Soft Sphere Detection with Bounded Search for High-Throughput MIMO Receivers

We propose a soft sphere detection algorithm where search-bounds are determined based on the distribution of candidates found inside the sphere for different search levels. Detection accuracy of unbounded search is preserved while significant saving of memory space and reduction of latency is achieved. This probabilistic search algorithm provides significantly better frame-error rate performance than the soft K-best solution and has comparable performance and smaller computational complexity than the bounded depth-first search method. Techniques for efficient and flexible architecture design of soft sphere detectors are also presented. The estimated hardware cost is lower than the hardware cost of other soft sphere detectors from the literature, while high detection throughput per channel use is achieved.