Broadcast scheduling in packet radio networks by Hopfield neural networks

Abstract

A multihop packet radio network is a packet switching network where nodes communicate by radio signals. When used in real-time multimedia or military communication networks, it is very important to broadcast a packet from a source node to all other nodes as quickly as possible. Unfortunately, the problem of finding an optimal schedule to minimize the time needed to complete the broadcast for a general multihop packet radio network is NP-complete. We propose a Hopfield neural network solution for one-source-node broadcasting in a multihop packet radio network. The proposed neural network model always finds a good solution by trying to maximize the number of correctly receiving nodes in every time slot. Experiments indicate good results. The model requires n processing elements (neurons) for an n-node radio network.