Abstract

International audience; In this paper, we introduce a novel framework for the off-line flow allocation problem for traffic engineering in MPLS networks. We use a multi-stage optimization that considers minimizing successively congestion, resource consumption and operational costs. The framework is based on linear programming and presents an efficient solution for the problem both in terms of computational complexity and approximation guarantees. Preliminary simulation results prove that we can easily tune congestion and resource consumption while keeping low operational costs, expressed in the number of Label Switched Paths (LSPs) in use. Moreover, the flow allocation solution is obtained in a polynomial time and results can be further extended to larger topologies with a big number of demands.