Dimensioning computer network using tabu search

Abstract

This paper outlines a tabu search approach for dimensioning computer networks. Such an operation consists of finding a network topology configuration that minimizes the total communication cost, while taking into account some performance and reliability constraints. The proposed approach consists of applying some moves to a starting topology in order to reduce its total link cost and/or to improve its average delay. Based on simulation results, it generally provides better solutions than alternative heuristics such as cut saturation and genetic algorithm (GA).