An Application of ICA to Blind DSCDMA Detection: A Joint Optimization Criterion

Abstract. This paper studies the application of the theory and algo-rithms of blind signal extraction to solve the problem of the detection of the desired users in a DS-CDMA communications system. Typically, the uplink in this system is characterized by users that transmit asyn-chronously and propagation channels that are multipath. We address inverse filter criterion introduced by Tugnait in [1] and we show that a prewhitening preprocessing approach together with the joint combina-tion of several higher order statistics improves the detector performance.