Using mobile agents and maximum path flow algorithms to maximize network utilization

Abstract

The purpose of this thesis is to show how the use of one or more of the shortest augmenting path (polynomial time) algorithms are used to maximize the bandwidth usage and minimize the cost and latency in Multiprotocol Label Switched (MPLS) networks. The thesis will also show how providers that offer services such as Virtual Private Networks (VPNs) based on Border Gateway Protocol (BGP) and MPLS, or Virtual Private LAN Services (VPLS) can use such an algorithm to offer premium services to higher paying customers. This thesis extends the problem of adaptive routing in communications networks: It focuses on maximizing the bandwidth usage in a MPLS network by using a modified augmenting path algorithm. The algorithm is influenced by Resource Reservation Protocol (RSVP) to reserve traffic flows on routers. The algorithm proposed is inspired by previous research in the area of indirect communication among agents through modifications induced in their environment or simply known as stigmergy. The agents implement the Edmonds Karp Algorithm rather than the pheromone tables in stigmergy. By using a mobile agents approach, the Augmenting Path Algorithm is adapted to find the maximum flow through a network.