1. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 1 11ac Explicit Feedback Format Authors: Date: 2010-11-11

2. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 2

3. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 3 Outline Beamforming feedback frame format –VHT MIMO control field Angle resolution in Codebook info –Compressed beamforming report Tone mapping and grouping –MU-Exclusive beamforming report Per-Tone SNR subfield

4. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 4 Feedback Frame Format Proposal Propose to define compressed V matrix FB, which is based on the 802.11n subclause 20.3.12.2.5, 7.3.1.29, and 7.4.10.8, with appropriate changes for 11ac [1]. Propose to define the 11ac compressed V FB frame for SU and MU be the Action-No- ACK format based on 7.4.10.8, with appropriate extensions: –Category: VHT –Action: Compressed Beamforming –Define VHT MIMO Control Field, as in the next slide –Define MU exclusive beamforming report that contains per-tone SNR Per-tone SNR details TBD OrderInformation 1Category (=VHT) 2Action (=Compressed Beamforming) 3VHT MIMO Control 4Compressed Beamforming Report 5MU-exclusive Beamforming Report Needs codebook info definition Needs tone mapping/grouping Needs to define TBD subfields for MU

5. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 5 VHT MIMO Ctrl Field Bit definition: –Nc: 0~7 correspond to 1~8 columns of V The number of maximum space-time streams that beamformer can use for beamforming. If the SU-type feedback frame (indicated by MU-type bit = 0 in VHT MIMO CTRL field) includes MFB, Nc shall be the same as Nsts in MFB field –Nr: 0~7 correspond to 1~8 rows of V The number of the transmit antennas to be used for beamforming –BW: 0: 20MHz 1: 40MHz 2: 80MHz 3: 160MHz –Ng: 0: Ng=1 1: Ng=2 2:Ng=4 for tone grouping –Codebook Info: Bit resolution for angles, (b , b  ) : (In SU Mode, when MU-type bit is not set): 0: (2,4), 1: (4,6) (In MU Mode, when MU-type bit is set): 0: (5,7), 1: (7,9) –MU-type: 0: SU-FB 1: MU-FB –Sounding Sequence Number NcNrBWNg Codebook Info TBD 33221 Sounding Seq # 8 MU-type 14 3 Octets in total

6. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 6 Codebook Info - Angle Resolution -

7. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 7 Samples of Simulation with SU-BF for Codebook Info (1) SU-MIMO, 40MHz, 4x3, 4STA, 11ac Chan D, Ng=1, 200 Channel Realizations Loss in % relative to first line 13dB 16dB19dB22dB25dB28dB31dB34dB V – 6bits Avg Tput (Mbps) 161214262317362388424467 V – 5bits0.10.30.10.2 000 V – 4bits0.50.80.61.10.60.20.1 V – 3bits2.52.82.13.71.90.80.60.2 V - 2bits7.68.76.39.86.243.40.9

8. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 8 Samples of Simulation with SU-BF for Codebook Info (2) SU-MIMO, 40MHz, 8x3, 4STA, 11ac Chan D, Ng=1, 200 Channel Realizations Loss in % relative to first line 13dB16dB19dB22dB25dB28dB31dB34dB V – 6bits Avg Tput (Mbps) 233 284349388442485509535 V – 5bits0.30.20.10.0.1000 V – 4bits1.20.70.40.20.40.20.1 V – 3bits3.72.61.40.81.50.90.3 V - 2bits9.910.47.34.87.13.1.31.6

9. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 9 Samples of Simulation with SU-BF for Codebook Info (3) 4x3, Nss=2, Ng=2, MCS9, DNLOS 40MHz

10. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 10 Samples of Simulation with MU-BF for Codebook Info (1) (2 users: Each user has 4x4 H with 2 streams) H feedback –16 elements –Each element has Q and I. –Feedback bits = 16 x 2 x B + 3 H H H feedback –4 + 6 elements –4 elements are real, but 6 elements has I and Q –Feedback bits = (4 + 6 x 2) x B + 3 V feedback –12 angles –Feedback bits = 12 x B

11. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 11 Samples of Simulation with MU-BF for Codebook Info (2) Slide 11 Avg of 8 bits per angle are used

12. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 12 Tone Mapping/Grouping

13. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 13 Background The tone location for the feedback in 11n does not differentiate data tones and pilot tones. However, it is not trivial to estimate the channel for pilot tones in 11ac –For efficient offset tracking, pilot tones has a single stream in 11ac –We need to redefine the tone mapping for feedback, especially with tone grouping –We propose to have a uniform distribution of tone mapping when grouping is used

14. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 14 FB Tone Mapping (20/40/80MHz) BWGrouping NgNsCarriers for which matrices are sent 20MHz152All data subcarriers: -28:1:28 (skip -21, -7, 0, 7, 21) 230-28,-26,-24,-22,-20,-18,-16,-14,-12,-10,-8,-6,-4,-2,-1 28, 26, 24, 22, 20, 18, 16, 14, 12, 10, 8, 6, 4, 2, 1 416-28,-24,-20,-16,-12,-8,-4,-1 28, 24, 20, 16, 12, 8, 4, 1 40MHz1108All data subcarriers: -58:1:58 (skip -53, -25, -11, -1, 0, 1, 11, 25, 53) 258-58:2:58 (same as 11n) 430-58:4:58 (same as 11n) 80MHz1234-122:1:122 (skip -103, -75, -39, -11, -1, 0, -1, 11, 39, 75, 103) 2122-122:2:122 462-122:4:122 For the case of 160MHz, the 80MHz tone mapping in the above table indicates the subcarrier indices to be fed back for each 80MHz segment.

15. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 15 Tone Mapping Interpretation Define the compressed V matrix FB, report field, which is based on the 802.11n subclause 20.3.12.2.5 and 7.3.1.29, with appropriate changes for 11ac : –Sub-Channel Feedback Tone Interpretation at the AP/BFMer: In 40/80/160 MHz BSS, when receiving a feedback with BW=20MHz, the AP/BFMer interprets the FB to correspond to the tones in the Primary channel. In 80MHz and 160MHz BSS, when receiving a feedback with BW=40MHz, the AP/BFMer interprets the FB to correspond to the tones in the Primary and Secondary channels. In 160MHz BSS, when receiving a feedback with BW=80MHz, AP/BFMer interprets the FB to correspond to the tones in the Primary, Secondary and Secondary 40 channels.

16. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 16 Exclusive-MU Feedback Report - PT-SNR -

17. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 17 Per-tone-SNR for MU-type Feedback Unlike SU-MIMO where feedback of the pair {V,MCS} is sufficient, known MU-MIMO precoding algorithms benefit from knowledge of the transmit correlation VS 2 V H with more accurate S per tone. We propose to add a new field (MU-exclusive Beamforming Report field) at the end of Compressed Beamforming Report field, in order to define the feedback of per-tone-SNR for each space time stream when MU-type feedback is indicated. We propose to define the per-tone-SNR for each space time stream as the deviation in dB relative to the fed back average SNR per space- time stream (1 to Nc) The new field has additional 4xNc bits per tone: –This new field (per-tone-SNR) is added in MU-exclusive Beamforming Report field It is signaled by beamformee using MU-type bit in MIMO control field (slide 4) –Whether the report field includes per-tone-SNR depends on the SU/MU-type feedback which is requested by the beamformer. Sli de 17

18. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 18 SNR per tone Calculation For MU-BF, it is recommended to specify how to calculate SNR at BFee, so that BFer can unify the meaning of SNR to calculate the steering matrix fairly for participating MU-clients Add following text in the spec (in 7.3.1.61) –SNR per tone for subcarrier k and space-time stream i is found by where V k,i is the i th column of the feedback beamforming matrix at subcarrier k, and N is the noise plus interference power measured at the beamformee [Note] PT-SNR encoding for the feedback report is presented in the next slide [Note] How to find SNR_average using SNR per tone is already described in 7.3.1.61

19. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 19 Per-Tone-SNR Subfield in MU-Exclusive Beamforming Report The proposed subfield for Per-Tone-SNR FieldSizeMeaning Delta-SNR for carrier at negative band edgeNd x NcNext slide* Delta-SNR for carrier at negative band edge + Ng’Nd x NcNext slide* … Delta-SNR for carrier at negative band edge + (m-1)*Ng’Nd x NcNext slide* Delta-SNR for carrier at negative DC edgeNd x NcNext slide* Delta-SNR for carrier at positive DC edgeNd x NcNext slide* Delta-SNR for carrier at positive DC edge – (m-1)*Ng’Nd x NcNext slide* … Delta-SNR for carrier at positive DC edge - Ng’Nd x NcNext slide* Delta-SNR for carrier at positive band edgeNd x NcNext slide* Each field for Per-Tone-SNR at each tone is in the order of the columns of corresponding V matrix: Nd bits for the first column of corresponding V are followed by Nd bits for the second column and so on, up to the last Nc_th column of corresponding V at each tone. Corresponding V is reported in Compressed Beamforming field (see slide 3)

20. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 20 Meaning of PT-SNR Delta-SNR –The deviation in dB at each tone relative to the fed back average_SNR per space-time-stream,1 to Nc (reported in Compressed Beamforming Report field), from -8 dB to 7 dB with 1 dB granularity –Nd = 4 bits –Ng’ (grouping for PT-SNR) is 2xNg Ng’ is not signaled in VHT MIMO control field – where is a function to find the minimum integer that is not smaller than the argument x.

21. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 21 Simulation Results for 8x3, 4 STAs Slide 21 The effects of Per-Tone-SNR for grouping are captured Tone mapping without tone augmentation in slide 8 is used

22. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 22 Simulation Results for 4x1, 4 STAs The effects of Per-Tone-SNR for grouping are captured Tone mapping without tone augmentation in slide 8 is used Slide 22

23. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 23 Summary Propose to define explicit feedback format for 11ac SU and MU. –Codebook Info (angle resolution) is defined for SU-BF and MU- BF –Tone mapping/grouping is modified from 11n –MU-Exclusive feedback report is introduced only for MU-type feedback SNR per tone is defined with equation

24. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 24 Reference [1] IEEE 802.11-10/1105r1 Explicit sounding and feedback.ppt by H. Zhang and J. Kim et. al.

25. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 25 Pre-Motion #1 Do you support the VHT MIMO Control Field in the beamforming report frame as described in slide 5 and to update the VHT MIMO Control Field in the spec framework document accordingly? –Yes: –No: –Abs:

26. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 26 Pre-Motion #2 Do you support feedback tone mapping as shown on slide 14 and to define it in the spec framework document accordingly? –Yes: –No: –Abs:

27. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 27 Pre-Motion #3 Do you support Sub-Channel Feedback Tone Interpretation as in slide 15 and add the text in the spec framework document? –Yes: –No: –Abs:

28. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 28 Pre-Motion #4 Do you support to include per-tone-SNR field in the MU-exclusive beamforming report as described in slide 19 and 20, and to add the proposed field in the spec framework document? –Yes: –No: –Abs:

29. doc.:IEEE 802.11-10/1227r0 Submission Nov, 2010 Joonsuk Kim, et al Slide 29 Pre-Motion #5 Do you support to define SNR per tone for PT-SNR subfield as described in slide 18 and to add the proposed text in the second bullet in the spec framework document? –Yes: –No: –Abs: