A soft tree pruning based fixed-complexity sphere decoder for interference-limited MIMO systems

In wireless networks, interference from adjacent base stations is usually a dominant factor for user performance degradation. A maximum likelihood (ML) detection can provide superior performance by jointly detecting the serving and interfering signals. However, the complexity is extremely high. This paper presents a soft tree pruning based fixed-complexity sphere decoder. It uses reliability information on bits to estimate a search radius and thereby prunes unpromising nodes in the early stage of the tree search detection to reduce complexity. Based on the soft tree pruning method, the fixed-complexity sphere decoder can still work in a fully pipelined mode. Simulation results show that the soft tree pruning based fixed-complexity sphere decoder can provide a better trade-off between complexity and performance in interference-limited scenarios.