Blind adaptation of a matched filter using the constant modulus algorithm coupled with an optional correction method

Abstract

The matched filter (MF) with the whitening filter (WF) is the optimum receiver front end for the maximum likelihood sequence detection (MLSD). The MF+WF followed by the decision feedback equalizer (DFE) is an approximation for the maximum likelihood (ML) receiver. The estimation of the MF is of utmost importance in a receiver. The WF and the MF are treated separately and the MF is estimated blindly using the constant modulus algorithm (CMA) in this work. In this way, a very simple and blind adaptive way of the MF estimate is obtained instead of the methods like the singular value decomposition (SVD), which is complex and computationally expensive. Moreover, a correction method on the MF is introduced to obtain faster convergence and better performance. Simulations prove that the method is very effective and successful.