Multi-user Detection Technique Based On Optimization Algorithm For MIMO-OFDM System

Multiple-Input-Multiple-Output (MIMO) system has received extensive attention, because it could exponentially improve the capacity of wireless channels. Orthogonal Frequency Division Multiplexing (OFDM), with its strong advantage of high spectrum efficiency and resistance to frequency-selective fading, has been favored by scientists. The combination of MIMO-OFDM is regarded as the core technology of next generation high-speed wireless LAN. Many users could communicate with other at the same frequency for MIMO-OFDM, inevitably interfering with each other. For reliable communication, Multi-user Detection (MUD) is used for signal separation and detection of individual user. In a number of multi-user detection methods, detection method based on optimization algorithms because of its extensive use and good performance has been concerned.This paper describes the optimal detection of NP hard problems equivalent to the optimization problem. Deeply researches and analyses about multi-user detection technologies of multiuser MIMO-OFDM system based on Genetic Algorithm (GA) and Repeated Weighted Boosting Search (RWBS) respectively are done in this paper. Main contents of this paper are as follows:1. Fundamental principles of MIMO-OFDM system in the implementation of space-time block coding (STBC) are analyzed and simulations of the system’s BER performance are made. Then several basic multiuser detection algorithms, such as: Zero Forcing (ZF) Detection, Minimum Mean Square Error (MMSE) Detection, Ordered Successive Interference Cancellation Based on Minimum Mean Square Error (MMSE-OSIC), Maximum Likelihood Detection (MLD) are simulated and analyzed.2. GA aided MMSE multi-user detection algorithm (MMSE-GA MUD) has been improved. At the receiver GA is used for joint optimization of multiuser detection and signal demodulation (MMSE-GA MJD), in which accurate estimates brought by signal demodulation are valued as the initial population of GA, which makes the simulation results more reliable. Simulation results show that compared with MMSE-GA MUD, the algorithm could make a better system performance without increasing the complexity.3. RWBS aided MMSE multi-user detection algorithm has been proposed. This method values the weights of the individual to be optimized as a guide to strengthen the good individuals’role in the evolutionary process, thus accelerating convergence rate of the optimal solution. It also values MMSE detection estimates as the initial population of the algorithm. Compared with GA, this algorithm could find the optimal solution faster and more accurate. Thus the computational complexity is further reduced. Simulation results show that under the similar conditions, the algorithm could achieve a better performance than MMSE, MMSE-OSIC, MMSE-GA MUD and MMSE-GA MJD.