Distributed Multipath Routing with Handling of Congestion in TCP/IP Networks using Neural Network

Abstract

Unlike traditional routing schemes that route all traffic along a single path, multipath routing strategies split the traffic among several paths in order to ease congestion. In this paper, one method is proposed to solve multipath routing problem in TCP/IP networks. A feed forward neural network is included at each node of network to make local decision by finding all neighbor nodes for the source node that reach to destination node, also it take in consideration handling of congestion problem. The proposed method is applied for typical examples of computer networks. The recurrent neural network is trained. Results of testing proof on its good performance.