Genetic Algorithm- and Cuckoo Search Algorithm-Based Routing Optimizations in Network-on-Chip

On-chip networks have come to the fore as a reliable solution to counter the problems of flexibility and scalability in many-core architectures, which persist in earlier generations of SoCs. One of the major ingredients in this field is routing data packets through the network of IP cores. The state-of-the-art routing mechanisms deal with a high trade-off between two principal performance parameters, viz. throughput and latency, by optimizing one parameter and becoming a performance bottleneck for the other. This research work is an endeavor to propose an optimized all-round routing algorithm to overcome the aforementioned problem. We have introduced two nature-inspired routing mechanisms based on genetic algorithms and cuckoo search algorithms, respectively, to provide the optimality over the state-of-the-art XY algorithm under diverse traffic scenarios. Simulations done for evaluating the performance of proposed schemes validate a significant improvement in the network throughput while incurring a lesser network delay over the XY routing scheme.