Research On Multiuser Detection In DS-SS System

3G moblile communications industry has been developed rapidly in the last fewyears. As one of core technology in3G, CDMA has advantages of high capacity, finesecurity and high spectrim efficiency. Considering that CDMA system belongs toDS-SS system, there’s MAI between signals of multiple users due to the random accessof users and zero correlation between spreading codes. As a result, system performancedeteriorates and system capacity is limited. Different from AWGN, MAI is, in somesense, predictable since it contains the mutual information between multiple users.Taking advantage of this feature, MUD can mitigate MAI. In this paper, algorithmresearch focus on MUD in DS-SS systems,and main contributions are listed as follows:(1) A synchronous signal model and an asynchronous signal model for DS-SSsystem are established. And then performance measure indicators and the principlesand features of classical MUD algorithm are introduced.(2) On the basis of analysis of the principles of the existing interferencecancelling detection algorithm, a modified user reliability based successiveinterference cancelling detection algorithm(RB-SIC) is intruduced and applied toCDMA system. In this detector, a novel user reliability measurement was establishedfor interference cancelling priority, and the conventional match filtering in SIC isreplaced by MMSE detection. The experimental simulation shows that RB-SICdetector outperforms the other detectors over Rayleigh fading channel. Even when thenumber of users K is greater than the spreading gain factor N, it has a better BERperformance.(3)With deep study of the principles and algorithms of subspace blind detection, ahigh-perfomance antinoise fast approximated power interations based modified blindMMSE detector（AFAPI-ModifiedMMSE） and a simplified fast approximated powerinterations based modified blind MMSE detector （SFAPI-ModifiedMMSE）areproposed in the light of modified FAPI algorithm and modified blind MMSEalgorithm.The experimental simulation shows that these two novel algorithms showbetter performance over subspace tracking error, output SINR and average BER.