Adaptive decision-directed quantized-state algorithms for multiuser detection in CDMA systems

The commercial IS-95 CDMA systems employ matched filter (MF) as detectors. Its capacity is limited by multiaccess interference (MAI) as the MF detector regards MAI as additional background noise. Multiuser detection (MUD) techniques such as the decorrelating detector and minimum mean-square error linear detector were recently developed to increase system capacity by reducing effects of multiaccess interference. These detectors however are limited to complex practical implementation constraint for reasonable number of users.; This thesis proposes an adaptive decision-directed linear MUD system that incorporates novel quantized-state (QS) algorithms which addresses both MF and Linear MUD limitations. It also improves on the adaptive linear MUD by eliminating the requirement of knowledge and transmission of training sequences. Simulations of the resulting system with QS algorithms indicate an overall bit error rate performance improvement over conventional algorithms and MF detector.