(An) efficient tree-traversal algorithm for soft-output MIMO detection in high-performance wireless communications

Abstract

The multiple-input multiple-output (MIMO) technology has been employed in wireless communication systems as a promising technique to increase data rate more than hundreds of megabits or several gigabits per second. To approach the channel capacity, in addition, recent communication standards mandate iterative decoding which requires the log-likelihood ratios of all the transmitted bits. As a result, the efficient computation of such values has drawn considerable attention, and the importance of soft-output MIMO detection has been enlarged. In this thesis, an efficient tree-traversal strategy is proposed to reduce the computational complexity of soft-output MIMO symbol detection. To minimize unnecessary computations, the proposed algorithm never visits a node more than once. Meanwhile, it repetitively reorganizes the set of nodes which are likely to be the maximum-likelihood solution or counter-hypotheses that need to be identified in the detection. As the reorganization directs the algorithm to search only meaningful nodes for the required symbols, the complexity can be effectively mitigated while preserving the max-log optimality. Experimental results claim that the proposed algorithm reduces the number of visited nodes by 29% and 36% over the single tree search for a MIMO system equipped with 4 $\times$ 4 16-QAM and 4 $\times$ 4 64-QAM, respectively.