Critical analysis of hopfield's neural network model and heuristic algorithm for shortest path computation for routing in computer networks

Abstract

Shortest path routing and its computation is a crucial point in computer networks, and has significant impact on overall network's performance. Being an issue of salient importance, many algorithms were proposed for shortest path computation and are still under research for more enhancements. Hopfield proposed a neural network based architecture for such optimization problems. Mehmet and Park Keum suggested improved energy functions for this neural network to implement it for routing in computer networks. A* search algorithm is a heuristic based approach, with the properties of Dijikstra algorithm and is used for same purpose. Performances of both approaches are compared and results are analyzed.