Sorted QR Decomposition Based Tree Search Detection Algorithm for MIMO System

QR decomposition with M-algorithm (QRD-M) is a breadth-first tree search algorithm, which has been proposed to provide near Maximum likelihood (ML) detection performance with lower computation complexity in Multiple-Input Multiple-Output (MIMO) system. To accomplish near-ML detection performance in QRD-M detection, M should be set at a large value, and the increasing of parameter M will lead to the increasing of complexity. In order to trade off the detection performance and calculating complexity, an improved detection algorithm is presented here. Householder transformation is performed for sorted QR decomposition of channel matrix firstly, and ML algorithm is used to detect the bottom several layers, then the partial accumulated metrics are calculated and sorted to produce an ordered set, QRD-M algorithm is performed serial to each branch of the ordered set with terminal condition, at last. The proposed detection algorithm can get better trade-off between performance and complexity. Simulation results show the validity of proposed algorithm.