System and method for blind identification of multichannel finite impulse response filters using an iterative structured total least-squares technique

1. A system for blind identification of multichannel finite impulse response filters, comprising:

a decomposition generator configured to construct, from noisy symbols containing data, a block-Hankel structured matrix and a decomposition thereof;

an iterative matrix re-estimator coupled to said decomposition generator and configured iteratively to re-estimate singular vectors corresponding to a scalar that satisfies said block-Hankel matrix until a sufficient convergence occurs;

a structured low-rank estimate matrix generator coupled to said iterative matrix re-estimator and configured to generate a block-Hankel structured low-rank estimate matrix related to said block-Hankel structured matrix; and

a structured total least-squares (STLS) channel estimator coupled to said structured low-rank estimate matrix generator and configured to obtain STLS channel estimates from said block-Hankel structured low-rank estimate matrix.

2. The system as recited in claim 1 wherein said decomposition is a QR decomposition.

3. The system as recited in claim 1 wherein said iterative matrix re-estimator is further configured to compute an eigenvector corresponding to the smallest eigenvalue of said block-Hankel matrix to yield initial estimates for said singular vectors.

4. The system as recited in claim 1 wherein said iterative matrix re-estimator is further configured to use arbitrary vectors resulting from said decomposition to re-estimate said singular vectors.

5. The system as recited in claim 4 wherein said iterative matrix re-estimator is further configured to use said re-estimated singular vectors to re-estimate symmetric positive definite matrices related to the decomposition until said sufficient convergence occurs.

6. The system as recited in claim 5 wherein said decomposition is a QR decomposition.

7. The system as recited in claim 1 wherein a channel compensator associated with said system is configured to use said STLS channel estimates to remove at least some channel-induced noise from said noisy symbols.

8. A method of blind identification of multichannel finite impulse response filters in an apparatus, comprising:

constructing a block-Hankel structured matrix and a decomposition thereof from noisy symbols containing data;

iteratively re-estimating singular vectors corresponding to a scalar that satisfies said block-Hankel matrix until a sufficient convergence occurs;

generating a block-Hankel structured low-rank estimate matrix related to said block-Hankel structured matrix; and

obtaining structured total least-squares (STLS) channel estimates from said block-Hankel structured low-rank estimate matrix.

9. The method as recited in claim 8 wherein said decomposition is a QR decomposition.

10. The method as recited in claim 8 further comprising computing an eigenvector corresponding to the smallest eigenvalue of said block-Hankel matrix to yield initial estimates for said singular vectors.

11. The method as recited in claim 8 wherein said iteratively re-estimating comprises using arbitrary vectors resulting from said decomposition to re-estimate said singular vectors.

12. The method as recited in claim 11 wherein said iteratively re-estimating comprises using said re-estimated singular vectors to re-estimate symmetric positive definite matrices related to the decomposition until said sufficient convergence occurs.

13. The method as recited in claim 12 wherein said decomposition is a QR decomposition.

14. The method as recited in claim 8 further comprising using said STLS channel estimates to remove at least some channel-induced noise from said noisy symbols.

15. A method of blind identification of multichannel finite impulse response filters in an apparatus, comprising:

computing a decomposition of a block-Hankel structured matrix containing noisy data from multiple channels;

computing an eigenvector corresponding to the smallest eigenvalue of said block-Hankel matrix to yield initial estimates for singular vectors corresponding to a scalar that satisfies said block-Hankel matrix;

using re-estimated singular vectors to re-estimate symmetric positive definite matrices related to said decomposition until sufficient convergence occurs; and

generating a block-Hankel structured low-rank estimate matrix related to said block-Hankel structured matrix.

16. The method as recited in claim 15 wherein said decomposition is a QR decomposition.

17. The method as recited in claim 15 further comprising constructing said block-Hankel structured matrix.

18. The method as recited in claim 15 wherein said scalar is a smallest scalar.

19. The method as recited in claim 15 further comprising using arbitrary vectors resulting from said QR decomposition to re-estimate said singular vectors.

20. The method as recited in claim 15 further comprising obtaining structured total least-squares (STLS) channel estimates from said block-Hankel structured low-rank estimate matrix.