Research On Algorithms For Near-Optimal Detection For V-BLAST Systems

Vertical Bell Laboratories layered space-time (V-BLAST) architecture is a promising multiple-input multiple-output (MIMO) system, which mainly focuses on multiplexing and can provide high speed data rate. Thus, it is an important technique in future broadband wireless communication system.The overall performance of V-BLAST system is crucially related to the detection algorithm at the receiver. As is well known, the maximum likelihood detection (MLD) yields the optimal detection performance. However, it suffers from an exponential complexity with the number of transmit antennas, making it infeasible for high-dimensional V-BLAST systems. Therefore, it is of great significance to develop low-complexity detection algorithms for near-optimal detection for V-BLAST system.This thesis mainly deals with three types of near-optimal detection algorithms: parallel detection (PD), fixed-complexity sphere decoder (FSD) and adaptive selection of surviving symbol replica candidates (ASESS) algorithm.Nevertheless, the existing PD algorithms have a common drawback in equally handling all the branches obtained via partial MLD (PMLD) till reaching the final minimum Euclidean distance (MED) decision. In the PD part, two simplified algorithms are developed from a recently proposed generalized parallel interference cancellation (GPIC) algorithm, by means of early termination of the inferior parallel branches and only dealing with error patterns with high probability. Then, another novel PD with partial decision feedback is developed. Simulation results demonstrate that all the three novel algorithms can approach the near-optimal performance with low complexity.In the FSD part, a simplified fixed-complexity sphere decoder (SFSD) with much lower computational cost than that of the original FSD is developed for signal detection in V-BLAST system, by pruning the inverted tree layer by layer. Simulation results on a 16-QAM system with 8 transmit and 8 receive antennas show that the SFSD can attain the near-optimal performance with a cost that is about 37% of the counterpart of the original FSD.In the ASESS part, an improved ASESS detector achieving much performance gain compared with the original ASESS detector at the cost of almost the same computational complexity is investigated. The superiority of the improved ASESS algorithm over the original one is because it always finds out the M optimum paths while the original one can pick out merely M fine paths at each level.