An adaptive fuzzy logic-based routing algorithm for networks-on-chip

Abstract

In this paper, we propose an adaptive routing algorithm based on fuzzy logic in which each link cost is dynamically determined based on the current network condition. Using this algorithm, the traffic is distributed through the nodes that are less congested or have a spare capacity. The technique using a fuzzy controller takes advantage of two factors that are the number of empty spaces in the buffer of each neighbor and the waiting time for the previous packet. The output of the fuzzy controller is the link cost so that in each router, the link with the lowest cost is chosen as the optimal route. To evaluate the proposed routing method, we have used two multimedia applications and a random traffic profile. The experimental results show that the proposed routing scheme improves the performance up to 30% with a negligible hardware overhead.