Near-ML Lattice Reduction-aided Detection Scheme for Low Complexity MIMO-OFDM Systems

摘要

In this paper, we propose a novel lattice reduction (LR) algorithm for the low-complexity multiple-input multiple-output (MIMO) detection with near-maximum-likelihood (ML) performance. The proposed LR algorithm is designed considering both the hardware complexity and the power consumption. First, a modified column traverse strategy is proposed to reduce the worst-case complexity (hardware complexity). Also, in order to reduce the average complexity (power consumption), we focus on the joint optimization by employing the early termination (ET) criterion in the context of MIMO detection, whereas the conventional approaches are based exclusively on channel characteristics. In order to make it possible for the LR-aided fixed-complexity sphere detector (FSD) to perform the partial detection, the LR process is thoroughly modified so that the ET criterion is able to be employed. Furthermore, we perform the joint optimization of these two approaches. The experimental results demonstrate that the worst-case and average complexity is reduced considerably maintaining the near-ML BER performance at the BER of 10~(-5).