Identification of Parallel Mimo Channels in Wireless Communication Using Power Iterations

Identification of the channel matrix is of main concern in wireless MIMO (Multiple Input Multiple Output) systems. To maximize the SNR, the best way to utilize a MIMO system is to communicate on the top singular vectors of the channel matrix. Here, we present a new approach for direct blind identification of the main independent singular modes, without first estimating the channel matrix itself. The top right and left singular vectors are determined using payload data, and are continuously updated while at the same time being used for communication. The feasibility of the approach is demonstrated by simulating the performance over a noisy, fading time-varying channel. Mathematically, the technique is related to the iterative numerical Power method for finding eigenvalues of a matrix as well as the “time reversal mirror” technique developed within acoustics.