A fast convergence LLL algorithm with fixed-complexity for SIC-based MIMO detection

In this paper, we propose a fixed-complexity variant of LLL (Lenstra-Lenstra-Lovasz) algorithm. LLL algorithm is widely used in MIMO signal processing to obtain full diversity gain with low complexity increase. However, because of its non-deterministic nature with varying complexity and high worst-case costs, the real implementation of original LLL algorithm is difficult. Although some fixed-complexity variants of LLL algorithm has been proposed, but still their complexity in large MIMO system is high. The proposed algorithm uses column selection method based on threshold which leads to the fast convergence in fewer iterations. Simulation result shows that the proposed algorithm converges faster than other fixed-complexity variants of LLL algorithms while it saves about 30% complexity in 8 × 8 MIMO system compared to existing fixed-complexity LLL (fc-LLL) algorithm.