Adaptive interference suppression for DS-CDMA in multipath channels

The decentralized minimum mean square error-zero forcing (MMSE-ZF) receiver which can be implemented as a precombining interference canceler followed by coherent combining lends itself to a particular disjoint adaptation of the interference canceling filter and the channel. A least mean square (LMS) based adaptation of this receiver is presented. The quadratic cost-function viz. the output variance is quadratic in coefficients of the interference canceling filter leading to global convergence. On the other hand, the channel coefficients are separately optimized based upon the interference canceler. We explore decision directed strategies to improve the performance of the overall receiver. It is shown that significant performance gains can be achieved if decisions are reused in a soft fashion to influence the adaptation procedure.