ADAPTIVE INTERFERENCE CANCELLATION IN DS-CDMA

Abstract

Multiple Access Interference Cancellation in DS-CDMA systems has been a research focus in recent years due to its potential to substantially increase the system capacity. Among many DS-CDMA detectors that have been proposed, adaptive detectors based on the Minimum Mean Output Energy (MOE) criteria are promising candidates for practical application because of their effectiveness and simplicity. This thesis studies the adaptive MMSE and MOE detectors. Apart from describing the general theory and implementation of them, it concentrates on tackling the issues arising in practical usage. The ordinary adaptive MMSE detector often loses tracking of the desired user in fading channels. To solve the problem, the behaviour of the MMSE detector in fading channels is studied and a modified MMSE detector is proposed which is shown to work very effectively in tracking the faded signal and dealing with multipath signals. The ordinary adaptive MOE detector suffers great performance loss in the multipath channel and we present a modified MOE detector which surmounts the "mismatch" problem posed by the presence of multipath signals. To apply the adaptive detector to the situation where the processing gain is very large, we present a partial adaptive MOE detector based on a novel subspace tracking algorithm (PAST) for reducing the number of filter taps needed. A comparison study is also made between the MMSE and MOE detectors and some properties therein are uncovered.