Traffic engineering in MPLS networks with multiple objectives: modeling and optimization

With the increasing usage of QoS-based applications, traffic engineering in communication networks has become more crucial. For the optimal utilization of network resources, service providers need to consider multiple criteria that can be customeror traffic-oriented. Multiprotocol Label Switching (MPLS) has been developed to apply more convenient traffic engineering in autonomous systems. This thesis focuses on the multiobjective optimization of Label Switched Path design problem in MPLS networks. Minimal routing cost, optimal load-balance in the network, and minimal splitting of traffic form the objectives. The problem is formulated as a zero-one mixed integer program and aims at exploring the trade-offs among the objectives. The integer constraints make the problem NP-hard. In the thesis first both the exact and heuristic multiobjective optimization approaches are discussed, and then a heuristic framework based on simulated annealing is developed. Various search strategies within the framework are investigated and experimental studies are carried out for a performance comparison.