Research On Blind Multi-user Detection Algorithms Based On Independent Component Analysis

Code Division Multiple Access (CDMA) System is a developed spread spectrum wireless communication system and distributes mutually orthogonal code to different users for signal modulation in order to realize multiple users to access the system at the same time and frequency. Multi-user detection technology can reduce the Multi-access interference existing in the CDMA system, solve the Near-far problem and improve system capacity. It has become one of the key techniques for3G. For blind multi-user detection technology do not require sending training sequences and interfering user information, it has good prospects for development with low complexity, less system requirements etc.This thesis focuses on studying in blind multi-user detection based on independent component analysis (ICA). It summarizes the previous work, and proposes some new improved algorithms. Major work and innovation are as follows:Firstly, a survey on multi-user detection methods is given, and then the basic theory of the spread spectrum communication and ICA and its application in the field of multi-user detection are introduced.Secondly, comparing the DS-CDMA and ICA signal models, ICA can be applied to blind multi-user detection of DS-CDMA system. A Noise-ICA algorithm is proposed for multi-user detection, meanwhile overcome the weaknesses of the traditional ICA algorithm ignoring noise. The simulation results show that the Noise-ICA algorithm has better detection performance in the number of users increase and near-far effect serious cases.Thirdly, this thesis introduces KICA algorithm which is used to solve the current popularity nonlinear mixed problem, and brings in a new combination of kernel function. Then combined with Noise-ICA algorithm, an improved-KICA blind multi-user detection algorithm is proposed and compared with the traditional KICA algorithm. The simulation results show that the improved-KICA blind multi-user detection algorithm is good at separation performance.Finally, the summary and the prospect on this research are given.