Probabilistic Fault-Tolerant Robust Traffic Grooming in OTN-over-DWDM Networks
Dimitrios Michael Manias, Joe Naoum-Sawaya, Abbas Javadtalab, Abdallah Shami
TL;DR
The paper addresses resilient traffic grooming in OTN-over-DWDM networks under demand uncertainty and fault scenarios. It introduces a probabilistic robust optimization model that uses independent backup paths and a robustness parameter set, including $\Gamma$, $\hat{B_d}$, and $\mu_r$, to control overprovisioning. The objective minimizes the number of optical channels and OTN switches while maintaining capacity constraints even under worst-case deviations. Experiments on 6-node topologies show that the robust solution preserves capacity and service continuity under $20\%$ demand deviations of up to $10\%$, outperforming deterministic approaches and supporting SLA guarantees for future networks.
Abstract
The development of next-generation networks is revolutionizing network operators' management and orchestration practices worldwide. The critical services supported by these networks require increasingly stringent performance requirements, especially when considering the aspect of network reliability. This increase in reliability, coupled with the mass generation and consumption of information stemming from the increasing complexity of the network and the integration of artificial intelligence agents, affects transport networks, which will be required to allow the feasibility of such services to materialize. To this end, traditional recovery schemes are inadequate to ensure the resilience requirements of next-generation critical services given the increasingly dynamic nature of the network. The work presented in this paper proposes a probabilistic and fault-tolerant robust traffic grooming model for OTN-over-DWDM networks. The model's parameterization gives network operators the ability to control the level of protection and reliability required to meet their quality of service and service level agreement guarantees. The results demonstrate that the robust solution can ensure fault tolerance even in the face of demand uncertainty without service disruptions and the need for reactive network maintenance.