Efficient Soft MIMO Detection by QR Decomposition of Channel Matrix

This paper presents a new soft detection scheme for multi-input multi-output (MIMO) system with turbo codes, which inherently requires soft input information from the MIMO detection part. The MIMO system operates for spatial multiplexing, in which an exhaustive search is the only option for a maximum likelihood (ML) soft detection. Although the maximum performance can be achieved using the ML soft detection scheme, the complexity increases exponentially with the number of antennas and modulation orders. In this paper, we first detect the signal in a symbol level with QR decomposition of the channel matrix, and then apply soft demapping scheme with proper channel gains. By this way, the complexity is greatly reduced, and combined with turbo code, the proposed method can achieve a close performance to that of ML detection.