A nature inspired optimized application mapping technique for On-Chip Networks

Abstract

Network-on-Chip (NoC) is a promising communication infrastructure that packs multiple cores into one chip for efficient data exchange. In such NoC architecture, application mapping is the process of assigning tasks to processing cores. Optimized application mapping technique improves chip performance and reduces overall chip power consumption. Optimizing application mapping is essential to NoC design. Mapping the task graph of the application onto a Network-on-Chip (NoC) Intellectual Property (IP) core is an NP-hard problem. The performance of the NoC network depends primarily on an effective and efficient mapping technique and optimization of performance and cost metrics. These metrics primarily include communication cost, energy, throughput, latency, power dissipation, and simulation time. A state-of-theart nature-inspired mapping technique for NoC's called Sparrow Search Algorithm (SSA) has been introduced in this work which has never been applied with NoC. The proposed algorithm minimizes NoC power consumption by a cognitive base utilizing shared K-nearest neighbor clustering method based on six standard available benchmarks including VOPD, MPEG-4, MWD, MP3enc MP3dec, 263enc MP3dec and 263dec MP3dec. The analysis of the experimental results demonstrate that the proposed technique outperforms as compared to other existing nature-inspired meta-heuristic application mapping approaches in terms of various performance metrics, such as energy consumption, communication cost, and average packet latency.