Method and detector for a novel channel quality indicator for space-time encoded MIMO spread spectrum systems in frequency selective channels

• US 7,606,319 B2
• Filed: 07/15/2004
• Issued: 10/20/2009
• Est. Priority Date: 07/15/2004
• Status: Expired due to Fees

• Current Assignee: Nokia Corporation
• Original Assignee: Nokia Corporation
• Inventors: Wang, Yan, Zhang, Jianzhong, Raghothaman, Balaji, Mandyam, Giridhar
• Primary Examiner(s): Liu; Shuwang
• Assistant Examiner(s): Guarino; Rahel
• Application Number:
  US10/892,833
• Publication Number:
  US 20060013328A1
• Time in Patent Office: 1,923 Days
• Field of Search: 375/267, 375144-148, 375/260, 375232-234, 375/142, 370/278, 370/290, 370/335, 370/473, 455/83, 455/570
• US Class Current: 375/267
• CPC Class Codes: H04B 1/7105 Joint detection techniques, e.g. linear detectors
• CPC Class Codes: H04B 1/71055 using minimum mean squared error [MMSE] detector
• CPC Class Codes: H04L 1/0026 Transmission of channel quality indication
• CPC Class Codes: H04L 1/0656 Cyclotomic systems, e.g. Bell Labs Layered Space-Time [BLAST]
• CPC Class Codes: H04L 1/20 using signal quality detector
• CPC Class Codes: H04L 2025/03426 transmission using multiple-input and multiple-output channels
• CPC Class Codes: H04L 2025/03617 Time recursive algorithms H04L2025/03643 takes precedence

Abstract

In a JE, spread spectrum communication MIMO system where a demultiplexed packet is transmitted over multiple streams, two versions of a single CQI are disclosed: generalized SNR or constrained mutual information CMI between the transmitted and received chip vectors. In the receiver, the CMI is constrained so that filtering is suboptimal. The filter bank is preferably LMMSE or MVDR filters that convert the multi-path channel into a single path channel so that joint sequence detection is unnecessary. Detection is by a per-Walsh code architecture, wherein a plurality of Walsh-code specific detectors in parallel detect bits or symbols from the single-channel chips after downconverting chip to symbols. Link to system mapping is realizable using the disclosed CQI for a JE MIMO system in that the CQI or related information is returned to the transmitter, which adapts coding rate and/or modulation for the channel as represented by the CQI.

Claims

1. A method comprising:

receiving a jointly encoded signal over a multi-path channel by N receive antennas, wherein N is an integer greater than one;

estimating a transmitted chip vector within a chip interval for the jointly encoded signal;

for each of the N receive antennas, sampling the received signal to resolve an antenna-wise received chip vector for each of the N receive antennas;

filtering a block of antenna-wise received chip vectors, the block comprising each of the N antenna-wise chip vectors, using a channel quality indicator that maximizes constrained mutual information between the estimated transmitted chip vector and the block of antenna-wise received chip vectors;

downconverting the filtered block to one of bits and symbols;

detecting in parallel, for each spreading code by which the jointly encoded signal is spread, the downconverted bits or symbols,

wherein the channel quality indicator describes the entire multi-path channel over which the signal was received by the N antennas and wherein the channel quality indicator comprises a generalized signal to noise ratio that represents all channel uses of the multiple input/multiple output multi-path channel over which the jointly encoded signal was received.

2. The method of claim 1 wherein the generalized signal to noise ratio is computed by:

${\mathrm{GSNR}}_k<br><br>\widehat{=}<br><br>{\mathrm{\beta <br><br>}}_k<br><br>\frac{\mathrm{Trace}<br><br>\left({\mathrm{\sigma <br><br>}}_d^2<br><br>{\stackrel{\_}{I}}_M\right)}{\mathrm{Trace}<br><br>\left({\stackrel{\_}{R}}_{\mathrm{zz}}<br><br>\left({\stackrel{\_}{W}}_{\mathrm{LMMSE}}\right)\right)};<br><br>$

where

β

_k is a scalar factor that translates a chip-level signal to noise ratio to a symbol-level signal to noise ratio for the user;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas from which the encoded signal was sent;

R_zz is an error covariance matrix; and

W_LMMSE is the filtered block of antenna-wise chip vectors.

3. The method of claim 1 wherein filtering comprises filtering such that W_MC= R⁻

1 H₀Ā

;

where

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

σ

² is noise with respect the block of antenna-wise received chip vectors;

Ī

is an identity matrix

H₀ is a memoryless multi-path channel estimate matrix, and superscript^H indicates a Hermitian operation; and

Ā

is an arbitrary invertible matrix.

4. The method of claim 1 wherein the constrained mutual information I( d_i y_i−

Fi+F| H) is maximized such that I( d_i y<sub>i−

F;

i+F</sub>| H)=log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀);

wherein

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas from which the encoded signal was sent;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

² is noise vaiance with respect to the block of antenna-wise received chip vectors; and

Ī

is an identity matrix.

5. The method of claim 1, wherein filtering comprises employing a bank of linear minimum mean square error filters W_LMMSE that filter the block of antenna-wise chip vectors according to W_LMMSE=σ

_d² R<sup>−

1</sup> H₀;

where;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix; and

R<sup>−

1</sup> is an inverted covariance matrix of the received block of antenna-wise received chip vectors.

6. The method of claim 1, wherein filtering comprises employing a bank of minimum variance distortionless response linear filters W_MVDR that filter the block of antenna-wise chip vectors according to W_MVDR= R⁻

1 H₀( H₀^H R<sup>−

1</sup> H₀)<sup>−

1</sup>;

where;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

σ

² is noise vairance with respect to the block of antenna-wise received chip vectors and

Ī

is an identity matrix.

7. A method comprising:

receiving a jointly encoded, spread spectrum signal from a multi-path channel over at least two receive antennas in a chip interval;

estimating a transmitted chip vector for the signal;

sampling the signal to achieve a chip-wise signal vector from each receive antenna;

storing the chip-wise signal vectors as a block;

estimating the multi-path channel by determining mutual information between the block of chip-wise signal vectors and the estimated transmitted chip vector, wherein the estimated multi-path channel describes the entire multi-path channel over which the signal was received by the N antennas and wherein the estimated multi-path channel comprises a generalized signal to noise ratio that represents all channel uses of the multiple input/multiple output multi-path channel over which the jointly encoded signal was received;

using the estimate of the multi-path channel, filtering the block of chip-wise signal vectors to restore orthogonality to spreading codes used to spread the signal;

down-converting, descrambling and despreading the filtered block of chip-wise signal vectors to yield parallel outputs of symbol-level signal vectors, each parallel output corresponding to a spreading code; and

for each of the parallel outputs, spatially detecting one of bits or symbols using one spreading code.

8. The method of claim 7 wherein filtering the block of chip-wise signal vectors comprises passing the block of chip-wise signal vectors through a linear minimum mean square error filter.

9. The method of claim 8, wherein the linear minimum mean square error filter operates on the block of chip-wise signal vectors according to W_LMMSE=σ

_d² R<sup>−

1</sup> H₀;

where;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix; and

R<sup>−

1</sup> is an inverted covariance matrix of the block of antenna-wise received chip vectors.

10. The method of claim 7 wherein filtering the block of chip-wise signal vectors comprises passing the block of chip-wise signal vectors through a minimum variance distortionless response filter.

11. The method of claim 10 wherein the minimum variance distortionless response filter operates on the block of chip-wise signal vectors according to W_MVDR= R⁻

1 H₀( H₀^H R<sup>−

1</sup> H₀)<sup>−

1</sup>;

where;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

σ

² is noise variance with respect to the block of chip-wise signal vectors and

Ī

is an identity matrix.

12. The method of claim 7 wherein spatially detecting one of bits or symbols comprises spatially detecting bits for the case where the received signal is jointly encoded with a binary code.

13. The method of claim 7 wherein filtering the block of chip-wise signal vectors to restore orthogonality comprises equalizing sub-channels of the multi-path channel.

14. The method of claim 7 wherein determining mutual information comprises determining a maximized constrained mutual information I( d_i y_i−

F;

i+F| H), such that I( d_i y<sub>i−

F;

i+F</sub>| H)=log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀);

wherein

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas from which the encoded signal was sent;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

² is noise variance with respect to the block of chip-wise signal vectors; and

Ī

is an identity matrix.

15. A method comprising:

in a first transceiver,

jointly encoding a first signal to be transmitted at a first coding rate,

modulating the first signal to be transmitted with a first modulation, and

transmitting the jointly encoded and modulated first signal over a spread spectrum multi-path wireless channel by at least one transmit antenna;

in a second transceiver,

receiving the jointly encoded and modulated first signal by at least two receive antennas over the multi-path channel;

converting the multi-path channel over which the first signal was received to an effective single-path channel;

determining a single channel quality indicator that characterizes the effective single-path channel by constraining mutual information between an estimate of a transmitted chip and a block of antenna-wise received chip vectors;

from the effective single-path channel, detecting in parallel one of bits and symbols, each parallel detection being according to one spreading code by which the first signal is spread over the spectrum; and

transmitting to the first transceiver a feedback based on the channel quality indicator,

wherein the channel quality indicator describes the entire multi-path channel over which the signal was received by the N antennas and wherein the channel quality indicator comprises a generalized signal to noise ratio that represents all channel uses of the multiple input/multiple output multi-path channel over which the jointly encoded signal was received; and

in the first transceiver,

receiving the feedback;

jointly encoding a second signal to be transmitted,

modulating the second signal to be transmitted, and

transmitting the jointly encoded and modulated second signal over a spread spectrum multi-path wireless channel by the at least one transmit antenna,

wherein, in response to the feedback, the second signal is at least one of encoded at a second coding rate and modulated with a second modulation.

16. The method of claim 15 wherein the feedback is an estimated error rate that is calculated from the channel quality indicator.

17. The method of claim 15 wherein the feedback is an instruction to change at least one of coding rate and modulation format based on the channel quality indicator.

18. The method of claim 15 wherein the channel quality indicator is a generalized signal to noise ratio for the second transceiver such that

${\mathrm{GSNR}}_k\; <li></li>\widehat{=}<br><br>{\mathrm{\beta <br><br>}}_k<br><br>\frac{\mathrm{Trace}<br><br>\left({\mathrm{\sigma <br><br>}}_d^2<br><br>{\stackrel{\_}{I}}_M\right)}{\mathrm{Trace}<br><br>\left({\stackrel{\_}{R}}_{\mathrm{zz}}<br><br>\left({\stackrel{\_}{W}}_{\mathrm{LMMSE}}\right)\right)};<br><br>$

where

β

_k is a scalar factor that translates a chip-level signal to noise ratio to a symbol-level signal to noise ratio for the user;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas in the first transceiver from which the encoded signal was sent;

R_zz is an error covariance matrix; and

W_LMMSE is the filtered block of antenna-wise received chip vectors.

19. The method of claim 16 wherein the feedback is the generalized signal to noise ratio.

20. The method of claim 17 wherein the channel quality indicator is maximized constrained mutual information I( d_i y_2F+1| H)=log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀).

21. The method of claim 20 wherein the feedback is the maximized constrained mutual information.

22. A receiver comprising:

at least two receive antennas;

a filter bank of linear filters having a first input coupled to an output of each receive antenna and a second input, said filter bank for equalizing signal vectors received over sub-channels of a multipath channel into signal vectors of a single channel by constraining mutual information between an estimate of a transmitted chip and a block of antenna-wise received chip vectors;

a channel estimator having an input coupled to an output of each receive antenna and an output coupled to the second input of the filter bank, wherein channel estimator is configured to estimate the multi-path channel which describes the entire multi-path channel over which the signal was received by the at least two receive antennas and wherein the estimated multi-path channel comprises a generalized signal to noise ratio that represents all channel uses of the multi-path channel over which the jointly encoded signal was received;

a plurality of joint detectors in parallel with one another, each joint detector having an input coupled to an output of the filter bank and an output coupled to a decoder, each joint detector for detecting one of bits or symbols according to one spreading code;

a chip-to-symbol down-converter, a de-scrambler, and a de-spreader, each disposed between the filter bank and the plurality of joint detectors.

23. The receiver of claim 22 wherein the filter bank comprises a bank of linear minimum mean square error filters that operate according to W_LMMSE=σ

_d² R<sup>−

1</sup> H₀;

where;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix; and

R<sup>−

1</sup> is an inverted covariance matrix of the antenna-wise received chip vectors.

24. The receiver of claim 22 wherein the filter bank comprises a bank of minimum variance distortionless response filters that operate according to W_MVDR= R⁻

1 H₀( H₀^H R<sup>−

1</sup> H₀)<sup>−

1</sup>;

where;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

σ

² is noise variance with respect to the block of chip-wise signal vectors and

Ī

is an identity matrix.

25. The receiver of claim 22 wherein each of the plurality of joint detectors is a spatial detector.

26. The receiver of claim 22 having N receive antennas, wherein the filter bank operates to maximize constrained mutual information I( d_i y_i−

F;

i+F| H), such that I( d_i y<sub>i−

F;

i+F</sub>| H)=log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀);

wherein

σ

_d² is noise variance with respect to the estimated transmitted chip vector;

H₀ is a memoryless multi-path channel estimate matrix, and superscript ^H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

² is noise variance with respect to the block of chip-wise signal vectors; and

Ī

is an identity matrix.

27. An apparatus comprising:

a processing unit configured to compute a channel quality indicator of a channel comprising multiple streams; and

a transmitter configured to transmit the channel quality indicator,

where the channel quality indicator averages constrained mutual information between an estimated transmitted chip vector within a chip interval for a jointly encoded signal received over a multi-path channel by N receive antennas, wherein N is an integer greater than one and a filtered block of antenna-wise received chip vectors comprising each of the N antenna-wise chip vectors,

wherein the channel quality indicator describes the entire multi-path channel over which the signal was received by the N antennas and wherein the channel quality indicator comprises a generalized signal to noise ratio that represents all channel uses of the multiple input/multiple output multi-path channel over which the jointly encoded signal was received.

28. The apparatus as in claim 27, where the channel quality indicator is further computed according to log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀), wherein

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas from which the encoded signal was sent;

σ

_d² is a scalar factor;

H₀ is a memoryless multi-path channel estimate matrix, and superscript H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

² is noise variance with respect to the block of chip-wise signal vectors; and

Ī

is an identity matrix.

29. The apparatus of claim 27, where the channel quality indicator is further computed according to

${\beta }_{}$

k ⁢

Trace ⁡

( σ

d 2 ⁢

I _ M ) Trace ⁡

( R _ zz ⁡

( W _ LMMSE ) ) ;

where

β

_k is a scalar factor that translates a chip-level signal to noise ratio to a symbol-level signal to noise ratio for the user;

σ

_d² is a scalar factor;

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas in the first transceiver from which the encoded signal was sent;

R_zz is an error covariance matrix; and

W_LMMSE is the filtered block of antenna-wise received chip vectors.

30. The apparatus as in claim 27, where the channel is a transmission channel originating from a transmitter, the channel quality indicator being fed back to the transmitter.

31. An apparatus comprising:

a receiver configured to receive a channel quality indicator of a channel comprising multiple streams; and

a processing unit configured to adjust, on the bases of the channel quality indicator, a transmission of the multiple streams on the channel,

where the channel quality indicator averages constrained mutual information between an estimated transmitted chip vector within a chip interval for a jointly encoded signal received over a multi-path channel by N receive antennas, wherein N is an integer greater than one and a filtered block of antenna-wise received chip vectors comprising each of the N antenna-wise chip vectors,

wherein the channel quality indicator describes the entire multi-path channel over which the signal was received by the N antennas and wherein the channel quality indicator comprises a generalized signal to noise ratio that represents all channel uses of the multiple input/multiple output multi-path channel over which the jointly encoded signal was received.

32. The apparatus of claim 31, where the channel quality indicator represents the channel according to log det(Ī

_M+σ

_d² H₀^H R<sup>−

1</sup> H₀), wherein

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas from which the encoded signal was sent;

σ

_d² is a scalar factor;

H₀ is a memoryless multi-path channel estimate matrix, and superscript H indicates a Hermitian operation;

R{circumflex over (=)}σ

_d²H₀H₀^H+σ

²Ī

;

σ

² is noise variance with respect to the block of chip-wise signal vectors; and

Ī

is an identity matrix.

33. The apparatus of claim 31, where the channel quality indicator represents the channel according to

${\beta }_{}$

k ⁢

Trace ⁡

( σ

d 2 ⁢

I _ M ) Trace ⁡

( R _ zz ⁡

( W _ LMMSE ) ) ;

where

β

_k is a scalar factor that translates a chip-level signal to noise ratio to a symbol-level signal to noise ratio for the user;

σ

_d² is a scalar factor;

Ī

_M is an identity matrix of size M×

M, wherein M is a number of transmit antennas in the first transceiver from which the encoded signal was sent;

R_zz is an error covariance matrix; and

W_LMMSE is the filtered block of antenna-wise received chip vectors.

34. The apparatus of claim 31, further comprising a transmitter for transmitting the multiple streams.