Signal Detection Based On Sphere Decoding In MIMO System

Without additional bandwidth and extra power, Multiple Input Multiple Output(MIMO) communication system can increase the channel capacity and spectrum efficiencysignificantly, which makes it becomes a hot spot in the field of wireless communicationresearch. However, it use a plurality of antennas simultaneously transmit signal, whichincreases the difficulty of the receiving signal detection. Since the quality of signaldetection is directly related to the performance of the system, the signal detectiontechnology is a focus of research in this field.The MIMO communication system model has been described in chapter2. And thesame time, the diversity gain and multiplexing gain of the MIMO system also has beenanalyzed and the simulation. The space-time coding technology and analyze the system ofchannel capacity has briefly presented. In chapter3it mainly analyzed the detectionalgorithm in MIMO communication system, and discussed the maximum likelihooddetection algorithm, zero forcing detection algorithm, the minimum mean square errordetection algorithm, serial interference cancellation detection algorithm, QR decompositiondetection algorithm. Sphere decoding (SD) detection algorithm is also discussed in detail.In the paper，we do research from the following two aspects in SD detection algorithm:The computational complexity of SD varies with the initial search radius. To solve theproblem, a novel SD based on Ant Colony Optimization (ACO) has been presented in thispaper. Firstly, we get a new radius with taking advantage of the ant colony algorithm,which has a global search ability of the self-organization, then we use SD detectionalgorithm with a new radius. The theoretical analysis and experimental results show thatthe new algorithm has a lower complexity than the previous SD detection algorithm in thelow signal to noise ratio, and they can be approximation ML detection performance effectin the high signal to noise ratio.Ill-conditioned channel matrix can effect decision domain in multi-dimensionalconstellation, and make it more irregular distortions. To solve the problem, a novel SDdetection algorithm based on lattice reduction has been proposed. The channel matrixwould be closer orthogonal when it has been processed by the lattice reduction algorithm.The nearly orthogonal channel matrix has a relatively good signal judgment domain.Theoretical analysis and simulation results show that compared with the original algorithm,the novel algorithm can acquire a more approximate ML detection performance with increase a slight complexity, which confirmed the feasibility of the proposed algorithm.