FastICA Based Blind Multiuser Detection Of DS-CDMA

Since the development of the third generation mobile communication technology,code division multiple access technology has become one of the main access methods of mobile communication technology.In CDMA systems,there are always four kinds of problems such as far-near effect problam,multipath interference,inter-symbol interference,and multiple access interference.Multiuser detection is the most effective method of solving these problems.However,effective multiuser detecton requires a lot of system parameters in advance,which is almost impossible to implement in the processing of downlink signal.With the in-depth study of blind source separation and independent component analysis by domestic and foreign scholars,some new methods used on the system of short-spreading code and direct-sequence code division multiple access downlink signal processing were provided.In this paper,we introduce a Newton iteration method with third-order convergence to improve the update process of Fast Fixed-Point ICA algorithm,and propose an improved FastICA algorithm.Based on this,we derive the improved noise FastICA algorithm.We apply the improved FastICA algorithm on DS-CDMA multi-user detection.And the accuracy of detection is improved.The specific work is as follows: Firstly,the basic concepts of BSS and ICA and their development and application issues are briefly introduced.The implementation principle and process of ICA algorithm and the relationship between ICA and BSS are discussed in detail.Then,the Newton iterative method for solving nonlinear equations proposed in recent years is briefly introduced and analyzed.Based on this,taking the computational complexity,convergence stability and sensitivity to outliers into comprehensive consideration,we introduce an improved Newton iteration method with third-order convergence.In view of the excellent convergence performance of the Newton iterative method,we substitute it into the update process of the traditional FastICA algorithm,and then provide the improved FastICA algorithm.The simulation results verify the convergence performance of the improved algorithm.Secondly,we introduce the principle of multi-user detection,the performance measurement of three commonly used evaluation and detection algorithms,the principle of DS-CDMA signal generation and the mathematical model of synchronization signal generation.According to the generation model of the received signal at the DS-CDMA receiving end,we establish the relationship between the ICA model and the mathematical model of DSCDMA synchronization signal generation,and provide a multi-user detection method based on ICA.Finally,we use the improved FastICA algorithm proposed in this paper to DS-CDMA multiuser detection.Simulation results show that the improved FastICA algorithm can effectively perform multi-user detection.Compared with traditional multi-user detection,the proposed algorithm has better detection effect.However,due to the uncertainty in ICA,the bit error rate generated during the detection process is highly volatile when the number of users is small.In the actual use environment,the number of users is large enough.And as the number of users increases,the volatility of the bit error rate gradually decreases.So this does not affect the practical application of the proposed algorithm in this paper for multiuser detection.