 10 dB. Further, it is assumed that macrocell has allocated 80% of total power to HSDPA, resulting in HS-PDSCH Ec/Ior of approx. -1 dB.
Interference Scenario I.1, HSDPA 650 600 Pmax = 10 dBm Pmax = 15 dBm Pmax = 20 dBm
10
Figure 15-4: Maximum macrocell-MUE separation as a function of femtocell-MUE separation, for reception of HSDPA.
If the femtocell MUE separation is fixed at 5 m, the macrocell MUE separation should not be more than 185 m - 360 m in order to decode the HS-PDSCH at the specified rate. It is well known that a macrocell allocates highest HSDPA data rates only when UEs are located close to the cell site. Thus, it is not apparent whether interference from the femtocell will significantly deteriorate HSDPA performance at the MUE.
A simple analysis of the interference Scenario I has been carried out based on link-budget type calculations and 3GPP specifications. Adjacent channel deployment for the macro- and femto-layers has been assumed. The analysis considers impact of interference on two services AMR 12.2kbps voice, and 14.4Mbps HSDPA. In terms of AMR service, a minimum separation of 5 m between the femtocell and MUE can be achieved if the macrocell site is within 1.0 km, and the femtocell is not transmitting above 10dBm. It is recommended to implement adaptive control of maximum transmit power level at the femtocell and restrict maximum transmit power to 10 dBm, in order to achieve a good trade-off between femtocell coverage and adjacent channel deadzone. We have also analysed HSDPA performance under this interference scenario using link-budget type
2010 Femto Forum Limited | www.femtoforum.org
15.3 Conclusions
page 75
In terms of AMR service, it was found that femtocell downlink interference can block macrocell signal if the MUE is located close to the macrocell edge, and the femtocell transmit power is above 10 dBm. In terms of HSDPA performance, it is not clear that femtocell interference will significantly deteriorate HSDPA performance at the MUE.
Assuming dedicated spectrum deployment for the macro and femto cellular layers, the adjacent channel deadzone created by the femtocell can be adjusted by performing adaptive control of maximum femtocell transmit power. For example, femtocell should reduce the maximum transmit power level when it detects a weak macrocell signal, and vice versa.
16 Scenario J: Femtocell UE Uplink Interference to the adjacent channel Macrocell NodeB Receiver 16. 1 Introduction
This document provides an analysis of Femtocell Uplink Interference from femtocell mobiles (FUEs) to a Macrocell NodeB Receiver on the adjacent channel. The scenario being investigated is as follows: An FUE is located next to the apartment window that is in the sight of an adjacent channel rooftop macrocell (approx 1,000m distance), as shown in Figure 16-1. At the same time the FUE is connected to the femtocell at the edge of its range, and is transmitting at full power.
page 76
In this analysis the impact to the macro Node B is measured by the sensitivity degradation also referred to as noise rise (or relative increase in uplink Received Total Wide Band Power (RTWP)), experienced by the macro Node B due to the femto UE. In Section 16.2 analysis of Scenario J described in [Law08] is presented, including the assumptions used. The analysis shows that the femto UEs impact on the macro Node B is negligible.
An analysis of this scenario is presented based on link budget calculations. The analysis looks at the noise rise at the Macro Node B antenna connector due to the femtocell UE in the described scenario.
A macro Node B with a noise floor derived based on the assumption that the sensitivity of the Wide macro Node B for 12k2 voice service at the time is equal to -121 dBm (ie. the 3GPP reference sensitivity level for a 12k2 voice service on a Wide Area Node B at the antenna connector [TS25.104]). This sensitivity captures both the loading and noise figure of the micro Node B. The noise floor calculation is shown in Table 16-1.
page 77
Units
dBm
kbps MHz dB
Required EbNo
8.30
dB
EbNo
(see [FF09])
Noise floor
-104.32
dB
nf_ant
Next, the factors that could lead the femto UE to transmit at a power higher than expected are considered. This will occur if the femto UE is at the femtos cell edge, and the femtocell experiences a noise rise or its receiver is experiencing a blocking effect, caused by one of the following: An adjacent channel macro UE. Another femto UE located very close (~1m Free Space Loss) to the femtocell eg. a laptop with a 3G data card doing a data upload on the same desk as the femtocell.
Subsequently, for the purposes of this scenario, the following assumptions are made: The femto is operating under extreme conditions, experiencing a total noise rise equivalent to 70% loading in the uplink. A 21 dBm class femto 2 is used in the scenario that can provide a coverage path loss of up to 120dBs (path loss estimate based on minimum RSCP sensitivity of UE of -111 dBm and a 11 dBm CPICH transmit power and assumption of negligible downlink interference from surrounding Node Bs).
Based on these assumptions, the link budget in Table 16-2 estimates the likely femto UE uplink transmission power at the femtocell edge of coverage for a 12K2 voice service and a 2Mbps HSUPA service.
Under the same RF conditions, a 21 dBm class femtocell will provide larger downlink coverage than a 15dBm class or a 10dBm class femto.
2010 Femto Forum Limited | www.femtoforum.org
page 78
Femto Receiver Noise Floor Femto UE Service Rate Chip rate Femto UE Processing Gain 24.98 dB PG = 10*log(W/R) -94.93 12.2 3.84 -94.93 dBm kbps MHz trnp R W =rnp +IM
DCH performance without rx diversity Required EbNo 8.30 dB EbNo [FF09] EbNo PG for 12K2 Typical EcNo to achieve HSUPA rates Required EcNo Minimum Required Signal Level for Femto UE Femto UE Path loss to femto Femto UE Tx Power 120 8.39 120 21 dB dBm DLcov Pfue = min(21, max ((Pfmin + DLcov), -50) -111.61 -94.93 dB Pfmin = trnp +EcNo -16.68 0 dB of ~ 2Mbps [Hol06]
page 79
The noise rise caused to the adjacent channel macro by a femto UE transmitting at 8.39dBm for a 12K2 voice service and 21dBm for a 2Mbps HSUPA service was calculated, using the link budget in Table 16-3 as 8.6104
Table 16-3: Noise rise calculation for Scenario D1 (femto UE is transmitting at 8.39dBm and 21dBm 1000m from a macro Node B for a 12K2 service and 2Mbps HSUPA service).
Value 12K2 Voice Node B Antenna Gain Feeder/Connector Loss Noise Floor at antenna connector -104.32 -104.32 dBm nf_ant Table 16-1 17 3 HSUPA 17 3 Units dBi dB Gant Lf Comments [FF09]
8.39 0 8.39 5
21 0 21 5
Pfue Gmant Pfue_eirp Lw Ltot =1000m Okumura-Hata(Node B at30m and mobile at 1.5m) +Lw Adjacent Channel selectivity (+/=Pue Gmant +m
Path loss to Macro Node B Adjacent Channel Selectivity Femto UE Interference @ macro antenna connector Rise above noise floor Noise rise
130.77
130.77
33
33
dB
ACS
5MHz)
dB dB dB
Pfue_rec R NR
page 80
The following conclusions can be drawn: It is unlikely that a femto UE will be transmitting at maximum power due to the relatively smaller coverage of the femto compared to the macro. When the femto is operating under extreme loading conditions, the analysis for a 12k2 voice service has shown that a femto UE in the described scenario will be transmitting in the region of 8.39 dBm, and will cause a negligible noise rise of approximately 8.6 10-4dB. When the femto is operating under extreme loading conditions, the analysis for a femto UE with 2Mbps HSUPA data service has shown that a femto UE in the described scenario will cause a negligible noise rise amounting to approximately .02 dB. The general conclusion is that a femto UE operating on the adjacent channel to a macro Node B will not cause an impact to such an adjacent channel macro Node B.
16.3 Conclusions
page 81
17. Downlink and Uplink Scenarios Modelling Power Control Techniques for Interference Mitigation
In [FF08], system level simulations were presented for the downlink and uplink under deployment of
femtocells for 2 GHz carrier frequency. In this section, HNB deployment in 850 MHz is discussed vis a vis a deployment in the 2 GHz band done in Section 17 of [FF08] and system level simulations are provided. It is shown that simple modification to the parameters setting for power calibration can be used in 850MHz to achieve nearly the same performance (Coverage and Throughput statistics) as 2GHz deployment. It is also shown with simulations that the uplink interference mitigation technique of adaptive attenuation continues to work well in 850MHz as well. All results presented in this section are under the same set-up and simulation conditions as Section 17 of [FF08], except the propagation model. We restrict our attention to the femtocell deployment in the dense urban settings.
The propagation loss models specified in [FF09] (from [ITU1238]) identify the frequency dependent term for propagation in indoor environment and for small distances as 20*log10(f) , where f is the carrier frequency and the path loss is expressed in dB. This term suggests that the typical path loss between two points will be 20*(log10(2000/850)) ~= 7.4 dB higher in 2GHz than in 850 MHz. This is the major component of difference in the propagation loss seen in the two bands.
We apply this frequency dependent path loss offset of -7.4 dB to the path losses from 2 GHz system simulations using the simulation framework described in Section 17 of [FF08]. Specifically, all the path loss values from 2 GHz modelling (outdoor to outdoor, outdoor to indoor, indoor to indoor in same or different apartment) are reduced by the path loss offset to model 850 MHz propagation. Other components, such as outdoor to indoor wall penetration loss, are observed to be not as sensitive to this frequency difference 3, and are left unchanged.
As identified in [FF08], the coverage of a femtocell for a given transmit power differs based on its location within a macrocell, and hence it is crucial to calibrate the transmit power of the femtocell. A reference power calibration algorithm that attempts to strike a balance between increasing the femtocell coverage and reducing the interference to the macro network was specified in [FF08, Section 17.1.2.4, and TR25.820]. Various studies over the years have produced inconclusive and sometimes contradictory trends in the behaviour of outdoor to indoor penetration loss with change in frequency (eg. see [Kob92, Stav03, Dav97]).
2010 Femto Forum Limited | www.femtoforum.org
3
page 82
This simple change in the parameter for HNB power calibration ensures that the algorithm works well in 850 MHz as well.
In this section we show illustrative results and compare with 2 GHz deployment to show that outage and throughput performance in 850 MHz band does not significantly differ from that in 2 GHz band, provided the power calibration of femtocells takes into account the impact of the frequency band. We show the results for dense urban model depicted in Section 17 of [FF08]. Similar to Section 17 of [FF08], we assume 2000 apartments per cell with 4.8% HNB penetration giving 96 HNBs per cell. Out of these, 24 HNBs are simultaneously active (have HUEs in connected mode). If an HNB is active it transmits at full calibrated power, else it transmits only the pilot and overhead channels.
Similar to Section 17 of [FF08], we assume co-channel deployment where HUEs and MUEs share the same carrier. Closed subscriber group is assumed throughout. We say a UE is unable to acquire the pilot if the CPICH Ec/No is below Tacq. We use Tacq=-20dB for our analysis. For this analysis, the MNBs are assumed to transmit at 50% of the full power (ie. 40dBm). The CPICH Ec/Ior for MNBs and HNBs are set to -10dB (ie. 33dBm). In addition, we take into account idle cell reselection procedure to determine whether a HUE is
2010 Femto Forum Limited | www.femtoforum.org
page 83
Table 17-1 Parameters for the co-channel idle cell reselection procedure.
Macro -18 dB 10 dB NA HNB cells: -50 dB Macro cells: 3dB HNB cells: -12 dB Macro cells: not needed
HNB -18dB 4dB NA HNB cells: 3dB Macro cells: 5dB Not needed
In this section we analyse the coverage statistics of UEs with calibrated HNB transmit power algorithm described in previous sections. Table 16-2 and Table 16-3 show the pilot acquisition and outage statistics for dense-urban model, with calibrated HNB transmit power. We compare three cases: i) ii) iii) Calibrated HNB transmit power with Pmin=-20dBm and Pmax=20dBm Calibrated HNB transmit power with Pmin=-10dBm and Pmax=20dBm Calibrated HNB transmit power with Pmin=0dBm and Pmax=20dBm.
17.3.2 Coverage Statistics at 850 MHz for Calibrated HNB Transmit Power
The results show the expected trade off between good HNB coverage and interference to Macro UEs as a function of the HNB transmit power. Results corresponding to Pmin=-10 dBm and Pmin=0 dBm were presented in [FF08] for 2 GHz. Additionally, this section presents results for Pmin=-20 dBm. It can be readily seen that the statistics corresponding to Pmin=-10dBm and Pmin=0 dBm in Table 16-2 and Table 16-3 closely matche those in Table 17.7 of [FF08]. Each point on the cell sees a lower path loss in 850 MHz from both macro and femtocells and, consequently,
2010 Femto Forum Limited | www.femtoforum.org
page 84
3.9%
1.9%
0.5%
0.6%
0.2%
0.2%
2.7%
5.2%
12.0%
Table 17-3: Coverage statistics for dense-urban model with 24 active HNBs and calibrated HNB transmit power.
Pmin=-20dBm, Pmax=10dBm MUEs moved to another carrier HUEs unable to camp on own HNB HUEs switched to macro on shared carrier HUEs moved to another carrier Pmin=-10dBm, Pmax=20dBm Pmin=0dBm, Pmax=20dBm
9.7%
13.5%
25.5%
9.6% 7.7%
4.9% 3.6%
2.4% 1.1%
1.9%
1.3%
1.3%
In these simulations the possible calibrated transmit powers for HNBs are assumed to take a continuous range of values. In practice, these values will be quantised with a given granularity.
2010 Femto Forum Limited | www.femtoforum.org
page 85
CDF
0.5 0.4 0.3 0.2 0.1 0 -10 HNB Tx Power: 2 GHz,PL Edge 80 dB, PMin 0 dBm HNB Tx Power: 2 GHz,PL Edge 80 dB, PMin -10 dBm HNB Tx Power: 850MHz ,PL Edge 72.6 dB, PMin 0 dBm HNB Tx Power: 850 MHz ,PL Edge 72.6 dB, PMin -10 dBm
-5
15
20
In this section we study the performance of HSPA+ DL on 850 MHz under HNB deployment by system level simulations. The assumptions for the simulation are the same as those in Section 17 of [FF08]. In the denseurban model, blocks of apartments are dropped into the three centre cells of a macrocell layout with ISD of 1 km. We drop 2,000 apartment units in each macrocell that corresponds to 6,928 households per square kilometre. This represents a dense-urban area. Taking into account various factors such as wireless penetration (80%), operator penetration (30%) and HNB penetration (20%), we assume a 4.8% HNB penetration, which means 96 of the 2,000 apartments in each cell have a HNB installed from the same operator. Out of these, 24 HNBs are simultaneously active (have a HUE in connected mode). We assume cochannel performance for all HUEs and MUEs. All UEs have one receive antenna. We assume that the power transmitted for the overhead channels, including CPICH pilot is 25% and the transmit power for the pilot, is 10%. The transmit power of HNBs is calibrated using the algorithm specified in Section 16.2. We assume a Rician channel with Rician factor K=10 and 1.5 Hz Doppler frequency. Macrocells are loaded with HNBs, HUEs and MUEs. There are 10 MUEs per cell, and 96 HNBs, of which 24 are active. Each active HNB has one
page 86
0.5 0.4 0.3 0.2 0.1 0 All UEs: No HNB present All UEs: HNB Present, Pmin = -10 dBm All UEs: HNB Present, Pmin = -20 dBm
0.2
0.4
1.6
1.8 x 10
2
7
page 87
0.8
0.6 All UEs: No HNB present All UEs: HNB Present, Pmin = -10 dBm All UEs: HNB Present, Pmin = -20 dBm
CDF
0.4
0.2
6 x 10
5
It is seen that deployment of HNBs helps all users. The users served by HNBs see very good RF conditions and dedicated Node B and, hence, see very high throughputs. The users on macrocells see a reduced load on the network and, hence, experience better throughputs. Even when the lower limit on the transmit power to HNBs is reduced to -20 dBm, the HUEs continue to experience high user throughputs. Figure 16-3 shows a magnified version of the lower range of throughputs to identify the impact of Pmin on outage.
17.3.4 Conclusions
To summarise, HNB deployment continues to provide the benefits identified in Section 17 of [FF08] in 850 MHz. The small change in parameters of power calibration enables the same algorithm to be used in 850 MHz, and results in nearly the same transmit power distribution on HNBs as that in 2 GHz.
In this section we study the HNB and macro uplink throughput performance in a co-channel deployment of HNBs for 850 MHz. In [FF08] the benefits of uplink adaptive attenuation at an HNB were identified. This section carries out the uplink throughput analysis and comparison of HNB deployment with and without adaptive attenuation in 850 MHz in a dense urban scenario. The layout and deployment scenario is the same as those in [FF08] and Section 16.2.
We assume a Rician channel with K factor of 10 dB and 1.5 Hz Doppler fading. The MUEs and HUEs are assumed to transmit full-buffer traffic using 2ms TTI HSUPA. The maximum number of transmissions is set to
page 88
In Baseline 1, the NF setting at HNB is similar to MNB. In Baseline 2, a fixed NF of 20dB is assumed at the HNB. This is similar to the 19dB NF used in local area basestation class specified in [TS25.104]. The Enhanced case uses adaptive attenuation (or noise figure), which means additional attenuation is added only when needed, depending on out-of-cell and in-cell signal strength. We run uplink simulations for the scenario described in the previous section. Figure 16-4 and Figure 16-5 show the HUE and MUE uplink throughput CDFs for Baseline 1, Baseline 2 and Enhanced cases. The HUE and MUE transmit power distributions are shown in Figure 16-6 and Figure 16-7. It is seen from Figure 16-4 that the HUE Baseline 1 uplink throughput performance is poor, due to intra-HNB, inter-HNB and Macro-to-HNB interference. Adding 15dB fixed attenuation at HNBs (ie. Baseline 2) improves the HUE performance significantly, but there are still some HUEs that have poor uplink throughput. This is because 15dB fixed attenuation does not solve inter-HNB interference problem. In addition, in some cases, more than 15dB attenuation is needed to overcome Macro-to-HNB interference. With fixed uplink attenuation (ie. Baseline 2), the HUE transmit powers are higher compared to adaptive attenuation. As seen in Figure 16-4, adaptive UL attenuation completely eliminates HUE throughput outage and achieves good throughput performance. It is also seen from Figure 16-5 that the MUE uplink performance is not impacted by adding attenuation at HNBs. In addition, Figure 16-6 and Figure 16-7 show that the transmit power in 850MHz is roughly 7 to 10dB lower than that in 2GHz. The reduced power will both reduce interference and improve battery life.
page 89
CDF
500
1000
3000
3500
4000
CDF
0.5 0.4 0.3 0.2 0.1 0 Baseline 1 Baseline 2 Enhanced 0 50 100 150 200 250 300 350 MUE Throughput [kbps] 400 450 500
page 90
CDF
-40
-30
10
20
30
10 MUEs + 24 HUEs per macro cell 1 0.9 0.8 0.7 0.6 Baseline 1 Baseline 2 Enhanced
CDF
-40
-30
10
20
30
page 91
0.5 0.4 0.3 0.2 0.1 0 34 MUEs + 0 HUEs per macro cell 10 MUEs + 24 HUEs per macro cell with adaptive uplink attenuation 0 500 1000 1500 2000 UE Throughput [kbps] 2500 3000
Figure 17-8 UE uplink throughput distributions in 850 MHz. There are, in total, 34 UEs per macrocell, of which 24 UEs migrate to MNB in the No HNBs case. HNB deployment increases the system capacity significantly.
page 92
Simple adjustment of Power Calibration settings, namely changing the HNB target coverage path loss, is sufficient to make HNB deployments nearly equivalent in different frequency bands. Similar DL throughput performance is seen in Dense Urban deployment of HNBs in 850 MHz and 2 GHz. UL throughputs are higher in Dense Urban deployments of HNBs in 850 MHz, compared to 2GHz. The UE transmit powers are seen to be smaller for 850 MHz compared to 2 GHz.
17.3.6 Conclusions
In summary, HNB deployment continues to provide expected benefits in 850 MHz band as well.
page 93
18 Summary of Findings
Scenario
A - Macrocell Downlink Interference to the Femtocell UE Receiver
Conclusions
When a strong macro signal is present, customers already obtain excellent service; adding a co-channel femtocell offers little additional coverage gain. Assuming standard models and parameters, it is shown that even at 10 dBm transmit power, the femtocell is able to comfortably provide voice to the UE when the femtocell is located as far as 100 m away and maximum HSDPA throughput can be expected up to 25 m away.
Impacts
Low, but a way of identifying customers who are unlikely to benefit from femto because of already high macro coverage would be desirable. If the macro is dominant, the consequence for the customer is that they will be provided service by the macro carrier so the impact of this scenario is mainly on zonal-based propositions.
page 94
Scenario
B - Macrocell UE Uplink Interference to the Femtocell Receiver
Conclusions
The analysis results showed that in order to be able to maintain the uplink connection between the FUE and femtocell, the transmitted power requirements are within the capability of the UE. Additionally, the performance of HSUPA on the femto FUE link has been analysed in the presence of uplink interference from the Macro UE. By simulation, it has been found that in order to obtain HSUPA throughput of at least 2.8Mbps with a category 6 UE, the FUE needs to be near to the femtocell (5m) and transmit at a power level greater than 15dBm, if the MUE is within 15m of the femtocell.
Impacts
From the point of view of the MUE, the femtocell is a source of interference to the macrocell. However, the macro network can already cope with re-directing UEs to other WCDMA frequencies, or RAT, if a user is affected by high interference. Those locations with no coverage from alternative WCDMA frequencies, or RATs, may be adversely affected by poor Eb/No levels, leading to dropped calls. Due to femtocells, the macrocell may also be affected by an increase of uplink interference, as femto-UEs increase power levels in order to achieve required quality levels. This may be limited by capping the maximum power level transmitted by FUEs, or by limiting uplink throughput. The minimum separation between MUE and femtocell has a
However, such analysis must take into account the downlink deadzone created by the femtocell. High power from the femtocell in order to maintain the downlink will interfere with the macrocell signal at the MUE, and will force the macrocell to handover the call to another WCDMA frequency or RAT; or, if none of these are possible, the MUE call may be dropped.
strong effect on the capability to offer the required QoS to the femtocell user. However, the FUE has enough power to sustain a voice call while the MUE is in the coverage range of the femtocell. The downlink deadzone sets a minimum separation between MUE and femtocell, meaning that the FUE transmit power is always within its capability. For HSUPA, the user is required to go closer to the femtocell in order to be provided with the best throughput. Simulation has shown that at 5m from the femtocell, good throughput can be achieved for MUEs further away than 12m. Availability of alternative resources (a second carrier, or underlay RAT) for handing off or reselecting macro-users is the best way to provide good service when macro-users are in the proximity of femtocells.
page 95
Scenario
C - Femtocell Downlink Interference to the Macrocell UE Receiver
Conclusions
In the scenario presented in this section, the performance of MUE attached to the macrocell is shown to be affected by the femtocell in some locations. This can be mitigated by the use of adaptive power control on the femto. Results show that in some cases the MUE might experience deadzone when in close proximity to the femto. One firm conclusion from this analysis is that adaptive power control is necessary for the femtocells; another is that femtocells will require higher output power when the femtocell is deployed in locations near the centre of the macrocell. Adaptive power control on the femtocell mitigates interference by offering just the required transmit power on the femto based on level of interference from macro. However, it is shown that a macrocell UE (MUE) might not receive adequate signal level from the macro to compensate for the femto interference. This is evident in all places in close proximity to the femto when the macro and femtocells share the same carrier. It is also concluded that there is no apparent and fundamental performance change between the case when 850 MHz or 2100 MHz is used for the carrier. In general, if a macro network is designed to provide fixed coverage in terms of cells radius, then the macrocell requires lower output power when operating at 850 MHz. Therefore, the interference level seen by a femto is the same, regardless of the carrier frequency.
Impacts
For operators without a dedicated carrier on which to deploy femto, adaptive power control is essential for the success of the network Even though the intrinsic coverage of the macro network is reduced by the deployment of femto, other studies have shown (eg. Section 17) that the total capacity of the network (macro + femto) may increase a hundredfold.
It is shown that the femto is an effective vehicle for delivering a good carrier re-use. Furthermore, femtocells are an efficient technique for delivering highspeed data offered by HSPA to the femto users. This should be compared to the macrocell case where cell radius is larger resulting in the effect of distributing the potential bandwidth of the HSDPA to a larger number of users. It is also a well known that HSPA throughput is affected by the location of the UE, the closer the UE to the centre of the cell the higher the throughput. This lead us to conclude that small cells like femto cells are an optimum complimentary technique to macro cells for addressing high data usage.
page 96
Scenario
Conclusions
It is unlikely that a femto UE will be transmitting at maximum power, due to the relatively smaller coverage of the femto compared to the macro. The analysis for a 12k2 voice service has shown that a femto UE in the described scenario will be transmitting in the region of 8.39 dBm, and will cause a noise rise of approximately 0.07dB. Further, a macro UE at the same location as the femto UE will cause a 0.09dB noise for the same 12k2 voice service. The analysis for a femto UE with 2Mbps HSUPA data service has shown that a femto UE in the described scenario will cause a noise rise amounting to approximately 1.09dB; however, it should be noted that a macro UE operating at the same position and on the same service (with the same service requirement) is expected to cause the same amount of noise rise.
Impacts
The maximum allowed femto UE transmission power can be limited appropriately, such that the noise rise caused by a femto UE when transmitting at its maximum allowed power is limited based on the femtocells proximity to the surrounding Macro Layer Node Bs. This is important, especially when one considers the cumulative affect of multiple femto UEs spread across a network. A similar approach is suggested in [R4-071578]. The femtocell could also handover a femto UE to a macrocell if an in-service femto UE is at the verge of the femtocell; thereafter, uplink interference to a macrocell from this UE is avoided.
page 97
The downlink throughput of the UE connected to the femtocell is shown to be affected by downlink of neighbouring femtocells. This case shows that driving femtocells to provide coverage to adjacent location deemed to be covered by other femtocells yields performance degradation.
If the femto coverage is controlled through mechanisms such as adaptive power control, then this scenario will generally result in the visiting UE being handled by a Macro Layer. These impacts exist when a UE femtocell experiences interference levels in the order of -50dBm. Consequently, there is a risk that for adjacent apartment
The closer the femtocells are, the higher the mutual interference and performance degradation. It is therefore strongly recommended that femtocells use effective power control to confined coverage to their premises, and where the UE can not get service from the its femto, this UE should be supported by the macro network. There is a need to make sure that the pilot and transmit power of the femto is carefully adjusted to provide coverage to UEs within the intended area. It can be concluded that the femto coverage should aim to be restricted to a single apartment/ house only in order to limit any undue interference between femtos. Adaptive power control is one method to help this. This leaves the issue of supporting visiting UEs to be under the control of the macrocell.
deployments coverage may not be assured from the femtocell under all circumstances.
page 98
The following conclusions can be drawn: The closer from UE2 to AP1, the greater interference from UE2 to AP1. The interference reaches maximum at the point when UE2 is disconnecting from AP2 (call is dropping). However, the analysis is based on the extreme scenarios. Usually, UE2 will handover to a macrocell before call drop, which will avoid the interference to AP1. The following recommendations are made, which will help ensure harmonious coexistence of co-channel femtocells: It is desirable to limit the allowed maximum transmission power of UE2 to avoid a noise rise to the nearby AP1, when UE2 is at the verge of AP2.
In typical cases, both wanted and Aggressor femtocells should have dynamically optimised coverage to their respective UE; hence, this co-channel scenario is unlikely to occur.
If this femtocell power optimisation does not occur, the cochannel interference can indeed occur, and range reduction is the consequence. This range reduction can be mitigated to an extent by the normal dynamic power control of the wanted UE. Consequently, this is manageable as long as minimum performance requirements for adaptive power control are agreed.
The AP2 could also handover a UE2 to a macrocell (macrocell on another frequency channel preferred) if in-service UE2 is in the vicinity of the AP1; thereafter, uplink interference to AP1 from this UE2 is avoided.
page 99
Both theoretical analysis and simulation results show that femtocell UE experiences little adjacent channel interference from an outdoor macrocell in most cases.
There is no impact.
page 100
It was found that if the MUE is transmitting at the maximum power of 21 dBm, it needs to be separated from the femtocell by around 3.2 m. This separation can be reduced further by employing Automatic Gain Control (AGC) at the femtocell receiver. It has been shown that the minimum MUE to FAP separation can be reduced to 1.5 m if a reduction in gain of 10 dB is applied by AGC. The resulting loss in receiver sensitivity will not deteriorate femtocell coverage of voice, as there is sufficient power headroom available at the UE. The performance of HSUPA has been analysed in the presence of uplink interference from the macro UE, which is operating on the adjacent frequency. The femtocell MUE separation is fixed at 2 m and 5 m. The FUE femtocell path loss is fixed at 90 dB, representing the coverage edge scenario. It was seen that in order to obtain 70% of nominal HSUPA bit rate with a category 6 UE, the MUE transmit power should be below 7.5 dBm and 18.5 dBm, respectively. In both cases minimum transmit power required for HSUPA transmission is equal to -3 dBm. As the likelihood of MUE transmitting at high power increases at the macrocell edge, HSUPA throughput at femtocell is expected to deteriorate in this interference scenario.
If the minimum separation between the MUE and femtocell is not maintained, the femtocell receiver may not be able to decode the wanted speech signal at the required QoS level. Similarly, the HSUPA performance will deteriorate gradually as the MUE transmit power is increased for a given separation between the MUE and femtocell receiver.
page 101
In terms of AMR service, a minimum separation of 5 m between the femtocell and MUE can be achieved if the macrocell site is within 1.0 km, and the femtocell is not transmitting above 10dBm. It is recommended to implement adaptive control of maximum transmit power level at the femtocell and restrict maximum transmit power to 10 dBm, in order to achieve a good trade-off between femtocell coverage and adjacent channel deadzone. We have also analysed HSDPA performance under this interference scenario using link-budget type calculations and UE specifications. At the minimum supported femtocell MUE separation of 5 m, it was found that the macrocell MUE separation should not be more than 185 m - 360 m, in order to decode the HSPDSCH at the specified rate. Analysis was performed for a fully loaded femtocell transmitting at 10 dBm, 15 dBm and 20 dBm. It is well known that a macrocell allocates highest HSDPA data rates only when UEs are located close to the cell site. Thus, it is not apparent whether downlink interference from femtocell will significantly deteriorate HSDPA performance at the MUE.
In terms of AMR service, it was found that femtocell downlink interference can block macrocell signal if the MUE is located close to the macrocell edge and the femtocell transmit power is above 10 dBm. In terms of HSDPA performance, it is not clear that femtocell interference will significantly deteriorate HSDPA performance at the MUE. Assuming dedicated spectrum deployment for the macro and femto cellular layers, the adjacent channel deadzone created by the femtocell can be adjusted by performing adaptive control of maximum femtocell transmit power.
page 102
It is unlikely that a femto UE will be transmitting at maximum power, due to the relatively smaller coverage of the femto compared to the macro. The analysis for a 12k2 voice service has shown that a femto UE in the described scenario will be transmitting in the region of 8.39 dBm and will cause a negligible noise rise of approximately 3.4 10-5dB. The analysis for a femto UE with 2Mbps HSUPA data service has shown that a femto UE in the described scenario will cause a negligible noise rise amounting to approximately 6.2 10-4dB. The general conclusion is that a Femto UE operating on the adjacent channel to a macro Node B will not cause an impact to such an adjacent channel macro Node B.
The uplink noise rise experienced by the macro nodeB from the adjacent channel femto UE is likely to be significantly less than the noise rise experienced by the macro Nodes Bs own UE transmitting from the same location. Consequently, there is negligible impact to the adjacent channel macro.
page 103
A simple adjustment of Power Calibration settings namely, changing the HNB target coverage path loss is sufficient to make HNB deployments nearly equivalent in different frequency bands. Similar DL throughput performance is seen in Dense Urban deployment of HNBs in 850 MHz and 2 GHz. UL throughputs are higher in Dense Urban deployments of HNBs in 850 MHz compared to 2GHz. The UE transmit powers are seen to be smaller for 850 MHz compared to 2 GHz. In summary, HNB deployment continues to provide expected benefits in 850 MHz band as well.
The conclusions depend on the operation of important techniques, such as adaptive CPICH power setting, adaptive attenuation (AGC) in the femto receiver, and UE transmit power capping. With these techniques in play, the impact on the performance of the networks is total available data capacity gain of two orders of magnitude for the simulated conditions.
page 104
19 Overall Conclusions
By examining a series of scenarios, building on the work of 3GPP RAN4 as well as the previous Femto Forum work at 2 GHz, we have reached and confirmed the following conclusions: Femtocell performance at 850 MHz is very much similar to that at 2 GHz. Power management of the UE is important to manage the noise rise in the macro network. o o In normal operation, the noise rise contribution from the UE is small (a decibel or less). Power capping of the UE when operating in the femto environment ensures that, even in difficult radio conditions, the UE hands-off to the macro network before its transmit power increases to the point where macro noise rise is a problem. o Dynamic receiver gain management in the femto (AGC or adaptive attenuation) ensures that femtos can offer good service to both near and far UEs, without unnecessarily increasing the UE transmit power, and, therefore, keeping the noise rise contribution to a minimum. o An increase in the dynamic range specifications is required to accommodate femto operation in both near and far cases. Downlink power management is equally key in managing the tradeoff between service range (in the closed user group cases), and deadzone. o By measuring its environment, the femto can set its transmit power appropriately for both dense urban and suburban deployment, even in shared carrier situations. o Given a reasonable distribution of indoor and outdoor users, the link budget indoors with femto is so good in comparison with the corresponding macro link budget that the total air interface capacity can be a hundred times greater with femto than without it. With these power management techniques in place, femto operation in the co-channel deployment with macro is possible. A second carrier is preferred, to give macro users service even within the deadzones of the femtocells.
Some of these factors (adaptive attenuation, power capping, and downlink power management) are becoming widely available in the industry. Others (increased receiver dynamic range) are already approved in standards. All of them will deliver the performance and capacity gains required for next-generation cellular networks.
page 105
20 Further Reading
Title: Macrocell Downlink Co-Channel Interference to the Femtocell UE Receiver 3GPP Analysis References: [R4-071941] R4-071941, "Simulation results for Home NodeB to Home
20.1 Scenario A
NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080409] [R4-080149] R4-080149, Ericsson, "Simulation assumptions for the block
of flats scenario, 3GPP TSG-RAN Working Group 4 (Radio) meeting #46, February 2008. [R4-080150]
Title: Macrocell Uplink Co-Channel Interference to the Femtocell Receiver 3GPP Analysis References: [R4-070825] [R4-070969] R4-070969, Home B output power, Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #43bis, R4-070969, June 2007. [R4-070970 [R4-071619] [R4-071941] R4-071941, "Simulation results for Home NodeB to Home
20.2 Scenario B
NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080097] [R4-080409] [R4-080153]
Title: Femtocell Downlink Co-Channel Interference to the Macrocell UE Receiver 3GPP Analysis References: [R4-071231] [R4-071253] [R4-071263] [R4-071540] [R4-071554] [R4-071578] [R4-071660] [R4-071661] R4-071661, "Impact of HNB with controlled output power on macro HSDPA capacity", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #44bis, October 2007. [R4-072004] R4-072004, Huawei, "Performance Evaluation about HNB coexistence with Macro networks", 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-071941] R4-071941, "Simulation results for Home NodeB to Home NodeB downlink co-
20.3 Scenario C
existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working page 106
Title: Femtocell Uplink Co-Channel Interference to the Macrocell NodeB Receiver 3GPP Analysis References: [R4-070969] R4-070969, Home B output power, Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #43bis, R4-070969, June 2007. [R4-070970 [R4-071231] [R4-071578] [R4-071619] [R4-071941] R4-071941, "Simulation results for
20.4 Scenario D
Home NodeB to Home NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080409] [R4-080154]
Title: Femtocell Downlink Interference to Nearby Femtocell UE Receivers 3GPP Analysis References: [R4-071617] R4-071617, HNB and HNB-Macro Propagation Models,
20.5 Scenario E
Qualcomm Europe, 3GPP TSG-RAN Working Group 4 (Radio) meeting #44bis, October 2007. [R4-071618] [R4-080409] [R4-080151] [R4-080149] R4-080149, Ericsson, "Simulation assumptions for the block of flats scenario, 3GPP TSG-RAN Working Group 4 (Radio) meeting #46, February 2008. [R4-080150] R4-081344
Title: Femtocell Uplink Interference to Nearby Femtocell Receivers 3GPP Analysis References: [R4-070971] [R4-071185] [R4-071617] #44bis, October 2007. [R4-071618] [R4-080409] [R4-080152] [R4-080153] R4-071617, HNB and HNB-Macro Propagation Models, Qualcomm Europe, 3GPP TSG-RAN Working Group 4 (Radio) meeting
20.6 Scenario F
page 107
Title: Macrocell Downlink Adjacent Channel Interference to the Femtocell UE Receiver 3GPP Analysis References: [R4-071941] R4-071941, "Simulation results for Home NodeB to Home NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080409] [R4-080149] R4-080149, Ericsson, "Simulation assumptions for the block
20.7 Scenario G
of flats scenario, 3GPP TSG-RAN Working Group 4 (Radio) meeting #46, February 2008. [R4-080150]
Title: Macrocell Uplink Adjacent Channel Interference to the Femtocell Receiver 3GPP Analysis References: [R4-070825] [R4-070971] [R4-071185] [R4-071941] R4-071941,
20.8 Scenario H
"Simulation results for Home NodeB to Home NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080097] [R4-080409]
Title: Femtocell Downlink Adjacent Channel Interference to the Macrocell UE Receiver 3GPP Analysis References: [R4-071211] [R4-071231] [R4-071263] [R4-071540] [R4-071554] [R4-071660] [R4-071661] R4-071661, "Impact of HNB with controlled output power on macro HSDPA
20.9 Scenario I
capacity", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #44bis, October 2007. [R4-072004] R4-072004, Huawei, "Performance Evaluation about HNB coexistence with Macro networks", 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-071941] R4-071941, "Simulation results for Home NodeB to Home NodeB downlink co-
existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-072025] [R4-080409] [R4-080151]
page 108
Title: Femtocell Uplink Adjacent Channel Interference to the Macrocell NodeB Receiver 3GPP Analysis References: [R4-070971] [R4-071185] [R4-071231] [R4-071619] [R4-071941] R4-071941, "Simulation results for Home NodeB to Home NodeB downlink co-existence considering the impact of HNB HS utilization", Ericsson, 3GPP TSG-RAN Working Group 4 (Radio) meeting #45, November 2007. [R4-072004] [R4-080409] [R4-080152]
20.10
Scenario J
Title: Downlink and Uplink Scenarios Modelling Power Control Techniques for Interference Mitigation 3GPP Analysis References: [R4-081344] [R4-081345] [R4-081346]
20.11
Scenarios Section 16
page 109
21 References
[FF08] Femto Forum, Interference Management in UMTS Femtocells, December 2008. [FF09] Femto Forum Working Group 2, Recommended Simulation Parameters 850 MHz, April 2009. [COST231] Commission of the European Communities, Digital Mobile Radio: COST 231 View on the Evolution Towards 3rd Generation Systems, L-2920, Luxembourg, 1989.
[ITU1238] International Telecommunication Union, ITU-R Recommendations P.1238: Propagation data and prediction models for the planning of indoor radiocommunications systems and radio local area networks in the frequency range 900MHz to 100GHz, Geneva, 1997. [ITU1411] International Telecommunication Union, ITU-R Recommendations P.1411-3: Propagation data and prediction methods for the planning of short range outdoor radiocommunication systems and radio local area networks in the frequency range 300 MHz to 100 GHz, Geneva, 2005. [Hol06] H. Holma and A. Toskala, HSDPA/HSUPA for UMTS: High Speed Radio Access for Mobile Communications, J. Wiley & Sons, Ltd, 2006. [Kob92] H. Kobayashi, G. Patrick, Preliminary Building Attenuation Model, NTIA Technical Memorandum 92-155, 1992. [Stav03] Stavrou, S. Saunders, S.R., Factors influencing outdoor to indoor radio wave propagation, Intl Conference on Antennas and Propagation (ICAP), 2003. [Dav97] Davidson, A. and Hill C., Measurement of Building Penetration into Medium Buildings at 900 and 1500 MHz, IEEE Transactions on Vehicular Technology, February 1997. [Kee90] J. M. Keenan, A. J. Motley, Radio coverage in buildings, British Telecom Technology Journal, vol. 8, no. 1, Jan. 1990, pp19-24. [Lai02] J. Laiho, A. Wacker and T. Novosad, Radio Network Planning and Optimization for UMTS, J. Wiley & Sons, Ltd, 2002. [Oku68] Y. Okumura, E. Ohmori, T. Kawano and K. Fukuda, Field strength and its variability in VHF and UHF land mobile radio service, Rev. Electr. Commun. Lab., Vol. No 16, pp825-73, 1968. [Sha88] K. S. Shanmugan and A. M. Breipohl, Random Signals: Detection, Estimation and Data Analysis, J. Wiley & Sons, Ltd, 1988. [Law08] A. Law, Interference Management Evaluation Scenarios, April 2008.
page 110
page 111
page 112
page 113
page 114
This section provides a set of recommended values and path loss models for the interference studies at 850
Table 21-1 lists the simulation parameter values that were used in this paper unless otherwise stated in the text. Table 21-1: Recommended simulation parameters.
Parameter External Wall Loss Window Loss Maximum Macro Node B Tx Power Maximum Micro Node B Tx Power Macro Node B Antenna Gain Macro Node B Feeder/Cable Losses Micro Node B Antenna Gain Micro Antenna Feeder Loss Node B sensitivity Femtocell Noise Figure Macro Node B Loading Femto Loading Downlink/Uplink Channel performance (ie. EbNos & EcNos for various services) UE transmission power range Femtocell Maximum DL powers Value 10dB [COST231] 5dB 43dBm 38dBm 17dBi 3dB 2dBi 1dB Based on reference sensitivity in 3GPP Spec [TS25.104] 8dB (and 12dB) 50% 50% Minimum performance requirements based on 3GPP specs [TS25.101][TS25.104] Based on 3GPP spec [TS25.101] Up to 21dBm. Analysis to cover 10dBm, 15dBm & 21dBm power levels Maximum co-channel DL deadzone created by femto for non-femto UEs [R4070969] Maximum adjacent DL deadzone created by femto for non-femto UEs Height of mobile Height of femto Height of macro basestation Frequency 1.5 m 1m 30 m 850 MHz 60dB for 10dBm Femto DL Tx Power 65dB for 15dBm Femto DL Tx Power 70dB for 21dBm Femto DL Tx Power
page 115
Building dimensions (width by length) Indoor to indoor path loss modelling Indoor to outdoor path loss modelling Outdoor to outdoor path loss modelling Outdoor to indoor path loss modelling
Apartment 10m by 10m House 15 by 15m ITU P.1238 [ITU1238] Okumura-Hata [COST231] + Wall/Window loss (d > 1 km) Okumura-Hata [COST231] (d > 1 km) Okumura-Hata [COST231] + Wall/Window loss (d > 1 km)
Several path loss models are used within the study to calculate the signal attenuation as it propagates within different environments. These have been chosen from the range of models in the public domain that are widely accepted within the industry. They are, therefore, not tuned to a specific environment or set of measurements. The models should, however, be indicative of the realistic range of path loss values that are likely to be encountered in a realistic deployment. The path loss models are described in this section.
Although the Okumura-Hata (OH) model is a fully empirical model, entirely derived from the best fit of measurement data without real physical basis, the model remains widely used and is well-accepted by the mobile cellular community. It is the most widely implemented model and is available as the main model in most radio planning tools.
22.2.1 Okumura-Hata
L = 69.55 + 26.16 log( f ) 13.82 log(hB ) + (44.9 6.55 log(hB )) log(d ) F (hM )
(1.1 log( f ) 0.7) hM (1.56 log( f ) 0.8) medium to small cities F (hM ) = 2 for large cities 3.2 (log(11.75 h M )) 4.97
The parameters in the above expressions stand for:
f : frequency [MHz] hB : base station height above ground level [m] hM : mobile station height above ground [m] d: distance from basestation [km]
page 116
150 MHz < f < 1000MHz 30m hB 200m 1m hM 10m d > 1km
This model predicts path loss between two indoor terminals assuming an aggregate loss through furniture, internal walls and doors represented by a power loss exponent N that depends on the type of building (residential, office, commercial, etc.). Unlike other site-specific models (such as Keenan and Motley 0), this method does not require the knowledge of the number of walls between the two terminals, and therefore offers a simpler implementation.
where:
In the frequency range 900 MHz, P.1238 suggests using the following power loss coefficients N: Residential: Office: Commercial: --33 20
page 117
P.1238 doesnt provide power loss coefficient or floor penetration loss for residential buildings at 900 Mhz, but does say that for the power loss coefficient it is acceptable to use the value given for office buildings. After some discussion among the members of the simulation team it was decided to use a value of 28, which is slightly less than that for office buildings but consistent with measured data. It was also decided by the members of the simulation team that a floor penetration loss factor of 4 dB per floor penetrated would be used, since that is consistent with measured data. For fading, a log-normal distribution is assumed with a standard deviation of 8 dB.
In Section 17 the following simplified path loss models were used: The free-space component for the micro-urban model is given by
page 118
23 Contact Information
The Radio and Physical Layer working group (WG2) of the Femto Forum would be pleased to respond to further queries on the aspects examined in this paper.
Contact details: Email: info@femtoforum.org Web: www.femtoforum.org Postal: The Femto Forum PO Box 23 GL11 5WA UK
page 119