ROUTING OF COMMUNICATION NETWORKS WITH RELIABLE & UNRELIABLE COMPONENTS USING ANN

Abstract

Inspired by the human neural system, several artificial neural network models have been studied. These models are composed of many non linear computational elements operating in parallel. This massive parallelism makes- these models an ideal tool for computational work in constrained optimization problems. The dissertation is aimed at how the neural-network models are effective for the routing of communication network with reliable and unreliable components- and comparing the convergence of control vectors used in these models toward optimal solution. Initially the minimization process is implemented using a modified version of neural network TRAVELLING SALESMAN PROBLEM (TSP) Algorithm considering all nodes and links reliable i.e. without failure (Rauch-Winarske model). A modified version of the above model, with the capability to handle dependent node failure has also been considered (Su-Ling & Shyang model). Su-ling & Shyang model does not consider the link failure therefore a mathematical model for link failure has to be incorporated in it. Three cases has been taken i.e. network with link and node both perfect, network with node dependency and reliability and the network with both link failure and dependent node failure. The speed of convergence and the effectiveness of the above models for the three cases have been compared.