A Higher Order Statistics based Multiuser Detection and its Performance Analysis

The conventional higher order statistics based multiuser detections suffer the local minima in cost function. A blind adaptive algorithm using least mean fourth (LMF) criterion is introduced here. The criterion is proved to be globally convergent, and the weights of the stable extremum satisfy the requirement of multi-access interference decorrelation, i.e. zero-forcing solution. Closed-form expression for the steady-state excess mean-square-error (EMSE) and signal-to-interference- plus-noise ratio (SINR) are derived, and verified by simulations in AWGN channel.