잡음증가현상을 최소화하는 Lattice Reduction 기반의 신호검출 알고리즘

Abstract

In Multiple Input Multiple Output (MIMO) detection systems, Lattice Reduction(LR)-based detectors show considerably good performance. Still, many researches are struggling to catch up the performance gap between Maximum Likelihood (ML) and LR-based detectors. In this paper we propose a new LR-based detection algorithm based on singular value decomposition (SVD). Noting that the least singular value of channel matrix is a major factor of noise enhancement, we perform the LR-based detection using the projection of received signals obtained with singular vectors corresponding to the other singular values but the least singular value. The key idea is to project the received signals in a space spanned by singular vectors that are orthogonal to the one corresponding to the least singular value. Through computer simulations, it has been found that the proposed technique provides at least 3dB improvement compared to the conventional LR-based ordered Successive Interference Cancellation (LR-SIC) detector for Bit Error Rate (BER) = 10-3 with a comparable computational load. Also, the detection performance of the proposed technique is almost the same as that of the ML detector for both correlated and uncorrelated channels.