Iteration Optimized Layered Tabu Search for large scale MIMO detection

Higher computational complexity in large multiple-input multiple-output (MIMO) systems motivates to development of an efficient algorithm for detection. Layered tabu search (LTS) has been considered a potential solution to achieve op-timum performance with lower computational cost. However, the complexity of LTS is dependent on the channel condition complexity is very large for lower SNR scenarios. Motivated by this, we propose a block-wise iteration allocation technique to reduce the excess number of iterations with almost the same detection performance. The proposed algorithm significantly reduces the required iterations at lower SNR, while the number remains unaltered at higher SNR. A training-based approach is considered to decide the number of iterations. As a case study, we have simulated 64 × 64, 32 × 32 and 16 × 16 MIMO systems for 64QAM and 16QAM modulations. The proposed optimization can achieve upto 40% of iteration reduction and upto 10% and 22% of complexity reduction for 64 × 64-64QAM and 32 × 32-64QAM systems, respectively.