Methods for opportunistic multi-user beamforming in collaborative MIMO-SDMA

• US 20080075058A1
• Filed: 09/27/2006
• Published: 03/27/2008
• Est. Priority Date: 09/27/2006
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Mundarath, Jayakrishnan C., Kotecha, Jayesh H.
• Granted Patent: US 8,073,486 B2
• Time in Patent Office: Days
• US Class Current: 370/342
• CPC Class Codes: H04B 7/0417 Feedback systems
• CPC Class Codes: H04B 7/043 using best eigenmode, e.g. beam forming or beam steering
• CPC Class Codes: H04B 7/0456 Selection of precoding matrices or codebooks, e.g. using matrices antenna weighting
• CPC Class Codes: H04W 16/28 using beam steering
• CPC Class Codes: H04W 24/02 Arrangements for optimising operational condition
• CPC Class Codes: H04W 24/06 Testing, supervising or monitoring using simulated traffic
• CPC Class Codes: H04W 24/10 Scheduling measurement reports ; Arrangements for measurement reports
• CPC Class Codes: H04W 4/00 Services specially adapted for wireless communication networks; Facilities therefor
• CPC Class Codes: H04W 72/21 in the uplink direction of a wireless link, i.e. towards the network
• CPC Class Codes: H04W 72/23 in the downlink direction of a wireless link, i.e. towards a terminal
• CPC Class Codes: H04W 84/12 WLAN [Wireless Local Area Networks]
• CPC Class Codes: H04W 88/08 Access point devices

Abstract

A system and method for opportunistically designing collaborative beamforming vectors is disclosed for a wireless multiple input, multiple output (MIMO) space division multiple access (SDMA) communication system by sequentially designing beamforming vectors for ranked channels in order to exploit the instantaneous channel conditions to improve per user average SNR performance. Each subscriber station independently transmits information to a base station that allows the base station to determine beamforming vectors for each subscriber station by ranking the subscriber stations by channel strength. Using sequential nullspace methods, the ranked channel matrices are then used to select the channel matrix H_i for the best subscriber station, to design the w_i, v_i for the best subscriber station as the left and right singular vectors of the MIMO channel matrix H_i, to transform the remaining channels and to continue the process until beamforming vectors are designed for all channels.

Claims

1. A method for designing beamforming vectors for a collaborative space division multiple access system, comprising:

(a) assembling at a base station m channel matrices for m subscriber stations;

(b) ranking the m channel matrices by left singular value to select a top-ranked channel matrix H from the m channel matrices;

(c) jointly designing weighting and combining beamforming vectors w and v for the top-ranked channel matrix H as left and right singular vectors, respectively, of the top-ranked channel matrix H;

(d) transforming any non-selected channel matrices into a nullspace of the beamforming vectors w and v;

(e) selecting a next highest ranked channel matrix as the top-ranked channel matrix; and

(f) repeating steps (c), (d) and (e) until beamforming vectors for all m channel matrices are designed.

2. The method of claim 1, where the weighting beamforming vector w is computed as a left singular vector corresponding to a maximal singular value of channel matrix H and the combining beamforming vector v is computed as a right singular vector corresponding to a maximal singular value of channel matrix H.

3. The method of claim 1, where the transforming step comprises computing transformed channel matrices for any non-selected channel matrices H_i such that H_i=T_tT_t^HH_iT_rT_r^H, where T_t=null({H v}) and T_r=null ({H^Hw}).

4. The method of claim 1 further comprising transmitting each combining beamforming vector v or an index representative thereof to a corresponding subscriber station for use as a combining vector v.

5. The method of claim 1 where the step of jointly designing weighting and combining beamforming vectors comprises:

dividing the m channel matrices into a P set and an S set, if any subscriber station has a receive antenna array with k antennas, where k<

m, where the P set contains k of the highest ranked subscriber stations and the S set contains all remaining subscriber stations;

designing combining vectors v_s for the S set;

transforming channel matrix information for the P set;

designing weighting vectors w_p and combining vectors v_p for the P set;

transforming channel matrix information for the S set; and

designing weighting vectors w_s for the S set.

6. The method of claim 5, where the step of designing combining vectors v_s for the S set comprises computing, for each subscriber station in the S set, a right singular vector corresponding to a maximum singular value of a channel matrix between the base station and the subscriber station.

7. The method of claim 6, where the step of transforming channel matrix information for the P set comprises projecting the channels matrices of subscriber stations in the P set onto a null space corresponding to the combining vectors v_s for subscriber stations in the S set.

8. The method of claim 7, where the step of designing weighting vectors w_p and combining vectors v_p for subscriber stations in the P set comprises:

ranking channel matrices for the subscriber stations in the P set by left singular value to identify a highest ranked channel matrix,

computing a weighting vector w as the left singular vector of the highest ranked channel matrix;

computing a combining vector v as the right singular vector of the highest ranked channel matrix;

removing the highest ranked channel matrix from the channel matrices for the subscriber stations in the P; and

transforming any remaining channel matrices for the subscriber stations in the P set before computing any additional weighting and combining vectors for subscriber stations in the P set.

9. The method of claim 8, where the step of designing weighting vectors w_s for the S set comprises using a linearly constrained minimum variance process after transforming channel matrix information for the S set by projecting the channels of the subscriber stations in the S set onto a null space corresponding to the combining vectors v_p for subscriber stations in the P set.

10. The method of claim 1, further comprising transmitting over a feed forward channel an index representative of a combining vector v_i to a corresponding subscriber station i, where the index is used to retrieve the combining vector v_i from a predefined codebook at the corresponding subscriber station i.

11. The method of claim 1, further comprising receiving over a feedback channel an index representative of an estimated channel matrix H for a channel between the base station and a first subscriber station, where the index is used at the base station to access the estimated channel matrix H from a codebook for use in jointly designing the weighting and combining beamforming vectors w and v.

12. A base station comprising:

a plurality of antennas for transmitting one or more signals to m subscriber stations over m channels represented by a corresponding m channel matrices (H₁, . . . H_m), respectively; and

a module for jointly designing combining and weighting beamformers for each subscriber station in a set of unprocessed subscriber stations, where each subscriber station is ranked by left singular value of the channel matrix associated with subscriber station, where the module is configured to;

select from the set of unprocessed subscriber stations a top-ranked subscriber station having an associated channel matrix H_i;

jointly design a pair of combining and weighting beamformers (v_i, w_i) for the top-ranked subscriber station by computing left and right singular vectors, respectively, of the channel matrix H_i associated with the top-ranked subscriber station;

remove the top-ranked subscriber station from the set of unprocessed subscriber stations; and

compute transformed channel matrices for all remaining subscriber stations in the set of unprocessed subscriber stations by projecting each channel matrix for the remaining subscriber stations onto a nullspace of the combining and weighting beamformers (v_i, w_i).

13. The base station of claim 12, where the module is further configured to feed forward to a subscriber station i a combining beamformer v_i or information representative thereof by forwarding an index representing which of an ordered set of possible combining beamformers extracted from the channel matrix for the subscriber station i is the combining beamformer v_i., where the combining beamformer v_i is used as a combining or weighting beamformer.

14. The base station of claim 12, further comprising a first codebook for storing an indexed plurality of channel matrices, such that the base station receives over a feedback channel an index representative of an estimated channel matrix H_i for a channel between the base station and subscriber station i, where the index is used at the base station to access the estimated channel matrix H_i from the first codebook for use in jointly designing the combining and weighting beamformers w and v.

15. The base station of claim 12, where the module is further configured to receive an estimated channel matrix H_i from a subscriber station i for use in jointly designing the combining and weighting beamformers w and v.

16. The base station of claim 12, where each channel matrix for the remaining subscriber stations is projected onto a nullspace by computing transformed channel matrices for any non-selected channel matrices H_i such that H_i=T_tT_t^HH_iT_rT_r^H, where T_t=null({H v}) and T_r=null({H^Hw}).

17. The base station of claim 12, where the module is further configured to:

divide the m subscriber stations into a P set and an S set, if any subscriber station has a receive antenna array with k antennas, where k<

m, where the P set contains k of the highest ranked subscriber stations and the S set contains all remaining subscriber stations;

design combining beamformers v_s for the S set;

transform channel matrix information for the P set;

design weighting beamformers w_p and combining beamformers v_p for the P set;

transform channel matrix information for the S set; and

design weighting beamformers w_s for the S set.

18. The base station of claim 17, where the module is configured to design combining beamformers v_s for the S set by computing, for each subscriber station in the S set, a right singular vector corresponding to a maximum singular value of a channel matrix between the base station and the subscriber station.

19. The base station of claim 18, where the module is configured to transform channel matrix information for the P set by projecting the channels of the subscriber stations in the P set onto the null space corresponding to the combining beamformers v_p for subscriber stations in the S set.

20. The base station of claim 19, where the module is configured to design weighting beamformers w_p and combining beamformers v_p for subscriber stations in the P set by repeating, until all weighting beamformers w_p and combining beamformers v_p for subscriber stations in the P set have been computed, the following steps:

ranking channel matrices for the subscriber stations in the P set by left singular value to identify a highest ranked channel matrix,

computing a weighting beamformer w as the left singular vector of the highest ranked channel matrix;

computing a combining beamformer v as the right singular vector of the highest ranked channel matrix;

removing the highest ranked channel matrix from the channel matrices for the subscriber stations in the P; and

transforming any remaining channel matrices for the subscriber stations in the Pset.

21. The base station of claim 20, where the module is configured to design weighting beamformers w_s for the S set by using a linearly constrained minimum variance process after transforming channel matrix information for the S set by projecting the channels of the subscriber stations in the S set onto the null space corresponding to the combining beamformers v_p for subscriber stations in the P set.

22. The base station of claim 12 used in a collaborative, multiple input, multiple output (MIMO) space division multiple access (SDMA) system comprising one or more subscriber stations, each of which feeds back an estimated channel matrix information H_i or information representative thereof to the base station for use in jointly designing combining and weighting beamformers.

23. The base station of claim 22, where the base station feeds forward to each subscriber station a corresponding combining or weighting beamformer or an index representative thereof, where the corresponding combining or weighting beamformer has been designed for said subscriber station.

24. A wireless communication method using beamforming for a collaborative, multiple input, multiple output (MIMO) space division multiple access (SDMA) system, the method comprising:

ranking m subscriber stations by left singular value of a channel matrix associated with each subscriber station; and

designing weighting beamformers w and combining beamformers v for each of the ranked m subscriber stations by sequentially selecting, in descending order of rank, each subscriber station and computing therefore left and right singular vectors, respectively, from a channel matrix associated with the selected subscriber station, where any channel matrices associated with any non-selected subscriber stations are transformed onto a nullspace of the designed weighting beamformers w and combining beamformers v.