A Self-Adaptive Mapping Approach for Network on Chip With Low Power Consumption

Abstract

Application mapping of disseminated intellectual property into Network on Chip (NoC) is a well-defined NP-Hard problem. Improvement of network performance in NoC is purely based on an effective mapping approach with cost and performance metrics optimization which includes area, power, delay, reliability, and thermal distribution. A self-adaptive mapping approach for NoC is proposed in this paper. In this method, the self-adaptive chicken swarm optimization algorithm (SCSO) is used for an effective mapping, which has never been applied with NoC. The proposed method reduces the power consumption of NoC through a cognitive base using shared K-nearest neighbor clustering method and it offers faster mapping over standard and randomly generated benchmarks. The experimental results indicate that the proposed method outperforms existing bio-inspired metaheuristic algorithms, especially for large application graph.

Keywords

• Network on chip
• self-adaptive chicken swarm optimization
• shared K-nearest neighbor clustering
• bio-inspired metaheuristic algorithm