Applying Fuzzy Logic to Codesign Partitioning

This article deals with the problems connected to the Hardware/Software Codesign partitioning phase. The novelties of our proposal lies in the approach to choosing between hardware and software implementation. To achieve this, the framework proposed benefits from the simultaneous use of Soft Computing, i.e. Fuzzy Logic, Genetic Algorithms and Neural Networks, which allow the performance of single modules to be evaluated without having to actually perform a low-level simulation of the synthesized device. Finally, we propose an algorithm to choose a good solution for partitioning based on a fuzzy resolver which uses the data provided by the previous analysis.