Methods for fast blind identification and equalization of communication channels

Abstract

We propose new methods for fast blind identification and equalization of communication channels based on channel outputs sampled at symbol rate. These methods exploit the FA property of the transmitted sequence as well as the algebraic structures and relationships of the symbol and channel parameter matrices to achieve blind equalization. The block based methods proposed herein are fast converging and data efficient. Simulated studies have illustrated the proposed methods capable of achieving reliable blind equalization with a sequence length as short as 20 samples over a wide-range of SNR.