One kind of optimization smoothing singular value decomposition of the unitary transformation matrix

The present invention discloses a multi-input multiplexing system optimization smoothing SVD unitary transform matrix orthogonal frequency division multiple output, starting from the second active subcarriers weight of the k-1 sub-carrier channel matrix around configuration singular matrix U '(k-1) and V' (k-1) respectively unitary transform to obtain approximately reconstructed k-th subcarrier channel matrix singular matrix U '(k) and V' (k); first randomly generating a left singular matrix Wu unitary transform matrix, and accordingly give Wv orthogonalization, then successive iterations of the optimization Wu and orthogonal optimization convergence until finally obtain the reconstructed left and right singular matrix U '(k) and V' (k). The relationship between the presence of unitary transformation method of the present invention contemplates the left and right singular matrix of a singular value decomposition of any two left and right singular matrix of a singular value decomposition is not unique properties and the same matrix, to improve the iterative optimization adjacent left and right singular matrix of the unitary transformation matrix subcarrier channel correlation matrix, thereby improving the performance of the communication system based on packet channel state information feedback.