Algorithm of neuron-reduced Hopfield neural network routing in communication networks

A feedback neural network is utilized in communication networks to solve the routing problem. In order to overcome the shortcoming of local stability when using Hopfield neural network, an algorithm of simulated annealing is introduced to the make the Hopfield net iterate to the global optimal state. However, the traditional Hopfield net is a large scale net, which needs a large quantity of neurons and consumes much time for calculating. So we propose a neuron^_ reduced Hopfield neural network routing algorithm that has fewer neurons in order to speed-up the routing selection. We compare this new routing algorithm with the traditional one in various aspects of iteration, operating speed and stable network-energy when the communication network´s and neural net´s parameters change, and summarize the relationship of three restraint coefficients. The neuron-reduced Hopfield neural network routing algorithm has better validity, reliability and application adaptability compared with the HNNR algorithm.