Memory-constrained ML-optimal tree search detection

In this paper, we propose a memory-constrained tree search (MCTS) algorithm for the detection in multiple-input multiple-output (MIMO) systems. The MCTS algorithm offers a wide range of trade-offs between computational complexity and memory requirement, and is guaranteed to achieve the exact maximum-likelihood performance. By tuning the memory size, the MCTS algorithm ranges from being memory-efficient to being computation-efficient with abundant choices in between.We show that the MCTS algorithm visits slightly fewer nodes and requires slightly less memory than the sphere decoding (SD) algorithm in the memory-efficient case, and visits similar number of nodes and requires significantly less memory than the stack algorithm in the computation-efficient case.