Study Of Blind Multi-user Detection For DS-CDMA Systems

Multi-user detection(MUD) is one of the key techniques for 3G systems. It can supress multiaccess interference, mitigate the near-far effect and improve the system capacity. While the blind MUD requires only the information of the desired users, it has attracted considerable attention and interests.Firstly, blind MUD based on signal subspace and reduced-rank subspace are analysed. The performance of subspace method is improved by projecting the received signals to the signal subspace. And it can be executed by subspace tracking algorithms with low computational complexity. The energy of the desired user may lie on a lower dimensional subspace, so the detection can be constrained on a reduced-rank subspace. Simulations have approved that the reduced-rank detection achieves better performance.Secondly, further analysis have been made to group blind multi-user detection. The MUD based on signal subspace and reduced-rank signal subspace are both blind algorithms. When the receiver in the base station is taken into consideration, the information of some interferers is already known additionally. It is a topic that how to deal with the intercell interferences using the already known information. The group blind MUD is discussed to deal with the interferences of the intergroup users. And simulation results have proved the effectiveness of group blind MUD.Thirdly, the multiuser detection based on reduced-rank subspace and least square constant modulus algorithm is proposed. Simulation shows that there is a reduced-rank subspace existing, on which the LSCMA performs superiorly.Finally, the multipath combining based on the subspace detection is proposed to deal with the multipath fading, with the assumption that all paths have chip based delay. The simulation indicates that this method defends the multipath fading effectively.