Research On The Technology Of CDMA Multi-user Detection Based On Independent Component Analysis Algorithm

In the CDMA communication system, the various user signals are usually independent each other, and there is some relevance between the users’ signals, which is the source of the multiple-access interference (MAI) exists. The traditional way of detection is to test multi-access(MAI) as noise to deal with, which can not fully use multi-user information, thereby reducing communication quality and system capacity; but The multi-user detection take the MAI as an effective information processing, and it can overcome the distance-effect (Near-Far Effect) effectively, and increase the system capacity. However, the frequent sending training sequences which constrain the real-time multi-user detection technology’s real-time and practicality. Digitally modulated signals typically meet with the distribution of the independent non-Gaussian signals. Therefore independent component analysis can be applied to blind multi-users detection of CDMA communication, and can using the feature of the various users which are independent in CDMA systems, then it can eliminate the interference of multiple access and ease the distance-effect effect effectively.Firstly, this paper introduces the principle of multi-user detection, which includes the principle of CDMA system、the disturbance type of system、the main measure of overcoming the multiple access interference、the classification of multi-user detection and its performance measurement indicator. Secondly this paper give the basic theory of independent component analysis(ICA) and the typical algorithms, the typical ICA algorithms mainly include the maximum entropy、mutual information minimization algorithm and Fast-ICA algorithm.This paper compares the relationship between the ICA model and the CDMA model; the simulation shows that the Fast-ICA algorithm can be applied to CDMA multi-user detection. On this basic theory, this paper proposes Fast-ICA adaptive MMSE multi-user detection algorithm, combines the independent component analysis and the existing adaptive MMSE detector, and make the independent component analysis as post-processing of MMSE detector to improve the system performance. The simulation shows that the performance of improved algorithm is better than traditional adaptive MMSE multi-user detection.