The Improved Maximum-Likelihood Detection Algorithm for MIMO Systems Based on Differential Metrics

Abstract

The MIMO detection algorithm is the key for design of MIMO receiver. The MLD-DM can achieve the optimal BER performance and does not need QR decomposition and matrix inversion. However, the initial sequence is constant vector, which will cause the big space of tree search. Based on this problem, we propose an improved MLD-DM algorithm, where the initial sequence with many bits will be adjusted in the light of their differential metrics of order one and then is more similar to the maximum-likelihood sequence. After calculating all differential metrics of order one, we will change the bit with the biggest metric value and then use recursive function to update the differential metrics of all bits. Compared to MLD-DM, our improved algorithm effectively reduce the complexity without any performance loss. The simulation verify our proposal.