1. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 1 HEW SG PHY Considerations For Outdoor Environment Date: 2013-05-12 Authors:

2. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 2 Abstract This document includes PHY considerations for HEW development especially for Outdoor Environment.

3. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 3 Introduction IEEE 802.11a/b/g/n/ac have been developed focusing on indoor usage cases. Even though the indoor usage case will be one of the major environment for HEW project, we also need to consider outdoor environment as well. [1] Outdoor channel environment is quite different from indoor channel model as measured and modelled in [2][3]. In this contribution, we analyze major characteristics of outdoor channel model (urban macro cell (UMa)).

4. Submission doc.: IEEE 11-13/0536r0 Delay Spread Following figures shows CDF of the maximum excess delay for UMa channel model at 2.4GHz Slide 4Wookbong Lee, LG Electronics May 2013 User Location Average SNR (dB) 95% Max Excess Delay (ns) 30m42.6806 100m27.31330 500m0.31540

5. Submission doc.: IEEE 11-13/0536r0 Delay Spread Following figures shows CDF of the maximum excess delay for UMa channel model at 5GHz Slide 5Wookbong Lee, LG Electronics May 2013 User Location Average SNR (dB) 95% Max Excess Delay (ns) 30m36.5804 100m211344 500m-41546

6. Submission doc.: IEEE 11-13/0536r0 Impact of Larger Delay Spread Performance loss due to larger inter-symbol-interference. There is no other solution than increasing CP length in this case. Just increasing CP length brings throughput loss due to larger PHY overhead. (See Appendix for more details) During the development of IEEE 802.11af or.11ah, members come up with increasing CP length by increasing FFT size for a given bandwidth to maintain physical layer overhead. ─For example, the subcarrier spacing of.11af is 41.7kHz (in case of U.S. channel) and that of.11ah is 31.25kHz while the subcarrier spacing of.11a/n/ac is 312.5kHz. With same logic, we can increase CP length by increasing FFT size for a given bandwidth (20MHz, 40MHz, 80MHz or 160MHz). Slide 6Wookbong Lee, LG Electronics May 2013

7. Submission doc.: IEEE 11-13/0536r0 Impact of Larger Delay Spread Slide 7Wookbong Lee, LG Electronics May 2013 User Location @ 100m Performance gain over FFT 64 (%) FFT 12822.2% FFT 25639.4% FFT 51247.7%

8. Submission doc.: IEEE 11-13/0536r0 Impact of Larger Delay Spread Slide 8Wookbong Lee, LG Electronics May 2013 User Location @ 100m Performance gain over FFT 64 (%) FFT 12816.0% FFT 25627.2% FFT 51232.7%

9. Submission doc.: IEEE 11-13/0536r0 Channel Variation In outdoor channel, per tone channel is varying faster than that in indoor channel due to faster environmental change as well as more channel tap (more multi-path) Slide 9Wookbong Lee, LG Electronics May 2013

10. Submission doc.: IEEE 11-13/0536r0 Impact of Larger Channel Variation Slide 10Wookbong Lee, LG Electronics May 2013 Channel variation of one tone per one OFDM symbol ─Observation during the PPDU max Time In 802.11 ac, aPPDUMaxTime is 5.484 ms

11. Submission doc.: IEEE 11-13/0536r0 Impact of Larger Channel Variation Slide 11Wookbong Lee, LG Electronics May 2013 Channel variation of one tone per 1ms ─ It is possible to check that channel is very fast change in outdoor

12. Submission doc.: IEEE 11-13/0536r0 Design Consideration Points for HEW Slide 12Wookbong Lee, LG Electronics May 2013 Observed features Things to be resolvedEnabling technologies Larger delay spread Need to minimize performance loss due to larger inter-symbol- interference Larger CP length by increasing CP portion versus symbol duration - Pros: Robustness against delay spread in outdoor - Cons: Throughput loss from increased CP portion Larger CP length by increasing FFT size - Pros: Robustness against delay spread in outdoor and no direct throughput loss - Cons: Possible impact on PPDU design by diff. OFDM numerology Larger channel variation Need to compensate performance loss due to distorted channel information Channel estimation improvement - Accurate channel estimation (e.g. pilot, midamble) Channel feedback improvement - Efficient feedback (e.g. Effective SINR (ESINR) feedback, codebook based channel information, fast feedback channel) Techniques to mitigate fading - Enhancement of link quality (e.g. effective error correction and retransmission mechanism)

13. Submission doc.: IEEE 11-13/0536r0 Appendix May 2013 Wookbong Lee, LG ElectronicsSlide 13

14. Submission doc.: IEEE 11-13/0536r0 Performance result for just increasing CP Slide 14Wookbong Lee, LG Electronics May 2013

15. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 15 Channel Model Scenario UMa LoSNLoS Delay Spread (log 10 (s))(-7.03,0.66) * (-6.44,0.39) * K-factor (K) (dB)(9,3.5) * N/A Delay distributionExp Delay scaling parameter2.52.3 Number of cluster1220 Per cluster shadowing std ζ (dB)33 LoS probability as a function of distance, d(m) * (μ,σ) For SNR evaluation, we assume noise figure 5dB, cable loss 2dB, signal power 1W for 20MHz. See page 30-41 of reference [2]

16. Submission doc.: IEEE 11-13/0536r0 Channel Model Slide 16Wookbong Lee, LG Electronics May 2013 Scenario Path loss (dB) Note: fc is given in GHz and distance in m! Shadow fading std (dB) Applicability range, antenna height default values Urban Macro (UMa) LoS PL = 22.0 log 10 (d) + 28.0 + 20 log 10 (f c )  = 4 10m< d 1 < d′ BP (1) PL = 40 log 10 (d 1 ) + 7.8 – 18 log 10 (h′ BS ) –18 log 10 (h′ UT ) + 2 log 10 (f c ) d′ BP < d 1 < 5 000 m (1) NLoS PL = 161.04 – 7.1 log 10 (W) + 7.5 l og 10 (h) – (24.37 – 3.7(h/h BS ) 2 ) log 10 (h BS ) + (43.42 – 3.1 log 10 (h BS )) (log 10 (d )  3) + 20 log 10 (f c ) – (3.2 (log 10 (11.75 h UT )) 2  4.97)  = 6 10 m < d < 5 000 m h = avg. building height (20 m) W = street width (20 m) h BS = 25 m, h UT = 1.5 m (1)h BS = 25 m, h UT = 1.5 m, d′ BP = 4 h′ BS h′ UT f c /c, h′ BS = h BS – 1.0 m, h′ UT = h UT – 1.0 m

17. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 17 Delay profile vs. CP length And we need to have proper modeling on how channel and CP impact performance. One of possible modeling is as follows [4]: T FFT is FFT period CP is CP period |α m | 2 is power of m-th tap τ m is delay of m-th tap including OFDM symbol timing

18. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 18 Delay profile vs. CP length Modified SINR by equations in previous slide matches performance of without ISI

19. Submission doc.: IEEE 11-13/0536r0 Simulation Assumption for Slide 7 and 8 Slide 19Wookbong Lee, LG Electronics May 2013 FFT1FFT2FFT4FFT8 FFT size64128256512 Used data subcarriers52104208416 Symbol length with CP4us8us16us32us CP length0.8us1.6us3.2us6.4us Center frequency2.4GHz/5GHz Bandwidth20MHz Channel modelUMa 3km/h Packet structurePreamble +Data frame PreambleL-STF/LTF/SIG+VHT-SIG-A/STF/LTF/SIG-B =40 us Data frame SERVICE(2bytes) +MAC header(40bytes) +Data +Tail(6bits) +Padding (Data =8kbyte)

20. Submission doc.: IEEE 11-13/0536r0 HTTP traffic [4] Main Object Size (Truncated Lognormal: μ=8.37, σ=1.37, Min=100byte, Max=2Mbyte) Embedded Object Size (Truncated Lognormal: μ=6.17, σ=2.36, Min=50byte, Max=2Mbyte) Number of Embedded Objects per page (Truncated Pareto: α = 1.1, k=2, Max = 53) Slide 20Wookbong Lee, LG Electronics May 2013 Mean: 8,728byte

21. Submission doc.: IEEE 11-13/0536r0 May 2013 Wookbong Lee, LG ElectronicsSlide 21 References [1] IEEE 802.11-13/0331r5- Laurent Carious et al., “High- efficiency WLAN,” March 2013 [2] Report M.2135, “Guidelines for evaluation of radio interface technologies for IMT-Advanced, ” available at http://www.itu.int/pub/R-REP-M.2135 http://www.itu.int/pub/R-REP-M.2135 [2] IST-4-027756 WINNER II D1.1.2 V1.2, “WINNER II channel models,” available at http://www.ist-winner.org/deli verables.htmlhttp://www.ist-winner.org/deli verables.html [3] Mickael Batariere, Kevin Baum, and Thomas P. Krauss, “Cyclic Prefix Length Analysis for 4G OFDM Systems,” VTC 2004 Fall [4] IEEE 802.16m-08/004r5, “IEEE 802.16m Evaluation Methodology Document (EMD),” Jan. 2012