Providing a Routing Algorithm in Network on Chip to Reduce Energy Consumption and Increase Reliability with Fuzzy Neural Network and Genetic Programming

Network On-Chip (NOC) includes a set of techniques and tools in the field of network design and their application to design System On-Chip (SOC). The performance, type, and dimensions of the topology are largely dependent on the switching method and the routing algorithms used in it, including deterministic or adaptive. In this paper, using service scheduling techniques and genetic programming in order of entry, round robin of data on selected chip-based networks and using a defined neural fuzzy system, an almost adaptive routing algorithm for two-dimensional NOC is presented. It is able to increase efficiency, reduce power consumption and increase reliability. In the proposed algorithm, the average latency in different traffic load conditions faced by each packet is much less than similar algorithms. The results show a relatively good improvement in routing modification, which makes this method usable and even competitive with other existing and successful routing methods, so that when the nodes are fixed, about %6 life network has been added.