Research On Blind Multi-user Detection Algorithms In DS-CDMA Communication Systems

The capacity and performance of a code division multiple access (CDMA) system is essentially dominated by multiple access interference (MAI). The well cancellation of MAI that degrades the capacity and performance could depend on multiuser detection technology. In recently years, the main research work may focus on how to present a high performance and low complexity algorithm. Blind multiuser detector has become an important study point as it has no needs of training sequence and the prior knowledge of interferential users.This paper has studied the blind multiuser detection based on subspace, and PASTd algorithm and OPAST algorithm .Both of the two algorithms have the same flaw, i.e.divergence. The divergence of PASTd algorithm is because of the eigenvectors estimated with weak orthogonality. This paper has adopted Gram-Schmidt method to orthogonalize the eigenvector set. The divergence of OPAST algorithm is because of the error of the actual computation, which make the positive definte matrix no positive definte. According to the character of the positive definte matrix, this paper has improved it. Simulation results show that the two improved algorithms are of great convergence and accurate tracking performance.And this paper also has studied the blind multiuser detection based Kalman filter, and compared with the two improved algorithms, the MOE algorithm and CMA algorithm by computer simulating. The simulation results show that the Kalman filter algorithm has a good capability of convergence and tracking, also the simulation results show that it has strong capability of suppressing MAI.