Co-evolutionary reliability-oriented high-level synthesis

The main contribution of this paper is utilizing bio- inspired evolutionary algorithm for reliability oriented high level synthesis. In this paper genetic algorithm is used to schedule a data-flow graph considering latency and resource allocation considering resource constraints and area overhead. Then a co-evolutionary strategy merges the results of these solutions to find the RT level design of the circuit which satisfies both performance and area constraints. To satisfy the user- defined reliability, another genetic algorithm is developed to insert some hardware redundancies to the resulted data-path. Experimental results show using the proposed approach results in an acceptable reliability with a considerable area reduction in comparison with TMR.