A Study On Blind Multiuser Detection Algorithms For DS-CDMA System

Multiuser detection has become one of the key techniques of CDMA, which can effectively supress multi-access interference(MAI) and improve the system capacity. Blind multiuser detection (?)as become a research focus for its less need of prior knowledge.Firstly, this paper foucs on several key technologies in comunication system. And then, system model of DS-CDMA and the basic principle of multiuser detection was introduced. Several important performance index of multiuser detection is discussed respectively, which provide the important evaluation standard for theoretical study later.Secondly, several classical algorithms in multiuser detection, such as optimum multiuser detection, decorrelating multiuser detection, MMSE multiuser detection etc, are theoretically analyzed and simulated. The advantages and disadvantages of the algorithms mentioned above were analyzed according to the simulation results, which provide a research direction.Thirdly, we study and simulate several constant modulus algorithm such as basic constant modulus algorithm(CMA), least square constant modulus algorithm (LSCMA), linearly constrained least square constant modulus algorithm(LCLSCMA). Then, minor component analysis constant modulus algorithm(MCACMA) is proposed, and optimized weight vector corresponding to the minimum of Rayleigh entropy is derived. Simulation results show that the proposed MCACMA algorithm has better performance in signal to interference plus noise ratio(SINR) and bit error rate(BER) than LSCMA and LCLSCMA. Furthermore, this proposed algorithm has more extensive application prospect because it effectively reduces the computation complexity by avoiding the matrix inversion calculation.Finally, several subspace blind multiuser detection algorithms were discussed, performance of those algorithms on multiuser detection was simulated and analyzed. On this basis, we present a blind multiuser detection approach based on subspace minor component analysis constant modulus blind multiuser detection algorithm(SUB-MCACMA), which can effectively eliminate the effect of noise subspace. Simulation results show that, compared with LCLSCMA and MCACMA, the proposed SUB-MCACMA algorithm can improve the performance on SINR and BER dramatically, is robust to MAI and has good detection performance.