May 2007

H. Zhang., et al.Slide 1

doc.: IEEE 802.11-07/0612r1

Submission

Comment Resolution on TX Beamforming CSI/Steering Feedback Quantization Bitwidth Nb

Notice: This document has been prepared to assist IEEE 802.11. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.

Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE 802.11.

Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures <http:// ieee802.org/guides/bylaws/sb-bylaws.pdf>, including the statement "IEEE standards may include the known use of patent(s), including patent applications, provided the IEEE receives assurance from the patent holder or applicant with respect to patents essential for compliance with both mandatory and optional portions of the standard." Early disclosure to the Working Group of patent information that might be relevant to the standard is essential to reduce the possibility for delays in the development process and increase the likelihood that the draft publication will be approved for publication. Please notify the Chair <stuartk@ok-brit.com> as early as possible, in written or electronic form, if patented technology (or technology under patent application) might be incorporated into a draft standard being developed within the IEEE 802.11 Working Group. If you have questions, contact the IEEE Patent Committee Administrator at <patcom@ieee.org>.

Date: 2007-05-02

Name Company Address Phone email Hongyuan Zhang

Marvell 5488 Marvell Ln Santa Clara, CA 95054

408-222-1837

hongyuan@marvell.com

Douglas Chan Cisco Systems, Inc. 170 West Tasman Drive, San Jose, CA 95134

408-527-9344 dougchan@cisco.com

Rohit Nabar Marvell 5488 Marvell Ln Santa Clara, CA 95054

408-222-0865 rnabar@marvell.com

Xiantao Sun University of Delaware

140 Evans Hall, Newark, DE 19716

xsun@udel.edu

Peter Loc Marvell 5488 Marvell Ln Santa Clara, CA 95054

408-222-9148 ploc@marvell.com

May 2007

H. Zhang., et al.Slide 2

doc.: IEEE 802.11-07/0612r1

Submission

CID344

Since the antenna configuration is constraint to 4x4, we don't need N_b to be as high as 8 bits.

Remove 8, and leave the max to be 6 bits.

CID857

Currently the coefficient size Nb in MIMO control field in Figure n21 has 4 options {4,5,6,8}. Would be good to allow more resolution choices for steering matrix precision (for example Nb = 3 or Nb = 7).

Use additional reserved bit in MIMO control field so that Nb field is 3 bits long. Then Nb = number represented by 3 bits + 1.Then shift "Codebook information" and "remaining matrix segment" subfields to the right by 1 bit. Also the Nb values in Clause 20.3.11.2.1 and 20.3.11.2.2 at page 286 need to be modified.

May 2007

H. Zhang., et al.Slide 3

doc.: IEEE 802.11-07/0612r1

Submission

Overview • Based on our simulations for different Nb:

– For CSI feedback, Nb = 4 bits performs optimal or near optimal (in terms of goodput) in most of the cases.

– For non-compressed steering feedback, Nb = 2 bits may be preferable (in terms of goodput) for some scenarios.

• We suggest to interpret the Nb values in MIMO control field differently for CSI feedback and non-compressed steering feedback, given that their performance changes on different feedback bitwidth are NOT similar.

– Keep the current set of Nb values for CSI feedback.

– Reinterpret the set of possible Nb values for non-compressed steering feedback in the “Coefficient Size” subfield of MIMO control field, to more appropriately reflect actual performances.  This will also provide more flexibility towards balancing the feedback overhead with performances over different configurations, eg. SNR range of a location, Doppler effects, different MCS, steering calculation approaches, etc.

• Nb = 8 may be desirable in some applications requiring high resolution feedback (e.g. Calibration exchange for implicit TxBF), therefore it should be kept as one option.

May 2007

H. Zhang., et al.Slide 4

doc.: IEEE 802.11-07/0612r1

Submission

Proposed Solutions• CID 344: reject (refer to 11-07-0613r1)

• CID 857: counter (refer to 11-07-0613r1)

• Summary of the resolution (refer to 11-07-0613r1 for details)– In D2.0, 7.3.1.29, table n14, keep the current 4 possible Nb values for CSI feedback, and

reinterpret the Nb=5 to Nb=2 for non-compressed steering feedback.

– Update Nb values in 20.3.11.2.2 correspondingly.

Coefficient Size Indicates the number of bits in the representation of the real and imaginary parts of of each element in the matrix:

For CSI feedback:Set 0 for Nb=4Set 1 for Nb=5Set 2 for Nb=6Set 3 for Nb=8For Non-compressed beamforming matrix feedback:Set 0 for Nb=4Set 1 for Nb=2Set 2 for Nb=6Set 3 for Nb=8

May 2007

H. Zhang., et al.Slide 5

doc.: IEEE 802.11-07/0612r1

Submission

Simulation Settings

• Channel D/B NLOS, MCSs with Nss=1 and 2

• Explicit TxBF with CSI Feedback or with Non-compressed Steering, Ng=1 (no tone grouping)

• RF Impairments and actual channel estimation for receiving sounding PPDUs are added

• PERs are plotted

• PHY layer goodput is plotted: – PHY preamble and CSI feedback treated as overhead

– Lowest rate (MCS 0) on CSI feedback—assume error-free

– Burst transmission (SIFS interval) assumed for steered packets

– Fast fading case (fd=6 Hz) : TxBF sounding interval is set in the range where the channel Doppler does not affect PER significantly.

May 2007

H. Zhang., et al.Slide 6

doc.: IEEE 802.11-07/0612r1

Submission

I. CSI Feedback

May 2007

H. Zhang., et al.Slide 7

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 6, Nrx=1, Ntx=3, Channel D NLOS

SNR (dB)

PE

R

May 2007

H. Zhang., et al.Slide 8

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 6, Nrx=1, Ntx=3, 10ms sounding interval

* Nb=3 wins out at most SNRs,

* Nb=2 performs well at high SNR.

May 2007

H. Zhang., et al.Slide 9

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 9, Nrx=2, Ntx=3, Channel D NLOS

May 2007

H. Zhang., et al.Slide 10

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 9, Nrx=2, Ntx=3, Channel D NLOS, 10 ms sounding interval

* Nb=4 wins out at most SNRs,

* Nb=3 preferred at high SNR.

May 2007

H. Zhang., et al.Slide 11

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 12, Nrx=2, Ntx=3, Channel D NLOS

May 2007

H. Zhang., et al.Slide 12

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 12, Nrx=2, Ntx=3, Channel D NLOS, 10 ms sounding interval

* Nb=4 wins out for most SNRs,

May 2007

H. Zhang., et al.Slide 13

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 15, Nrx=2, Ntx=3, Channel B NLOS

May 2007

H. Zhang., et al.Slide 14

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 15, Nrx=2, Ntx=3, Channel B NLOS, 10 ms sounding interval

* Nb=4 is favorable

May 2007

H. Zhang., et al.Slide 15

doc.: IEEE 802.11-07/0612r1

Submission

II. Non-compressed Steering Feedback

May 2007

H. Zhang., et al.Slide 16

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 1, Nrx=1, Ntx=3, Channel D NLOS

May 2007

H. Zhang., et al.Slide 17

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 1, Nrx=1, Ntx=3, Channel D NLOS, 10 ms sounding interval

* Nb=2~4 perform similar, and are favorable through all SNR

May 2007

H. Zhang., et al.Slide 18

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 6, Nrx=1, Ntx=3, Channel D NLOS

May 2007

H. Zhang., et al.Slide 19

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 6, Nrx=1, Ntx=3, Channel D NLOS, 10 ms sounding interval

May 2007

H. Zhang., et al.Slide 20

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 14, Nrx=2, Ntx=3, Channel D NLOS

May 2007

H. Zhang., et al.Slide 21

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 14, Nrx=2, Ntx=3, Channel D NLOS, 10 ms sounding interval

* Nb=3 desirable until intermediate SNR

* Nb=2 desirable at high SNR

May 2007

H. Zhang., et al.Slide 22

doc.: IEEE 802.11-07/0612r1

Submission

PER v.s. SNR for MCS 23, Nrx=4, Ntx=4, Channel D NLOS

May 2007

H. Zhang., et al.Slide 23

doc.: IEEE 802.11-07/0612r1

Submission

Goodput v.s. SNR for MCS 23, Nrx=4, Ntx=4, Channel D NLOS, 10 ms sounding interval

* Nb=3 desirable until intermediate SNR

* Nb=2 desirable at high SNR

May 2007

H. Zhang., et al.Slide 24

doc.: IEEE 802.11-07/0612r1

Submission

Conclusions• For feedback limited cases (i.e., fast-fading), smaller Nb (4, 3, or 2) may achieve higher goodput than larger

Nb, especially for non-compressed feedback.

• Non-compressed steering FB is less sensitive to reduced Nb values, Nb=2 could be desirable, especially for one-stream case.

• Different range of Nb values (2,4,6,8) will give more design flexibility for non-compressed steering feedback (w.r.t. channel Doppler, SNR range, MCS, the steering calculation approaches, etc).

• When optimal link adaptation is involved, goodput results could be different, but smaller Nb values will be still desirable in some cases (e.g. non-compressed steering feedback with one-stream case)

May 2007

H. Zhang., et al.Slide 25

doc.: IEEE 802.11-07/0612r1

Submission

Discussions• The proposed change is a minor modification buying significant

performance improvement (regarding the feedback overhead) in some specific applications. The other applications may still apply Nb= 4, 6, 8.

• Coex with CSI feedback implementation on interpreting Nb subfield:

– For BFmer or BFmee that only implements CSI FB, nothing needs to be changed.

– For BFmer or BFmee that implements both CSI and non-compressed steering feedback, the additional complexity is trivial (while the overhead reduction might be significant):

• If (Non-compressed-FB-Flag is on) && (Nb subfield in MIMO Control Field = “01”), then Nb=2 bits (rather than 5 bits)

• Otherwise, Nb is determined by the rules defined for CSI FB.