Efficient Ordering Schemes for Sphere Decoder

In this paper, we propose an efficient ordering scheme that can be embedded in the QR decomposition to reduce the average and worst-case computational complexity of the sphere decoding (SD) algorithm. Our proposed ordering scheme takes into account both the channel matrix and the received signal (noise), and is designed based on a study of the data collected from worst-case scenarios. Simulation results show that our proposed ordering scheme leads to lower average and worst-case computational complexity for the SD algorithm than the optimal vertical Bell Labs layered space-time (V-BLAST) ordering scheme. We also propose a simplified ordering scheme when the unconstrained zero-forcing (ZF) estimate is unavailable.