Determination of dependent and independent communication paths using neural networks.

Abstract

Efficient and timely computation of routing algorithms is very important for proper operation of multihop packet radio networks. When a network operates in the presence of jammers (a hostile environment) additional constraints must be considered when computing the routing tables. Insertion of those constraints in routing algorithms enables the determination of dependent and independent paths between source and destination nodes, thus fixing lower limits on the number of external jamming sources required to sever a transmission. An important measure of network reliability is also provided. This paper proposes a solution to the routing problem through the implementation of a Hopfield network and demonstrates that by proper selection of an energy function we can solve the problem rapidly and reliably.