A tabu search heuristic for the offline MPLS reduced complexity layout design problem

Abstract

MPLS paths can be calculated on-line as demands arrive or off-line for an estimate of the traffic demand over time. Off-line calculation has the advantage of allowing a globally optimal network design. Prom the operational standpoint, a layout design for large networks should consider minimizing the layout complexity, thus reducing the cost of operation. In this paper, we formulate an optimization problem whose objective is to minimize the number of required hops. An end-to-end path delay constraint provides quality of service (QoS) guarantees. The resulting Minimum Path Set and Flow Allocation Problem (MPSFAP), formulated as a Mixed Integer Non Linear Program (MINLP) is A/'P-complete. As such, it is only possible to solve it exactly on small size networks. To overcome this limitation, we designed a Tabu Search (TS) based algorithm that computes good quality approximate solutions for the MPSFAP problem. Solutions for medium to large size networks in reasonable CPU time and within acceptable optimality tolerances are found using the proposed TS based algorithm. © IFIP International Federation for Information Processing 2004.