GSASRA: A Globally Self-Adaptive and Scalable Routing Algorithm for Network-on-Chips Architecture Using Particle Swarm Optimization

Adaptive algorithms are able to intelligently adjust their behavior in light of the changing situation to achieve the best promising results. The adaptive routing algorithms have been employed in multi-chip interconnection networks in order to get better network performance. In this paper we propose a new Globally Self-adaptive and Scalable Routing Algorithm for Network-on-Chip (NoC) architectures namely (GSASRA). Our algorithm makes global routing decisions using particle swarm optimization technique to find global congestion information and to efficiently and intelligently decide the path to route the packet in each direction from source to the destination nodes. We implemented our proposed approach using SystemC and compared our approach with the SystemC based Fully-Adaptive and XY routing algorithms under matrix-transpose traffic pattern. Results from our experiment and simulations show that, our approach proves to be efficient in terms of throughput, latency and energy consumption for self-adaptive and scalable network-on-chip architectures.