Regularized sphere decoding techniques for data transmission systems

Abstract

Recently, the idea of regularization received a lot of interest in signal processing with ill-conditioned problems. In this paper, instead of using the traditional sphere decoding (LSD) technique, regularized LSD detection techniques have been used for block data transmission systems (BDTS). The first regularization scheme that has been used is Tikhonov regularization (L2-regularization) which is a summation of consistent norms while the other is L1-regularization which is a summation of mixed norms. Regularization shows a significant impact on the condition number. In performance analysis, simulation results clearly indicate the advantage of the regularized LSD techniques for BDTS over others. LSD technique using L2-regularization performance is very near to optimum performance. On the other hand, it has been shown that L2-regularization provides better performance than L1-regularization.