In the released standards of many mobile communication systems, multiple-input multiple-output (MIMO) technology has already been applied. Therotical work has shown that in sufficient rich scattering environments, MIMO systems hold the potential of enhancing spectral efficiency-hence capacity comparing with single input single output (SISO) systems. But due to the high complexity of receiver, the practical MIMO schemes are still very simple with a few antennas. To promote the application of more research achievements about MIMO, the complexity of detection algorithm must be cut down firstly. The subject of this dissertation is the investigation of the detection algorithms in MIMO systems with multiplexing scheme, and the objective is just to improve the detection performance and to reduce the computation complexity.The first part of this thesis deals with some issues of the traditional linear detection. We firstly derive the probability density function of the output SNR for MMSE detection, which is used to analyze the performance and capacity after the different detections. For the performance of ZF and MMSE at high SNR, we give an explanation for the contradiction between the intuition of their same performance from the convergent weighting matrix and the two parallel curves in the simulation. Besides, we give another definition of the diversith order, i.e. the slope of the cumulative distribution function of the output SNR.The detection based on sphere decoding (SD) is investigated in part two. One is SD with hard decision. After summarizing the methods used in SD, we propose an optimal SD detector with square bound, which complexity is proved to be lower than that of the traditional V-BLAST detection at high SNR for a large antenna configuration. Then, we firstly utilize the condition number of the matrix to describe the channel's characteristic and analyze it's influence on detection. Aiming at the problem of high complexity in the channel with a large condition number, an idea of combined-metric at high layer is developed for SD. In addition, a method named split-tree SD, is presented for providing the tradeoff between performance and complexity. It's application in the receiver for MIMO-CDM-OFDMA system, shows the best performance with the lowest complexity, comparing with other receivers which detect the signals of multiple code channel and multiple antennas respectively. The other is SD with soft decision for the coded MIMO system. We give a... |