The Research Of Blind Multi-user Detectionin MIMO-OFDM System

MIMO-OFDM system is widely regarded as a new generation of wirelesscommunication system, which consists of Multiple-Input Multiple-Output (MIMO)technology and the Orthogona l Frequency Division Multiplexing (OFDM) technique. Blindmultiuser detection technology is worth researching, which can effectively improve theutilization of spectrum resources.This paper firstly analyzes the model of MIMO-OFDM system and translates blindmultiuser detection of MIMO-OFDM system into N points BSS problem. Secondly, the paperdeeply research classic ICA algorithm, analyzing four different ICA independent standards -minimiza tion mutual information (MMI) criterion, information ma ximiza tion criterion,ma ximizing non-ga ussia n criterion, ma ximizing likelihood estimation (MLE) criteria.Finally,the paper study ICA method based on the natural gradient algorithm. it can be seenthrough simulation that the natural gradient algorithm has good BER performa nce tha n thetraditiona l MMSE algorithm, but the selection of learning rate has great influence onalgorithm convergence. In addition, FastICA algorithm is applied to MIMO-OFDM systemfor detecting the signa l of multiuser. The signa l detection algorithm theory analysis andsimulation results show that FastICA algorithm has a good BER performance tha n naturalgradient algorithm in equal SNR. Furthermore, the performa nce of algorithm is improvedobviously with more antennas. It is also show that FastICA algorithm has a faster convergencerate , sma ller steady-state value and reliability tha n natural gradient algorithm.