1. C30-20080417-005 1 3GPP2 TSG-C WG3 TITLE : UMB performance results SOURCE: TSG-C WG3 EMAH Contact to: Satoshi Konishi, Vice-chair of EMAH sa-konishi@kddi.com ABSTRACT: This contribution provides performance results of UMB (Ultra Mobile Broadband) system. In addition, the EMAH (Evaluation Methodology Ad-Hoc) activities for the UMB characterization are also summarized in this contribution. RECOMMENDATION: Review and adopt for the submission from TSG-C ©2008 3GPP2. All rights reserved.

2. C30-20080417-005 2 Outline 1.Introduction 2.Background  Activities in TSG-C WG3 EMAH for performance evaluation of UMB system  Calibration process among companies 3.Simulation conditions for UMB performance evaluation 4.Simulation results 5.Conclusion 6.Appendix: A.Examples of link level simulation results B.System level simulation results for calibration C.Backup slides for simulation results under NGMN setup

3. C30-20080417-005 3 Introduction Standardization of UMB system  March 2006 to February 2007: Decision of framework and components  April, 2007: Publication of UMB Air Interface Specification Version 1.0  May to August, 2007: Revision of UMB Air Interface Specification  September, 2007: Publication of UMB Air Interface Specification Version 2.0 However, UMB characterization has not been completed yet … Necessity of Performance evaluation for UMB system!

4. C30-20080417-005 4 Background (1) ~ History on Activities for Performance Evaluation of UMB ~ History:  June, 2007: agreed upon to start UMB characterization  July to August, 2007: preparatory work (e.g. outline of simulation setting, parameters, timeline)  August to December, 2007: Calibration of link curves  December 2007 until now: Calibration of system level simulation results

5. C30-20080417-005 5 Background (2) ~ Calibration Processes for UMB Performance Evaluation ~ Link level simulations:  From Aug., 2007, each company began to provide simulation results  Oct, 2007: confirmed that three companies (Samsung, Huawei, and KDDI) were aligned  Nov. to Dec., 2007: another two companies (Qualcomm and Alcatel-Lucent) were also aligned with the three companies Appendix A: examples of link level simulation results

6. C30-20080417-005 6 Background (3) ~ Calibration Processes for UMB Performance Evaluation ~ System level simulations:  From Dec., 2007, companies (KDDI, Samsung, Nortel, Qualcomm, Huawei, and Alcatel-Lucent) began to provide simulation results  In total, five rounds have been performed for calibration among companies so far  March, 2008: confirmed that all the companies involved in the calibration (KDDI, Samsung, Nortel, Qualcomm, and Huawei) were aligned on the full-buffer traffic model in the forward link (FL-FB)  April, 2008: confirmed that all the companies involved in the calibration were aligned on the VoIP traffic model in the both forward and reverse link (FL-VoIP and RL-VoIP)  Up to now: most of the companies have been aligned in the remaining simulation scenarios as follows Full-buffer traffic model in the reverse link (RL-FB) Full-buffer traffic model in the forward link with MIMO (FL-MIMO) Appendix B: examples of system level simulation results

7. C30-20080417-005 7 Toward UMB Performance Evaluation For the purpose of calibrations among companies, we use a set of simplified simulation models such as  Simplified power control model compared to the UMB air interface specification  No power control in the forward link  No rank adaptation in the FL-MIMO simulations  Etc. This presentation provides actual performances both  taking account of more realistic scenarios in contrast to above and  using simulation parameters in the NGMN’s evaluation methodology document

8. C30-20080417-005 8 Simulation Parameters

9. C30-20080417-005 9 FL overhead OverheadValue10MHz full buffer5MHz VoIP Pilot18 out of 12814.1% Guard band32 subcarriers (992 used subcarriers for 10MHz and 480 used subcarriers for 5MHz with 9.6KHz subcarrier spacing) 4.77% (with respect to 10MHz) 7.84% (with respect to 5MHz) Control overhead 96 subcarriers for full buffer and 64 subcarriers for VoIP 9.68%13.3% CP and windowing 6.51+3.26 μs out of 113.93 μs 8.58% Superframe preamble 1 out of 26 frames3.85% Total35.1%40.0%

10. C30-20080417-005 10 RL overhead OverheadValue10MHz full buffer5MHz VoIP Pilot18 out of 12814.1% Guard band32 subcarriers (992 used subcarriers for 10MHz and 480 used subcarriers for 5MHz with 9.6KHz subcarrier spacing) 4.77% (with respect to 10MHz) 7.84% (with respect to 5MHz) Control overhead ACKC H 32 subcarriers3.23%6.67% CDMA control One CDMA segment (128 subcarriers) every 8 frames for full buffer, and one CDMA segment every frame for VoIP 1.61%26.7% CP and windowing6.51+3.26 μs out of 113.93 μs 8.58% Superframe preamble 1 out of 26 frames3.85% Total31.6%53.6%

11. C30-20080417-005 11 Simulation Conditions for Full Queue Traffic Model System bandwidth – 992 subcarriers over 10MHz Overhead  Guard band – 32 subcarriers  CP - 6.51 us  Control overhead – 6 tiles (96 subcarriers) in FL with 10% power overhead, one CDMA subsegment (128 subcarriers) every 8 frames and 2 tiles (32 subcarriers) for R-ACKCH every frame  Superframe preamble – one out of 26 frames (3.85%) Scheduler – Equal resource scheduler (similar to proportional fair) Power control – No power control for FL. OSI-based power control for RL with DataCtoI range [-3dB,10dB] and IoT target 7.5dB Antenna configurations  SIMO1x2 for FL and RL  MIMO2x2 with precoding (64 precoding matrices used) and rank selection for FL

12. C30-20080417-005 12 ~ Full Queue Traffic Model ~ Simulation Results Average sector throughput and spectrum efficiency Average user throughput and spectrum efficiency Cell-edge user throughput and spectrum efficiency Non-MIMO (1x2) 12.5 Mbps1.25 Mbps0.38 Mbps MIMO (2x2) 13.3 Mbps1.33 Mbps0.50 Mbps  Forward link (Downlink) Average sector throughput and spectrum efficiency Average user throughput and spectrum efficiency Cell-edge user throughput and spectrum efficiency Non-MIMO (1x2) 9.46 Mbps0.95 Mbps0.36 Mbps  Reverse link (Uplink)

13. C30-20080417-005 13 Simulation Conditions for VoIP Traffic Model System bandwidth – 480 subcarriers over 5MHz Overhead  Guard band – 32 subcarriers  CP - 6.51 us  Control overhead – 4 tiles (64 subcarriers) in FL with 26% power overhead, 160 subcarriers in RL, including one CDMA subsegment (128 subcarriers) and 2 tiles (32 subcarriers) for R-ACKCH  Superframe preamble – one out of 26 frames (3.85%) Power control – Target termination based power control. Antenna configurations – SIMO1x2 for FL and RL

14. C30-20080417-005 14 ~ VoIP Traffic Model ~ Simulation Results VoIP capacityAverage latency95 percent tile of 98% packet delay Non-MIMO (1x2) 350+12ms45ms  Forward link (Downlink) VoIP capacityAverage latency95 percent tile of 98% packet delay Non-MIMO (1x2) 30015ms50ms  Reverse link (Uplink) * Since the VoIP system is RL limited, though the FL VoIP capacity is far more than 350, we did not simulate to find the actual number.

15. C30-20080417-005 15 Conclusions This contribution summarizes activities in TSG-C WG3 EMAH for UMB characterization  Calibration of both link- and system-level simulations has been almost completed UMB performance results are provided using NGMN simulation conditions. Some limitations in this set of simulations  Subband scheduling is not used in the full buffer simulations  Suboptimum link curves are used with Max-Log-Map decoding and suboptimum demapper  Rate prediction loop is not optimized  Control overhead can be optimized

16. C30-20080417-005 16 References [1]3GPP2, “Ultra Mobile Broadband (UMB) Air Interface Specification,” C.S0084-000~009 v.2.0, Sep. 2007. [2]NGMN Alliance, “NGMN Performance Evaluation Methodology” Version 1.2, June 2007.

17. C30-20080417-005 17 Appendix A ~ Examples of link level simulation results ~

18. C30-20080417-005 18 Link level results (1) FL, BRCH, 8 tiles, PF#10 H: Huawei S: Samsung K: KDDI Q: Qualcomm A: Alcatel-Lucent AV: Averaged values

19. C30-20080417-005 19 Link level results (2) RL, 1 tiles, PF#5 H: Huawei S: Samsung K: KDDI Q: Qualcomm A: Alcatel-Lucent AV: Averaged values

20. C30-20080417-005 20 Appendix B ~ System level simulation results for calibration~

21. C30-20080417-005 21 System level simulation results (1) ~ FL, Full-buffer Traffic, Ped-B w/ 3km/h ~

22. C30-20080417-005 22 System level simulation results (2) ~ FL, VoIP Traffic, Ped-B w/ 3km/h ~

23. C30-20080417-005 23 System level simulation results (3) ~ RL, VoIP Traffic, Ped-B w/ 3km/h ~

24. C30-20080417-005 24 Appendix C ~ Backup slides for simulation results under NGMN setup~

25. C30-20080417-005 25 ~ Full Queue Traffic Model ~ Simulation Results (1)

26. C30-20080417-005 26 ~ Full Queue Traffic Model ~ Simulation Results (2)

27. C30-20080417-005 27 ~ Full Queue Traffic Model ~ Simulation Results (2)

28. C30-20080417-005 28 ~ VoIP Traffic Model ~ Simulation Results (1)

29. C30-20080417-005 29 ~ VoIP Traffic Model ~ Simulation Results (2)

30. C30-20080417-005 30 ~ VoIP Traffic Model ~ Simulation Results (2)