Hybrid Techniques for Fast and Low Complexity MIMO Detection

Sphere Decoding (SD) can minimize the complexity of Maximum Likelihood Decoding (MLD) without loss of performance, but its complexity and execution time are still high for applying it into practical Multiple-Input Multiple-Output (MIMO) systems. In this paper, a new technique for MIMO signal detection using a hybrid combination between SD and linear detectors is proposed. The proposed technique is intended to reduce the complexity and execution time of the detection process. The main idea is based on splitting the vector of transmitted symbols and the channel matrix into two equal size subsets, where the received symbols detection process will be performed in two separate successive stages. It is proposed to use the SD for the first stage for accurate estimation of the first subset of symbols, which has a critical influence on the next stage performance. Low complexity detectors such as Zero Forcing (ZF) and Minimum Mean Square Error (MMSE) are utilized to estimate the second subset of symbols in order to reduce the complexity. In this case, the SD search in the whole lattice is avoided. Only a half search burden is required which will lead to a significant saving in the computational complexity and time.