On the optimization of lattice reduction-based approximate MAP detection using randomized sampling in MIMO systems

For iterative detection and decoding (IDD) in multiple-input multiple-output (MIMO) systems, the maximum a posteriori probability (MAP) detection is desirable to maximize the performance. Unfortunately, the MAP detection usually requires a prohibitively high computational complexity. In this paper, a lattice reduction (LR)-based MIMO detection method is proposed to achieve near MAP performance with reasonably low complexity in IDD, where the a priori information (API) is taken into account during list generation using randomized sampling to improve the performance. The sampling distribution is optimized to maximize the probability of sampling the MAP solution. It is shown that the proposed method outperforms conventional LR-based ones, where no API is considered during the list generation. Furthermore, a trade-off between performance and complexity is exploited with different list lengths.