A cross-relation based affine projection algorithm for blind SIMO system identification

A new time-domain adaptive algorithm is proposed for blind identification of single-input multiple-output systems that is based on the cross-relation (CR) method. The proposed algorithm novelly exploits the affine projection principle to minimize the CR error. As a result, a cost function is minimized that differs from the one used in existing CR based adaptive algorithms. A major advantage of the proposed multichannel affine projection algorithm (MCAPA) is that the affine projection order can be used to control the tradeoff between the rate of convergence and computational complexity. In an experimental study, MCAPA is compared with two recently developed adaptive algorithms, i.e., the low-cost multichannel least-mean-square (MCLMS) and high-performance multichannel Newton (MCN) algorithms. The results show that MCAPA converges faster than MCLMS with a computational complexity that is significantly lower than MCN, thereby increasing the applicability of the CR method.