Research On Blind Multi-user Detection Algorithm Based On MIMO-OFDM System

With the rapid development of mobile communication technology, telecommunication industry is stepping into 4G field fast. MIMO-OFDM system can effectively combat multi-path effect and frequency selective fading in communication process by combining Multiple Input Multiple Output (MIMO) technology and Orthogonal Frequency Division Multiplex (OFDM) technology, multiply increasing system capacity and spectrum efficiency, therefore, it has become a hotspot in research of communication.With the increasing of users, the Multi-site interference (MAI) phenomenon is becoming more serious between multiple users in system. The blind multi-user detection (MUD) technology on Blind Source Separation (BSS) can provide an effective solution. Because the signal generated from MIMO-OFDM system is complex signal, so this thesis selects Complex Independent Component Analysis (CICA) algorithm of BSS for blind MUD, the main research works are as follows:Firstly, based on the introduction of the basic concepts and principles about MUD, this thesis converts blind multi-user detection to BSS problem in N subcarriers based on MIMO-OFDM system by analyzing MIMO-OFDM system model.Secondly, this thesis applies Complex FastICA (CFastICA) algorithm into blind multi-user detection in MIMO-OFDM system, N times CFastICA algorithm separations in the N subcarriers, analysing and revising the uncertainty of CFastICA algorithm by correlation function. The simulation results show that:compared with Complex Natural Gradient algorithm (CNGA), CFastICA algorithm has the advantages of fast convergence rate, low bit error rate and good stability.Thirdly, this thesis selects Complex Joint Approximative Diagonalization of Eigenmatrix (CJADE) for blind multi-user detection in MIMO-OFDM system, the method process signal with batch mode. It is different from adaptive signal processing way in CFastICA algorithm. The simulation results show that:the separation effect of CJADE and CFastICA algorithm is basically the same, both bit error rates are an order of magnitude lower than the CNGA algorithm. CJADE algorithm runs faster than CFastICA algorithm, but its programing complexity is relatively high.Finally, the summary and the prospect on this research are given.