Rearch Of Blind Multiuser Detection Technique In CDMA Communication System

Code Division Multiple Access (CDMA) system has been widely applied in the third generation mobile communication system (3G) due to its soft capacity, soft switching, high spectrum efficiency and strong immunity to interference. In CDMA system, the signals of multiple users divide themselves from one another by depending on their different spread spectrum codes. But in actual applications, the spread spectrum codes are impossible to achieve the completely orthogonal state, therefore when multiple users correspond simultaneously, system will generate Multiple Access Interference (MAI). With the increasing of users, this kind of interference will be the main factor that influences the system performance and the system capability. How to beter eliminate multi-access interference has draw much attention recently years. Multiuser detection can effectively eliminate multi-access interference, improve capability, solve the problem of the Near-far effect which make it to be one of the key technologies of 3G. Its main purpose is to detect the interest user from multiple CDMA signals, and eliminate fundamentally the existence of multi-access interference. Blind multiuser detection which directly makes use of the interest userinfo to detect is a kind of blind adaptive multiuser detection. It has been a pop research subject in the field of communications recently.This paper takes blind multiuser detection as research object. In this paper, we first systematically introduce the basis principle and classification on the multiuser detection and some linear blind multiuser detection algorithms. On this basis, a new algorithm which is based on the orthonormal projection approximation and subspace tracking of deflation (OPASTd) algorithm is introduced to overcome the deficiencies of PASTd algorithm. The simulation results indicate that the algorithm is a better subspace tracking algorithm. According to the disadvantage that the convergence speed and precision of the LMS algorithm is greatly influenced by iteration step size, an improved LMS algorithm with its variable step size is introduced. The simulation results show that the algorithm is adaptive, and has higher signal interference noise ratio output. We also expatiate on the basic principle and mathematical model of independent component analysis (ICA), and mainly analyze several FastICA algorithms and their processes of the algorithms. In order to improve the convergence of FastICA algorithm, an improved FastICA algorithm based on the sixth order convergence of Newton's method is proposed. The simulation results are presented to demonstrate the superiority of the proposed algorithm. Because ICA algorithm and the improved FastICA algorithm ignore the noise so they will brought a bias. In order to remove the bias, we introduce the wavelet transform to the improved FastICA. A combined algorithm of wavelet transform and the improved FastICA (WICA) is raised, through the simulation results show that the algorithm has good performance. Through the discussion on the feasibility of independent component analysis algorithm in practical applications, we will make use of the WICA algorithm to optimize the blind multiuser detection. Through the two groups experimental results show that the WICA algorithm is more effective than FastICA algorithm. Especially when system has been lower SNR and more subscribers, the blind multiuser detection algorithm based on WICA has the higher detection performance.In conclusion, this paper has finished the following works:(1)On the basis of compression approximation subspace tracking algorithm, a new algorithm which is based on the orthonormal projection approximation and subspace tracking of deflation (OPASTd) algorithm is introduced.(2)On the basis of LMS Algorithm, we introduce an improved LMS algorithm with its variable step size.(3)We give an improved FastICA algorithm based on the sixth order Newton iteration format.(4)We will introduce the wavelet transform to the improved FastICA algorithm and give a blind multiuser detection algorithm based on WICA.