A near-ML sphere constraint stack detection algorithm with very low complexity in VBLAST systems

The stack algorithm is a promising tree-search algorithm with relatively low computation complexity for multi-input multi-output (MIMO) systems. Recent researches show that it obtains low detection complexity at the price of performance degradation. To achieve a better compromise between computational complexity and detection performance, a sphere constraint stack detection algorithm (SC-Stack) is proposed in this paper. With the aid of sorted QR decomposition based on the MMSE criterion (MMSE-SQRD), the proposed algorithm constrains conventional stack algorithm by a sphere radius obtained from partial serial interference cancellation (PSIC) algorithm. The SC-Stack algorithm avoids abundant metric computation by excluding a large number of nodes from the stack according to the sphere radius. The simulation results of computational complexity and detection performance presented in this paper show that the SC-Stack algorithm improves detection performance with lower complexity than the conventional stack algorithm. Moreover, the proposed algorithm achieves almost the same performance as sphere decoding algorithm while expanding far fewer nodes. So it is more feasible in practical systems.