Self-similar QoS Manager: a bio-inspired approach

This work presents a Self-similar QoS Manager using a immunological algorithm to optimize a MPLS network. The system receives a set of requirements, imposed by many different client applications, and creates a set of paths that satisfy all the capacity constraints necessary for each specific traffic. The solution also keeps the load balance in the network and considers the self-similar characteristics of the traffic, which lead to improved results. The ns2 (network simulator 2), working with real traffic traces injected in source nodes, was used to validate the approach. Results showed that all simultaneous client application requirements were satisfied, while optimizing the traffic allocation in the MPLS network. The paper also analyzes the influence of different bio-inspired algorithms (genetic and immunological) in the solution of the MPLS path allocation problem.