An algorithm for multi-layer channel routing problem using chaotic neural networks

Abstract

In this paper a novel algorithm for the multi-layer channel routing problem in VLSI design using a chaotic neural network (chaotic NN) is proposed. For this problem, Funabiki and Takefuji (1992) proposed a parallel algorithm using the maximum neural network. However it is often caught in a local minimum because the maximum neural network is based on the Hopfield neural network. On the other hand, the chaotic NN has the characteristic of escaping from a local minimum. A novel algorithm using the chaotic NN is proposed. In order to confirm the effectiveness of the algorithm, numerical experiments are carried out, and it is confirmed experimentally that the proposal is more effective than the Funabiki and Takefuji algorithm.