A hybrid algorithm and its re-configurable architecture for MIMO detector

It has been demonstrated that MIMO detectors using sphere decoding (SD) and K-best tree search both can exhibit close to optimum error rate performance. Sphere decoding using depth first search has less average complexity than K-best tree search, but the complexity varies greatly among different symbols, thus creating problems for systems implementation requiring constant throughput. In this work, we propose a hybrid algorithm that uses both sphere decoding and K-best tree search to provide better performance with invariable throughput. The corresponding re-configurable architecture is also designed to implement the hybrid algorithm. From the experimental results, it is shown that a 25% enhancement in throughput is obtained compared with the previous block-based fixed throughput SD detector.