A4SDN - Adaptive Alienated Ant Algorithm for Software-Defined Networking

Abstract

This paper introduces A4SDN, a distributed, adaptive, load-balancing algorithm for traffic engineering on Software-Defined Networks. A4SDN, i.e. the Adaptive Alienated Ant Algorithm for Software-Defined Networking, is based on the Alienated Ant Algorithm (AAA), a stochastic-based, heuristic approach used to solve combinatorial and multi-constraint optimisation problems. Based on a non natural ants' behaviour, the AAA forces the ants in searching for food to distribute themselves over all the available paths rather than converge to a single one. This strategy, applied to the packets to route, supplies an autonomic dynamic routing and leads to a better exploitation of the network bandwidth, enforcing best effort traffic and improving network performance. A comparison between A4SDN with two Dijkstra-based routing solutions clearly shows that A4SDN is able to guarantee a higher throughput associated with a lower delay and packet loss rate.