ICT-248891 STP FREEDOM - cordis.europa.eu · [Huawei-ePico] ePico3801 MML Command...

ICT-248891 STP FREEDOM

Femtocell-based network enhancement by interference management and coordination of information for seamless connectivity

D6.2.1

Trial report

Contractual Date of Delivery to the CEC: 31 December 2011

Actual Date of Delivery to the CEC: 16 January 2012

Author(s): Hadi Hariyanto, Retno Wulansari, Toha Nugraha, Hazim Ahmadi (TELK), Julien Stéphan, Yoann Corre, Antoine Cordonnier, Romain Charbonnier (SIR)

Participant(s): TLK, SIR

Workpackage: WP6

Est. person months: 21

Security: Internal

Dissemination Level: PU

Version: a

Total number of pages: 121

Abstract: This document presents the measurement results and analysis as part of task 6A2. It contains interference characterization, backhaul and femtocell mobility. The interference characterization and mobility measurement have been done mainly in BM1 environment; while xDSL measurement as femtocell backhaul in BM3. The report starts with testbed overview, environment and measurement methodology. The trial report is based on measured scenarios, relevant analysis are then detailed to derive the femtocell deployment rules reported in D6.2.2 document. Keyword list: testbed, trial methodology, interference, backhaul, mobility management

ICT-248891 STP Document number: D6.2.1 Title of Deliverable: Trial Report

Freedom_D621TELKa.doc 1



Executive Summary

This deliverable summarizes the trial activities carried out in task 6A2 in frame of WP6. The main focus is on interference characterization, backhaul observation and mobility management in order to derive engineering rules for femtocell deployment. A small-scale WCDMA test bed compliant with business model 1 (BM1) has been designed and a specific measurement campaign was specified in order to address interference characterization. The test-bed is composed of 5 FAPs (Femto-Acces Points) distributed into the two first floors of the TELKOM RDC main building, which is partially covered by an operational 3G urban macro network. It is thus relevant for on-air characterisation of femto-based interference. Measurements were conducted in August and September 2011 in TELKOM RDC premises (Indonesia). Several scenarios have been collected along a common trajectory in order to characterize the impact of interferences generated by a single-FAP or multi-FAPs (up to 4 in the single floor) in closed access mode indoors and outdoors. The measurements are collected by using SCANNER and TRACE mobile receivers as well as Network Management System for FAPs and MBS’s (Macro Base Stations). Impacts of interference are thus evaluated by comparing the statistics got from the different scenarios on network KPIs (Key Performance Indicators), interference power and SINR levels to those obtained in presence of the macro network alone. Main conclusions drawn from this analysis are exploited, together with the 5A2 simulations outcomes, to derive the radio-planning engineering rules reported in Deliverable [FREEDOM-D622].

Backhaul characterization activity is performed in order to observe the properties of xDSL technologies (ADSL, ADSL2, VDSL2) and corporate LAN as femtocell backhaul. The performance of xDSL over distance, transmission delay, pro-cons between interleave vs fast mode have been reported in order to give feedback to technical WPs about backhaul quality model. The femtocell performance over xDSL and corporate LAN are reported. According to the measurement, if mobile network operator and ISP do not have agreement to manage the network it is most likely the femtocell performance will be suffered due to background traffic in xDSL modem using a single PVC. The same situation is experienced by femtocell in enterprise network, the bottleneck may be occurred in the internet link due to competition between femtocell traffic and internet activities from the corporate’ users. In order to maintain femtocell performance we have observed the bandwidth minimum requirement to support mix traffic according to FREEDOM traffic model [FREEDOM-D21]. This minimum bandwidth study is performed in order to address the limitation of xDSL under FTTE (fiber in the exchange). Besides, by knowing the minimum bandwidth both for xDSl and corporate LAN, it will allow ISPs to offer the optimum bandwidth for their customer according to the xDSL penetration rate and affordable price. The discussion of bandwidth requirement for femtocell and QoS management isfurther elaborated in [FREEDOM-D622]. Mobility management is observed in order to understand the impact of handover parameter setting namely signal quality threshold, hysteresis margin, delay timer, and neighbouring cell offset. It will give an indication about the optimum parameter to be used in femtocell deployment.



DISCLAIMER

The work associated with this report has been carried out in accordance with the highest technical standards and the FREEDOM partners have endeavoured to achieve the degree of accuracy and reliability appropriate to the work in question. However since the partners have no control over the use to which the information contained within the report is to be put by any other party, any other such party shall be deemed to have satisfied itself as to the suitability and reliability of the information in relation to any particular use, purpose or application.

Under no circumstances will any of the partners, their servants, employees or agents accept any liability whatsoever arising out of any error or inaccuracy contained in this report (or any further consolidation, summary, publication or dissemination of the information contained within this report) and/or the connected work and disclaim all liability for any loss, damage, expenses, claims or infringement of third party rights.



Table of Contents

1 INTRODUCTION ............................................................................................... 10

2 TEST-BED OVERVIEW .................................................................................... 11

2.1 TEST-BED CONFIGURATION .................................................................................... 11 2.1.1 Selected Environment for Business Model 1 (BM1) ........................................... 12 2.1.2 Selected Environment for Business Model 3 (BM3) ........................................... 14

2.2 TEST EXECUTION TIMELINE ................................................................................... 16 2.3 CONSTRAIN AND RISK .............................................................................................. 16

3 KEY OBSERVATION & METHODOLOGY ................................................. 18

3.1 INTERFERENCE CHARACTERISATION ..................................................................... 18 3.1.1 Key Observations ................................................................................................ 18 3.1.2 Interference Scenario Overview ......................................................................... 19 3.1.3 Measurement Equipment .................................................................................... 24 3.1.4 Measurement Methodology ................................................................................ 26

3.1.4.1 Specification ................................................................................................... 27 3.1.4.2 Preparation ...................................................................................................... 27 3.1.4.3 Installation and Localisation ........................................................................... 27 3.1.4.4 Measurements ................................................................................................. 28 3.1.4.5 Save ................................................................................................................. 29

3.2 BACKHAUL CHARACTERIZATION............................................................................ 29 3.2.1 Key Observations ................................................................................................ 30 3.2.2 Backhaul Observation Scenario ......................................................................... 31

3.2.2.1 xDSL Access Network Characteristics ........................................................... 31 3.2.2.2 xDSL as Femtocell Backhaul .......................................................................... 31 3.2.2.3 Corporate LAN as FAP Backhaul ................................................................... 33

3.2.3 Measurement Tools ............................................................................................. 36 3.3 MOBILITY MANAGEMENT CHARACTERIZATIONS ................................................. 38

3.3.1 Key Observations ................................................................................................ 38 3.3.1.1 Performance Parameters ................................................................................. 40 3.3.1.2 Mobility Scenarios .......................................................................................... 40

3.3.2 Measurement Tools ............................................................................................. 42 3.3.3 Measurement Methodology ................................................................................ 42

3.3.3.1 Specification ................................................................................................... 42 3.3.3.2 Preparation ...................................................................................................... 42 3.3.3.3 Test Setup........................................................................................................ 42 3.3.3.4 Measurement ................................................................................................... 43 3.3.3.5 Data Collection ............................................................................................... 44

4 MEASURED SCENARIOS ................................................................................ 45

4.1 INTERFERENCE CHARACTERISATION ..................................................................... 45 4.1.1 Overview .............................................................. ¡Error! Marcador no definido. 4.1.2 Scenario 0 ........................................................................................................... 50 4.1.3 Scenario A1 – DL FAP → MUE ........................................................................ 51 4.1.4 Scenario A2 – DL MBS → FUE ......................................................................... 52 4.1.5 Scenario A3 – DL FAP → FUE ......................................................................... 52 4.1.6 Scenario B2 – UL MUE → FAP ........................................................................ 54 4.1.7 Scenario B3 – UL FUE → FAP ......................................................................... 56 4.1.8 Generic scenario................................................................................................. 57



4.2 XDSL BACKHAUL CHARACTERISATION ................................................................. 59 4.2.1 xDSL Access Network Characteristics ............................................................... 59 4.2.2 xDSL Scenario 0 ................................................................................................. 60 4.2.3 xDSL Scenario 1 ................................................................................................. 61 4.2.4 xDSL Scenario 2 ................................................................................................. 63

4.3 CORPORATE LAN AS FAP BACKHAUL ................................................................... 64 4.3.1 LAN Scenario 0 ................................................................................................... 64 4.3.2 LAN Scenario 1 ................................................................................................... 65 4.3.3 LAN Scenario 2 ................................................................................................... 66 4.3.4 LAN Scenario 3 ................................................................................................... 66 4.3.5 LAN Scenario 4 ................................................................................................... 68

4.4 MOBILITY MANAGEMENT TEST GUIDELINE .......................................................... 70 4.4.1 Test Overview ..................................................................................................... 70 4.4.2 Scenario FF (FAP –to- FAP) ............................................................................. 70 4.4.3 Scenario FM (FAP –to- MBS) ............................................................................ 71

5 POST PROCESSING & ANALYSIS ................................................................ 72

5.1 INTERFERENCE CHARACTERIZATION ..................................................................... 72 5.1.1 Macro-only Network (scenario 0.1) ................................................................... 72 5.1.2 Single-FAP Deployment (scenario 0.2) .............................................................. 74

5.1.2.1 Indoor Coverage .............................................................................................. 74 5.1.2.2 FAP Coverage Radius ..................................................................................... 77 5.1.2.3 Outdoor Analysis ............................................................................................ 80 5.1.2.4 Throughputs .................................................................................................... 82

5.1.3 Impact of Inter-FAP Distance (Scenarios A3 and B3) ....................................... 82 5.1.4 Impact of Generic Multi-FAPs Deployment (scenarios G1 & G2) .................... 86

5.1.4.1 Indoor Coverage .............................................................................................. 86 5.1.4.2 Outdoor Coverage ........................................................................................... 87 5.1.4.3 Indoor Throughput .......................................................................................... 90

5.1.5 Summary ............................................................................................................. 91 5.2 XDSL AS FEMTOCELL BACKHAUL .......................................................................... 92

5.2.1 xDSL Access Network Characteristics ............................................................... 92 5.2.2 xDSL Network Performance (Scenario 0) .......................................................... 95 5.2.3 Femtocell Bandwidth Requirement (Scenario 1) ................................................ 97 5.2.4 Femtocell Service Performance (Scenario 2) ................................................... 102

5.3 LAN AS FEMTOCELL BACKHAUL .......................................................................... 103 5.3.1 TELKOM RDI LAN Characteristics ................................................................. 103 5.3.2 Femtocell Bandwidth Observation ................................................................... 105 5.3.3 Femtocell Service Performance ........................................................................ 106

5.3.3.1 Voice Performance in the Presence of LAN traffic (Scenario 2) .................. 106 5.3.3.2 Video Conference Performance in the Presence of Internet Traffic (Scenario 3) 107

5.3.4 E2E QoS (DiffServ) Observation ...................................................................... 110 5.3.4.1 DiffServ in Femtocell System ....................................................................... 110 5.3.4.2 DiffServ in Metro Ethernet ........................................................................... 113

5.4 MOBILITY MANAGEMENT REPORTS ..................................................................... 114 5.4.1 Mobility Scenario (FAP to FAP) ...................................................................... 114

5.4.1.1 Scenario FF1 ................................................................................................. 117 5.4.1.2 Scenario FF2 ................................................................................................. 119 5.4.1.3 Scenario FG1 ................................................................................................ 120

6 CONCLUSION .................................................................................................. 123



References & Standards [Becvar10] Z. Becvar, P. Mach, “Adaptive Hysteresis Margin for Handover in

Femtocell Networks”, International Conference on Wireless Mobile Communication, Spain, 2010

[Epitiro08] Epitiro Technologies, Ltd, “Femtocell Deployment Guide: An Operator-focused Strategy for a Successful Femtocell Rollout”, 2008

[FREEDOM-D21] G.Vivier, A.Agustin, J.Vidal, O.Muñoz, S.Barbarossa, L.Pescosolido, M. Omilipo, E. de Marinis, G.Imponente, Z.Becvar, P.Mach, Y.Corre, H. Hariyanto, A.K.Widiawan, L.Simamora, “Scenario, Requirements and First business model analysis”, June 2010, available at www.ict-freedom.eu

[FREEDOM-D41] Future FREEDOM deliverable document: D4.1: “Advanced procedures for handover in femtocells”.

[FREEDOM-D622] Refined Engineering Rules for femto deployment based on trials

[Golden06] P. Golden, H. Dedieu, K. Jacobsen, “Fundamental of xDSL Technologies”, Auerbach Publications, Taylor & Francis Group, New York, 2006

[Inacon06] UMTS Network Optimisation & Trouble Shooting

[Saunders09] Simon R. Saunders, Stuart Carlaw, Andrea Giustina, Ravi Raj Bhat, V. Srinivasa Rao and Rasa Siegberg.”Femtocell Opportunities and Challenges for Business and Technology”, John Wiley & Sons Ltd. 2009.

[Telecoms-Academy] Informa Telecoms & Media. ” Radio Planning and Optimization”, December 2011”

[Huawei-ePico] ePico3801 MML Command Reference(V200R010C00_03)



List of abbreviations & symbols 3G Third Generation 3GPP 3rd Generation Partnership Project ADSL Asynchronous Digital Subscriber Line BM Business Model CAD CPICH

Computer-Aided Design Common Pilot Channel

CSG Closed Subscriber Group CW Continuous Wave dB Decibel DL DBM

Downlink Digital Building Model

DXF Drawing eXchange Format FAP Femto Access Point FAP GW FAP Gateway FC FEXT

Central Frequency Far End Cross Talk

FUE FTTx

Femto User Equipment Fiber to the x (x=exchange, curb/street cabinet, building, home)

GGSN H(e)NB

Gateway GPRS Support Node Home (e)Node B

HSPA IPR JVM

High Speed Packet Access Intelectual Proprietary Right Joint Vendor Meeting

KPI Key Performance Indicators Lat Latitude Long LMA LTE

Longitude Local Mobile Anchor Long Term Evolution

MBS Macro Base Station MHz Mega Hertz MIMO Multiple Input Multiple Output MNO Mobile Network Operator MSAN Multi-Service Access Node MUE Macro User Equipment NEXT Near End Cross Talk NMS Network Monitoring System OFDMA Orthogonal Frequency-Division Multiple Access ONU Optical Network Unit POTS Plain Old Telephone Service PRTG Paessler Router Traffic Grapher QoS Quality of Service PE Provider Edge RRC Radio Resource Controller RSCP Received Signal Code Power RSSI Received Signal Strength Indicator RFI Request for Information SGSN Serving GPRS Support Node SC Scrambling Code SecGW Security Gateway SHDSL Single-pair High-speed DSL



SINR or Ec/Io Signal to Interference plus Noise Ratio SIR SLA

SIRADEL Service Level Agreement

SLG TELK

Service Level Guarantee TELKOM Indonesia

UE User Equipment UARFCN UTRA Absolute Radio Frequency Channel Number UL Uplink UMTS Universal Mobile Telecommunication System UTRA UMTS Terrestrial Radio Access VDSL Very high-bit-rate DSL WiFi Wireless Fidelity WP Work Package



1 INTRODUCTION This document is the second deliverable report of Work Package 6 in the European ICT FREEDOM project. It focuses on trial report. The trial has been conducted using TELKOM 3G HNB testbed in Bandung. Section 2 gives the overall testbed configuration for BM1 and BM3 including timeline for executing the trial activities. Key observation and methodology are detailed in section 3 as a basis to derive measurement scenarios. All measured scenarios are reported in section 4. Section 5 presents the post-processing and analysis of these scenarios. Outcomes of this study are used as a basis of D6.2.2 “Refined engineering rules for femto deployment based on trials”.



2 TEST-BED OVERVIEW

WP6 research activities produce two kinds of testbeds

a. A reduced-scale laboratorial environment consists of standalone WiMAX-based FAP, WiFi Access Point connected to access gateway, router and local mobile anchor (LMA). This testbed is considered as a demonstrator testbed to testify prototypes implementing selected techniques from WP3 (fast scanning, uplink collaborative MIMO) and from WP5 (routing optimization extension to Proxy Mobile IPv6). The prototype description and development status are reported in D6.1 deliverable, therefore beyond the scope of this document. The standalone testbed which contains of FAP prototype and optimized routing for Mobile IPv6 will be demonstrated at the closing of project.

b. A small-scale test bed consists of 3-5 FAPs connected to FAP-GW and 3G core networks. The testbed has been developed in TELKOM RDC premises and TELKOMSEL testbed Indonesia, allowing partners to observe the characteristics of interference, handover, backhaul and basic 3G service performance. This tested will be used as observational environment in order to provide input to business model (WP2) and technical WPs including WP3, WP5 and deliverable D6.2.

The decision of developing 3G-based femtocell testbed is based on the following reasons:

The fact that FAP prototypes been developed are WiMAX based in one hand, and on the other hand TELKOM does not have any commercial Mobile WiMAX networks. Therefore the integration of WiMAX FAP with the WiMAX macrocell & core network is not possible.

According to RFI outcomes as part of 6A1 activity, most vendors will have LTE Femtocell prototypes for trial in 2012 (3 out of 4 vendors), the earliest plan will be in Q4 2011. In addition, most vendors prefer to start developing HeNB in 2600 MHz unless there is significant demand in other spectrum band. Eventhough TELKOM has conducted LTE trial, it used 2.1 GHz band. The fact that LTE FAP will be available after FREEDOM period, it is not possible to have LTE FAP in TELKOM network.

According to RFI & JVM activities conducted in 6A1 and reported in D622 the most mature FAP products are 3G femtocell system. TELKOM has initiated FAP system testbed and conduct trial as part of 6A2 activity. TELKOM can provide test bed which is able to proof most FREEDOM requirement indicated as short term implementation. The testbed will be used in order to refine engineering rule for femtocell deployment (deliverable D6.2.2, Month 24). TELKOM focuses to initiate testbed based on HNB system operating in 2.1 GHz.

This section presents TELKOM testbed configuration for femtocell 3G. It accommodates some scenarios and network configurations for BM1 (corporate environment) and BM3 (residential environment).

2.1 Test-Bed Configuration

TELKOM test bed consists of - HDSPA-based femtocell system (FAPs, FAP Gateway, Provisioning and OAM server) - Metro ethernet and local area network (LAN) to support BM1 scenarios - xDSL system including DSLAM and BRAS to support BM3 scenario - 3G (HSPA and W-CDMA) full system including 3G MBS, MSC, SGSN, GGSN connected to

the global internet. While MSC and SGSN is part of TELKOMSEL testbed, 3G MBS and GGSN are commercial network.

Figure 1 show overall testbed configuration used in 6A2 activities; while Figure 2 shows TELKOM RDC testbed locations in Bandung, Indonesia.



The environment of BM1 for corporate customer is the TELKOM RDC main building which has three floors. In this environment both of RF and backhaul is setup to represent real femtocell implementation in corporate environment. Whereas for the residential environments (BM3), TELKOM uses two-storey building called OASIS. Physically the OASIS building may not fully represent the RF conditions for typical residential houses (BM3) because the building is located in office area. However, it have representative backhaul links for BM3 (xDSL and metro Ethernet backbone), that is used in order to observe xDSL backhaul characteristics and femtocell minimum bandwidth requirement.

Figure 1: Testbed configuration to support corporate and residential scenarios

Figure 2: Testbed buildings and relation with Business Models (BM1 and BM3)

2.1.1 Selected Environment for Business Model 1 (BM1)

The environment for BM1 is TELKOM RDC main building mainly in the first floor as can be seen at Figure 3. The building surroundings is suburban, then it is not urban or dense urban as expected in BM1 scenario. This might have an impact on the characterisation of FAP outdoor leakage, i.e. the



interference from femtos on MUE. Nevertheless, most of the FAPs (i.e. 4/5 FAPs) are located in the first floor in order to partially avoid this drawback and be more adapted to reproduce realistic BM1 scenario. Remark that the last FAP is located in the second floor with approximately the same 2D location that FAP n°1.

Figure 3: BM1 environment – Aerial image (from DigitalGlobe) with superimposed indoor layout of the first floor

Main characteristics of this environment (corporate building) are summarized in Table 1.

76m

63m



Characteristic Description

Category Suburban Corporate

Building height 12m / 3 floors

Floor number First floor

Surface ~2000 m2

Material Type External walls: concrete walls with large windows or glass walls Internal walls: plasterboards until mid-floor height and large windows above Glass doors

Surrounding buildings type Mix of corporate and residential Surrounding buildings density

Corporate buildings: low Residential buildings: medium

Surrounding buildings height

Corporate buildings: 10m (one building at 20m) Residential buildings: 5m

Backhaul LAN + Metro-Ethernet Leased channel

Table 1: Main characteristics of BM1 environment

2.1.2 Selected Environment for Business Model 3 (BM3)

The selected environment for BM3 is in the OASIS building (see Figure 4). It is a two-storeys building where femto systems are installed. In the second floor there is a showcase for home environment, however 3G signal is relatively strong. We already setup the residential environment connected to the FAP GW through xDSL. The surrounding environment is suburban mix between office area and residential. In terms of house density, it will not fully represent typical residential environment such as real estate area, since it is located in office area of TELKOM RDC, however within range 50-100 meters away from OASIS building, there are real estate (up to two floors houses) and university areas which likely make 3G MBS are highly occupied.

Figure 4: Selection of Test Bed buildings for BM3.

We will mainly use this environment to observe xDSL backhaul characteristics in supporting femtocell deployment. Table 2 summarizes the main characteristics of BM3 environment considered in WP6.



Characteristic Description

Category Suburban Corporate

Building height 8m / 2 floors

Floor number 2 floors

Material Type External walls: concrete walls with large windows or glass walls Internal walls: plasterboards until mid-floor height and large windows above Glass doors

Surrounding buildings type Mix of corporate and residential

Surrounding buildings density

Corporate buildings: low

Residential buildings: medium

Surrounding buildings height

Corporate buildings: 10m (one building at 20m)

Residential building: 5m

Backhaul xDSL, Metro Ethernet

Table 2: Main characteristics of BM3 environment.

As illustrated in Figure 5, RSCP power was collected in first floor of OASIS building, with intermediate and very good coverage, while in second floor is mostly very good as shown in Figure 6.

Figure 5: RSCP power collected in first floor of OASIS building.

Figure 6: RSCP power collected in second floor of OASIS building.



2.2 Test Execution Timeline

As part of 6A2 activity TELKOM has initiated HNB testbed. This testbed, initially was developed to support TELKOM technology assessment program; however it has been used to perform several studies in FREEDOM context including interference, backhaul and mobility management issues. Figure 7 is the overall timeline which includes testbed preparation and trial execution.

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

M611 M612 D61

M62 D62

6A1 HW feasibility study and prototyping

step 1 RFI and JVM (HNB)

step 2 JVM with selected vendor H(e)NB

6A2 Integration and proof of concept

step 8 post processing and trial report (D6.2.1)

step 9 deployment guideline (D6.2.2)

step 3 testbed installation and environment preparation

step 4 xDSL characterization

step 5 Interference Measurement

step 6 test guideline and trial preparation progress

step 7 femto backhaul and mobility measurement

Figure 7: Test Bed Preparation and Trial Execution Timeline

TELKOM conducted three trial phases. The first trial (step 4) conducted to derive xDSL characteristics by using TELKOM data and additional trial controlled in the testbed. In this trial, we have characterized the xDSL performance without femtocell in order to derive xDSL backhaul model including throughput versus distance, latency versus xDSL bandwidth profile and some QoS observation. In parallel, TELKOM has prepared the environment for BM1 trial (interference measurement trial, step 5). In this trial, TELKOM and SIRADEL conducted measurement of RF characteristics mainly in corporate environment. The second trial had been collaboratively done by TELKOM, SIRADEL with support from Huawei and TELKOMSEL. The third trial is performed to observe femtocell performance over xDSL and LAN. This trial is supported by Huawei, NEC, Spirent and Cisco. In order to observe femtocell backhaul and mobility issues in WP6, TELKOM has prepared and performed backhaul and mobility trial starting from month 21st to month 24th. Due to interoperability problem between NSN 3G core network, commercial MBS and Huawei femto system, FAP-to-MBS trial cannot be performed in the commercial MBS, therefore TELKOM with the support from Huawei reports FAP-to-MBS observation which is based on trial conducted in the testbed (will be updated as the trial finished).

2.3 Constrain and Risk

In general, testbed development can be done and fits with WP6 objectives. Even though the trial is based on HNB system, TELKOM believes that the research activities to derive deployment rule related to interference characteristics, backhaul and mobility management are representative and achievable. The main constrain in the test bed initiation is that TELKOM should deal with multiple partnerships and comitment between TELKOM - vendors and TELKOM – SIRADEL. TELKOM should manage them carefully and smoothly in order to avoid conflict of interest and IPR issues (backward and/or forward as stated in the consortium agreement). Integration issues between femto systems and 3G core network have taken significant time and effort that has affected the trial execution timeline. TELKOM has considered this as high priority issue, most of integration between femto system and PS core network has been done successfully. However integration between femtocell system in the testbed and commercial MBS cannot be done by mid December 2012, due to unsolved interoperability



issue between Huawei (femto system Bandung city) and NSN (MBS-MSS Bandung and MSS TELKOMSEL testbed in Jakarta). TELKOM considered this task is important and the observation will be delivered and reported in the new version of this document. Overall risk/constrain and resolution can be seen as in Table 3.

Risk Item

Risk Impact Category Resolution

Multiple partnership

The trial is initially only between TELKOM and vendor interest. The synchronization problems between vendor, TELKOM/TELKOMSEL, FREEDOM partner.

Moderate TELKOM coordinates the schedule. The interference measurement can be done successfully.

Integration problem between Femtocell system, MSS testbed and 3G MBS live network

Integration and IOT affected the trial execution for FAP-to-MBS handover.

Moderate Femto-handover observation will be done before March 2012.

Table 3: Risk and resolution.



3 KEY OBSERVATION & METHODOLOGY

3.1 Interference Characterisation

3.1.1 Key Observations

As reported in [FREEDOM-D21], the most critical interference situations occur in closed access mode. They were identified by 3GPP and Femto Forum for UMTS and LTE networks: 1) FAP towards MUE on downlink, which occurs especially when MUE is close to the FAP and

MUE gets poor signal level from the macro network layer. 2) MUE towards FAP on uplink, which occurs especially when MUE transmit power is high (MUE

is at the macro cell edge) and MUE is close to the active FAP. 3) Neighbor FAP to FUE on downlink, which occurs especially when FUE is at the edge of the femto

cell and FUE is close to the neighbor active FAP. 4) FUE to neighbor FAP on uplink, which occurs especially when FUE transmit power is high (FUE

at the femto cell edge) and FUE is close to the neighbor active FAP.

All these interference situations have been characterized by 3GPP and Femto Forum based on simulation case studies. The investigations lead as part of 6A2 activities of the FREEDOM project aims at characterizing them into a testbed composed of several FAPs, with a special focus on the interference likely to be problematic in a dense femto deployment, i.e. interference situations 1, 3 and 4. The interference power levels are growing simply because the average distance between victim and aggressors become smaller, and the number of aggressors is growing. This may generate significant degradation on the macro network layer (interference situation 1) and greatly reduce the benefits expected from FAPs (interference scenarios 3 and 4). Freedom project works on several techniques to mitigate the destructive impact of the interference rising from a large femto deployment in an OFDMA network: Decentralized algorithms for power control and resource allocation;

Coordination methods for dynamic interference management, based on messages on control-plane level.

Cooperation among nodes, based on messages exchanged at the data-plane level.

6A2 trial allows characterizing this interference by measuring its impact on the network interference power and SINR levels. As the available testbed relies on 3G network, these levels are collected from scanning the downlink pilot power (RSCP) and total received power (RSSI), then providing statistics that may be relevant for other cellular network technologies (OFDMA network in particular): statistical distribution of the best-server received power, best-sever FAP received power, interference power, SINR level. Statistics are obtained in several macro+femto scenarios with FAPs in closed access mode. They are compared to reference statistics got from the macro network layer alone. The statistical analysis is completed with some local observations to highlight cases where: FAP deployment allows received power improvement without significant additional interference. FAP deployment generates highly destructive interference levels.

It is necessary to note that this first characterization does not deal with user traffic and dynamic interference. It only analyses the interference power levels on 3G pilot downlink. 6A2 trial also provides interference characteristics related to the 3G technology: analysis of trace measurements (based on trace mobile devices) gives statistics on network KPIs (key performance indicators) such as downlink and uplink throughput, call failure, call drop and handover. These trace measurements are realized in the same scenarios than pilot power scanning, and are accompanied by a monitoring on the user traffic within the macro and femto cells. All collected data are used in a multi-

76m



parameter analysis, where the KPIs evolutions can be correlated to the evolution of the useful signal power, interference signal power and cell load. Worse case in terms of interference occurs when user traffic is high into the macro network layer and/or FAP network layer. Besides, the femto deployment is expected to permit an increase in the data traffic. Thus some of scenarios are realized with high user traffic. This part of 6A2 trial generates very relevant results for femto deployment into 3G network, but might not be generalized to OFDMA. Other issues related to large-scale femto deployment are addressed in FREEDOM WP4: handover procedure must include femto specificities (in terms of priority and speed); and scanning techniques must cope with a much larger number of detected cells. These issues are not strictly speaking linked to the interference characterization, but measurements are also exploited to get statistics on the cell overlapping, cell size and handover zones within the dense FAP area and neighbor streets. Note that same kind of results is also analyzed by simulation in 5A2 activity for different environments. Finally, the human activity is expected to have a strong impact on the variations of the signal and interference powers, then on the link performance, resource allocation and handover, even when the user terminal is static. Characterization of its impact is addressed based on simulation case studies in FREEDOM WP3 but not analyzed within 6A2 trial as it would have been to complex and time consuming. Nevertheless, to avoid any bias of the analysis due to an uncontrolled human activity, measurements have been mostly collected out of working hours. Trial scenarios, measurement equipment and methodology attached to all the above topics are described respectively in sections 5.1 3.1.3 and 3.1.4 of this document.

3.1.2 Interference Scenario Overview

As aforementioned, the testbed air interface is not fully appropriate to BM3 as located into a corporate environment instead of residential environment. Thereby, it appeared relevant to limit the interference characterisation to scenario BM1 and thus permit a deeper investigation of this particular BM scenario. BM1 environment is composed of several FAPs distributed into a corporate multi-floors building covered by an operational 3G urban network. It is thus relevant for on-air characterisation of femto-based interference. Figure 8 illustrates the generic scenario that has been collected during the interference characterization measurements. The measurement equipment is moving along a pre-defined trajectory within the FAP area and beyond, then allowing investigating the impact of the various FAP properties given inTable 4. The measurements into this generic scenario are then compared to reference measurements in presence of macro BS only (see scenario 0.1 in section 5.1). The objective is to characterise the impact of the femto deployment through the evolution of following metrics: Received power distribution (based on downlink UE pilot power or RSCP). SINR distribution (based on downlink Ec/I0). Inter-cell interference distribution (based on downlink RSSI). QoS statistics (call failure, call drop) for MUE or FUE. Statistics on the number of neighbours. Handover statistics (e.g. average distance between two handover).



Figure 8: Generic interference scenario –BM1 environment

FAP property under test Range

FAP distribution

Medium-density distribution (less than one FAP over 200m²)High-density distribution (about one FAP over 100m²)Within two different floors (to especially investigate the impact of several FAPs located close to the window apertures at different floors)

FAP locations

Mix between FAPs located in good macro-coverage and poor macro-coverage Mix between FAPs located on desktop and above mean furniture

Access mode

Closed with FUE subscribing to only one FAP Closed with FUE subscribing to all FAPs (configuration that may occur in a corporate unit that deploys several FAPs)

Transmit power Without power control (to characterize interference in the worse case). A constant power of 10dBm is set on all FAPs.

Traffic No traffic, except from the trace mobile High traffic generated by high data rate communications

Service requested by the UE under test

High throughput data transfer (e.g. HDTV streaming) (smaller service coverage)

Human activity (generating time-variant obstruction on radio paths)

Highly limited and controlled human activity (to avoid bias of the measurement

Table 4: FAP properties under test

As illustrated in Figure 8, the measurement area is limited to few rooms and offices in the first floor of the building and outdoor areas in close vicinity. Typical pictures of this environment are reported in Figure 9.

70m



The measurement area is mainly covered by three MBS (and notably by the MBS with SC258, i.e. the nearest). Main parameters of these MBS are summarized in Table 5. Remark that the macro network uses two frequencies: 10663 for the W-CDMA technology and 10638 for the HSDPA technology.

Figure 9: Typical pictures of the measurement environment



Figure 10: MBS – Location

Parameters GEGERKALONG KARANGSETRA SETRASARI

Frequency (DL) 10663, 10638 (HSDPA) 10663, 10638 (HSDPA) 10663, 10638 (HSDPA)

Scrambling Codes 258 183 344

Location Long: 107.589

Lat: -6.8702

Long: 107.5882

Lat: -6.8799

Long: 107.5815

Lat: -6.8776

Antenna height 31, tower 40 28, tower 40 28, tower 40

Transmitted power 20 W, 20 W, 20 W,

Antenna reference Katherine K742215 Katherine K742215 Katherine K742215

Antenna gain 18 dBi 18 dBi 18 dBi

Antenna tilt 30 , mec 2 20 , mec 0 20 , mec 0

Antenna azimuth 190 30 40

Antenna horizontal beamwidth

64 64 64

Antenna vertical beamwidth

65 65 65

Table 5: MBS – main parameters

Two different kinds of FAP are used: FAPs n°1, 2 and 3 are Huawei ePico whereas the FAPs n°4 and 5 are Huawei UAP (See Figure 11). Because of the measurement needs, all FAPs were set at a common frequency (in order to evaluate impact of interferences in the worst case) and at a fixed transmit power (i.e. 10dBm to keep a homogeneity on the FAP parameters). Remark that; with the equipments installed on the testbed, only handover between ePico FAPs and/or between a FAP

400m

BM1 environment



(whatever the model) and a MBS is possible. Handover between an ePico FAP and an UAP FAP or between a MBS and a FAP are not possible. This technical limitation cause impacts on the measured scenarios (see section 3.1.4.4).

Huawei ePico

Huawei UaP

Figure 11: Femto Access Points – Overview.

Parameters Value

Frequency (DL/UL) DL: 2110 to 2170MHz

UL: 1920 to 1980MHz

Transmitted power Max: 20dBm for corporate FAPs; 10dBm for residential FAPs

Power were set at 10dBm for all FAPs during the measurements

Number of carriers One carrier per cell (5 MHz)

Simultaneous User (Voice)

Up to 16 for corporate FAPs; up to 4 for residential FAPs

Handover Between FAPs only

Access method Open and Closed

Table 6: FAPs – main parameters

Analysis of these measurements is not only realised on global statistics but also on local analysis based on the comparison between co-localised measurements in different scenarios. As the measurements are carried out along a pre-defined trajectory inside the building but also in the streets around the building (see Figure 8), the impact of the FAP deployment is characterized: Within the FAP areas. At the boundary of FAP areas. In overlapping areas between two FAPs.



In streets adjacent to FAP areas. Note that the same trajectory has been repeated several times for most of the scenarios with a fixed set of FAP properties. The objective is to cope with measurement fluctuations that result from the network traffic irregularities. It has been also precisely repeated from one set of FAP properties to another in order to get fair comparison. The measurements are conducted by using two different kinds of equipment: Scanner equipment, which scans the downlink channel without establishing any communication. Trace mobile (used as FUE or MUE), which collects QoS metrics related to an established

communication. The basic methodology is as follows: The scanner collects characteristics of the downlink pilot channel for a fine characterisation of the

power and interference levels, cell limits and handover zones; Whereas the trace mobile collects QoS metrics function of the FAP access method, traffic from

other-users and service demanded by the user under test.

Obviously, this generic scenario is complex and involves many FAP-UE links and many interference links: From MBS to FUE DL. From neighbour-FAP to FUE DL. From FAP to MUE DL. From MUE to FAP UL. From FUE to neighbour-FAP UL. It is considered as the target scenario into our analysis process. Its characterisation has been greatly simplified by analysis of simpler basic scenarios dedicated to the investigation of a particular set of FAP properties. Each of them may be considered as a brick of the generic scenario and supports a part of the analysis. Scenario 0: Power levels from MBS and each individual FAP. Scenario A1: DL interference from FAP to MUE. Scenario A2: DL interference from MBS to FUE. Scenario A3: DL interference from FAP to FUE’s attached to a neighbour FAP. Scenario B2: UL interference from MUE to neighbour FAP. Scenario B3: UL interference from FUE to neighbour FAP. Section 5.1 gives detailed scenario definition.

3.1.3 Measurement Equipment

As aforementioned, two measurement solutions (receiver plus software) are used at the receiver side: SCANNER based solution TRACE mobile based solution

TRACE mobile is a Sony Ericsson W995 handset and SCANNER is a PCTEL EX scanner as illustrated in Figure 12. Both are monitored by TEMS software (resp. TEMS investigations 10 and 9).



(a) (b)Figure 12: Receiver used: Trace mobile (a) and Scanner (b).

Both measurement solutions (SCANNER and TRACE mobile) provide complementary information useful for interference characterization. TRACE measurements provide an 3G-based interference characterization by getting statistics on network KPIs (throughput, call failure, call drop, handover) whereas SCANNER provide accurate statistics on BS received-power, interference powers, SINR levels, cells size and cells overlapping. Remark that TRACE mobile can also scan 3G pilot channels but SCANNER is faster, more flexible, more accurate and allows the decoding of greater cells number simultaneously. Moreover, utilization of a SCANNER allowed us to use an external antenna (i.e. mobile antenna PROCOM MU 901/1801/UMTS-LX) for which we have a fine characterisation of the radiation pattern. This antenna is connected to the SCANNER with an extra-low loss coaxial cable (i.e. 1dB). All these equipments are installed on two different trolleys (See Figure 13).

Figure 13: Description of trolleys

(1) Trolley n°1: measurement with TRACE mobile; (2) Trolley n°2: measurement with SCANNER (2a) and TRACE mobile (2b)

Remark that the trolley n°1 is only used during the first day of measurements. Height for the antenna (SCANNER or TRACE mobile) in Rx side is respectively 1.40m when trolley n°1 is used and 1.35m when trolley n°2 is used.



Measurement device Available metric Comments

Scanner: all metrics defined for all scanned couples UARFCN/SC

UARFCN Frequency SC Scrambling code

Aggregated Ec Associated to RSCP Aggregated Ec/Io

Io Associated to RSSI

Trace mobile

UTRA carrier RSSI Ec/No Associated to Ec/Io

SC Scrambling code UARFCN Frequency

RSCP TrCh Throughput DL TrCh Throughput DL

NMS FAP DL Throughput

Meanwhile, no precise correspondence with

Trace mobile DL Throughput

EcNo For comparison only RSCP For comparison only

NMS MBS Total CS traffic

Concerns only voice throughput

RRC setup attempts Represents network

load Table 7: Exploited parameters at UE.

NMS (Network Management System) are also used in the transmitter side to characterize interference impact on UL and control traffic load respectively on FAP and MBS. Besides, Huawei USB 3G Modems E398 are used to emulate controlled traffic on FAPs (requirements for some scenarios, see section 5.1). Table 7 summarizes all metrics collected during the interference characterization measurements.

3.1.4 Measurement Methodology

Figure 14 shows the overall process followed during the measurement campaign.



Figure 14: Measurement overall process.

3.1.4.1 Specification

Main specifications are summarized in section 3.1.3 of this document. Next sub-sections describe the methodology regarding the remaining of the process, from preparation to reporting.

3.1.4.2 Preparation

Preparation includes: Definition of the agenda and synchronisation between partners. Selection of the measurement equipment: scanner and trace mobile. Getting datasheets on measurement equipment and FAP equipment. Selection of the best candidate FAPs location and UE trajectory (i.e. drive-test) based on MUE

location and detailed building drawings. Definition of a candidate UE trajectory. Logistics (based on equipment checklist, which consists in taking inventory of the equipment

needed). Audit of the testbed: characterisation of testbed equipment; correction of the building architect

plan; etc. This preparation was done in collaboration by both partners. Definition of FAPs location and UE trajectories was notably helped by the characterization of the MBS coverage through scanning of the downlink pilot power (RSCP) in indoor and outdoor.

3.1.4.3 Installation and Localisation

This installation and localisation step occurs during the measurement stage. It consists in: Setting and geo-locating the FAPs. Geo-locating the indoor UE trajectory. Geo-locating the outdoor UE trajectory. Remark that a particular attention was given to finely set and locate both the FAPs and the UE trajectories as it is major requirements for: Comparison between the different scenarios. Comparison between measurements and simulations.



Estimate of FAP-UE distances. The FAP locations were characterised by distances to geo-located objects represented in the digital building plan, e.g. walls, window edges or pylons. The location of the measurement equipment following the UE trajectory in indoor was done by following the methodology hereafter:

1) Define the trajectory; 2) Localise and mark on ground all the turnings along the trajectory; 3) Report these landmarks on the digital floor plan; 4) Import the digital floor plan with landmarks in the measurement acquisition software (e.g.

TEMS acquisition); 5) Both, physic and numeric landmarks are then used during the measurement:

a. The operator moves on a first “physical” landmark; b. Point it as the departure position on digital floor plan (by clicking on the equivalent

“digital” landmark); c. Move toward a second “physical” landmark with velocity the most constant as

possible, stop on it and point the route track on the digital floor plan (by clicking on the equivalent second “digital” landmark);

d. Repeat the same process for each portion of the trajectory; 6) All UE locations between two pointed positions are interpolated (by the measurement

equipment software, or from post-processing operations). Remark that the outdoor UE trajectory was measured either by collecting GPS coordinates and time or by using same process than in indoor. Remark also that, if required, localisations of each portion of the trajectory have been corrected by post-processing afterwards (see section 5.1).

3.1.4.4 Measurements

During the measurement campaign, a specific importance was given to first days of the measurements, as it generally permits constructive adjustment of the methodology. These days were divided in three phases:

Presentation and Overview: Come to grips with measurement equipments and specific methodology (audit, installation and localisation);

First Measurements: Collect the first data; Post-processing and Analysis: Post-processing and analysis on the first data collected during

the day is recommended. The goal of this last phase is to check the accuracy and the relevance of the data collected. Several outcomes and observations were extracted from these preliminary analyses. A first observation concerns the macro network coverage and by the way the frequency selection for the measurements. As aforementioned, the macro network in Bandung uses two carriers in downlink: 10663 and 10638. The second carrier is used to transmit HSDPA whereas the first carrier is used to transmit W-CDMA but also includes the PS setup R99. Despite MBS properties are identical for both signals, it is observed that the W-CDMA coverage is better than the HSDPA coverage thanks to a lower interference level (see Figure 15).



Figure 15: Coverage by HSDPA and W-CDMA technology along the measured trajectories in indoor and outdoor

The W-CDMA carrier is selected for the FAP deployment in order to avoid handover between FAP on HSDPA carrier and macro W-CDMA carrier. Remark that a third and free carrier (i.e. 10688) has been used scenarios that evaluate inter-FAP interference. An unexpected behavior of the TRACE mobile is observed. As the TRACE mobile supports both the W-CDMA and HSDPA technologies, it has been locked on the W-CDMA carrier in order to measure FAP to MUE interference in closed-access mode scenarios. In spite of this precaution, we observed that the TRACE mobile initiates a handover procedure. This procedure always fails, but prevent from collecting data along portions of time. Thereby, only few TRACE data has been collected in high-interference areas, i.e. basically near the active FAP. No bypass was found. Thus measurements have necessarily been collected in that way. But their interest is reduced. As aforementioned, handover between MBS and FAP is not possible. The characterization of the MBS to FUE interference by TRACE measurements is restricted to the seamless FAP coverage area. The amount of data for this scenario will be limited. In addition, human crowd activity is observed to be very dense during working hours. Thereby, it has been decided to collect most of the measurements out of working hours in order to avoid uncontrolled fading on measured radio links but also uncontrolled MUE interference levels in FAP vicinity.

3.1.4.5 Save

All measurements are saved on two hard disks during the measurement campaign, with expressive filenames, and with all comments necessary to keep trace of the scenario conditions.

3.2 Backhaul Characterization

In terms of backhaul, a FAP is designed to use a broadband IP backhaul which is usually a shared IP connection with other broadband users in residential and office areas. It is contrary to MBS backhaul which uses a dedicated leased line in order to guarantee non-blocking configuration. This is not the case for femtocell, over-provisioning is hard to achieve due to physical link limitation, affordability of broadband service in the market and the existence of background traffic. As a result, femtocell performance depends not only on radio channel quality condition but also on under-lined backhaul quality. The existence of bottleneck or congestion in the backhaul link will affect overall femtocell performance.




The quality of broadband IP connection highly depends on the following issues [Epitiro08]: Allocated / available bandwidth per ISP plan (Mbps); Additional IP-based activities in the home/office; ISP traffic management policies (QoS, priority traffic, etc) according to SLA between ISPs,

customer and mobile network operators; Network load by time-of-day; Efficiency of routing from the subscriber premises to the core network.

One of 6A2 objective is to address backhaul types which can represent last mile-connection for femtocell in BM1 and BM3. Except routing efficiency and network load, femtocell performance will be observed with related to the above issues. The objectives of backhaul characterization in 6A2 activity are

a. to observe xDSL access network characteristics in order to derive backhaul quality model for BM3 environment such as performance over distance, delay over different bandwidth profiles and peak data rate supported by various xDSL/FTTx configuration (FTTE, FTTC, FTTB),

b. to study minimum bandwidth requirement of xDSL link to support a FAP deployment considering mix services, simultaneous calls/sessions supported from a single FAP. This study is important to set minimum backhaul bandwidth required in case only limited backhaul bandwidth available e.g. xDSL where DSLAM is located in the local exchange (FTTE),

c. to observe femtocell E2E QoS performance over various backhaul links including xDSL (BM3) and corporate LAN (BM1), considering background traffic from xDSL/LAN users where applicable.

Table 8 summarized the key observations for individual backhaul type.

Key Observations Backhaul Types

Corporate LAN xDSL

Environment BM1 BM3

Access Network Characteristics It is based on ethernet access, the characteristics is well known

Yes xDSL transmission performance to derive backhaul quality model for BM3

Mimum Bandwidth of link to support a FAP deployment

Yes Considering mix service, simultaneous users

Yes Considering mix services, simultaneous users

Femto Service Performance Real background traffic in Intranet LAN and Internet link

Non SLA and SLA Network

Non SLA Network in access , background traffic in a xDSL link (from modem to DSLAM)

Table 8: Summary of backhaul key observations

As performance indicator, we observe delay, throughput, jitter and packet loss experienced by a 3G packet-based (PS) services in HSDPA terminology. According to [Saunders09] and [Epitiro08], most femtocell technologies provide good quality voice calls and sufficient support to data services when the broadband IP link provides a minimum performance of:

1) Less than 150 ms round-trip delay (more than 200 ms will not be practical for two ways conversation);

2) Less than 40 ms jitter;



3) A general packet loss of 3% or less is acceptable; however, packet loss is typically “bursty” by nature, and, as such, average rates below 0.25% should be maintained;

4) At least 1 Mbps in downlink, i.e. from the broadband IP provider network to the FAP GW; 5) At least 256 kbps in uplink, i.e. from the FAP GW to the broadband IP provider network.

We will use minimum performances point 1,2,3 as our reference. For items 4 and 5, we will observe them in the trial and provide optimum recommendation for bandwidth requirement associated with mix-services being used.

3.2.2 Backhaul Observation Scenario

As can be seen from Figure 1, TELKOM testbed consists of several FAPs connected through SecGW and other femtocell network element through various access technologies. Within OASIS building, Femto systems (SecGW, FAP management, NMS) were installed and connected through xDSL and corporate LAN in the access part. While in the core network part, FAP GW is connected to TELKOMSEL 3G core network using metro-ethernet which both bandwidth and QoS are maintained according to carrier grade SLG (100 MB, layer 2 connection). In the following subchapter, individual backhaul observation will be explained further.

3.2.2.1 xDSL Access Network Characteristics

In order to derive backhaul quality model for xDSL as femtocell backhaul, we will report TELKOM existing study result on TELKOM’s copper cable performance in supporting several xDSL technologies including ADSL2, ADSL2+, and VDSL2. Even though the study initially performed to study IPTV implementation, we assumed the bandwidth requirement for the service is relevant also to support femtocell implementation. The study of xDSL performance could not be separated with FTTx technologies implementation, since both are complementary to each other. Performances of transmission technologies over copper have been evaluated for the following reference architectures: 1) FTTE (DSLAM or MSAN in Exchange) which use technology: ADSL2/2+ technology as the last

mile access 2) FTTCab (total replacement with MSAN at cabinet) which uses technology: ADSL2/2+, VDSL2

(profile 8b, same Tx level as ADSL2/2+) 3) FTTB (ONU or MSAN at building) which uses technology: VDSL2 (profiles 17a and 30a)

The studies are conducted by TELKOM under Telecom Italy supervision. The overall scenario is defined in Section 4.2.1.

3.2.2.2 xDSL as Femtocell Backhaul

According to [FREEDOM-D21] document (scenario, requirements and first business model analysis), for residential customer, femtocell will use FTTx/xDSL backhaul. 6A2 will mainly focus on xDSL characterization, while FTTx is used as complement to the xDSL. This is done for the reason that xDSL usually couple with FTTx deployment which follows FTTE, FTTC or FTB configuration as been discussed in the previous chapter. Femtocell deployment uses FTTH is assumed to be more reliable, higher throughput and symmetrical between downstream/upstream, therefore it will not be observed in the trial and not be presented in this document. Furthermore the penetration rate of xDSL deployment with FTTE configuration is higher compared to FTTB and FTTH. Due to the fact the xDSL performance relies on copper cable quality and length (attenuation as function of cable length) it is high likely that the bottleneck will exist in FTTE configuration compared to FTTB/FTTH.



Figure 16: Femtocell Testbed Configuration for Residential Scenario

Figure 16 shows xDSL and femto system installed in TELKOM RDC. ADSL2/ADSL2+ technology is used. The DSLAM node connected to metro-ethernet located in TELKOM OASIS building. The SecGW and FAP GW is connected to metro-ethernet (ME) backbone in order to connect to SGSN in 3G core network side (through Iu-PS interface). Iu-CS connection (FAP GW to MSC in TELKOMSEL Jakarta) is also available for CS voice call. The connection between xDSL and Femto System and between Femto System and SGSN are considered as controlled environment. Since GGSN is a commercial network, the network load may affect the femtocell performance, especially during busy hours. Another drawback, it is impossible to put measurement tool in the GGSN side, so that the measurement tool or application server is put in internet cloud. In positive way, the femtocell trial performed in this activity, will reflect the real condition. Below are the scenarios to be considered in the femtocell over xDSL measurement campaign: Scenario 0: check initial E2E network performance, without FAP, Scenario 1: observe minimum bandwidth requirement of a single FAP considering mix services,

various UE types and 4 simultaneous users access Scenario2: verify femtocell performance in the existence of background traffic in xDSL modem. Section 4.2 gives detailed scenario definition. In order to derive bandwith requirements and femtocell performance, we define mix traffic composition according to [FREEDOM-D21] with some modification. Besides HTTP, FTP and voice, we use video streaming as the forth traffic type. Individual application content is defined according to the survey result conducted by TELKOM. There are several internet content accessed by subscribers of two xDSL providers in Indonesia. As can be seen from Figure 17; facebook.com, detik.com, youtube.com, 4shared.com are among the contents which very popular in Indonesia. Among top 10 internet content we choose detik.com/facebook.com to represent HTTP traffic, youtube.com for streaming and 4shared.com for file download. Traffic mix definition being used in xDSL and LAN observation is shown in Table 9.



xDSL Provider A xDSL Provider B

Figure 17: Top 10 most visited internet content by subscriber of two ISPs in Indonesia

Traffic Mix

Traffic Source WWW File transfer Video

streaming Voice Notes

Smartphone mix 4 CS

Call Real - - - AMR scenario 1

Smartphone mix all traffic

Real m.detik.com 4shared.com

(5 MB) youtube AMR scenario 1

Tablet PC mix all traffic

Real detik.com &

facebook.com 4shared.com

(13 MB) youtube -

scenario 1,2

PC Background

Real Detik.com FTP up to 40 MB youtube - Scenario 2

Mix Traffic Generated and

Real detik.com &

facebook.com 4shared.com

(13 MB) Video

conference AMR

scenario 1, 2

Table 9: Summary of traffic mix for femtocell performance using xDSL as the backhaul

3.2.2.3 Corporate LAN as FAP Backhaul

Femtocell observation in TELKOM’s corporate LAN can be considered as a field trial, therefore some aspects of the environment are uncontrollable, for example background traffic from LAN caused by intranet users along a day and day of week. It is impractical to control user behavior, network topology, network policy, etc. While interference measurement in BM1 environment observed radio behavior and its fluctuation, in this activity we will focus on the impact of intranet traffic fluctuation to the femtocell performance. In order to achieve this objective, we limit the impact of radio channel fluctuation due to interference and mobility. In this case, FUEs will access the FAPs in such away; the FAPs’s signal quality is very good and stable.



Figure 18 shows femtocell systems and corporate LAN installed in TELKOM RDC. The femto system and 3G core network are connected using carrier class metro-ethernet as in xDSL+Femtocell performance observation. In the access network, several FAPs will be installed in TELKOM RDC’s LAN. The FAPs will be located in the workspace area, so that pre-registered users can access 3G service using close access mode.

Figure 18: Intranet + Femtocell trial configurations

TELKOM RDC is inhabited by about 150 employees separated into six different departments. Each department is assigned different network (dedicated gateway 10.14.x.1 and vlan). IP and VLAN assignment are based on department rather than location or floor where the employees working or moving around. Therefore in every switch access there may be different vlans are active according to the people online activities. TELKOM RDC’s LAN provides intranet services to its employees such as email, knowledge management online, project management online, information portal, various paper-less office applications, IP Phone, video streaming and file sharing. In addition, the employees from different R&D department may have its own client-server application for research purpose. Therefore it is very difficult to identify, model and monitor individual traffic traverse in the network. For the reason that the LAN users behaviour cannot also be controlled, we will only monitor the utilization of network and the amount traffic traversed in a certain node (switch/router/gateway) especially in department or research & development of infrastructure (RDI). Since the measurement activity must have limited impact to the live network, TELKOM can only perform trial in the TELKOM RDI department which can be monitored and considered safe for the measurement. For the purpose of network monitoring, switch access and 3rd floor GW are monitored. By monitoring 3rd floor GW, both traffic from WiFi and wired LAN can be obtained using PRTG IP network monitor. In terms of leased line, TELKOM RDC has different connection i.e 512 kbps internet link for VPN access, 2 Mbps for internet proxy and several direct internet link for research purposes. To limit the impact to the live network, we use RDI internet link which used for internet access. This link can be considered as un-controlled link, where all internet traffic from users connected to switch access and WiFi access point will merge into this link. As a result the femtocell traffic may be affected by internet traffic in the switch access. The femto traffic will also be affected by fluctuation in ISP since the RDI link may be shared with other TELKOM users in the backbone. Unfortunately it is impossible to monitor fluctuation in the backbone. As a result, we can only see the impact of RDI leased line performance to the femtocell. We have conducted a survey of TELKOM RDI LAN users to get their behaviour in accessing intranet and internet services.



Figure 19: Percentage of Terminal and Network Access Point Type

Figure 19 shows that 63% of users connected to the network using laptop, and most of them use WiFi as the access point. In terms of application as shown in Figure 20, the top three most accessed applications are email, social media and intranet portal. 36% - 32 % - 21 %

Figure 20: Proportion of Intranet/Intranet Application Types

Figure 21: Proportion users accessing intranet portal and internet streaming



Intranet portal is a web-based office application which includes web-mail, knowledge management, news, collaboration tools, human resource database, e-learning, project management online, etc. Half of employees access the portal more than 3 times a day. There are about one-third of users never access the portal. They are mostly outsourcing employees, research assistants who are temporary employees. According to Figure 21, streaming media is accessed by 7% of users with frequency once a day (67%).

The scenarios considered in the measurement campaign: Scenario 1: observe a FAP bandwidth consumption and its performance without background

traffic. The FAP will be used to access mix traffic from 1-4 FUEs Scenario 2: observe a FAP performance with background traffic from LAN users. Scenario 3: observe a FAP performance due to corporate leased line activity and compared with

the one without background traffic Scenario 4: observe DiffServ implementation in the femtocell, switch access and the metro

Ethernet. Section 4.3 gives detailed scenario definition.

3.2.3 Measurement Tools

In order to observe femtocell performance for BM1 and BM3 scenario, the following measurement tools are used:

Smartbit 600B from Spirent Avalanche (STC Avalanche from Spirent Test Center) IxChariot from Ixia CISCO Telepresence

SmartBits is used to check initial end-to-end network performance prior to attaching any FAPs, by sending layer 2 frames; delay, packet loss, throughput of the network can be recorded. Figure 22 shows the use of SmartBits in the measurement campaign.

Figure 22: an example of SmartBists measurement setup to capture initial xDSL network performance

STC Avalanche is a high-performance traffic generation and analysis application that can simulate real-world traffic scenarios at the transport and application layers. It will use to observe femtocell performance over xDSL network. The tool will be used to verify the Layer 4-7 performance of packet switch 3G services over femtocell in a controlled lab environment. Each port hosts STC Avalanche full TCP/IP protocol stack and can emulate a large number of users, creating thousands of sessions from a single test port. Using multiple ports, a single test configuration



can scale to millions of sessions, enabling the testing of large content aware devices and networks. In femtocell measurement context, STC Avalanche will emulate application servers which put it in internet cloud. In the access side, STC Avalanche will also generate traffic models for 3G routers to represents FUEs. Figure 23 shows an example of STC Avalanche measurement setup to observe femtocell performance accessed by a FUE with the presence of background traffic from an xDSL user.

Figure 23: An example of STC Avalanche measurement setup for femtocell+xDSL

observation

CISCO Telepresence and STC are collaboratively used to observe femtocell performance over xDSL, where H264 codec is invited as one of traffic type. By utilizing performance report of the video conference, we are able to get service performance metrics for video converencing as reference. Another tool we will use in the backhaul observation is IxChariot. The tool is a traffic generation and decision support tool that enables test engineer to emulate real-world application data without the need to install and maintain extensive client/server networks. IxChariot provides thorough performance assessment and device testing by emulating hundreds of protocols and applications across thousands of network endpoints (up to 10 endpoints currently available). IxChariot provides the ability to confidently predict the expected performance characteristics of any application running on wired and wireless networks. In the femtocell contects, the end points can be embedded in up to 10 commercial 3G phone (with Androids OS). The end points can also be put in laptop connected to 3G modems. Figure 24 shows an example of Ixchariot measurement setup. We use this tool in corporate LAN measurement. Since we use GGSN commercial network, the implementation of port isolation between 3G users and between application server – 3G users, prohibit testers to observer two ways traffic and delay between two 3G users. Because this policy as well the measurement can be done to observe traffic in downlink direction (only between application server in internet cloud and 3G FUE).

Intranet

SecGW 3G CN

Ixchariot Console

Ixchariot Endpoin1

Ixchariot Endpoint2FAP

Figure 24: IxChariot measurement setup for femtocell+intranet observation

Femto system is supported by NMS which has monitoring capabilities. This function will be used in observing individual FAPs status and events data. It will also be used to record throughput generated



from individual FUEs under tracking. In the measurement of femtocell+xDSL performance, we first record characteristics of 3G packet switch services generated by FUEs following scenario 1. The bandwidth occupied by smartphone or tablet PC (iPad) is recorded and analyzed in order to configure traffic model being generated by STC Avalanched as close as the real traffic. In case femtocell+LAN performance measurement, we uses real time IP network monitoring (i.e. bit meter and PRTG) in order to monitor intranet/internet traffics traverse through a specific switch/router/gateway interface. This data will be used to analyse the impact of network fluctuation to the femtocell performance either monitored via measurement tools (PRTG) and NMS.

3.3 Mobility Management Characterizations

Mobility management measurements will be performed in two aspects, cell reselection and handover. These measurement activities will be separated into two phases. In the first phase, TELKOM will conduct mobility measurements between FAPs for both cell reselection and handover on BM1 Scenario. In the second phase mobility measurement between FAP and MBS conducted in the lab (BM3). In this lab trial TELKOM mainly aims to perform FAP to MBS handover.


The 3GPP project has considered several cases of mobility for the FAPs. Based on [FREEDOM-D21], TELKOM adopted two mobility cases. Measurement scenarios would be configured into these situations;

FAP out of coverage of any MBSs in order to measure mobility between FAPs. That is to avoid interference influenced by MBS signal or co-channel interference. In order to avoid interference from MBSs which use UARFCN 10663 and 10638, FAP is assigned UARFCN 10688. By using BM1 environment 3 FAPs were used the same placement as on interference measurement. TELKOM used FAP 1 (SC 300), FAP 2 (SC 219), and FAP 3 (SC 375). Mobility measurement between FAP was conducted on Corridor 1 and Corridor 2 (C1 and C2).

FAP is located within MBS coverage. FAP to MBS mobility measurement was conducted in the lab. FAP were configured to use the same frequency as MBS.

The objective of mobility measurement is to know the optimum set of handover parameters in order to avoid redundant handover and minimize handover failure rate.



In general, the key parameters of mobility management are described in Table 10 and illustrate in Figure 25.

Mobility Parameter Parameter Overview

Key Parameter

Q-Hysteresis [dB]

A cell can be mobilized when the average signal level of the target cell is higher than that of the current serving cell by at least the amount defined by hysteresis margin values.

Delay Timer [ms]

Indicates the time delay to trigger. The value of this parameter is related to slow fading. The larger the value of this parameter, the lower the probability of incorrect decision and the slower the response of the event to measurement signal change.

Q-Offset [dB]

Indicates the Offset of the cell CPICH measurement value. The set value is added to the measurement quantity before the UE evaluated whether an event has occurred. In handover algorithms, this parameter is used for moving the border of a cell. A greater value indicates a greater possibility for the cell to be the target cell of the handover.

Table 10: Key Performances Parameter [Huawei-ePico]

Figure 25: Principle of Mobility



3.3.1.1 Performance Parameters

Based on [FREEDOM-D21] document and [FREEDOM-D41], TELKOM focuses to observe the performance metrics for evaluations listed in Table 11.

Key Parameter

Key Performance

Handover Rate [FREEDOM-D21]

Handover rate represents the overall number of mobility initiated within a time interval. This metric is used for evaluation of efficiency of the technique for the reduction of a number redundant mobility. This metric can be represented as an absolute number of handover initiations over a time interval or as an average number on handover initiations.

Handover Failure Rate [FREEDOM-D21]

Handover failure rate (HFR) represents the ratio between failed handovers (NHO_fail) and total handover attempts (NHO).

HFR = NHO_fail/ NHO

Distance (mobility process) This parameter represents the location of mobility process.

Table 11: Key Performances Parameter.

3.3.1.2 Mobility Scenarios

FAP-to-FAP mobility scenarios were conducted in BM1 environment. Figure 26 shows the process of FAP-to-FAP handover.

Figure 26: FAP-GW coordinated cell handover procedures between FAPs [Huawei-

ePico]



Two main mobility scenarios to be considered in the measurement campaign are described below: • Scenario FAP-to-FAP: This is mobility scenario between FAPs. Three active FAPs would be

installed closely in the selected environments. BM1 environment scenario is applicable for this mobility scenario given that operating frequency for the FAPs and MBS are different. For detailed information of test guideline will be described in test scenario in section 4.4.2.

• Scenario FAP to MBS: This is defined as mobility scenario from FAP to MBS. Due to interoperability issue between FAP and MBS which affect commercial networks, the mobility measurements between FAP and MBS will be conducted as part of lab trial. In this lab trial TELKOM mainly aims to perform FAP to MBS.

FAPs placement scheme for mobility measurement in the environment scenario BM1 can be seen in Figure 27 where three FAPs are placed on the first floor.

Figure 27: Possible implementation of FAPs mobility scenarios (Scenario FAP-FAP)

The parameter value for FAP evaluation is presented in Table 12.

Parameter Values

FAP transmit power 10 dBm

Route of Corridor 1 (C1) ~ 50 m

Route of Corridor 2 (C2) ~ 45 m

UE speed 1-3 m/s

UE mobility Based on decision point

Table 12: Parameters for Multi-FAPs deployments



3.3.2 Measurement Tools

TELKOM used TEMS Investigation (Trace Mobile) tools to measure mobility performances. The performance parameters to be measured using the tools are listed in Table 13.

Parameters TRACE Mobile

Ec/No X RSCP X UE TxPower X Handover X Cell Reselections X Handover Failure X

Table 13: Measured mobility performances at UE.

3.3.3 Measurement Methodology

Similar to the approach in interference measurements, we define a measurement methodology consisting of five main stages: Specification, Preparation, and Test Setup, Measurement, and Data collection.

3.3.3.1 Specification

Specification and mobility environment refer to mobility scenario. Type of environment is Business Model (BM1) or Corporate Building

3.3.3.2 Preparation

Preparation includes: • Placing FAPs according to the placement schemes (refer to mobility scenario). • Preparing the measurement equipment: TEMS (Trace Mobile) • Defining UE mobility route. • Defining parameters on mobility management.

Figure 28: Point of Measurements

3.3.3.3 Test Setup

Test setup will be done according to the following steps: • Setting and locating the FAPs in indoor areas



FAP is installed and placed according to placement scenarios (BM1). Make sure FAP has turned on and working properly. FAP status could be seen by seeing the physical indicator at FAP (i.e. status light).

• Draw mobility measurement route It’s needed to draw measurement route on the measurement environment map which show overlap area between FAP to other FAP and MBS. This is an estimated route in which mobility process most likely will happen (for both cell reselection and handover).

• Change mobility parameters values Parameters values will be changed according the measurement scenarios

3.3.3.4 Measurement

Measurement process includes some measurement steps. The measurement steps are: • Presentations and coordination

Before doing measurement, presentation and session will be done with the partners involved in trial. This session is done to make measurement process easier. This presentation will explain about the scenario and methods for mobility measurement. This presentation session will also estimate any necessary resources for the measurement campaign.

• Measurement The measurement campaign team will do their measurement task depending on the availability of resources. The measurement could be conducted either in series or in parallel. Measurements were committed with typical human walk, 1~3 meters/second.

Figure 29: Measurement Process

• Data collections

The collected data should be recorded according to a defined format. The format of the collected data will be described in the next sub-chapter.

Mast (1, 5 Meter)

Handset Test SE-W995 (TEMS)

Laptop + Software (TEMS Investigation

10.0.5)



• Post processing and analysis

After all data collection is complete, then post-process and analysis will be done. In this process, impact of setup parameters on mobility performance could be analyzed. Reporting measurement used post processing tolls (Map Info, Excel, and Actix Analyzer). The results will be reported as part of engineering rules.

3.3.3.5 Data Collection

In order to ease the organization of all collected data from the measurement, a uniform format for file naming needs to be defined. The naming format is defined as date, measurement description, Mobility scenario, area, etc.



4 MEASURED SCENARIOS In this section, measured scenarios definition will be presented.

4.1 Interference Characterisation

Table 14 summarizes all measured scenarios.



Scenario Sub‐scenario Equipments Active FAPs FAP Traffic TRACE mobile Modems Carrier

Scenario 0

Sub‐scenario 0.1 SCANNER No No Not used

Not used

10663 & 10688 TRACE Mobile No No MUE

Sub‐scenario 0.2 (FAP1 ‐ SC300)

SCANNER

Only FAP1

No Not used 10663 & 10688

TRACE Mobile No (excepted TRACE mobile)

FUE 10663

Sub‐scenario 0.2 (FAP2 ‐ SC 219)

SCANNER

Only FAP2

No Not used 10663 & 10688


FUE 10663

SCANNER

Only FAP2

No Not used

10688 TRACE Mobile

No (excepted TRACE mobile)

FUE

Sub‐scenario 0.2 (FAP3 ‐ SC 375)

SCANNER

Only FAP3

No Not used 10663


FUE 10663 & 10688

Sub‐scenario 0.2 (FAP4 SC 118)

SCANNER

Only FAP4

No Not used 10663


FUE 10663 & 10688

Sub‐scenario 0.2 (FAP5 SC 446)

SCANNER

Only FAP5

No Not used 10663


FUE 10663 & 10688




Scenario A1 DL FAP ‐> MUE

Sub‐scenario A1.1 (FAP 1)

TRACE Mobile Only FAP1 yes MUE

Used as FUE on FAP1 in poor

signal coverage (at the cell limit)

10663


TRACE Mobile Only FAP2 Yes MUE















Scenario A2 DL MBS ‐> FUE

& Scenario B2

UL MUE ‐> FAP

Sub‐scenario A2.1 & B2.1 (FAP 1)

TRACE Mobile Only FAP1 No (excepted TRACE mobile)

FUE

Used as MUE in poor signal coverage

10663



FUE



FUE



FUE



FUE




Scenario A3 DL FAP ‐> FUE

& Scenario B3 UL FUE ‐> FAP

Sub‐scenario 3.1 (FAP distance 1)

SCANNER

3 FAPs actived

No (1 FAP) / yes (2 FAPs)

Not used

Used as FUE at mid distance

between FAP1 & FAP3

10688

TRACE Mobile No (1 FAP) / yes (2

FAPs) FUE


SCANNER No (1 FAP) / yes (2

FAPs) Not used


FAPs) FUE



FAPs) Not used


FAPs) FUE



FAPs) Not used


FAPs) FUE



FAPs) Not used


FAPs) FUE




Scenario Generic

Sub‐scenario G.1 (MUE)

SCANNER All Yes Not used

Used as FUE on all FAPs

simultaneously in poor radio coverage

10663 & 10688

TRACE Mobile All Yes Two‐tier user

Used as FUE on all FAPs


10663

Sub‐scenario G.2 (FUE)

SCANNER All No Not used Not used 10663 & 10688

TRACE Mobile All No FUE Not used 10663

TRACE Mobile All Yes FUE

Used as FUE on N‐1 FAPs


10663

Table 14: Measured scenario – Overview.

Next subsections give a detailed description for each scenario.



4.1.1 Scenario 0

Scenario 0.1 consists in collecting data along the pre-defined trajectory in absence of any FAP deployment. Only the macro network provides coverage. Measurements from this scenario are used as a reference to evaluate the impact of tested FAP deployments in terms of coverage, interference, QoS, etc.

Ref Scenario 0.1.

Title No FAP deployment.

Number of activated FAP’s

0.

FAP locations. -

FAP access mode. -

UE location Along the whole pre-defined UE trajectory.

Measured UE status MUE.

Traffic Macro traffic monitored.

Measurements done at different times to address different traffic properties.

Usage All metrics: reference data for evaluation of the FAP deployment impact.

Table 15: Interference scenario 0.1 – specifications

Note that measurements of scenario 0.1 are collected on both MBS carriers (i.e. 10663 and 10638). These measurements are considered as the reference data for evaluation of the FAP deployment impact. This sub-scenario was measured four times and with both receivers (i.e. SCANNER and TRACE mobile) in order to check the reproducibility of outputs in time and with different equipments.

Scenario 0.2 consists in collecting data around each FAP individually (all other FAPs are turned off). The main goal of this scenario is to get a reference characterization of cell limits, handover zones with macro coverage and macro deadzone limits. Most of these measurements were collected at least two times in the first carrier (i.e. 10663). Measurements were also collected with FAP n°2 transmitting on a free frequency (i.e. carrier 10688) in order to get a reference data for evaluation of the multi-FAPs deployment impact.

FAP locations are the same as in generic scenario.



Ref Scenario 0.2.

Title One single FAP deployment.


1.

FAP locations All FAP locations used in generic scenarios.

FAP access mode Closed (to prevent from undesired traffic on the FAP).

UE location Along the pre-defined UE trajectory of interest for activated FAP.

Measured UE status FUE.


FAP traffic null (except from trace mobile).

Usage

- SINR (Ec/I0): determination of handower zone between FAP and MBS; characterisation of local variations.

- All metrics: comparison to scenario 0.1 for characterisation of the radio property enhancements (received power distribution, SINR distribution, etc.).

- All metrics: reference data for evaluation of the multi-FAPs deployment impact.

Table 16: Interference scenario 0.2 – specifications.

4.1.2 Scenario A1 – DL FAP → MUE

Within this scenario, measured data are collected by using TRACE mobile as MUE locked on a single carrier (i.e. 10663) along the pre-defined trajectory around each FAP activated individually (all other FAP’s are turned off). The active FAP transmits continually a high traffic (data only) mainly through a poor radio link in order to generate high DL interference levels in its vicinity. As aforementioned, the amount of data collected for this scenario is limited as some unexpected handovers are initiated by the TRACE mobile. Ref Scenario A1

Title DL neighbor-FAP → MUE interference scenario.


1.






FAP highest traffic.

Usage - All metrics: comparison to scenario 0.1 for characterisation of DL FAP to MUE interferences.

Table 17: Interference scenario A1 – specifications.



4.1.3 Scenario A2 – DL MBS → FUE

Measured data are collected by using TRACE mobile as FUE locked on a single carrier (i.e. 10663) along a trajectory limited to the FAP cell (as handover between MBS to FAP is not possible with the equipments installed on the testbed). Ref Scenario A2

Title DL MBS → FUE interference scenario.


1.



UE location Along a trajectory limited to the FAP cell.

Measured UE status FUE.


FAP traffic null (except from trace mobile).

Usage - All metrics: comparison to scenario 0.1 for characterisation of DL MBS to FUE interferences.


4.1.4 Scenario A3 – DL FAP → FUE

Objective of this scenario is to evaluate DL interference caused by a neighbour FAP onto the FUE. To that end, three FAPs are used: FAP n°1, 2 and 3. FAP n°2 is set at a fixed location (same location than in the generic scenario) and FAP n°2 and FAP n°3 are moved step by step at different distances from FAP n°2 in order to simulate several density of FAPs deployment. Each FAP transmits on a free frequency, i.e. 10688.

One drive-test is collected for each inter-FAP distance by using both receivers (TRACE mobile being only in the CSG of the FAP n°2). The measurement area is limited to the global coverage area of all three FAPs, i.e. receivers are mainly moved along the corridors at the right and below the FAP n°2. Several modems are placed at mid-distance respectively between FAP n°2 and FAP n°1 and between FAP n°2 and FAP n°1 to generate traffic on these FAPs. Five inter-FAP distances have been tested: 30m, 25m, 20m, 15m and 10m. Figure 30 illustrates the measurement set-up for an inter-FAP distance of 15 m.



Figure 30: Scenario A3 (inter-FAP distance: 15m) – Modems and FAPs location.

Main specifications of this scenario are summarized in Table 19.

Ref Scenario A3.

Title DL neighbor-FAP → FUE interference scenario.


3 FAPs.

FAP locations Different inter-FAP distances: from 10m to 30m for instance.

Poor macro coverage.

FAP access mode Closed.


Restricted to indoor trajectory.


Measured UE status FUE in CSG of one FAP only.


Traffic null on FAP under test (except from trace mobile).

Usage - All metrics: comparison to scenario 0.2 and scenario A1 for characterisation of the interference impact (degradation of the SINR and QoS parameters in the overlapping area).


15m



4.1.5 Scenario B2 – UL MUE → FAP

Scenario B2 is based on the same setup as scenario A2 but with the following requirements: controlled traffic in the FAP (i.e. NMS FAP data) and the measured FUE

is replaced by MUE (i.e. the modems). Locations of the modems are illustrated in

Table 21.

Ref Scenario B2.

Title UL MUE → FAP interference scenario.


1.

FAP locations All FAP locations used in generic scenario.


UE location Along a trajectory limited to the FAP cell.



FAP traffic controlled.

Usage - All metrics: comparison to scenario 0.2 for characterisation of UL MUE to FAP interferences.

Table 20: Interference scenario B2 – specifications.



FAP Modems location

FAP 1 – SC300

Modems location during measurements around FAP 1.

FAP 3 – SC375

Modems location during measurements around the FAP 3.



FAP 4 – SC118


FAP 5 – SC446


Table 21: Interference scenario B2 – modems location.

4.1.6 Scenario B3 – UL FUE → FAP

Scenario B3 is based on the same setup as scenario A3 but with controlled traffic in the FAP (i.e. NMS FAP data).



Ref Scenario B3

Title UL FUE → neighbor FUE interference scenario.


3 FAPs.

FAP locations Different inter-FAP distances: from 10m to 30m.



UE location Along the pre-defined UE trajectory of interest for activated FAPs.

Restricted to indoor trajectory.


Measured UE status FUE in CSG of one FAP only.


FAP controlled.

Usage - All metrics: characterisation of the inter-FAP interference impact on UL.

Table 22: Interference scenario B3 – specifications.

4.1.7 Generic scenario

Scenario G1 consists in collecting measurement with SCANNER and TRACE mobile as two-tier user locked on a single carrier (i.e. 10663). Modems were used to generate traffic on all FAPs simultaneously as described in Figure 31.

The TRACE mobile was able initiate a handover procedure between both 3G carriers (however we tried to lock it on the WCDMA carrier only). Thereby, the amount of data collected in high interference areas is limited.

Main specifications of the generic scenario are detailed in Table 23.



Ref Scenario G

Title Generic scenario


At least 4 FAPs.

FAP locations Low and Medium density of FAPs within two floors.


UE location Some FAPs located in good-quality macro coverage.

Some other FAPs located in poor-quality macro coverage.

Measured UE status In closed access mode:

- MUE.

- FUE subscribing to only one FAP.

Traffic

In closed access mode

- MUE status: High traffic on all FAPs (G1).

- FUE status: High traffic on all FAPs except the subscribed FAP (G2).

Usage Global interference characteristics from comparison to reference.

Table 23: Generic interference scenario - specifications.



Figure 31: Modems (represented by circles) and FAPs (represented by triangles) location - Scenario G1.

Scenario G2 consists in collecting measurement with SCANNER and TRACE mobile as FUE locked on a single carrier (i.e. 10663). It was measured twice: without and with extra traffic on FAPs. Modems used to generate traffic on FAPs were set as described in Figure 31. Remark that with the equipments installed on the testbed, only handover between ePICO FAPs is possible (i.e. between FAPs with SC 300, 219 or 375).

4.2 xDSL Backhaul Characterisation

4.2.1 xDSL Access Network Characteristics

In order to derive xDSL characteristics, the cable models have been defined according to TELKOM‘s installed cables. Structure of Telkom Indonesia cables can be summarized as following:

Diameter of conductors: 0.6mm-0.8mm Insulation: polyethylene Basic structure: quad (2 pairs) First level of aggregation: For cables up to 100 cp: Unit, consists of 5 quad (10 cp) For cables with 200 cp or more: Super Unit of 25 quad (50 cp)

Based on those information, cable model were constructed based on the following parameter:

Diameter of conductors: 0.6mm (worst case) Attenuation: as of CT 1341 Italian cable, polyethylene insulated quad cable Reference cable binder:

1st Floor

2nd Floor

15m



o Primary cable: 50 pairs (1 super unit, 25 quads) o Secondary cable: 50 pairs (5 units, 5x5 quads) o Drop cable: 20 pairs

Number of boxes per access binder: 3 Number of drops per building cable: 4

Crosstalk modelization is based on following assumptions:

NEXT reference value (@ 1 MHz) is considered the value @ 1% of the estimated statistical distribution

FEXT reference value (@ 1 MHz and 1 km) is assumed as in standard (Recommendation ITU-T G996.1, ETSI 101 524)

Crosstalk (statistical value @ 99% of confidence): o KNEXT (@ 1 MHz) = -52.2 dB, o KFEXT (@ 1 MHz @ 1 km) = -45.0 dB

Noise mix follows the assumption;

Present broadband systems: 95% ADSL2/2+ (over POTS), 5% SHDSL, no regeneration allowed in access network

Medium Term Broadband Services penetration assumed for performance estimation: 30% Long Term Broadband Services penetration assumed for performance estimation: 50% Noise mix composition for scenario (FTTE):

o Mix 30%BB: 1 SHDSL (@2.3Mbit/s), 14 ADSL2/2+, 35 POTS (or vacant) o Mix 50%BB: 1 SHDSL (@2.3Mbit/s), 24 ADSL2/2+, 25 POTS (or vacant)

xDSL Physical layer setting for performance evaluation follows

Frequency Plan for VDSL2: 998 Hz Physical layer setting for ADSL2/2+ Internet access services: NM=6dB, Channel Mode FAST IPTV services: NM=9dB, INP=2, Max Delay=8ms Physical layer setting for VDSL2 IPTV services: NM=9dB, INP=2, Max Delay=8ms ADSL2/2+ performance curves refer to systems implementing extended framing

The above assumptions are incorporated to the simulation to gain several performance properties.

4.2.2 xDSL Scenario 0

The purpose of this scenario is to check initial E2E network performance, with and without FAP, in order to capture initial network performance including access part (xDSL to SecGW), backbone part (SecGW to SGSN) and internet exchange (GGSN to Application Server). We split these scenario into several sub-scenario. Scenario 0.1 is defined to check xDSL network performance over different DSLAM mode namely interleave vs fast. The measurement is performed using SmartBits without FAP. Scenario 0.2 is defined to check end-to-end delay performance by using NMS between FAP to SecGW, SecGW to SGSN, and end to end ping test from FUE to Application Server in internet.



Ref Scenario 0.1 DSL Title Initial xDSL Network Performance Test

Objective Check initial network performance, without FAP, in order to capture ADSL/ADSL2 access network performance over different modes (interleave vs fast)

Access Network DSL

Network Component & Setting Configuration

Switch & Metro E : L2 Non Blocking Connectivity bandwidth 100 Mbps DSLAM : DSLAM 20 Mbps, Fast and Interleave xDSL Modem :UBR

Measurement Tools Smartbit SMB 600 Traffic Type Real time traffic generations are from SmartBits

Case

1. generate layer 2 frame to see latency, packet loss, throughput for fast and interleave

2. observe delay of ADSL/ADSL2, using fast and interleave mode for different bandwidth profiles in DSLAM

Usage Get statistics latency, throughput, and packet loss.

Table 24: xDSL scenario 0.1 – Initial xDSL Network Performance

Ref Scenario 0.1 DSL Title Initial End to End DSL Network Performance Test

Objective

Check initial delay network performance, with FAP, in order to capture E2E network performance of access part (FAP to SecGW), backbone part (SecGW to SGSN) and end-to-end delay from FAP to application server.

Access Network DSL-MetroE-3G Core Network


Switch & Metro E : L2 Non Blocking Connectivity bandwidth 100 Mbps DSLAM : DSLAM 20 Mbps, Fast and Interleave xDSL Modem :UBR

Measurement Tools NMS Femtocell Traffic Type Ping or response time Case Check end to end delay from FAP-SecGW-SGSN-Application Server Usage Get statistics latency

Table 25: xDSL scenario 0.2 – Initial E2E Network Performance


Scenario 1 will be used to capture bandwidth required by a single FAP considering mix services, various UE types and 4 simultaneous users access. This scenario will be used as a basis to observe minimum xDSL bandwidth requirement and also as a reference of configuring traffic models in STC. The applications will be accessed through various UE types including, smartphone and tablet PC. As been stated in [FREEDOM-D21] each user will access only one service type. Each user will access real internet services such as HTTP, FTP, VoIP and Video. The traffic traverse through a FAP will be recorded by using Femto NMS while xDSL profile set to maximum (20 Mbps).



Ref Scenario 1.1 DSL Title Bandwidth requirement of a single FAP.

Objective Observe bandwidth requirement of a single FAP considering mix services, various UE types and 4 simultaneous users access

Access Network DSL


Metro E : L2 Non Blocking Connectivity bandwidth 100 Mbps DSLAM : Default (Interleave) xDSL Modem : Default (UBR) FAP : Default 3G Router : NAT, DHCP UE Type : Smartphone, 3G enabled tablet PC

Measurement Tools NMS Femtocell, STC Avalanched

Traffic Type

SmartPhone : - Single Traffic : Voice (AMR)

HTTP (m.detik.com) FTP (4shared.com) 5 Mbps Streaming (video Youtube)

- Mix traffic : Voice only (4 subscriber voice AMR) Voice (AMR), HTTP (m.detik.com), FTP 5 Mbps (4shared.com), Streaming (video Youtube)

Tablet PC/Ipad :

- Single Traffic : HTTP (detik.com) HTTP ( facebook.com) FTP (4shared.com) 13 Mbps Streaming (video Youtube)

- Mix traffic : HTTP (detik.com dan facebook.com), FTP (4shared.com) 13 Mbps, Youtube

Case

No background traffic. Record the bandwidth requirement (BR) from NMS Femtocell when the profile

maximum. Analize BR.

Usage 1. Get statistics throughput from NMS femtocell. 2. Get statistics bitrate, packet loss, jitter, and FPS (frame per second) from STC.

Varian Test 1. Different UE Type (Smartphone and tablet PC/Ipad). 2. Single user (single traffic) and 4 simultanous (mixed service).

Table 26: xDSL scenario 1.1 – Bandwidth requirement of a single FAP

Scenario 1.2 will be used to verify mix-traffic bandwidth gained from Scenario 1.1. xDSL bandwidth profile is set as close as possible to the initial BR. By observing video conference quality, new bandwidth with accepatable video quality will be recorded.



Ref Scenario 1 DSL Title Bandwidth requirement of a single FAP.

Objective verify bandwidth requirement of a single FAP considering mix services, support 4 simultaneous users access

Access Network DSL


Metro E : L2 Non Blocking Connectivity bandwidth 100 Mbps DSLAM : Default (Interleave) xDSL Modem : Default (UBR) FAP : Default 3G Router : NAT, DHCP UE Type : Smartphone, 3G enabled tablet PC

Measurement Tools NMS Femtocell, STC Avalanched

Traffic Type

Generated Mix traffic from STC : HTTP (detik.com dan facebook.com), FTP (4shared.com) 13 Mbps, Real Traffic from Video Conference AMRvoice from a smartphone)

Case

No background traffic. verify the bandwidth requirement (BR) from NMS Femtocell when the

bandwidth profile is set to 20 Mbps, 650 kbps, 800 kbps, 900 kbps, 1 Mbps Verify the video conference performance (packet loss, jitter, frame per second)

for each bandwidth profile. Analize BR from Scenario 1.1 and compared with the one from scenario 1.2.

Usage Get statistics throughput from NMS femtocell. Get statistics bitrate, packet loss, jitter, and FPS (frame per second) from

STC/video conference.

Table 27: xDSL scenario 1.2 – Bandwidth verification of a single FAP


Scenario2 uses to verify femtocell performance in the existence of background traffic in xDSL modem. The background will use real DSL traffic and mixed traffic femtocell generated from the measurement tools. Verify xDSL bandwidth profile whether it can support bandwidth demand both from a FAP and xDSL user. The bandwidth profile is increased to the level that video quality is within key performance indicator target (packet loss < 3%, jitter < 40ms). This scenario is designed to show that if MNO and xDSL provider does not sign an agreement, the internet traffic and FAP traffic will be mixed into a single PVC/VLAN, so that regardless the QoS setting in the modem, both traffic will have a same priority and compete each other as best effort.



Ref Scenario 2 DSL Title Femtocell Performance with a presence of background traffic from a xDSL user

Objective

Verify xDSL bandwidth profile whether it can support bandwidth demand both from a FAP and xDSL user. Analyze the femtocell performance in the existence of background traffic in xDSL modem.

Access Network DSL


Metro E : L2 Non Blocking Connectivity bandwidth 100 Mbps DSLAM : Default (Interleave) xDSL Modem : Default (UBR) FAP : Default (Conversational Traffic) 3G Router : NAT, DHCP UE Type : Tablet PC/Ipad

Measurement Tools NMS Femtocell, STC Avalanched, Cisco Movie Telepresence.

Traffic Type

FAP Mix traffic : HTTP (detik.com dan facebook.com), FTP (4shared.com) 13 Mbps, Video Conference and voice AMR (smartphone) Background traffic: Mix traffic (HTTP, FTP 40 Mbps, Youtube)

Case

Use Dedicate PVC/VLAN (Femtocell and Background use single VLAN) . Real background traffic. Throughput of mixed service and background measure and analize while DSL

profile set to maximum bandwidth. Analize the performance of video conference (packet loss, jitter) with the

existence of background traffic, bandwidth profile set to BR Increase the bandwidth profile step by step until the video performance below

the threshold (packet loss, jitter)

Usage

Get statistics throughput from NMS femtocell. Get statistics bitrate, packet loss, jitter, and FPS (frame per second) from

STC. Analyze the impact of background traffic

Table 28: xDSL scenario 2 – Femtocell Performance with a presence of background traffic from a xDSL user

4.3 Corporate LAN as FAP Backhaul

4.3.1 LAN Scenario 0

In this scenario, TELKOM RDC LAN characteristics will be observed in terms of network utilization (days of week). The network utilization will be observed through IP Network Monitoring (PRTG) both in 3rd floor gateway and RDI internet link.



Ref Scenario 0 LAN Title Network Utilization without a FAP

Objective Check initial network, without FAP, in order to capture initial network utilization in gateway and internet link

Access Network Corporate LAN Network Component & Setting Configuration

LAN Port to Gateway 3rd floor : same VLAN LAN Port of Internet Gateway –SecGW : L3 Connectivity

Measurement Tools PRTG IP network monitoring Traffic Type Existing LAN traffic

Case bandwidth profile internet leased line = 9 Mbps, LAN = 100 Mbps. Layer 2 connection from 3rd floor gateway, switch distribution, leased line

to Sec GW using IP public

Usage Get load utilization, from Monday-Sunday


Scenario 1 is defined to observe a FAP bandwidth consumption and its performance without background traffic. The FAP will be used to access mix traffic from 1-4 FUEs

Ref Scenario 1 LAN Title Performance of a FAP without intranet background Objective Observe a FAP performance without background trafficAccess Network Corporate LAN


LAN Gateway - Metro E – SecGW : L2 Connectivity, 100 Ethernet FAP and Intranet : seperate VLAN, (100 MB)

Measurement Tools NMS Femtocell, Smartphone, PRTG IP Network Monitoring, Cisco Movi Telepresence

Traffic Type HTTP, FTP, VoIP (G.729/G.711u), Video Streaming (MPEG 4/H.264)

Case

1. The FAP and LAN’s Users are separated using different VLAN 2. No Background Traffic. 3. Observe individual traffic from a smartphone attached to a FAP is recorded 4. Observe occupied bandwidth of Cisco Movi Telepresence 5. A single FAP, and mix traffic from 4 FUEs.

Usage Get statistics, throughput, jitter, packet loss. Observe throughput in NMS femtocell Observe throughput in PRTG IP network management




Scenario 2 is defined to observe the FAP performance with the presence of intranet-internet traffic in LAN gateway 3rd floor.

Ref Scenario 2 LAN Title Performance of a FAP in the presence of background traffic

Objective Observe a FAP performance when the network is shared with LAN traffic as background

Access Network Corporate LAN – Intranet


FAP and Intranet network : same VLAN, (100 MB) Gateway - Metro E – FAP SecGW : L2 Connectivity, 100 Ethernet RDI internet leased line is not used, but connected using LAN backbone through SecGW in OASIS

Measurement Tools NMS Femtocell, IP Network Management, 1xChariot

Traffic Type VoIP (G.711u)

Case

1. The FAP and LAN’s users are in the same VLAN 2. A single FAP is activated, one FUE modem connected to the Ixchariot. 3. Ixchariot generates traffic model and measure the performance (delay, jitter,

packet loss, throughput) 4. Femto NMS observe individual UE throughput 5. IP network management capture the LAN’s traffic in the switch access

Usage 1. Observe FAP/UE throughput in NMS femtocell 2. Observe in-bound/outbound traffic in the switch access 3. Record G.711u performance and save the log file

Table 29: Intranet Scenario 2 Performance of a FAP with background in LAN GW 3rd floor.


This scenario is developed to adopt a common sense, that corporate customer may be reluctant to have additional bandwidth to accommodate the femtocell traffic. However it should be anticipated by using the same internet link, the bottleneck may occurred and affect the femtocell performance. Internet link available in TELKOM RDC is 512 kbps, 2 Mbps, 3 Mbps, 9 Mbps and 15 Mbps. The link with speed 512 kbps is currently active to support VPN, so the link is fully utilized with the traffic. While 2 Mbps link is used as proxy to run internet service only. This link can not be used for femtocell, since femtocell tunnelling cannot traverse through the proxy. RDI internet link 9 Mbps is used instead. These links basically serve internet users in RDI department which consists of 33 people. Another 15 Mbps link is used only for research and trouble shooting. Scenario 3 is defined to observe a FAP performance due to internet activity. The video performance will be compared if there is no background in the internet link (Scenario 3.1), means the corporate assigned different VLAN for FAPs traffic.



Ref Scenario 3.1 LAN Title Performance of a FAP without internet background traffic in corporate leased line

Objective Observe the FAP performance when connected to SecGW through corporate internet leased line

Access Network Corporate LAN


FAP and corporate LAN : same VLAN (100 MB) Gateway – internet link – SecGW : L3 Connectivity RDI Internet link : 9 Mbps

Measurement Tools NMS Femtocell, IP Network Management, Cisco Telepresence

Traffic Type Video Conferencing (MPEG 4/H.264)

Case

The FAP and LAN’s users use a same VLAN A FAP is authenticated by SecGW attached in Internet Cloud A video conference client (VC1) attached to a FAP. FUE video conference communicate with VC2 connected to ip public (15

Mbps) Two VC clients communicate for 1 hours. Femto NMS observe individual UE throughput IP network management capture the internet traffic in the gateway (internet

link)

Usage

Observe FAP/UEs throughput in NMS femtocell Observe in-bound/outbound traffic in the gateway using IP network

monitoring Observe video quality metrix

Table 30: Scenario 3.1 - Performance of a FAP without internet background.



Ref Scenario 3.2 LAN Title Performance of a FAP with internet background traffic in corporate leased line

Objective Observe the FAP performance when connected to SecGW through corporate internet leased line



FAP and corporate LAN : same VLAN (100 MB) Gateway – internet link – SecGW : L3 Connectivity RDI Internet link : 9 Mbps

Measurement Tools NMS Femtocell, IP Network Management, Cisco Telepresence

Traffic Type Video Conferencing (MPEG 4/H.264)

Case

The FAP and LAN’s users use a same VLAN A FAP is authenticated by SecGW attached in Internet Cloud video conference client (VC1) attached to FAP. FUE video conference communicate with VC2 connected to ip public (15

Mbps) Two VC clients communicate for 1 hours in the presence of background up

to 80% utilization of the internet link Femto NMS observe individual UE throughput IP network management will capture the internet traffic in the gateway

(internet link)

Usage

Observe FAP/UEs throughput in NMS femtocell Observe in-bound/outbound traffic in the gateway using IP network

management Observe video quality metrix

Table 31: Scenario 3.2 - femtocell performance observation in the presence of background traffic in internet link.


In this scenario 4 we will observe DiffServ implementation in the femtocell, and metro ethernet. In order to deliver end-to-end QoS, DiffServ marking is implemented in the FAP. We will see whether the network (access and backbone) can deliver the marking. We are not in the position to re-engineer the end-to-end network to implement DiffServ, since it will requires huge efforts and may affect live network from different organization (xDSL ISP, backbone, mobile network operator).



Ref Scenario 4.1 LAN Title E2E QoS mechanism in femto network Objective Observe the QoS implementation in the femtocell in order to support DiffServAccess Network Corporate LAN


Gateway – internet link – FAP SecGW : L3 Connectivity FAP and corporate LAN : same VLAN (100 MB) RDI Internet link : 9 Mbps

Measurement Tools NMS Femtocell, Wireshark

Initial Condition FAP is properly connected according to the topology. Each network element runs properly. IMSI of UE is added to the admission list of the FAP.

Case

Femto OAM configure QoS scheme used by FAP system Start QoS tracing in end-to-end route Start services from all UE Compare the tracing message to the configured QoS mechanism

Expected Result E2E QoS result in traced messages is similar to the configured QoS

Table 32: E2E QoS Observation Scenario in Femtocell System

Ref Scenario 4.2 LAN Title QoS mechanism in metro ethernet as corporate access or backbone

Objective Observe the QoS implementation in the metro ethernet in order to support DiffServ



PE router of metro ethernet both ingress dan egress direction

Measurement Tools LMT Metro Ethernet

Initial Condition NA

Case Capture the ME profile Analyze the QoS policy implementation

Expected Result

Table 33: QoS Observation Scenario in Metro Ethernet



4.4 Mobility Management Test Guideline

This section will describe measured test scenario. We will explain general scenario performed in the measurement campaign.

4.4.1 Test Overview

These are the test list for mobility measurement, individual scenario will be explained in the next sub-section

No Test ID Scenario and Directions

Mobility Applications

1 M.10x FF CR (FF1) Idle Mode

2 M.20x FF HO (FF2) CS/ Voice Call

3 M.30x FF HO (FF3) PS Data/ HTTP Download

4 M.40x FM CR (FM1) Idle Mode

5 M.50x FM HO (FM2) CS/ Voice Call

6 M.60x FM HO (FM3) PS Data/ HTTP Download

Table 34: Test Case List

*CR : Cell Reselections, HO: Handover * X : No Initial Parameter (defined by default or manual configure). (TBD)

4.4.2 Scenario FF (FAP –to- FAP)

FAP to FAP mobility scenarios requires minimum two FAPS to be placed with overlapping coverage. Mobility measurement uses environment in BM1 as in Figure 28. However, the FAPs should use different operating frequency from the one used by MBS. Mobility movement should be done multiple times for accurate measurements. For every parameter changes, measurement will be repeated again.



Ref Scenario Femto –to- Femto (FF)

Title FAP mobility measurements


2 FAPs

FAP locations - All FAPs within one single floor. - No MBS coverage environment (high priority) by setting different carrier frequency or scrambling codes - FAPs distributed on different floors. (medium priority)

FAP access mode - Open

UE location - Moving from FAP A to FAP B coverage (minimal one UE) - Can be done with back and forth route

Measured UE status

- FUE downlink signaling messages - FUE uplink signaling messages - Radio (RSCP, RSSI, Ec/No) parameters

Traffic Idle mode, CS (Voice), and PS (Data/ HTTP)

Usage Detail test procedures in the attachment

Table 35: Scenario FF

4.4.3 Scenario FM (FAP –to- MBS)

FAP to MBS scenarios requires minimum one FAP to be placed with overlapping MBS coverage. Mobility movement should be done multiple times but in one way (FAP to MBS). For every mobility parameter values changes, measurement must be repeated. For better measurement result, the FAP should ideally be placed in cell edge of MBS coverage. This is test guidelines for FAP to MBS mobility scenarios;

Ref Scenario FAP to MBS (FM)

Title MBS Cell Reselection measurement

Number of activated FAP’s or MBS

One FAP and one MBS

FAP locations - Placed FAP in first floors. (BM1 and BM3 Scenarios). - Placed FAP in cell edge MBS (high priority)

FAP access mode - Open (high priority).

UE location Moving from FAP to MBS coverage (one way)

Measured UE status

- FUE downlink signaling messages - FUE uplink signaling messages - Radio (RSCP, RSSI, Ec/No) parameters

Traffic Idle mode, CS (Voice), and PS (Data/ HTTP)

Usage Detail test procedures in the attachment

Table 36: Scenario FM



5 POST PROCESSING & ANALYSIS

5.1 Interference Characterization

As aforementioned, four types of measurement devices have been used during the measurements: SCANNER and TRACE mobile receivers, NMS’s FAP and MBS. The first step of the post-processing consisted in the synchronisation of all collected data and creation of one measurement file by sub-scenario and exploited metric (see respectively Table 14 and Table 7 for details). Spatial averaging was applied on some metrics (data collected with the SCANNER receiver in particular) to smooth out the local fluctuations. The development of routines specific to each equipment data format has been necessary for this first step. Furthermore, the comparison between the different scenarios is possible only if measured data are well collocated. All measurement routes must match with the pre-defined UE trajectory. As the outdoor coordinates collected with GPS were not accurate enough (actually most of the outdoor locations were close to buildings, which is not a favourable for GPS precision), several measurement portions have been manually relocated. Lastly, several specific routines have been developed in order to make all required comparisons and to calculate all statistics given in the next sections.

5.1.1 Macro-only Network (scenario 0.1)

In scenario 0.1, there is no active FAP. Both SCANNER and TRACE mobile receivers are used to get reference data for evaluation of the FAP deployment impact. Figure 32 illustrates the RSCP collected for the MBS with SC 258 (see Figure 32.a) and SC 183 (see Figure 32.b). RSCP is higher than -95dBm all along the trajectory with a maximum level at approximately -55dBm.

SC 258 SC 183

Figure 32: Collected RSCP for MBS – No FAP deployment.



Figure 33 gives the cumulative function of Ec/Io, RSCP and RSSI for the detected MBS’s indoors and outdoors. These results are used as reference in the analysis of following scenarios.

a. Indoor Ec/Io (dBm). The dotted lines are the fitted lognormal distributions.

b. Indoor RSCP / RSSI (dBm)

c. Outdoor Ec/Io (dBm). The dotted lines are the fitted lognormal distributions.

d. Outdoor RSCP / RSSI (dBm)

Figure 33: Scenario 0.1 – Ec/Io, RSCP and RSSI collected in indoor / outdoor for MBS.

The MBSSC258 (i.e. MBS with SC258) is the one that offers the best coverage: best server in 58.3% of indoor locations. Note that the indoor best-server Ec/Io and RSSI levels may be approached with very good accuracy by lognormal distributions. The lognormal parameters are reported in Table 37. They may be considered as a characterization of the macro coverage inside the building under study. This simple approximation is reused in engineering rules provided in [FREEDOM-D622].

Mean (dB) Std deviation

(dB) RSSI -75.17 3.88

Best-server Ec/Io

-8.50 1.82

Table 37: Approximation of the indoor Ec/Io distribution by a lognormal law.



The WCDMA indoor macro coverage may be considered as incomplete and interference-limited inside the TELKOM premises. Indeed, the best-server RSCP is always greater than -95dBm however the best-server Ec/Io is lower than -10dB for 25% of indoor locations. A FAP deployment in such situation may be justified by desire to extend both coverage and capacity.

5.1.2 Single-FAP Deployment (scenario 0.2)

Scenario 0.2 consists in collecting data around each FAP individually (all other FAPs are turned off). This permits to investigate the impact of one unique FAP deployment. Four visibility conditions (VC) have been defined to classify the different radio links between the active FAP and the measured UE locations:

- VC1: the direct-path is free (line-of-sight) and the first Fresnel ellipsoid is not obstructed or quasi-not obstructed (e.g. a radio link along a corridor).

- VC2: only one light partition is crossed. - VC3: one strong partition or several partitions are crossed. - VC4: the FAP and UE are located in two different floors.

VC distribution for each FAP is shown in Figure 34.

Figure 34: FAP Visibility to measurement route points.

5.1.2.1 Indoor Coverage

This section presents a basic analysis of the indoor coverage after activation of one FAP. Firstly the coverage are evaluated from the measurements collected with the SCANNER receiver indoors and according to three different criteria:

- Ec/Io >-20 dB, where -20dB is the SCANNER detection threshold; - Ec/Io > -14 dB and RSCP > -98 dBm, which are common thresholds in WCDMA coverage

determination; - Ec/Io > -10 dB and RSCP > -98 dBm, which are other common thresholds but more severe

thresholds. Table 38 gives the coverage that would have been experienced by a:

- FAP-only user (measured from FAP SC scanning) - Macro-only user (measured from MBS SC scanning) - Two-tier network user (i.e. the UE is assumed to connect the base-station having the best

Ec/Io)

The coverage measured in the macro-only network are also given as a reference.



Ec/Io > -20 dB Ec/Io > -14 dB

RSCP > -98dBm Ec/Io > -10 dB

RSCP > -98dBm

Macro user

FAP user

Two-tier user

Macro user

FAP user

Two-tier user

Macro user

FAP user

Two-tier user

FAP 1 on 93 % 54 % 100 % 86 % 44 % 100 % 47 % 36 % 77 %

FAP 2 on 87 % 62 % 100 % 68 % 60 % 100 % 46 % 50 % 93 %

FAP 3 on 83 % 57 % 100 % 67 % 39 % 100 % 58 % 38 % 94 %

FAP 4 on 99 % 50 % 100 % 92 % 27 % 99 % 62 % 22 % 80 %

FAP 5 on 100 % 17 % 100 % 98 % 10 % 100 % 70 % 7 % 76 %

Macro-only 100 % - - 100 % - - 77 % - -

Table 38: Indoor coverage from a single-FAP deployment.

These results are advantageously completed by Figure 35, which shows the cumulative distribution function (CDF) of the Ec/Io experienced by the different user types and for each active FAP individually. The dotted line is the reference CDF from the macro-only network. Figure 36 gives the CDF averaged from the four FAPs located in the first floor.

a. Access to FAP only b. Access to FAP or macro

c. Access to macro only Figure 35: CDF of indoor Ec/Io from a single-FAP deployment.



Figure 36: Averaged CDF of indoor Ec/Io from a single-FAP deployment.

It can be observed that:

- For a FAP-only user, the coverage area represented ~50-55% of the measured trajectory. A high data rate (i.e. at Ec/Io > -10dB) is offered for approximately 70% of the covered area.

- For a two-tier network user, the enhancement of the coverage for higher data rate is sensitive (+9%); the initial macro coverage was already 80%.

- For a macro-only user, one unique active FAP strongly degrades the coverage, by 22% in average at Ec/Io > -14dB and 24% at Ec/Io > -10dB.

- FAP provides much higher Ec/Io levels than MBS, however the Ec/Io decrease is sharper. - FAPs n°1, 2 and 3, which are the ones with higher proportion of line-of-sight radio links,

cause large degradation in the macro-user coverage. The coverage at Ec/Io > -14dB decreases from 100% (macro-only) to 65–85% (one active FAP). The impact is lower with the FAP n°4 and 5, as the coverage at Ec/Io > -14dB only decreases from 100% to 92–98%.

- Even if the FAP n°5 (located in the second floor) provides a very small coverage in the first floor, it generates a significant degradation on the MBS coverage, 7% degradation at Ec/Io > -10dB.

Figure 37 shows the cumulative distribution function (CDF) of the RSSI. As previously observed, the FAPs with high proportion of line-of-sight measured points (i.e. FAPs n°1, 2 and 3) generate the highest interference levels.

Figure 37: CDF of the RSSI levels.



Result in Figure 37 must be completed with observation of the traffic load conditions in the macro network, which indeed impacts the total interference level (RSSI). The NMS data provides the RRC (Radio Resource Controller) setup attempts on the MBS, i.e. all connection attempts realized during an hour. This information is associated to each measurement scenario in Figure 38 thanks to the time information. We assume the RRC setup attempts to be proportional to the traffic load. Thus we observe that the traffic load on the MBSSC258 was quite constant during measurements on FAP n°1, 2, 3 and 4. It was about twice higher during the reference macro-only measurement, which means that the reference macro-based interference level is slightly greater than macro-based interference levels during FAP evaluation. Note that the traffic load significantly increased during FAP n°5 activation. That means that part of the RSSI level increase with FAP n°5 (in upper floor) was due to higher macro traffic, the observed degradation is certainly pessimistic.

Figure 38: NMS MBS data – Average RRC setup attempts per hour during each

measurement.

All these observations on the impact of one active FAP are of course highly related to the measurement environment. In this case, most measurement points located in corridors; the indoor macro coverage is limited by interference; the mean macro Ec/Io is -8.5dB. Engineering rules given in [FREEDOM-D622] tends to provide a more generic view.

5.1.2.2 FAP Coverage Radius

We evaluate in this section the FAP coverage radius. The analysis is done from measurements collected along the three main corridors of the building floor: corridors C1, C2 and C3 as defined in Figure 39. This permits to separately investigate the coverage radius in different visibility conditions.



Figure 39: Presentation of the three main corridors.

Figure 40 shows the evolution of Ec/Io as a function of the distance to the active FAP, measured respectively from scanning the MBS and FAP. This evolution is averaged in order to determine:

- The coverage radius, defined as the minimum distance where the average FAP Ec/Io goes below the threshold;

- The dead-zone radius, defined as the maximum distance for which the average MBS Ec/Io is below the threshold;

- The femto-to-macro handover distance, defined as the minimum distance where the average FAP Ec/Io goes below the average MBS Ec/Io.



VC

1

a. FAP 2 ON – Ec/Io along C2 b. FAP 3 ON – Ec/Io along C2

VC

2

c. FAP 4 ON – Ec/Io along C3 d. FAP 2 ON – Ec/Io along C1

VC

3

e. FAP 1 ON – Ec/Io along C1 f. FAP 1 ON – Ec/Io along C3

VC

4

FAP 5 ON – Ec/Io along C3

Figure 40: Ec/Io versus distance, measured respectively from FAP or MBS scanning.



Averaged distances are summarized in Table 39.

Ec/Io > -14dB

RSCP > -98dBm

Ec/Io > -10dB

RSCP > -98dBm Femto-to-Macro HO

distance FAP coverage radius

Macro deadzone

FAP coverage radius

Macro deadzone

VC1: LoS in corridor FAP2/C2: >50m

FAP3/C2: >45m

FAP2/C2: >41m

FAP3/C2: >45m

FAP2/C2: >50m

FAP3/C2: >45m

FAP2/C2: >50m

FAP3/C2: >45m

FAP2/C2: >50m

FAP3/C2: >45m

VC2: crossing one light partition

FAP2/C1: >50m

FAP4/C3: >18m

FAP2/C1: 14m

FAP4/C3: 7m

FAP2/C1: 40m

FAP4/C3: >18m

FAP2/C1: 38m

FAP4/C3: >18m

FAP2/C1: 38m

FAP4/C3: >18m

VC3: crossing several partitions

FAP1/C1: >50m

FAP1/C3: 16m

FAP1/C1: 7m

FAP1/C3: 6m

FAP1/C1: 30m

FAP1/C3: 16m

FAP1/C1: 10m

FAP1/C3: 18m

FAP1/C1: 13m

FAP1/C3: 16m

VC4: two different floors

FAP5/C3: 7m FAP5/C3: <4m FAP5/C3: <4m FAP5/C3: <4m FAP5/C3: <4m

Table 39: FAP indoor coverage radius and macro coverage deadzone radius for different visibility conditions.

This leads to the following observations:

- The FAP coverage radiuses at Ec/Io > -10dB seem at least superior to 45–50m in VC1 whereas it is potentially limited at approximately 40m in VC2 and between 16 and 30m in VC3. Furthermore it is observed that the FAP inter-floors radius coverage is very limited (i.e. at least inferior to 4m).

- Femto-to-macro handover occurs generally when the collected FAP Ec/Io becomes lower than -10 dB.

The results presented in this section allowed us to derive a simple analytical model for the average FAP impact in a WCDMA network. This model is reported in [FREEDOM-D622]; it is used to simulate other situations (higher mean macro level, lower mean macro level, higher FAP transmit power) in order to derive global engineering rules for FAP deployment within an existing macro network.

5.1.2.3 Outdoor Analysis

The impact of a FAP deployment on the outdoor network coverage is also evaluated. Figure 41 shows the outdoor Ec/Io as a function of the relative distance to the FAP, when FAPs n°2 and n°3 are respectively activated. Table 40 reports the outdoor radius coverage. Remark that only a limited amount of data has been collected outdoors as illustrated in Figure 42.



a. Outdoor Ec/Io Vs distance to FAP 2 b. Outdoor Ec/Io Vs distance to FAP 3 Figure 41: Outdoor Ec/Io versus distance, measured respectively from FAP and MBS

scanning.

a. FAP 2 b. FAP 3

Figure 42: Measured points collected outdoors for FAP n°2 and FAP n°3.

Ec/Io > -14dB

RSCP > -98dBm

Ec/Io > -10dB

RSCP > -98dBm Femto-to-Macro HO

distance FAP coverage radius

Macro deadzone

radius

FAP coverage radius

Macro deadzone

radius

FAP 2 12m <4m 10m <4m <4m

FAP 3 23m <8m 21m 36m 21m

Table 40: FAP outdoor coverage radius and Macro outdoor deadzone radius.

A sharp decrease of the FAP RSCP level is observed at distance 10m for FAP n°2 and distance 20m for FAP n°3, which limits the FAP coverage whatever the Ec/Io threshold level is.

25m

20m



5.1.2.4 Throughputs

Downlink (DL) and uplink (UL) throughputs have been collected by the TRACE mobile receiver. Figure 43 shows the DL and UL throughputs as a function of the distance to the active FAP. Remark that the throughput data collected with NMS has not been currently exploited.

a. UL Throughput Vs distance to FAP b. DL throughput Vs distance to FAP

Figure 43: UL and DL throughputs Vs distance to FAP.

Macro-only network throughput is the average throughput measured along the reference route without any FAP deployment. Only the throughput of the FAP n°1 presents a decrease within the measured distance range, becoming under the mean throughput reference at approximately 12m to the FAP in UL and DL. We do not have sure explanation for this phenomenon but evolution of throughputs according the distance to the active FAP seems follow the same overall trend as Ec/Io levels (see section 5.1.2.2). FAPs n°3 and n°4 provide a quasi constant throughput about 4 times higher than the macro reference in UL, and about 6 times higher in DL. The FAP n°2 delivers a lower throughput, however we do not have any explanation for this observation.

5.1.3 Impact of Inter-FAP Distance (Scenarios A3 and B3)

The scenarios A3 and B3 characterize the radio coverage and interference as a function of the inter-FAP distance (IFD). A FAP with fixed position (i.e. FAP n°2 in Figure 39) is located in a room close to the corner of two perpendicular corridors where measurements are collected. Two other FAPs (namely FAPs n°1 and n°3) are installed in each corridor at the same distance from FAP n°2. Different IFD are investigated, as given in Table 41. Please refer to section 4.1.4 for a detailed description of the scenario.

Measured inter-FAP distances (IFD)

30m 25m 20m 15m 10m

Table 41: Inter-FAP distance in scenarios A3&B3.

The metrics of interest (RSCP, RSSI and Ec/Io) are characterized within the inter-FAP area only, i.e. within the area delimited by the FAP n°1 and FAP n°3. Remark that the size of this inter-FAP area depends on each specific investigated IFD. These metrics are compared to results of the reference scenario 0.2 where the FAP n°2 was the unique activated FAP. Only SCANNER measurements have been used here. Figure 44 and Figure 45 show RSCPFAP2 as a function of the distance to FAP n°2 respectively along the corridor C1 and C2. A 5m averaging window (about 3 wavelengths) is applied to remove the



small-scale fading (i.e. the dark blue lines). The signal decrease is faster in C1 because the scanner equipment is in non-line-of-sight. On the contrary, the corridor C2 is in line-of-sight. Remark the large signal variations in C3 at ranges 40-50m. This is because measurements are collected in the room at the South-West part of the corridor C2, where different radio link configurations are encountered for identical distances (i.e. line-of-sight and non-line-of-sight). Note that the average RSCPFAP2 is greater than -80dBm at range 30m, which is the higher investigated IFD. As propagation conditions are different, the analysis of inter-FAP distance is conducted separately along corridors C1 and C2.

Figure 44: FAP2 RSCP Vs distance along C1.

Figure 45: FAP2 RSCP Vs distance along C2.

Figure 46 show measured RSCP and Ec/Io from each FAP and best-server along corridor C1 for IFDs = 10m, 20m and 30m as well as a comparison between RSSI with FAP n°2 alone and all three FAPs together. Figure 47 presents the CDF of these metrics (best-server in dotted line and FAP n°2 only in solid line).



IFD=30m IFD=20m IFD=10m

RS

CP

Ec/

Io

RS

SI

Figure 46: RSCP, Ec/Io and RSSI along C1 for IFD = 10m, 20m and 30m.

RSCP Ec/Io RSSI

Figure 47: CDF of RSCP, Ec/Io and RSSI for different IFDs. Dotted lines relate to best

server curves and solid lines relate to FAP2 only curves.

We observe that all situations provide Ec/Io levels greater than -10dB, meaning that service is always available. In all cases, the FAP n°3 located in corridor C2 provides a RSCP lower than the two other FAPs, and acts only as an interferer. At IFD = 20m or 30m, the FAPs n°2 and n°1 are alternatively the best-server, thus the best RSCP and Ec/Io are optimised compared to the single-FAP deployment. Highest Ec/Io values are notably obtained with IFD = 20m, which appears as the most robust configuration to combat interference (remark that interference levels may be higher than measured here in case of a large traffic on the FAPs). On the contrary, at IFD = 10m, the RSSI is maximum and best-server is always provided by the same FAP. Thereby, this IFD does not generate RSCP optimisation and even less for Ec/Io optimisation.



Furthermore, we observed that the observations made on RSCP measured along corridor C1 are also valid along the second corridor (i.e. C2). Indeed the best RSCP is maximized for an IFD = 20m or 30m. Besides, all situations provide Ec/Io levels greater than -10dB meaning that service is available. Nevertheless, the most robust configuration to combat interference are obtained with IFD = 30m against 20m along corridor C1. Figure 48 shows the CDF of best-server Ec/Io and measured RSSI along corridor C2 for IFDs = 10m, 20m and 30m (best-server in dotted lines and FAP n°2 only in solid lines). Remark that we may expect that the optimal IFD is superior to 30m in that case. Furthermore, the best-server is not always provided by a unique FAP when the IFD=10m as illustrated in Figure 49. Ec/Io RSSI

Figure 48: CDF of Ec/Io and RSSI for different IFDs along C2. Dotted lines relate to best server curves and solid lines relate to FAP2 only curves.

Figure 49: Ec/Io levels along C2. IFD = 10m.

Table 42 summarizes the outage percentage 80% along both corridors for IFDs = 10m, 20m and 30m: all the metrics of interest are considered (i.e. best-server RSCP, Ec/Io and RSSI).



IFD = 30m IFD = 20m IFD = 10m Best-server RSCP outage

C1 -62 dBm -60 dBm -55 dBm C2 -62 dBm -60 dBm -54 dBm

RSSI outage C1 -55 dBm -55 dBm -50 dBm C2 -56 dBm -52 dBm -48 dBm

Best-server Ec/Io outage

C1 -9 dB -6 dB -6 dB C2 -7 dB -10 dB -7 dB

Table 42: RSCP, RSSI and Ec/Io for outage percentage 80% as a function of IFD.

It can be observed that the IFD influences strongly both RSCP coverage and interference level along each corridor. Whereas the best-server RSCP outage is comparable for each IFD, inference level is higher along the corridor C2. Lastly, as previously observed, the most robust configuration to combat interference are respectively obtained with IFD = 20m along C1 (non-line-of-sight) and IFD > 30m along C2 (line-of-sight).

5.1.4 Impact of Generic Multi-FAPs Deployment (scenarios G1 & G2)

The generic scenarios G1 and G2 aim at reproducing a realistic multi-FAPs deployment within a corporate building. All FAPs used in the scenario 0.2 are simultaneously activated in a closed access mode. There is no traffic generated on the FAPs in G2, while modems are used to generate traffic on all FAPs simultaneously in G1. The location of these modems is described in Figure 31. Measurements were collected with SCANNER and TRACE mobile receivers. The TRACE mobile receiver was used as two-tier user in scenario G1 and as FUE for scenario G2. Please refer to section 4.1.7for a detailed description of the scenarios.

5.1.4.1 Indoor Coverage

Table 43 summarizes the indoor coverage for both scenarios G1 and G2 that would have been experienced by a:

- Single-FAP only user (measured from FAP SC scanning) - Multi-FAPs only user (from the FAP best-server, i.e. i.e. the UE connects to the FAP with the

best Ec/Io) - Macro-only user (measured from MBS SC scanning) - Two-tier network user (from the best-server, i.e. the UE connects to the server with best Ec/Io)

The obtained results are compared to the macro-only network cover rate get in scenario 0.1.

Ec/Io > -20 dB Ec/Io > -14 dB

RSCP > -98dBm Ec/Io > -10 dB

RSCP > -98dBm

User profile No modem traffic No modem traffic No modem traffic

Single-FAP

FAP1-only 53% 39% 30%

FAP2-only 51% 37% 29%

FAP3-only 45% 32% 27%

FAP4-only 32% 23% 20%

FAP5-only 6% 0% 0%

Multi-FAPs 100% 100% 94%

Macro-only (FAPs activated) 73% 33% 7%

Two-tier network 100% 100% 96%

Macro-only (no FAP) 100% 100% 77%

Table 43: Indoor coverage for multi-FAPs scenario.



Figure 50 shows the CDF of the Ec/Io experienced by three different user profiles as well as the CDF of the measured RSSI. Green and blue lines are respectively the CDF with modem traffic (i.e. G1) and without traffic modem (i.e. G2). The dotted black line is the reference CDF from the macro-only network.

Ec/Io

RSSI

Figure 50: Ec/Io and RSSI CDF from a multi-FAPs deployment.

It can be observed that:

- The extra RSSI level generated by the modems in G1 is very limited. Actually the number of available modems was too small to produce significant data traffic in this scenario.

- A multi-FAPs user gets a quasi full coverage for both Ec/Io > -10dB and Ec/Io > -14dB thresholds. Besides, Ec/Io > -5dB is reached in 50% of measured locations.

- The cover rate of single-FAP user decreases compared to scenario 0.2 (by approximately 10%, see Table 38), resulting from inter-FAP interference. Note that the FAP n°2 is the most impacted (-20% on indoor cover rate): the repartition of measured points indoors as well as the proximity of FAP n°1 and FAP n°3 may explain this observation.

- The cover rate of a two-tier user is only slightly better than the one from a multi-FAPs user. Indeed, the best server is quasi exclusively from the FAP layout all along the measured trajectory.

- The cover rate of a macro user is strongly impacted. It decreases from 100% to 33% at threshold Ec/Io > -14dB, and from 77% to 7% at threshold Ec/Io > -10dB.

Thereby, the coverage and capacity is strongly improved for two-tier users. The four FAPs located in the ground-floor offer a quasi full coverage (the macro best-server area is very marginal). In return, the multi-FAPs deployment quasi-totally eliminates the service coverage initially offered by the macro network.

5.1.4.2 Outdoor Coverage

Impact of the multi-FAPs deployment is also studied outdoors. Collocated measured points for the three compared scenarios (i.e. macro-only network and macro+femto network with and without modem traffic on FAPs) are illustrated in Figure 51. Figure 52 shows the evolution of the Ec/Io experienced by a macro user. The cover rate at Ec/Io > -10dB decreases from 95% in the macro-only network to respectively to 70% and 75% with or without modem traffic. The mean Ec/Io decreases respectively of 1.9 dB and 1.7 dB.



Figure 51: Measured points outdoors (used for CDF computation)

Figure 52: Outdoor Ec/Io CDF for a macro user.



a. G1 Ec/Io Vs distance to FAP 1 b. G2 Ec/Io according to distance to FAP 1

c. G1 Ec/Io Vs distance to FAP 2 d. G2 Ec/Io Vs distance to FAP 2

e. G1 Ec/Io Vs distance to FAP 3 f. G2 Ec/Io Vs distance to FAP 3

g. G1 Ec/Io Vs distance to FAP 4 h. G2 Ec/Io Vs distance to FAP 4 Figure 53: FAP Ec/Io and macro Ec/Io Vs distance to FAP for G1 and G2.



As observed in the scenario 0.2, there is a sharp decrease of the Ec/Io level at short outdoor distance. Thus the FAPs offer an outdoor coverage only within a very limited range:

- Until ~10m at threshold Ec/Io > -10dB; FAPs n°2, 3 and 4 in G2 only. - Until ~15m at threshold Ec/Io > -14dB.

Remark that compared to the scenario 0.1, an average difference of approximately -2.5dB is obtained on the Ec/Io levels outdoors for comparable traffic load conditions in the macro network. Furthermore, the MBS’s are always the best-server at a distance from a FAP greater than 20m. We observe a strong impact at distance less than 10m (close to FAP n°4) or less than 20m (close to FAP n°3), which can lead to macro service interruption if low Ec/Io is required and access mode is closed. Remark that none of the FAP locations was close to an external window.

5.1.4.3 Indoor Throughput

Throughput measurements were collected from the TRACE mobile used either as two-tier user (scenario G1) or a FAP-only user (scenario G2). DL and UL throughput distributions are given in Figure 54. Note that no measurement was collected along the corridor C2 for the scenario G2 (because of a measurement issue). Thereby, data collected along this corridor for other scenarios have been filtered out in the present analysis. As expected, the multi-FAPs deployment offers globally larger throughput than the macro-only network: Throughput is increased in about 60% measured locations. - DL throughput for 30% measured locations increases from 600kbps in the macro-only network to

1500kpbs in the two-tier network; - UL throughput for 30%% measured locations increases from 35kbps in the macro-only network to

75kpbs in the two-tier network; However two unexpected behaviours have been observed. Firstly, even if the TRACE mobile was always connected either to a MBS or a FAP, a null throughput was collected for many measured points scattered along the trajectory (i.e. approximately 30% of measured points on DL). We do not have sure explanation for this phenomenon. Is it related to the equipment mobility and handover procedures? Is it related to a problem in the testbed installation or to a limitation in the TRACE mobile? Additional investigation would have been necessary but is not available. Furthermore, unexpected very high throughputs (until 8Mbps) were also collected locally for the MBS. Thereby, we recommend reading these results with the greatest caution, and comparing (if possible) them to other similar measurement trials. Note that the potential bias of the lowest and highest values of the throughput distributions motivates us to base our analysis on the quantile 30% instead of the mean or median.



a. DL throughput b. UL throughput Figure 54: DL and UL throughput CDF for two-tier users and FAP-only users.

Secondly, the throughput delivered to the two-tier user is globally lower than throughput delivered to the FAP-only user. This is probably due to the handover procedure, which favours connection to the MBS even when a FAP is the best-server. We do not have precise explanation. We can only raise this issue, and recommend the implementation of efficient handover procedures to benefit from the maximum available throughput.

5.1.5 Summary

Radio measurements collected in the ground-floor of TELKOM testbed building permitted us to characterise the impact of a single-FAP or multi-FAPs deployment into a typical corporate environment with only poor macro WCDMA coverage. Analysis started from a characterisation of the macro coverage state: mean RSSI interference level of -75dBm; and mean Ec/Io signal-to-interference level of -8.5dB. Following femto-based results are of course constrained by these macro coverage metrics. Single-FAP deployments with 10dBm maximum transmit power are analysed to evaluate the FAP coverage radius, in line-of-sight and several non-line-of-sight situations: from more than 50m to 18m only at threshold Ec/Io > -14dB. The macro deadzone radius in case of a closed-access mode FAP deployment is in the range from more than 50m to 10m at threshold Ec/Io > -10dB. In a similar way, the FAP-to-macro handover distance goes from more than 50m to 13m. The impact of a FAP located in the upper floor seems very limited, with FAP coverage radius and macro deadzone radius around 4m. The Outdoor leakage is quite limited as well. The FAP coverage radius for outdoor users was measured about 10m and 20m. The degradation on the outdoor macro coverage is very small, and even not obvious. Of course, higher impact would have been observed from a FAP with 20dBm maximum transmit power as also allowed in commercial products. Note also that all FAPs were all more than 4m away from the building external façade. Besides, throughput measurements illustrate the ability of the FAP to deliver a 6 times higher DL throughput than the original macro network (up to 3000kbps) and 4 times higher UL throughput (up to 150kbps). Impact of the inter-FAP distance has been evaluated as well (on a carrier different than the macro network). The optimum inter-FAP distance for a maximum Ec/Io is about 20m in light non-line-sight-



situation (Ec/Io=-8dB at outage 90%) and 30m or above in light-of-sight situation (Ec/Io=-6.5dB at outage 90%). A deployment of 4 FAPs in the ground-floor of the TELKOM testbed building (about 2200m²) offers 100% coverage and 96% coverage at respectively Ec/Io > -14dB and Ec/Io > -10dB along the measurement route (located mainly in the central corridor), while the cover rate from the macro only network was only 77% at Ec/Io > -10dB. The throughput delivered to an indoor FAP subscriber (or to any subscriber in open-access mode) is significantly increased: x2.5 in DL and x2.15 in UL for 30% of measured locations. However some unexpected behaviours have been observed on the measured throughput: null throughput at about 30% of measurement locations; much higher throughput delivered to a FAP-only user (i.e. no authorised access to the macro). This can be due to issues in the measurement methodology or equipment, but could also result from suboptimal handover procedures. Results need to be confirmed. Of course, the DL FAP-to-macro user interference reduces the macro coverage to only 29% and 6% at respectively Ec/Io > -14dB and Ec/Io > -10dB. That shows how strongly a closed-access mode FAP deployment degrades the service offered to non-subscribers. Note that the remaining macro coverage surface is scattered and consequently does not permit any mobility. We observed that the outdoor FAP coverage only extends to about 10m, and sometimes 20m from the closest FAP. In open-access mode, this multi-FAPs deployment would only provide a very limited gain on the outdoor coverage quality. So far, the scenarios A2/B2 (that aim at characterizing the impact of DL MBS on FUE and the impact of UL MUE on FAP for a single-FAP deployment) and A1 (that aims at characterizing the impact of DL FAP on MUE for a single-FAP deployment network with modem traffic) were not exploited due to the lack of time and as the priority was given to scenarios based on SCANNER measurements, which allow a more generic approach. Furthermore, limited outcomes are expected from the analysis of these remaining scenarios as only few data are available due to several issues with the TRACE mobile during the measurements.

5.2 xDSL as Femtocell Backhaul

5.2.1 xDSL Access Network Characteristics

This section presents xDSL characteristics which follow the parameter setting and assumption as discussed in section 4.2.1. Figure 55 and Figure 56 show the performance of ADSL2/ADSL2+ and VDSL2, where the DSLAM/MSAN is located in local exchange and street cabinet respectively.



Figure 55: ADSL2 & ADSL2+ Performance in FTTE (MSAN in Exchange)

Figure 56: ADSL2, ADSL2+, VDSL Performance in FTTC (MSAN at street cabinet)

In general, attainable data rate depends on xDSL technology. VDSL2 deliver higher data rate since it uses wider frequency plan which up to 12 MHz compared to ADSL2+ which uses 2.2 MHz and ADSL2 which uses 1.1 MHz. As the cable length increases the total impedance will also increase, it will add more attenuation to the signal. As a result the data rate will decrease as the cable length increase. Higher frequency will experience more sever attenuation as a function of distance, so that the higher data rate can only be maintained in shorter distance. As can be seen from the graph, the maximum attainable rate also depends on xDSL channel mode. DSL is designed to deliver internet access. By default it uses fast channel mode, since it will offer a



higher efficiency (more data, less error correction redundancy code in each packet). The fast channel mode will allow users to have faster and smaller ping times. However as the real time applications such as IPTV was introduced to the market, interleave channel mode is required. In video applications, there is no time to re-transmit data if errors are detected. In order to limit the impact of long burst errors, an interleaver device is used to spread the data out or shuffles the data after it is encoded by the Reed-Solomon code [Golden06]. By using Reed Solomon and interleaver as in ADSL and VDSL technology, long error bursts will be equally distributed, so that the errors can be corrected more easily using forward error correction. Since there are bits used for codeword, bits number for data in the interleave mode will be less, hence it will affect the total data rate. ADSL2/ADSL2+ has limitation in the upstream bandwidth which is up to 1.1 Mbps for FAST mode and maximum 900 kbps for INTERLEAVED mode. However this relatively high upstream bandwidth is only available if the cable length is less than 600m from MSAN/DSLAM location. To ensure the stability, 512 kbps bandwidth should be considered for both modes, since the bandwidth is available even when the cable length 6 km away from MSAN location. For higher upstream bandwidth, VDSL2 should be considered under FTTC configuration. With VDSL2, the upstream data rate can reach around 5 Mbps at the range up to 1 Km.

Figure 57: Performance of VDSL2 in FTTB (MSAN at Building, ONU-B Total Replacement)

Figure 57 shows the performance of VDSL2 if MSAN or ONU installed in the building. In the high rise building for instance, the ONU can be installed in the ground floor; in order to support both voice and packet data services, operator or building management can use copper cable to reach each floor in the building by using VDSL2. This is only an option in case there are plenty of PSTN customers in the building or existing copper cable already installed in the building. Other alternatives are using fiber or UTP cable (ethernet) which can be distributed from ONU to each floor. As can be seen from the graph, VDSL2 30a provides maximum data rate almost double compared to 17a. VDSL2 30a and 17a represent VDSL2 profile which can utilize 17.664 MHz bandwidth and 30 MHz bandwidth respectively.



5.2.2 xDSL Network Performance (Scenario 0)

Measurement results based on xDSL-Scenario 0 are presented in this section to give further information about xDSL characteristics and also to show that the testbed is functional and ready to support femtocell performance observation. Based on xDSL lastmile performance test which used Spirent Smartbit, the average transmission delay for various QoS setting in the DSLAM and modem can be shown as in Table 44.

Frame Size Fast Mode Interleave Mode

Latency Measurement (µs) 1024 2687,55 8562,3 1518 3003,3 8917

Throughput Measurement 1024 13,312 13,576 1518 13,656 13,656

Packet Loss 1024 0 % 0 % 1518 0 % 0 %

Table 44: xDSL Layer 2 Performance Matrix (Scenario 0)

DSLAM is configured using FAST and INTERLEAVE. SmartBit generated layer-2 traffic which transmits 1024 byte and 1518 frame size. We use D-link modem and ZTE ZXA10 DSLAM for the testbed. In terms of transmission delay, it can be seen from the table that the bigger the frame size, the higher the latency. Latency for interleave is higher than the one in fast mode. In order to characterize xDSL as a backhaul, we also test transmission delay of ADSL and ADSL2+ from modem to the DSLAM as can be seen in Figure 58. Since it is not possible to measure transmission delay against different cable length, the delay is measured over various bandwidth profiles i.e 64 kbps, 128 kbps, 512 kbps, 1 Mbps, 2 Mbps, 3 Mbps, etc. The result is shown in Figure 59.

Figure 58: Test Configuration for Scenario 0.1



0

5

10

15

20

25

30

35

40

45

50

64

128

384

512

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

12000

13000

14000

15000

16000

17000

18000

19000

20000

21000

22000

23000

24000

transm

ission delay (m

s)

data rate (kbps)

ADSL Delay (ms) Interleave ADSL Delay (ms) Fast

ADSL 2+ Delay (ms) Interleave ADSL 2+ Delay (ms) Fast

Figure 59: Average delay transmission of ADSL/ADSL2+ over various bandwidth

profile

According to the figure, the delay at 64 kbps bandwidth is about 40 ms for fast mode and about two times for inteleave. As the bandwidth increases the delay decreases since wider frequency bandwidth is required to produce more throughput. In order to obtain end-to-end delay experienced by a femto user, a FAP attached to xDSL modem within OASIS testbed as can be seen from Figure 16. Delay measurement is performed using Femto NMS and ping test for end-to-end delay. The round-trip delay experienced by FUE and contribution delay from FAP-SecGW and SecGW-SGSN are shown in Figure 60.

Figure 60: Round-trip delay experienced by a FUE accessing internet.



It can be inferred from the graph that the delay between FAP-SecGW and SecGW-SGSN is stable since it is provisioned for carrier grade class to guarantee the performance. Eventhough the OASIS is located in Bandung which 200 km away from TELKOMSEL 3G core in Jakarta, the performance is maintained using layer-2 separation in metro-ethernet. The delay fluctuation is experienced by FUE user, since GGSN is commercial 3G core and has traffic load varied overtime. Furthermore, since the ping target is a server in internet cloud, the internet segment may contribute to the delay fluctuation.

5.2.3 Femtocell Bandwidth Requirement (Scenario 1)

In this section effective bandwidth consumed by a FAP will be observed. Bandwidth profile for xDSL is set to 20 Mbps so that the original bandwidth required to send traffic will go smoothly without any congestion or queuing in the access network. Since we use commercial GGSN, the traffic will be affected by the GGSN load, however we anticipated by using 3G SIM Card with high priority QoS profile. Furthermore the individual test is repeated several times, i.e. 30 times effectively, but only the best 3 retries will be shown. As been stated in [FREEDOM-D21]; each UE accesses one service. Firstly, we captured individual bandwidth consumed by http, ftp, voice (AMR) and youtube streaming. The service was accessed by a smartphone connected to a FAP. Secondly, we observed the throughput for four simultaneous voice calls and mix traffic HTTP, FTP, youtube and voice from four different smartphone at the same time. The throughput for individual service and mix traffic can be seen in Figure 61. We repeated the similar observation for iPad. Since iPad is designed to access packet data only, hence voice AMR call cannot be tested. We replaced the voice call with facebook. We understood that most ipad users frequently download new applications, audio-video podcasts, mp3 music, ebooks from iTunes or Apple Store. For similarity with the observation in smartphone, we used ftp traffic from 4shared.com to download bigger file size compared to the one from smartphone. We also used the same page from detik.com. While from smartphone mobile page version was displayed, in the iPad full web page was displayed, hence the consumed bandwidth is different. The throughput of individual service accessed from an iPad can be seen in Figure 62, while the traffic mix from 4 different iPads is shown in Figure 63.



a. Single traffic – HTTP (m.detik.com)

b. Single traffic – FTP (4shared.com)

c. Single traffic – Voice (AMR)

d. Single traffic – Streaming Youtube

e. Mix traffic – 4 voice

f. Mix traffic – HTTP, Video, Voice, FTP

Figure 61: Throughput of (a) HTTP, (b) FTP, (c) Voice (AMR), (d) youtube, (e) 4

simultaneous voice call and (f) mix traffic accessed from smartphone



a. Single traffic – HTTP (detik.com)

b. Single traffic – HTTP (facebook.com)

c. Single traffic – FTP (4shared.com)

d. Single traffic – Streaming Youtube

Figure 62: Throughput of (a) HTTP-Web, (b) HTTP-facebook, (c) FTP (d) youtube from smartphone iPad

Figure 63: Throughput of mix traffic (detik.com, facebook, ftp and youtube) from smartphone iPad

The summary of statistics properties for both smartphone and iPad bandwidth requirement are shown in two tables below.



Traffic Content Strm Max

(bps) Min (bps)

Average (bps)

Variance Distribution

Cs voice AMR 12,2 Kbps DL 84,800 8,480 77,372.7 466701399 Lognormal UL 84,800 8,480 77,372.7 466701399 Lognormal

HTTP m.detik.com, first page DL 224,720 4,240 42,995.5 3116322837 Lognormal UL 106,000 0 29,400.4 886953056 Lognormal

FTP www.4shared.com “05.The Lazy Song”, 5MB

DL 576,640 4,240 185,196.8 4640833694 Normal UL 42,400 0 16,178.7 43352265 Lognormal

Streaming

m.youtube.com, “If you sleep in at my house, you are "Doom"ed”, 41s, 240p, 380 Kbps

DL 339,200 29,680 278,521.7 7042225332 Normal

UL 71,016 8,480 27,326.68 148142570 Lognormal

Mix 4 CS call DL 339,200 8,480 298,168.5 8044648128 Lognormal UL 339,200 8,480 298,168.5 8044648128 Lognormal

Mix All traffic DL 606,320 132,496 431,063.5 19280386460 Lognormal

UL 157,936

21,200

58,089.91

768972191

Lognormal

Table 45: Statistical properties of individual and mix traffic (Scenario 1.1.1 smartphone)

Table 46: Statistical properties of individual and mix traffic (Scenario 1.1.2 iPad)

Based on the figures and tables for mix traffic, we can see that the bandwidth required for a femtocell depends on the type of traffics. Assuming maximum value is used, in the downlink side it is required about 607 kbps to perfectly handle mix traffic from 4 smartphones while in uplink is about 158 kbps. In case of iPad being used, the downlink bandwidth consumption is about 857 kbps and the uplink is about 632 kbps. The uplink traffic for iPad is higher than smartphone because mix traffic mainly used by youtube. However this is happened not all the time, since the average throughput of the uplink traffic is about 89 – 106 kbps. Furthermore the http traffic can be considered as best effort. So the 632 kbps for uplink is should be further elaborate to get more reasonable picture. So far we have derived the bandwidth requirement for femtocell by monitoring the throughput from Femtocell NMS:

BR for smartphone = 607 kbps (DL) and 158 kbps (UL) BR for iPad = 857 kbps (DL) and 632 kbps (UL)

We further verified the BR by executing scenario 1.2. In this scenario, we generated traffic using Spirent Test Centre Avalanched, where HTTP (m.detik.com), FTP 5 Mbps can be emulated. Eventhough youtube steaming can be accessed by STC-A, according to the measurement log, youtube is delivered as HTTP protocol, so only response time can be measured; therefore we use H264 video streaming traffic as alternative. In this case, we can monitor the effect of reducing xDSL bandwidth

Traffic Content Strm Max (bps)

Min (bps)

Average (bps)

Variance Distribution

HTTP www.detik.com, first page DL 1,094,976 12,720 282,005.8 2,24269E+11 Normal UL 168,536 0 94,768.0 2549473195 Normal

HTTP www.facebook.com, home page

DL 561,800 0 109,618.1 18271227072 Lognormal UL 210,936 0 45,605.9 3675590487 Lognormal

FTP

www.4shared.com “divxim.net-kLite Codec Pack 4.9.5 FULL”, 13,6MB

DL 407,040 4,240 213,622.6 3926226329 Normal

UL 59,360 0 22,047.5 65333480 Lognormal

Streaming

m.youtube.com, “If you sleep in at my house, you are "Doom"ed”, 41s, 240p, 380 Kbps

DL 339,200 4,240 280,032.6 4284588520 Normal

UL 48,760 12,720 27,491.1 61155798 Lognormal

Mix All traffic DL 857,536 67,840 501,702 10293926961 Normal

UL 631,760 12,720 92,369,06 13009342028 Lognormal



profile from 20 Mbps to the level just below BR and another one just higher than BR. Thanks to H264 video streaming, we can observe the affect of link changes to the video and audio performance including bitrate, delay, jitter and frame per second. Since STC does not support AMR, we generated AMR traffic using a smartphone, so that we can keep the total bandwidth as close to the BR obtained from scenario 1.1 for both smartphone and iPad. The comparison between mix generated traffic and mix traffic from iPad can be seen in Figure 64.

Real Traffic Generated Traffic

max 860,720 1,234,368

min 67,840 4,240

average 482,017 526,394

Figure 64: The comparison between generated traffic and iPad mix traffic

Link Profile

Average Video Quality for 5 minutes observation Bitrate (kbps) Packet loss (%) Jitter (ms) FPS Tx Rx Tx Rx Tx Rx Tx Rx

20 Mbps 76.43 36.98 0.87 2.38 14.70 9.96 28.89 6.32650 kbps 90.73 50.88 3.28 3.06 16.49 24.68 28.73 6.26

Link Profile

Average Audio Quality for 5 minutes observation Bitrate (kbps) Packet loss (%) Jitter (ms)

Transmit Receive Transmit Receive Transmit Receive 20 Mbps 47.75 44.60 0.51 0.28 7.60 3.15650 kbps 24 25.75 1.49 0.63 9.86 13.98

Table 47: H264 Video Conference Performance (Scenario 1.2.1 smartphone)

According to the Table 47 and Table 48, when the xDSL bandwidth profile set to 650 Kbps, the packet loss of video quality is just above the limit 3%. By increasing to 800 Kbps or equal to BR for iPad, the video quality for iPad is maintained below the threshold; the packet loss below 3% and the jitter far below than 40 ms. In conclusion, 800 Kbps can be used for bandwidth reference for a FAP to accommodate mix service both for smartphone and iPad. Giving 20% more bandwidth will allow additional space to anticipate other header and load during busy hours. Hence, we conclude that 1 Mbps bandwidth can be used.



Mode Average Video Quality for 5 minutes observation

Bitrate (kbps) Packet loss (%) Jitter (ms) FPS Tx Rx Tx Rx Tx Rx Tx Rx

20 Mbps 83.58 23.53 0.53 2.45 13.30 19.53 28.25 10.681 Mbps 78.93 22.93 0.34 2.47 12.23 20.75 28.78 21.45900 Kbps 77.45 18.71 0.35 1.11 12.70 23.80 29.25 10.57800 Kbps 84.70 20.26 0.37 2.12 13.45 24.96 28.83 18.84

Mode Average Audio Quality for 5 minutes observation

Bitrate (kbps) Packet loss (%) Jitter (ms) Transmit Receive Transmit Receive Transmit Receive

20 Mbps 49.33 25.38 0.14 0.47 7.38 3,31 Mbps 49.33 25.38 0.14 0.47 7.38 3,3900 Kbps 48.00 48.35 0.05 0.38 7.11 8,3800 Kbps 54.81 27.21 0.16 1.25 6.68 9.9

Table 48: H264 Video Conference Performance (Scenario 1.2.2 iPad)

5.2.4 Femtocell Service Performance (Scenario 2)

In this section, the performance of a femtocell service is observed in the presence of background traffic in xDSL modem. This observation will effectively address the nature of FAP which is customer premises equipment. It is most likely that the user will plug the FAP to xDSL modem or home gateway on top of the existing broadband access in the home. Without prior notice, the femtocell service will be mixed with traffic from PC or other devices connected to the same modem. The femtocell and PC are connected to xDSL modem using single PVC so that the traffic will mix each other. We set the DSLAM to interleave mode, while the modem uses default UBR type of service. The PC generated mixed traffic HTTP (www.detik.com), FTP (rapidshared 40 Mbps) and youtube. Femtocell is attached to serve 4 FUEs simultaneously. We are referering to iPad case to inline with the BR obtained from previous observation. The mix traffic from femtocell and PC can be seen in Figure 65.

Background Mix traffic Mix+Background

max 2,839,920 1,234,368 3,280,096

min 1,093,248 4,240 1,285,244

average 1,703,366.8 526,393.8 2,229,760.6

variansi 158653351055 66753348160 280358769397

Figure 65: Aggregate Throughput in xDSL link, BW profile set to 20 Mbps



Mode Average Video Quality

Bitrate (kbps) Packet loss (%) Jitter (ms) FPS Tx Rx Tx Rx Tx Rx Tx Rx

20 Mbps 81.83 24.97 1.25 2.47 13.92 18.05 29.25 18.492 Mbps 84.25 23.73 0.29 2.99 13.62 35.22 28.80 11.131,5 Mbps 85.28 26.48 0.38 4.43 15.70 29.95 28.65 19.311,2 Mbps 82.26 23.19 0.33 5.94 14.64 41.11 28.75 10.991 Mbps 89.07 27.85 0.12 3.96 13.87 45.52 28.83 11.081 Mbps No bckgr 78.93 22.93 0.34 2.47 12.23 20.75 28.78 21.45

Mode Average Audio Quality

Bitrate (kbps) Packet loss (%) Jitter (ms) Transmit Receive Transmit Receive Transmit Receive

20 Mbps 56.67 58.92 0.82 0.66 7.02 3.952 Mbps 52.00 25.33 0.18 0.95 7.35 14.081,5 Mbps 27.43 25.02 0.07 1.22 8.45 16.881,2 Mbps 47.08 26.92 0.13 1.67 7.00 16.851 Mbps 28.57 25.45 0.05 1.32 10.20 17.571 Mbps No bckg 49.33 25.38 0.14 0.47 7.38 3,3

Table 49: H264 Video Conference Performance in the presence of background traffic from PC (Scenario 2 iPad)

We use video conference as reference in order to see the effect of background traffic from PC to the femto service performance, in this case video conferencing. It can be seen from Table 49. As soon the background traffic exist, the packet loss of video quality increase from 2.47% to 3.98% and the jitter double from 20.75 ms to 45.52%. In order to improve the video performance we increase the xDSL bandwidth profile from 1 Mbps to 1.2 Mbps, 1.5 Mbps and 2 Mbps. When it set to 2 Mbps, the packet loss is below 3%, and the jitter is also below 40 ms. In this case, when the femto and PC using the same PVC, ones can improve the performance of femtocell by subscribing more xDSL bandwidth about double than required bandwidth by a FAP. In D622 document we will discuss in more detail how to improve the femtocell performance by utilizing type of service supported by modem and how separating PVC between FAP traffic with other conventional internet service will improve the performance.

5.3 LAN as Femtocell Backhaul

5.3.1 TELKOM RDI LAN Characteristics

According to the survey conducted in RDI department, top three applications accessed by users are email, social media and intranet portal. TELKOM portal is a web-based office application which includes web-mail, knowledge management, news, collaboration tools, human resource database, e-learning, project management online, etc. The email portal can be downloaded to the email client using POP3. It is based on the same email server in TELKOM intranet network. Both portal and email are considered as intranet applications which count 57% population. The intranet traffic occupies LAN Gateway in the 3rd floor. LAN Gateway (3rd floor) utilization in RDI (Research Development & Infrastructure) Department is shown in Figure 66.

Streaming and social media with total population 39% will go through internet link. According to the survey result, streaming media is accessed by 7% of users with frequency



once a day (67%). Streaming media like youtube will add more load to the internet link, however given the low percentage of user access the media and the frequency only once a day, it will create low effect to the link utilization within a day.

0

500

1000

1500

2000

2500

3000

3500

Monday Tuesday Wednesday Thursday Friday

Category Monday Tuesday Wednesday Thursday Friday

Max 453.69 1688.284 3271.0035 1209.24 2583.234

Hour 2:00-3:00 PM 4:00-5:00 PM 9:00-10:00 AM 4:00-5:00 PM 11:00AM-12:00PM

Average 259.87 486.81 727.19 432.83 497.47

Var 13151 174667 732399 104340 377097

Figure 66: Gateway (3rd Floor) load on between Monday-Sunday, December 2011

It can be seen from Figure 66 that LAN GW utilization is below 10 Mbps (1% - 10%) out of 100 Mbps fast Ethernet speed. The LAN bandwidth is occupied during working hours 08.00 – 17.00 from Monday to Friday. In average busy hours occurred in 09.00 – 12.00 and 2.00 – 5.00. During the busy hours, employees access email, portal and internet application such as social media and streaming.

Category Monday Tuesday Wednesday Thursday Friday

Max 4885.21 4317.39 2498.55 3483.34 3623.42

Hour 4:00-5:00 PM 4:00-5:00 PM 4:00-5:00 PM 10:00-11:00 AM 4:00-5:00 PM

Average 1345.39 1150.73 691.36 913.75 940.41

Var 1494872 1346396 577925 1119003 1378383

Load 17% 14% 9% 11% 12%

Figure 67: RDI internet link load on daily basis Monday-Sunday, December 2011

Figure 67 shows the internet link load from Monday to Friday. The higest link utilization is about 17% (out of 9 Mbps link speed) compared to average bandwidth or 54% compared to



its maximum bandwidth. The busy hours occurred mostly between 4.00 and 5.00 in the evening.

5.3.2 Femtocell Bandwidth Observation

In this section the bandwidth taken by a single femtocell will be studied.

Traffic Content Strm Min Max Average Variance Modus Distri.

CS voice call

AMR 12.2 kbps DL 3.32 102.92 44.00095 658477.1 66.4 normal

UL 3.32 66.4 32.27145 675292.7 66.4 normal

CS video call

video 64 kbps DL NA NA 148.40 NA NA NA

UL NA NA 148.40 NA NA NA

HTTP m.detik.com, first page

DL 4.24 619.04 239.494204 31689586.78 305.28 lognormal

UL 4.24 136.736 41.8146938 664124.4463 38.16 lognormal

Streaming m.youtube.com, “Android 2.0 Official Video”, 1.54s, 360p

DL 4.24 949.76 309.859777 23218710.46 271.36 lognormal

UL 4.24 90.096 37.5066464 208368.2624 38.16 lognormal

FTP www.4shared.com “Aku Bukan Bang Toyib.mp3” ; 2.84MB,

DL 4.24 964.60 268.848704 15966460.15 305.28 normal

UL 4.24 235.32 32.9056238 267111.4455 38.16 normal

Video Conference

Movi Telepresence (high quality)

DL 283.50 579.60 555.10 1411.10 - lognormal

UL 306.30 625.50 601.00 1631.10 - lognormal

Video Conference

Movi Telepresence (medium quality)

DL 3.88 825.56 196.18 1631296.44 175.44 lognormal

UL 3.52 518.06 276.89 2639476.99 272.2 lognormal

Table 50: Bandwidth of individual traffic accessed through a FAP

Figure 68: Mix Traffic captured in switch access

The total throughput of 4 voice calls and mix traffic (http, ftp, youtube and voice AMR) can be seen in Figure 68. In terms of voice bandwidth, there is about 20 kbps different between observed bandwidth in the switch access and the ones recorded in femtocell NMS. The bandwidth usage shown in Table 51 is captured from all calls originated from FUEs attached to a FAP under monitor. The FEU call another FUE attached to the second FAP in the same building. It can be seen from the table, if the voice activity factor is more than 80% (by playing background music near the handset), the occupied



bandwidth of a single stream voice in ethernet is 60 kbps, while in the femto system it becomes 84.8 kbps. As more calls are added to the FAP, the total bandwidth increases linearly.

No Status Bandwidth usage (kbps)

Sniff by switch access port Bandwidth usage (kbps)

Monitored at NMS

1 Idle 3-6

2 1 voice call 60 84.8

3 2 voice call 120 169.6

4 3 voice call 180 254.4

5 4 voice call 240 339.2

Table 51: Bandwidth of 4 voice calls served by a FAP

5.3.3 Femtocell Service Performance

We observed femtocell performance in network, where FAPs are attached to local area network using a same vlan with LAN traffic. The femto traffic will traverse from switch access, distribution switch/router for RDI internet link (9 Mbps), ISP network and SecGW. From SecGW to 3G core network, it uses the same configuration as in Femto-xDSL testbed. Since there is no guarantee that DiffServ marking will traverse from femto to the SecGW and vice versa, hence we consider this network as un-managed network or non-SLA network.

5.3.3.1 Voice Performance in the Presence of LAN traffic (Scenario 2)

Since the traffic load in RDI’LAN is below 10% (out of 100 Mbps), the femtocell performance is not affected by the background traffic. In this case using a same vlan or separate vlan will be no different. However in corporate LAN where intranet activities exceed more than 50% the effect may be different. It is not possible to observe this in TELKOM RDC local area network, since most network load in gateway is below 10-30%.

Figure 69: LAN and FAP Traffic observed in switch access and NMS respectively

Figure 69 shows the LAN traffic in 3rd floor GW which potentially affect the voice traffic (G711u) over femtocell attached to the same vlan. Figure 70 shows G711u performance (downlink) in terms of jitter, packet loss and throughput. It can be seen from the graph that the background traffic does not have impact to the voice performance. The MOS is remain above 4, except in the minute of 67th, where the MOS is about 3.7. In that time, packet loss increase sharply from about 0.2 % in minute of 64th to 1.35% in minute 67th. However it is diffucult to justify whether the performance is affected by the LAN traffic, since in LAN segment the traffic just fall from 3.5 Mbps to 3 Mbps in the time



between 64th – 67th minute. The fact that MOS value is remain satisfactory (see Table 52), it gives a clear indication that the voice quality G711u in the network is acceptable.

Maximum for G.711 codec 4.4 Very satisfied 4.3 – 5.0 Satisfied 4.0 – 4.3 Some users satisfied 3.6 – 4.0 Many users dissatisfied 3.1 - 3.6 Nearly all users dissatisfied 2.6 – 3.1 Not recommended 1.0 – 2.6

Table 52: MOS value for G711 voice codec

a. Throughput of G711u (layer 7 and layer 2)

b. Pakcet loss of G711u

c. One way delay and jitter of G711u

d. MOS of G711u

Figure 70: Performance of G711u Voice from a FAP in the presence of background traffic

5.3.3.2 Video Conference Performance in the Presence of Internet Traffic (Scenario 3)

In this section, video conference performance will be observed under two test conditions: RDI Internet link is unoccupied, so fluctuation in the ISP network and 3G Core Network will

affect the performance. RDI Internet link is occupied from 10% to more than 80%, so that fluctuation in internet link,

ISP network and 3G Core Network will affect the performance. The internet link utilization can be seen in Figure 71. Since in the workday the fluctuation is below 50%, we execute this scenario in the night, so we were able to flood the network with such very high utilization.



Figure 71: Internet Traffic observed in switch distribution of RDI internet link

Two video conference clients are incorporated in this scenario. One video client attached to the femtocell (called VC1) and another client (VC2) attached to the IP public (bandwidth 15 Mbps dedicated link). VC1 connected to FAP which served by SecGW (using IP Public) through ISP network. The video conference setting can be seen in the table below: VC1 VC2

Transmit Receive Transmit Receive Max Allowed Bitrate 518 kbps 518 kbps 518 kbps 518 kbps Video Codec H.264 H.264 H.264 H.264 Video Resolution 320*240 176*144 176*144 320*240 Video Configured bitrate 98 kbps NA 98 kbps NA Audio Codec G.722.1 G.722.1 G.722.1 G.722.1 Audio Configured bitrate 24 kbps NA 24 kbps NA

Table 53: Video Conference Application Setting

Video conference performance using NON SLA network without background traffic from LAN can be seen in Table 54. Tx refers to the transmitted data by VC1, Rx means the received data by VC2.

Tx Rx Tx Rx Tx Rx Tx Rx Tx Rx Tx Rx Tx Rx

MIN 61 55 0.0% 0.0% 9 8 24.0 18.5 24 19 0.0% 0.0% 5 5

MAX 161 184 20.3% 19.0% 33 41 30.0 33.0 64 67 20.3% 21.4% 21 22

AVERAGE 77.55 77.76 1.33% 1.48% 16.52 17.44 28.92 28.10 32.45 31.55 0.02 0.02 11.17 11.39

VAR 269.59 363.78 0.08% 0.07% 27.08 39.62 1.54 4.85 270.42 247.85 0.07% 0.08% 10.14 11.24

MOVI NON SLA NO BACKGROUND

VALUEVIDEO VOICE

BITRATE (kbps) PACKET LOSS (%) JITTER (ms) FRAME RATE (fps) BITRATE (kbps) PACKET LOSS (%) JITTER (ms)

Table 54: Video Conference Performance using non-SLA network, without background



Video conference performance using NON SLA network with background traffic from LAN can be seen in Table 55.

Tx Rx Tx Rx Tx Rx Tx Rx Tx Rx Tx Rx Tx RxMIN 74 66 0.0% 0.0% 14 16 27.0 0.0 4 22 0.3% 0.0% 9 9

MAX 181 171 25.7% 14.8% 78 56 30.0 60.0 64 66 8.9% 12.1% 25 25

AVERAGE 97.30 94.28 7.0% 7.5% 27.94 28.23 29.17 25.18 31.61 31.10 4.3% 4.8% 15.06 14.75

VAR 1104.81 1013.29 0.1% 0.1% 87.46 56.55 1.06 43.35 267.04 227.03 0.0% 0.1% 12.37 11.19

VALUE PACKET LOSS (%) JITTER (ms)VIDEO

MOVI NON SLA WITH BACKGROUND

BITRATE (kbps) PACKET LOSS (%) JITTER (ms) FRAME RATE (fps)VOICE

BITRATE (kbps)

Table 55: Video Conference Performance using non-SLA network, with background

from Intranet (more than 80% utilization)

Figure 72: Performance Comparison of VC with vs without background in internet link



In order to easly analize, Figure 72 shows the performance comparison between the two cases. It can be seen from the graph that when the utilization of internet link reach above 80% the performance of VC in the receiver side will degrade sharply. Average packet loss perceived by VC2 increases from 1.48 % to 7.5%, while jitter also increase from average 17.44 ms to 28.23 ms. The audio quality shows the similar performance that in the presence of load in internet link will degrade the VC performance.

5.3.4 E2E QoS (DiffServ) Observation

In this section, DiffServ as end-to-end QoS management in femtocell network will be observed both in femtocell and metro ethernet. The simplified version of network diagram for this observation can be seen in Figure 73. AP is attached directly to SecGW using a switch Ethernet. RAN-GW or FAP-GW is connected to SGSN/GGSN to TELKOM/TELKOMSEL Metro Ethernet. We omit DiffServ observation in RDI LAN, since there are only VLAN implemented so far to differentiate access from one department to another. Priority traffic has not been implemented so far since there are only intranet/internet application considered. Therefore by replacing the whole LAN network into a single switch access for a FAP will not make any difference.

SGSN/GGSN

IP Address(Physical)

Physical connection

Optical

LC (MetroE in Telkomsel network)

LC (LBR in RAN GW)

? LC (MetroE in Telkomsel network)

1/1 (LBR in RAN GW)

RAN GW

Network Address Network Address

IP Address(Physical) IP Address(Physical)

IP Address(Logical) IP Address(Logical)

Next hop (Route to SGSN) Next hop (Route to SGSN)

0 0

VLAN ID VLAN ID

Network Address Network Address 172.20.40.0/24

IP Address(Physical) IP Address(Physical) 172.20.40.253

IP Address(Logical) Don't care IP Address(Logical) 172.20.40.254

Next hop (Route to AP) 172.20.40.15

192.168.5.0/24

IP parameter for U-Plane16207 Network Address 172.20.40.0/24

3 (National spare) IP Address(UCG#1)

172.20.40.17

172.20.40.15172.20.40.254

APIP parameter for U-Plane (GTP terminate)

Network Address 192.168.5.0/24

M3UA routing context

M3UA traffic mode

AP

IP Address(UCG#2)

Network Address IP Address(Logical)

IP Address(Blade#9) Next hop (Route to SGSN)

IP Address(Blade#10) 0

INC

SS7 Parameters

SeG

W

Point Code

NI (Network indicator)

IP parameter for C-Plane

Next hop

SCTP port No.(Blade#9)

SCTP port No.(Blade#10)

M3UA parameters

LB

R

IP parameter for U-Plane

Co

mm

un

ica

te

with

CN

Co

mm

un

ica

te

with

CN

Co

mm

un

ica

te

with

INC

Co

mm

un

ica

te

with

Se

GW

Cable type :

Connector type :

Physical port :

LB

R


IP Address(Logical)

Default GW Default GW

SCTP port No.

M3UA parameters

M3UA routing context

IP parameter for U-Plane

Network Address Network Address

Netmask Netmask

IP Address(Physical)

SG

SN

/GG

SN

SS7 parameter for C-Plan

Point Code(SGSN)

NI (Network indicator)


SG

SN

/GG

SN

IP Address(Logical)

M3UA traffic mode

Figure 73: Network Configuration for DiffServ Observation

5.3.4.1 DiffServ in Femtocell System

Table 56 shows default QoS setting in a FAP. The value of QoS parameter can be changed, but it could be applied only for uplink, the downlink will follow QoS parameter from core network side. For each uplink IP packet the DSCP value in the tunnelling inner IP header is equal to DSCP value in the tunnel outer IP header. According to wireshark capture, expedited forwarding (EF) class applied for both CS voice and CS video call as can be seen in Figure 74 and Figure 75. These EF applied to support conversational services based on circuit switch. Eventhough the figures represent the flow from FAP to SecGW (uplink), in the opposite direction, EF class is maintained.



Interactive class is implemented in the dwonlink direction as core network assigned QoS according to the SIM Card Profile. It can be shown from Figure 76 and Figure 77, within Message Type Activate Accept, there is QoS – Negotiated QoS which can be summarized as below :

AF21 is given when PDG traffic handling Priority = 2. This treatment applied for postpaid SIM Card subscribing unlimited data packet with uplink bandwidth up to 384 kbps and downlink bandwidth up to 2048 kbps.

AF11 is given when PDG traffic handling Priority = 3. This treatement applied for prepaid SIM Card subscribing unlimited data packet with uplink bandwidth up to 384 kbps and downlink bandwidth up to 512 kbps.

Regardless the downlink traffic such as 4shared DL, youtube, browsing, skpy; these traffic will be tag using interactive packet switch according to the subscription profile in SIM card.

Traffic Type PHB - DiffServ DSCP value PHB 802.1p Conversational (RTP) Expedited Forwarding (EF) 46 5 Streaming PS Assured Forwarding (AF41) 34 4 Interactive PS Priority 1 Assured Forwarding (AF31) 26 3 Interactive PS Priority 2 Assured Forwarding (AF21) 18 2 Interactive PS Priority 3 Assured Forwarding (AF11) 10 1 Background PS Background (BE) 0 0

Table 56: E2E QoS Setting in FAP

Figure 74: Wireshark capture to show traffic class applied in CS Video call (EF)



Figure 75: Wireshark capture to show traffic class applied in CS Voice call (EF):

Figure 76: Wireshark capture to show traffic class PS interactive priority 2 (AF21):



Figure 77: Wireshark capture to show traffic class PS interactive priority 3 (AF11):

5.3.4.2 DiffServ in Metro Ethernet

QoS implementation in Metro Ethernet aims to give a special treatment toevery traffic from individual services traverse thourough metro ethernet network. The QoS is usually configured in every PE router as can be shown in Figure 80.

Figure 78: QoS in Metro Ethernet

In general, QoS in Metro Ethernet is divided into two parts namely network QoS policy and service QoS policy which are defined in both ingress and egress. Network QoS policy is applied in IP interface. In the ingress side, the policy perform EXP bit mapping of received packet from core network to a specific forwarding class and then perform appropriete action associated with the policy. In the egress side, the policy will map the traffic (being transmitted to the core network) to specific forwarding class according to EXP bit value of each packet. Service QoS policy is applied to access point for a specific customer or Service Access Points (SAP). In the ingress side, the policy will map the received packet from a subscriber into a specific queue of related forwarding class and perform a treatment according to the policy. Traffic mapping can be based on QoS marking from customer side which can be based on IEEE 802.1p



bits, DSCP and IP precedence. In the egress side, the policy will map the traffic (being transmitted to customer side) into appropriate forwarding class and perform an action according to the policy. We observe QoS implementation in metro ethernet. Currently TELKOM ME is positioned to have layer 2 VLAN differentiation with dedicated bandwidth assignment 100 Mbps. QoS setting remains default as can be seen in Figure 79.

configure qos network 1 config>qos>network# info detail description "Default network QoS policy." scope template ingress default-action fc be profile out no ler-use-dscp dscp be fc be profile out dscp ef fc ef profile in dscp cs1 fc l2 profile in dscp nc1 fc h1 profile in dscp nc2 fc nc profile in dscp af11 fc af profile in dscp af12 fc af profile out dscp af13 fc af profile out dscp af21 fc l1 profile in dscp af22 fc l1 profile out dscp af23 fc l1 profile out dscp af31 fc l1 profile in dscp af32 fc l1 profile out dscp af33 fc l1 profile out dscp af41 fc h2 profile in dscp af42 fc h2 profile out dscp af43 fc h2 profile out lsp-exp 0 fc be profile out lsp-exp 1 fc l2 profile in lsp-exp 2 fc af profile out lsp-exp 3 fc af profile in lsp-exp 4 fc h2 profile in lsp-exp 5 fc ef profile in lsp-exp 6 fc h1 profile in lsp-exp 7 fc nc profile in exit

egress no remarking fc af dscp-in-profile af11 dscp-out-profile af12 lsp-exp-in-profile 3 lsp-exp-out-profile 2 dot1p-in-profile 2 dot1p-out-profile 2 no de-mark exit ---deleted ---- fc ef dscp-in-profile ef dscp-out-profile ef lsp-exp-in-profile 5 lsp-exp-out-profile 5 dot1p-in-profile 5 dot1p-out-profile 5 no de-mark exit -- deleted -- fc l1 dscp-in-profile af21 dscp-out-profile af22 lsp-exp-in-profile 3 lsp-exp-out-profile 2 dot1p-in-profile 3 dot1p-out-profile 3 no de-mark exit --- deleted ---- exit exit

Figure 79: Default Network QoS Policy in Metro Ethernet – Ingress & Egress

According to Figure 79, metro ethernet is in the position to forward all DSCP value from ingress to egress. Packet flow with AF11 for instance will be traversed from ingress to egress using forwarding class (fc) af. Currently service QoS policy does not implemented in the metro Ethernet. So in the presence of congestion, all traffic (regardless its traffic types) will have similar treatment in term of queueing handling. To fully accommodate DiffServ implementation for femtocell deployment, service QoS policy should be implemented.

5.4 Mobility Management Reports

In this section, the result of mobility measurements will be reported. The report includes FAP-to-FAP mobility in BM1 and FAP to MBS handover with BM3 environment.

5.4.1 Mobility Scenario (FAP to FAP)

The realization of FAP-to-FAP mobility measurement can be summarized below; Scenario FF1, measurement scenario FAP-to-FAP used 2 active FAPs, with FAP SC 300 and

FAP SC 219. The measurements were done on route A and B (Corridor 1) with length of route



approximately 50 meter. The measurement process and the detail result could be seen in chapter 5.4.1.1.

Scenario FF2, measurement scenario FAP-to-FAP used 2 active FAPs, with SC 219 and SC 375. The measurements were done on route B and C (Coridor2) with length route approximately 45 meter. The measurement process and the detail result could be seen in chapter 5.4.1.2

Scenario FG1 is a generic scenario which been done on route A, B and C (Corridor 1 and 2) with total length of route approximately 95 meter. The measurement process and the detail result could be seen in chapter 5.4.1.3.

As been discussed in the measurement methodology, we will see the impact of Q-Hysteresis, Q-Offset and Handover Delay Timer to the handover performance. Every measurement was repeated two times (in minimum) using the same methodology and key observations. Table 57 shows measurement result of FAP-to-FAP mobility with those parameters set up on every FAP. As can be seen from Table 57, handover between FAP to FAP with Q-hysteresis equal to 2 dB will cause more frequent handover or redundant handover which can be seen in Scenario FF2. Redundant occurred because signal fluctuation exceed Q-hysteresis threshold. Signal fluctuations were caused by BM1 environment which has a lot of walls and bulkhead. Those situations may also cause call drop or handover failure. According to the measurement result; when Q-Hysteresis 4dB, and handover delay timer 640ms, the handover shows optimum performance in terms of handover failure rate (0%). We keep Q-offset parameter to zero so that the handover or cell reselection will be instantly happened as soon as the signal quality exceed Q-hysteresis and delay timer threshold. Figure 58 shows measurement result of cell reselection using Q-Hysteresis 4dB and Reselection Delay Timer (RDT) varied from 0s, 1s, and 2s. Cell reselection event was occured if UE is in idle mode (not in conversation or perform any transactions).



Handover Rate (Number of Handover

Process)

Handover Failure Rate (%) -

Scenario FF1 (FAP1 ↔ FAP2)

Q-Hysteresis [4dB] , Q-Offset [ 0dB], HDT [320ms]



Q-Hysteresis [4dB] , Q-Offset [ 0dB],

HDT [640ms]

Total measurement;4

Handover Total;6, handover occurs 1-

3 times every measurement

Handover Rate ; 1,5

Total measurement;4

Handover Total;6 , handover occurs 1-

3 times every measurement.

Handover Rate ; 1,5

Total measurement;4 HFR : 0

Total measurement;4

HFR : 0





HDT [640ms]

Total measurement;4



Handover Rate ; 1,5

Total measurement;4



Handover Rate ; 1,5


Total measurement;4

HFR : 0

Scenario FF2 (FAP2 ↔ FAP3)





HDT [640ms]

Total measurement;4

Handover Total;6, handover occurs 1-3

times in every measurement

Handover Rate ; 1,5

Total measurement;2

Handover Total;3, handover occurs 1-2

times in every measurement

Handover Rate ; 1,5


Total measurement;2

HFR : 0

Scenario FG1

(FAP1 ↔ FAP2 ↔ FAP3)





HDT [640ms]

Total measurement;8 (1

call dropped) Handover Total;18, handover occurs 2-6

times every measurement)

Handover Rate ; 2,57

Total measurement;10

Handover Total;24 handover occurs 2 –

4 times every measurement)

Handover Rate ; 2,4

Total measurement;10

HFR : 5.26 %

Total measurement;8

HFR : 0 %

Table 57: Matrix of Performed Handover Measurement



Cell Reselections ( per route)

Failure Rate (%)

FF1 (FAP1 ↔

FAP2)

Q-Hysteresis [4dB] , Q-Offset [ 0dB] ], RDT [0s]






1 1 1

0 0 0

Table 58: Matrix Performance of Cell Reselections

According to the Table 58, the cell reselection matrix performance remains un-changes eventhough the RDT set to 0 second, 1 second, and 2 second.

5.4.1.1 Scenario FF1

In Scenario FF1, measurement was conducted by using 2 FAPs (ePico). The distance between two FAP is approximately 50 meter. The measurement was performed when UE in idle mode and in traffic. FAP transmit power was fixed at 10dBm. FAP 1 and FAP 2 were deployed in co-channel with UARFCN 10688. The FAPs’ UARFCN are different with the one used by MBSs (10663 and 10638). As a result, the interference between FAPs and MBS was limited to none.

Figure 80: Measurement in Corridor 1 (C1), FAP 1 and FAP2 active

The first measurement was conducted to observe cell reselection performance. Q-Hysteresis of both FAPs is set to [4dB], Q-Offset [0dB], and RDT [1s]. The measurements were performed by moving a FUE from A to B point and vice versa from B to A direction. The FUE moving speed was approximately 1~3 m/s (human walk speed). The measurement was performed out of working hours in order to avoid the effect of human activity factor. Figure 82 illustrates cell reselection between FAP1 to FAP2.

Figure 81: Handover point between FAP 1 and FAP2



Figure 82: Cell Reselections between FAP1→FAP2 from A – B point

The second measurement of FF1 scenario was conducted to observe handover between the two FAPs. HDT of both FAPs is set to 640m, while Q-Hysteresis is 4dB, Q-Offset 0dB. The measurements were done by moving a FUE from A to B direction and vice versa. Figure 83 illustrates cell reselection between FAP1 to FAP2.

Figure 83: Handover between FAP1 to FAP2 along A-B route

Based on Figure 82 and Figure 83 we can derive the location of cell reselection and handover occured. Given that the distance between FAP1 and FAP2 is 50 meter, the measurement for handover conducted from 08:25:10 to08:26:25 or last for 75 second. Handover from FAP1 (SC 300) to FAP2 (SC 219) was done on 08:26:00 or approximately 50 second from FAP1. Point of handover can be calculated as following.

HO = (50/75) × 50 = 33.33 meter (30 – 35) from the origin or FAP1. The similar calculation can be done for Cell Reselections (CR) which give the following reult;

CR = (115/140) × 50 = 41.07 meter (40 – 45) from the origin or FAP1.



In conclusion, FUE handoff from FAP1 to FAP2 at point 30 – 35 meter relative from FAP1. When FUE moving from FAP2 to FAP1; handover event occured in point 30 – 35 meters from FAP2. In case of cell reselection, If FUE moving from FAP1→FAP2 the CR occurred in 30-35 meters from FAP1 position. However when FUE moving from FAP2→FAP1, cell reselection was done in point 25-30 meter from FAP2 positions.

5.4.1.2 Scenario FF2

In Scenario FF2, measurement was conducted by using 2 FAPs (ePico). FAP2 uses SC 219 and FAP3 SC 375. The distance between two FAPs is approximately 45 meter along B – C corridor. The measurement was performed when UE in idle mode and in traffic. Scenario FF2 has similar scenario to FF1, except that the measurement done in different corridor (Figure 84). FAPs transmit power were fixed at 10dBm. FAP 1 and FAP 2 were deployed in co-channel with UARFCN 10688. The FAPs’ UARFCN are different with the one used by MBSs (10663 and 10638). As a result, the interference between FAPs and MBS was limited to none. Cell Reselection in this corridor is observed by assigning the same Q-Hysteresis [4dB], Q-Offset [0dB], and RDT [1s] to the FAP2 and FAP3. The measurements were conducted by moving a FUE from B to C and also C to B. The measurement was done in out of working hours to limit human activity factor with the moving speed about 1 – 3 m/s.

Figure 84: Measurement in Corridor 1 (C1), FAP 2 and FP 3 active

Handover measurement in this corridor was done by assigning Q-Hysteresis [4dB], Q- Offset [0dB], and HDT [640ms] to FAP2 and FAP3 by using FF2 scenario.

By appling the post processing in ACTIX, cell reselection is occurred within 25 – 30 meter from reference point FAP2. Figure 85 illustrates the cell reselection. In terms of handover, if FUE moving from FAP3 to FAP2, handover process was done within 40 – 45 meter from FAP2. However if FUE moving from FAP2 to FAP3, it was happened within 35 – 40 meters from reference point FAP2.

Table 59: Point of Mobility FAP2 - FAP3

FUE Moving (based on route)

Point of Mobility from origin FAP2

Handover Cell Reselections FAP2 to FAP3

Range 35 – 40 m Range 25 – 30 m



Figure 85: Cell Reselection FAP2 to FAP3

5.4.1.3 Scenario FG1

In Scenario FG1, measurement was conducted by activating 3 FAPs (ePico). The total route length of three FAPs is approximately at 95 meter (FAP to FAP2 to FAP3). Figure 86 illustrates the FAPs placement and the trajectory for walk test. All FAPs were set up in Q-Hysteresis [4dB], Q-Offset [0dB], and RDT [1s] for cell reselection and Q-Hysteresis [4dB], Q-Offset [0dB], and Handover Delay Timer [640ms] for handover. The measurements were conducted by moving FUE from A to B to C and vice versa C to B to A. The measurement results of cell reselection process are shown in Figure 87. As can be seen from the graph FUE initially served by FAP1 (SC 300) until 5second to 5:50:00, it selects FAP2 (SC 219). FAP2 serves FUE until it reach area where Ec/No equal to -5 dB, then it selects FAP3.

Figure 86: Measurement in Corridor 1 (C1), FAP 1 and FP 2 active



Figure 87: Cell Reselections along C1 and C2

Figure 88: Handover along C1 and C2

The measurement results of handover process are shown in Figure 88. The distance between FAP1 and FAP3 is approximately 95 meter. The measurement was conducted from 09:15:15 to 09:17:55 lasting 160 second. Handover from FAP1 SC 300 to FAP2 SC 219 was occured on 09:16:35 or approximately 80 second from reference point FAP1. Point of handover can be calculated as following:

HO1 = (80/160) × 95 = 47.5 meter (45 – 50) from the origin or FAP1. While the second handover FAP2 to FAP3 was done in 09:17:33 or 138 second from the origin FAP1. Handover point can be calculated as

HO2 = (138/160) × 95 = 82 meter from the origin (FAP1) or 32 meter (30 – 35) from FAP2. Table 60 shows summary of mobility point for FG1 scenario. It can be seen from the table that HO1 wasn’t done on the centre of FAP1 and FAP2 but rather close to FAP2, i.e. when the signal RSCP from FAP1 is no longer able to cope with the interference level, so FUE was handed to FAP2. The same behaviour also happened in HO2, where handover process from FAP2 to FAP3 occured about 32 meter from FAP2.



Table 60: Point of Mobility FF1 totoFF2 totoFF3

Figure 89: Handover points along C1 and C2

When UE is moving in direction of FAP3→ FAP2→FAP1, handover between FAP3 (SC375) and FPA2 (SC219) occurred in the corridor in point between 40 – 45 meter from FAP3. Along Corridor 1, handover occurred between 15 – 20 meter from FAP2. If FUE moves from FAP3 to FAP2, CR will happened in point 35 – 40 meter from FAP3. CR is also happened in 15 – 20 meter in Corridor 1 from FAP2.

FUE Moving (route based on scenario)

Point of Mobility from origin Corridor 1 (FAP1 to FAP2) Corridor 2 (FAP2 to FAP3)

Handover Cell Reselections Handover Cell Reselections FAP1 - FAP2 - FAP3 45 – 50 m 45 – 50 m 30 - 35 m 30 – 35 m



6 CONCLUSION Radio coverage and interference characterization has been conducted in the WCDMA TELKOM testbed, which is compliant with business model 1 (BM1): corporate building in urban environment. A specific measurement methodology was elaborated (section 3.1.4) for investigating the impact of: Single-FAP and multi-FAPs deployments (up to 4 FAPs in the floor); Inter-FAP distance; Outdoor leakage; Closed- and open-access modes; Visibility condition (LOS, NLOS, inter-floor); On the following radio metrics: Useful signal level, interference level and signal-to-interference statistics; FAP coverage radius; Macro deadzone radius; FAP-to-macro handover distance; DL and UL throughput. A large number of measurement scenarios, based on different FAP deployments and compared the ones to the others, were necessary to reach the objective and get relevant results (section 5.1). Several types of equipment were involved: up to 5 commercial FAPs, up to 5 modems, SCANNER, TRACE mobile and monitoring equipments. Consequently, high precision was required in the measurement realization (fine geo-location, synchronization, study of time reproducibility, measurement path reproducibility, etc.) and data post- processing. Measurement results and analysis are respectively reported in section 5.1. Main conclusions are summarized in section 5.1.5. Major conclusions drawn from radio measurement analysis, but also from 5A2 simulations, are exploited for derivation of radio-planning engineering rules: Impact of FAP deployments (FAP density, FAP locations, etc.) on the macro network coverage

and capacity. Impact of FAP deployments (FAP density, FAP locations, etc.) on the femto network coverage

and capacity. Recommendations on the FAP deployment, spectrum usage and required mitigation techniques to

assure good network performance. Some rules to identify problematic sources of interference in the macro+femto network.

Backhaul characterization has been done for both BM3 (xDSL as femtocell backhaul) and BM1 (corporate LAN as femtocell backhaul). According to the measurement result it requires about 800 Kbps downlink and 158-632 kbps uplink to support smartphone and tabletPC. Given that 20% overhead is given, 1 Mbps xDSL link will be enough to support a femtocell service. Femtocell performance under non-SLA between MNO and ISP, will be potentially affected by background traffic in customer side. A single PVC assignment in xDSL modem will cause performance degradation. In order to recover the performance, the customer should double its capacity to accommodate both femto and internet traffic. The same case for corporate LAN. Potentially the bottleneck may occur in the internet link which may degrade the femtocell performance.



Major conclusions drawn from measurement analysis and RFI result are exploited for derivation of backhaul engineering rules: Impact the use of under-laying backhaul technologies (xDSL, corporate LAN) to the femtocell

performance Impact the background traffic which exist in customer sides (BM1, BM3) to the femtocell

performance Recommendations about maintaining QoS for femtocell deployment in BM1 and BM3.

Based on available mobility parameters observation (signal quality threshold, hysteresis margin, delay timer, and neighbouring cell offset), the mobility performances has been characterised. The analysis of mobility characteristics various set of parameters has been discussed in section 5.4 and is further elaborated in [FREEDOM-D622]. The major conclusions drawn from measurement analysis and RFI are exploited for derivation of femtocell engineering rules including: The effect of mobility parameters to performed mobility efficiency The effect of mobility parameters to successful rate of mobility Recommendations of optimum parameter values

Derivation of these engineering rules are reported in Deliverable [FREEDOM-D622].

ICT-248891 STP FREEDOM - cordis.europa.eu · [Huawei-ePico] ePico3801 MML Command...

Documents

Transcript of ICT-248891 STP FREEDOM - cordis.europa.eu · [Huawei-ePico] ePico3801 MML Command...