The Multi-Commodity Flow Problem with Disjoint Signaling Paths: A Branch-and-Benders-Cut Algorithm

Jiachen Zhang, Youcef Magnouche, Sébastien Martin, Antoine Fressancourt, J. Christopher Beck

Abstract: Data routing in networks is required to be efficient and reliable. Fast detection and recovery from link or path failures play crucial roles in the reliability guarantee. In this work, we investigate a variant of the multi-commodity flow problem to address one formalization of reliability, where for each demand, a primary path transmits the demand without exceeding a jitter limit and an arc-disjoint secondary path signals the possible failure of the primary path. We first present a compact mixed-integer linear programming model and then we devise a Branch-and-Benders-Cut algorithm to solve this combinatorial optimization problem. On a diverse set of instances, we evaluate the algorithm's performance and discuss several numerical results.