Optimization of List Sphere Decoding Through Parallelization and Feedback Control.

摘要

The demand for higher-performance MIMO systems continues to grow with the increasing demand for Gb/s wireless requirements. Sphere decoding continues to be the algorithm of choice for MIMO detection, because it sufficiently reduces the complexity of optimal detection. This paper further improves the efficiency of List-Combining Soft-Output Sphere Decoding by parallelizing list construction for large speedup and no reduction in BER. In addition to a large time reduction, parallelism allows for statistical probing techniques for use with feedback controls, which offers additional speed improvement and a negligible reduction in BER.