July 200

doc.: IEEE 802.11-06/0979r0

MIMO-OFDM Beamforming for Improved

Channel Estimation in 802.11n WLAN
Date: 2006-07-18

Authors:
Name
Cong Shen
Michael P. Fitz

Company

Address

Phone

email

University California,
Los Angeles
University California,
Los Angeles

420 Westwood Plaza, Los

Angeles, CA 90066
420 Westwood Plaza, Los
Angeles, CA 90066

+310-206-6718

congshen@ee.ucla.edu

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 IEEEs name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEEs 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 <stuart
.kerry@philips.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>.

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Abstract
MIMO-OFDM Beamforming for Improved Channel
Estimation in 802.11n WLAN

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Problem Formulation
System model
r
X( k , l )

r
N(k , l )

Beamformer

MIMO Channel

V (k )

H (k )

r
Y(k , l )

Receiver
Detection

r
Z(k , l )

k-th subcarrier, l-th time interval,

2Tx, 2Rx

Beamforming: SVD of the channel matrix

BF & water-filling maximize the capacity
BF & U(k) simplify the receiver detection/decoding.

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Problem Formulation (contd)

Channel Estimation
Interpolation and Smoothing (Wiener Filtering)
the MIMO channel H(k) are smooth over subcarriers

What happens to channel estimation when BF is

incorporated?
Can we design BF such that the equivalent channel
after BF, are still smooth over subcarriers?
Yes: interpolation and smoothing, good performance
No: single-subcarrier-based CE, performance degradation

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Challenges and Hopes

SVD of a complex-valued matrix is not unique

Lemma: Valid Us (Vs) differ by unitary rotations
Hope: pick up the right ones
Challenge: how?

Several issues when singular values become close

How to determine the corresponding singular vectors?
In some extreme situations, there is no way to maintain the
smoothness, e.g., a famous example in matrix perturbation theory:

0
1
1 0
A
V

0 1
0
1

1
1 1
%
A E
V

2 1
1

Submission

Slid

1
1

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Our Solution
SVD of a complex-valued matrix is not unique
Smoothed SVD algorithm.

Several issues when singular values become close

How to determine the corresponding singular vectors?
SSVD algorithm can automatically deal with it.
In some extreme situations, there is no way to maintain the
smoothness.
When this happens, there is nothing we can do. BUT using the
IEEE 802.11 TGn models we found this happens very rarely.
Thus it is not dominant in performance.

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Smoothed SVD Algorithm

This is a piece-wise smoothing SVD algorithm for two
adjacent Hs.

(1) H(1) V (1) (1)

Assume H(1) has a valid SVD: U
We want a valid SVD of H(2) such that U(2) and V(2) are obtained
by some unitary rotations of U(1) and V(1):
U(2) U(1) WU , V (2) V (1) WV
The target is to make U(2) / V(2) close to U(1) / V(1). An easy way
to obtain one unitary matrix from another one is by unitary rotation;
Exhaustive search for Wv / Wu is infeasible
H

Too many variables

Real-time application

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Smoothed SVD algorithm (contd)

Add extra (reasonable) constraints on Wv and Wu
Special unitary matrix, SU(2)

A
*
*

where 1

is real.
Thus, Wv / Wu only has two degrees of freedom:
1

1
1

1 P*
Q*
*
*
2
U(2) U(1)
and V (2) V (1)
(1 P P) 2
(1 Q Q)

Q 1
P 1

H
Solve P and Q such that U(2) H(2)V (2) has zero offdiagonal elements

Two complex variables, two equations, closed-form solution

available.
Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Extension to more than 2 antennas

Possible ways
Use the representation theory of SU(n), and similarly reduce the
number of unknowns and solve them.
Use the same argument as before, but
I

WU =

WV =

Q (I + Q Q)

Q I

0
1

H
H
I P (I + P P) 2

P
I

0
H

1
2

1
H 2

(I + QQ )

1
(I + PP H ) 2

matrices instead of scalars

Complexity issue, only feasible for small number of antennas

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Frequency Smoothed Beamformer Design

Based on Smoothed SVD

Input: H(k ), k 1,L D p

(1), (1), (1)]H
svd
Initialize: [UV
Repeat:

(1)

for i 2 to D p do

[UV
(i ), (i ), (i )] Hssvd H
(i 1),U (i), V(i 1), (i 1) ;
end for

Output: beamforming matrices

Submission

V (k ), k 1,L D p

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Statistics of the equivalent channel

H(k) has very nice statistical behaviors
%(k ) loses almost all the
However, strictly speaking, H
nice properties
Not Gaussian
Spatially correlated
Frequency autocorrelation functions are difficult to get

%(k ) to be
For simplicity we assume H

Gaussian
No cross-subchannel correlation
Obtain the frequency autocorrelation functions via simulation

Then, Wiener Filtering CE can be directly applied.

Submission

Slid

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Simulation Results
64QAM BICM MIMO-OFDM, channel D

10

-1

Bit Error Rate

10

-2

10

-3

10

-4

2 X 2 MIMO-OFDM preamble
OFDM structure identical to
802.11a/g (i.e., 52 subcarriers)
BICM Spatial multiplexing, with ML
detection and Viterbi hard decoding
64 QAM
IEEE 802.11n channel model D
Wiener Filtering channel estimation

10

perfect CSI at the receiver

FSB with smoothing across all subcarriers
no FSB with single-subcarrier CE

-5

10

Submission

10

12
14
16
18
20
SNR per receive antenna, dB

22

24

Slid

26

Con

July 200
6

doc.: IEEE 802.11-06/0979r0

Simulation Results (2)

256QAM BICM MIMO-OFDM, channel D

10

-1

Bit Error Rate

10

-2

10

-3

10

-4

2 X 2 MIMO-OFDM preamble
OFDM structure identical to
802.11a/g (i.e., 52 subcarriers)
BICM Spatial multiplexing, with ML
detection and Viterbi hard decoding
256 QAM
IEEE 802.11n channel model D
Wiener Filtering channel estimation

10

-5

10

12

perfect CSI at the receiver

FSB with smoothing across all subcarriers
no FSB with single-subcarrier CE
14

Submission

16

18
20
22
SNR per receive antenna, dB

24

26

Slid

28

Con