The Research On MIMO Detection Algorithm

Multiple-Input Multiple-Output (MIMO) system, as one of the hot technologies for the third-generation (3G), is attracted considerable attention. However, the optimal maximum-likelihood (ML) algorithm suffers from exponential complexity in the terms of the number of transmit antennas and signal modulation. Therefore, the paper studies the MIMO detection algorithms with low complexity, and the main work includes:First, the paper studies some conventional MIMO detection algorithms for the V-BLAST systems and gives simulation results. These include maximum-likelihood detection, null and cancellation, linear detection and semi-definition relaxation detection, then some meaningful conclusions are drawn. These work provide a good basis for follow-up study.Second, we propose a novel MIMO detection algorithm. Using the novel algorithm, part of globally optimal solution for ML detector can be decided with low computational complexity, later, we can substitute them into the original MIMO detection problem and obtain a smaller-scale MIMO detection problem for the undecided information sequences. Hence, we can use some conventional sub-optimal detectors to solve the smaller-scale problem so that the novel algorithm can achieve better performance than that of conventional ones with lower computational complexity. In addition, we make a probability analysis for the symbols which satisfy optimal decision criteria.Finally, we do preliminary research and apply the optimality judgment criterion to the initial vector of LAS (Likelihood Ascent Search) algorithm. Through statistical analysis, the estimated soft-values using optimality judgment criterion is used as the initial vector of LAS algorithm. Then we propose a LAS algorithm based on soft-values. Simulation results show the novel algorithm can obtain a best tradeoff between error performance and computational complexity.