Computationally efficient near-optimal combined antenna selection algorithms for V-BLAST systems

Abstract

The Vertical Bell Laboratories Layer Space-Time (V-BLAST) system using ordered successive interference cancellation (OSIC) detection is a popular system architecture featuring high data rate and moderate computational complexity. The performance of the V-BLAST system can be further improved by exploiting the diversity and closed-loop gain provided by the technique of combined antenna selection (CAS). In this paper, we investigate the design of CAS algorithms with a goal to minimize the system@?s error rate. As the error rate of a system varies dramatically with the transceiver architecture, the proposed CAS algorithms explicitly take the VBLAST detection structure into account. Despite the optimal CAS can be achieved by exhaustively searching all the transmit and receive antenna subsets, the applicability of this brute-force approach is excessively limited due to its huge computational complexity. To this end, we propose two computationally efficient CAS algorithms by implementing a greedy strategy incrementally or decrementally to fully exploit the recursive structure of mean-square-error matrices between consecutive V-BLAST detection stages. Near optimal error rate performance can therefore be obtained with significantly lower complexity as verified by the extensive computer simulations.