A 2D Sorter-Based K-Best Algorithm for High Order Modulation MIMO Systems

Although the K-best algorithm is well-known as an efficient suboptimal version of the maximum likelihood detection (MLD), its complexity is still significantly affected by the constellation size W. Especially, its sorting task is known as the major bottleneck. In this paper we propose a two dimensional (2D) sorter-based K-best algorithm, whose complexity is negligibly affected by W. Instead of considering all of the constellation nodes as the full K-best does, our algorithm specifies the best nodes directly and processes these nodes only. In addition, the notion of 2D sorter is introduced to simplify the sorting task. The paper shows that the complexity of our algorithm is 87 times less than that of the full K-best. It is also less complex than the conventional works. In terms of BER performance, our algorithm outperforms the BLAST MMSE, LRA-MMSE, and is close to the full K-best.