Regular-choice Petri Nets for MIMO detectors

摘要

Communications terminals in Multi-Input Multi-Output wireless communication systems have to maintain high quality signal transmission in highly dynamic operating environments and across channels of vastly differing quality. This places extreme pressure on model-based system synthesis approaches used to realise the embedded DSP processing within these devices, since the majority of the most effective techniques address static systems. In particular, these approaches suffer the key drawback that behavioral modelling languages for such systems are sorely lacking, resulting in inefficient implementations from current languages. We present a new Petri Net based language which resolves this situation for decoder-type algorithms in these systems, and show when applied to Sphere Decoder algorithms, specifically Fixed Complexity Sphere Decoders, it enables Texas Instruments TMS320C64+ DSP-based realisations with 90% reduction in run-time overhead and 10% reduction in code memory as compared to current languages.