Abstract: We have adirected graph describing a network and an origindestination matrix for customer internet traffic demands. Our aim is to optimize the routing of the traffic by adjusting the weights of the graph links. Though the internal design of the routing protocol is unavailable, we have access to the simulator to model it. Given the link weights, the simulator provides the values for traffic flow on each link. If the flow on a link exceeds its capacity, this link is considered overloaded. The objectives of the problem are to minimize the total number of overloaded links and the distance from the initial weight vector. We have developed a scheme based on a novel integer linear programming model. It uses values of the traffic flow changes depending on the link weights modifications. In the two-stage approach, this scheme is used to provide the initial Pareto set approximation. The approach outperforms the state-of-the-art multiobjective evolutionary algorithms.

Cite: Yuskov A. , Kulachenko I. , Melnikov A. , Kochetov Y.
Two-Stage Algorithm for Bi-objective Black-Box Traffic Engineering
In compilation Optimization and Applications : 14th International Conference, OPTIMA 2023, Petrovac, Montenegro, September 18–22, 2023, Revised Selected Papers. – Springer., 2023. – C.110-125. – ISBN 9783031478598. DOI: 10.1007/978-3-031-47859-8_9 Scopus OpenAlex

Dates:

Published print:	Nov 10, 2023
Published online:	Nov 10, 2023

Identifiers:

Scopus:	2-s2.0-85177192319
OpenAlex:	W4388522860

Citing:

DB	Citing
OpenAlex	1

Altmetrics: