A Parallel Smart Candidate Adding Algorithm for Soft-Output MIMO Detection

Abstract:
Tree search schemes are an efficient means for solving the detection problem in multiple-input multiple-output systems. One option for implementing such a tree search is the smart candidate adding approach: using first an unconstrained search for the MAP estimate and then multiple constrained searches for counter-hypotheses. An inherent disadvantage of this strategy is that tree nodes might be visited multiple times, resulting in high detection complexity. This paper presents a parallel smart candidate adding algorithm, where the initial and the constrained searches are combined into a single detection step. Thus, complexity can be significantly reduced compared to previous proposals, at the cost of a minor performance loss. Furthermore, the algorithm has fixed detection complexity as opposed to most prior smart candidate adding proposals.