Fuzzy neural-based adaptive deterministic routing algorithm for network-on-chip

摘要

Network-on-chip (NoC), a new packet based design method is an integrated solution to the challenging design in electronics, multimedia, and telecommunication domain. The method can be easily used in a conventional design flow. This paper is representing the architecture-level methodology for modelling and design of networks on chip using recent trends of technologies like fuzzy logic and neural network. Fuzzy provide the better platform to design of the routing algorithm by removing any ambiguities in the ouput port selection by using well defined rules. The Neural networks being faster in comparison with other traditional methodologies, are preferred for the design of parallel multiprocessor architectures. The study shows a comparison of these two approaches over conventional routing algorithm to see the effects of using such learning methods in the design of on chip router.