A hybrid method to reduce the complexity of k-best sphere decoding algorithm

Abstract

One effective way to reduce the complexity of receiver is the utilization of sphere decoding algorithm. Sphere decoding algorithms, particularly sphere decoding algorithms which search in the depth first known as depth first sphere decoding algorithm (DF SDA), have a similar performance to maximum likelihood with almost an acceptable complexity. However, depth first sphere decoding algorithm has a variable complexity. K-best sphere decoding algorithms have a fixed bitrate and complexity but their performances degrade compared to maximum likelihood detector (MLD). Therefore, we proposed a hybrid sphere decoding algorithm which contains the desired features of both DF SD and k-best sphere decoding in this paper. In order to reduce the complexity of DF SD, we proposed the utilization of the initial radius setting and radius updating strategy. Using this approach, we are able to reduce the computational complexity of sphere decoding remarkably. Simulation results confirm that the performance of our proposed method is at least equal to the performance of k-best sphere decoding algorithm but with a lower computational complexity.