Evaluating and analyzing reliability over decentralized and complex networks

Abstract

In an increasingly interconnected and distributed world, the ability to ensure communications becomes pivotal in day-to-day operations. Given a network whose edges are prone to failures and disruptions, reliability captures the probability that traffic will reach a target location by traversing edges starting from a given source. This paper investigates reliability in decentralized and complex networks. To evaluate reliability, we introduce a multi-agent method that involves pathfinding agents to reduce the graph. Performance of this method is tested on scale-free and small-world networks as well as real-world spatial networks. We also investigate reliability score which aims to rank the capability of nodes in terms of traffic dissemination traffic across all nodes. Analysis over spatial networks indicates that the reliability score correlates with central and sub-central regions in a geographical region.