Abstract

In this thesis, we propose a modified multistage linear parallel interference cancellation (PIC) structure using the blind adaptive minimum mean-output-energy (MMOE) algorithm for DS-CDMA systems. The proposed receiver has a computational complexity that is linear in the number of users, and exhibits a performance close to the optimum linear minimum mean-squared-error (LMMSE) receiver whose complexity has a cubic dependence on the number of users. Moreover, the proposed blind adaptive MMOE-PIC receiver requires no side information on users' received signal amplitudes, which can further lower the complexity of the overall system. Another multistage detector with low complexity is introduced for DS-CDMA systems over both flat and frequency-selective fading channels. The proposed detector employs the blind adaptive multiuser receiver as the first stage followed by stages with interference cancellation (IC). The performance of the proposed detector is very close to the single user bound and is independent of system loads in flat fading channels. In multipath fading channels, from low to moderate SNR's, the proposed receiver suffers less than 1dB penalty compared with ideal detection without interference. At high SNR's, an error floor occurs for high loaded systems. Finally we consider a class of decision-feedback detectors (DFDs) based on the modified MMSE performance criterion in the presence of dynamic fading. (Abstract shortened by UMI.)