Decentralized optimal routing for packets flow on data networks

Abstract

This paper concerns optimization of data traffic flows on atelecommunication network, modelled using a fluid - dynamicapproach. Flows can be controlled adjusting traffic distribution andpriority parameters. Two cost functionals are considered, whichmeasure average velocity and average travelling time of packets.   First we address general optimal control problems, showing thatexistence of solutions is related to properties of packet lossprobability functions.   A direct solution of the general optimal control problem corresponds to acentralized policy and is hard to achieve, thus we focus on a decentralizedpolicy and provide solutions for a single node with two entering and twoexiting lines and asymptotic costs. Such solutions permit to simulate thebehaviour of decentralized algorithms for complex networks. Localoptimization ensures very good results also for large networks. The latteris shown by case study of a test telecommunication network.