Leveraging duality for the gamma-robust segment routing traffic engineering problem

Segment Routing (SR), a modern network architecture enhancing traffic engineering, offers flexibility by allowing traffic to be routed through intermediate nodes or links. This paper addresses the challenge of optimizing routing in uncertain traffic distribution scenarios. Rather than relying on a single traffic matrix for optimization, we take a unique approach, considering an infinite set of matrices defined by linear constraints. Our goal is to optimize routing strategies under the worst-case scenario within this set. Through novel formulations, we achieve significant improvements in optimization speed compared to traditional methods that explore all extreme points or use iterative constraint generation for these matrices. This research contributes to the field by enhancing the robustness of SR-based traffic engineering, an area previously explored using different methodologies.