The Research Of Multiuser Detection Based On Improved Particle Filter Algorithms

Multi-user detection in wireless mobile communications is not only an important anti-jamming technology, but also the key technology of code division multiple access(CDMA) system. According to CDMA, if spread codes of users are completely orthogonal,simple code match filters can be used to perfectly distinguish the data of multi-users.However, it is hard to realize the orthogonality of spread codes in the actual project. It seriously disturbed the estimated information of users that Cross-correlation between the spreading codes is not zero, which is the root cause of MAI(Multiple access interference).Multi-user detection depth analysis of the mechanism of the MAI to eliminate the cross-correlation for the purpose that it enhance the reliability of the system.First of all, this paper constructs the model of Direct Sequence Spectrum-Code Division Multiple Access(DS-CDMA) system, the MUD algorithm is integrated into this model.Using computer simulation, it in turn discussed classic MUD detection algorithm such as the best MUD algorithm, decorrelating MUD algorithm and MMSE MUD algorithm.Secondly, in order to easily analysis classical MUD algorithm on the theory, it is generally assumed that background noise is Gaussian noise that is well known. However,characteristics of Gaussian noise do not fully reflect the real characteristic of pulse noise. The presence of non-Gaussian noise led to the degradation of classic MUD performance.According to the defect of background noise in classical MUD algorithm, researching on the theoretical basis of the non-Gaussian noise, this paper selects strong practical significance Laplace noise and stable alpha noise as a prerequisite for studying the robustness of all kinds of MUD programs.Finally, in this paper, particle filter algorithm is introduced by theoretical analysis of several classical multi-user detection algorithm to make up the defect of single multi-user detection algorithm. Particle Filter(Particle Filter, PF) algorithm is applied to strengthen the signal estimation and adaptability in MUD. Then standard PF arithmetic and implementation steps are discussed on the level of mathematics. For decreasing of particle diversity,selecting of the importance function and a large amount of calculation, MUD takes five kindsof improve particle filter algorithm, which include important resampling particle filter,selection of importance weight particle filter, weight approaching particle filter, the improved un-scented particle filter and adaptive partial systematic resampling particle filter. The result of simulation show that improved particle filter algorithm can greatly enhance the performance of multi-user detection, and remains effectively detection in non-Gaussian noise.The System’s Performance of DS-CDMA that is a typical multiple access mobile communication is optimized by improving particle filter algorithm.