IMPROVED ALGORITHM FOR THE PACKET ROUTING IN TELECOMMUNICATION NETWORKS

Abstract

Packet routing efficiency is and will remain one of the major factors that have a major impact on a network performance and a quality of service. One of the most common routing protocols used in autonomous Internet systems is the Open Shortest Path First (OSPF) protocol. The OSPF protocol performs packet transmission using the shortest paths, calculated according to a certain, but only one criterion. Default criterion is channels bandwidth, but in general any criteria can be selected. The disadvantage here is that channels loading is not considered and thus packets are routed to a shortest but potentially overloaded path. In wider sense usage of single criteria may results in routing through paths suboptimal or even inappropriate by other criteria. The growth of the network traffic, and especially the increase of its multimedia component drives continuous improvement of routing methods in existing telecommunication networks. Network performance will benefit from more advanced routing methods, which would take into account several network characteristics, important for specific services quality. In given work a new algorithm of a gateway selection for the routing tables of autonomous Internet systems is proposed. The algorithm is based on the OSPF protocol but takes into account three criteria: channel bandwidth, the number of intermediate nodes in the path and the signal propagation delay. Weighting coefficient for each criteria is introduced. Besides that, the integral criterion is proposed based on optimal path weights for individual criteria and path weights found by two other criteria. The total relative deviation of the path weight from the path weights calculated according to individual criteria is used as the optimization criterion. The purpose of optimization is to minimize this deviation. Routing tables constructed using proposed algorithm ensure packets transmission through the optimal path, selected according to the integral criterion, taking into account the three criteria and their coefficients of importance. Further research may be conducted to expand the list of criteria used in the optimization process.