Study Of Multi-user Detection Algorithm To CDMA Wireless Communication Systems

With the application of computer techniques, especially digital signal processing technique, wireless communication technology is promoted. Code division multiple access (CDMA) is self-interfered system, where multi-access interference (MAI) and near-far effect existing as the main factors restrict CDMA system capacity. The multi-user detection (MUD) is proposed basing on the traditional detection techniques, which makes full use of information from multi-access interference caused by all user signals to detect the user's information. MUD can reduce multi-access interference and solve the near-far problem, so it can improve the system capacity remarkably. The blind adaptation MUD is an especially attractive research hotpot which requires less prior knowledge of signals.Firstly, signal subspace based blind adaptive multi-user detection algorithms are discussed. Through the simulations and studies on the subspace tracking module in OPAST algorithm, the divergence point of the subspace tracking error is found under the condition of different forgetting factors. Based on it an improved method is proposed to reduce the error accumulation in iterative procedure, with global convergence as its simulation result. Secondly, the LCLSCMA MUD algorithm is discussed combining with the subspace method. An improved algorithm is proposed, which improves the SINR and BER performance of the system under the low SNR and strong MAI condition comparing with the LCLSCMA and LSCMA algorithm. The simulation results show that the improved algorithm can eliminate the interference of subspace effectively and improve the performance of the detection. View of the algorithm complexity, the LCLSCMA_sub algorithm increases the part of the matrix eigenvalue decomposition comparing with LCLSCMA algorithm, but its matrix inverse can be greatly simplified because of the orthogonality of the eigenvectors. Lastly, the improved algorithm and antenna arrays are combined to be discussed, which result offers better output SINR and BER performance. It indicates that the improved algorithm is accurate and more efficient.