Fault localization algorithm in computer networks by employing a genetic algorithm

Abstract

Fault localization is an important part in communication networks. Faults are unwanted and unavoidable in communication systems, and hence, their quick detection and localization is essential for sustaining the health of the network. This paper proposes an end-to-end approach that uses passive measurements for fault localization in communication networks. We formulate the fault localisation issue as an optimisation problem and then employ the genetic algorithm technique to solve it. Extensive simulation results show that although our algorithm needs to test only a small set of network components to localise all faults, it can infer the faulty nodes in at least 97% of cases. This simulation shows that the proposed algorithm, called genetic algorithm-based fault localisation is superior to the other approaches to localise all faults in a network.