Abstract
This paper proposes a resilient traffic engineering framework for Software-Defined Networks (SDN) using multi-commodity flow optimization. By predicting potential link failures and pre-computing backup paths, our method minimizes packet loss and maximizes network throughput under single and multiple link failures. Evaluated on real topologies (GEANT, Abilene), it achieves 99.8% resilience with only 12% overhead — significantly better than traditional methods.
Publication
Conference
IEEE ICC 2025
Year
2025