Knowledge-Based Autonomous Service Provisioning in Multi-Domain Elastic Optical Networks

Abstract

Optimization of end-to-end service provisioning across multi-domain EONs is critically important due to the restricted intra-domain visibility and the agile spectrum allocation mechanism of EONs. Previously demonstrated schemes apply fixed service provisioning policies (e.g., heuristic routing and spectrum assignment and threshold- based anomaly detection) and fail to fully address the inherent principle and dynamic conditions of multi-domain EONs. This article brings in data analytics and presents a knowledge-based autonomous service provisioning framework for multi-domain EONs. The proposed framework employs a new broker plane working with domain managers while supporting the autonomy of each management domain. This can be seen as a hierarchical multi-domain network control and management architecture in support of enhanced resource efficiency and network scalability where each hierarchy employs an observe-analyze-act cycle. Specifically, the broker and domain managers can observe the network state through comprehensive network monitoring (monitoring of optical paths' performance, resource utilization, and traffic dynamics), analyze the network behavior and form the knowledge repository of network rules using machine learning techniques, and act intelligently according to the obtained knowledge. We assess the proposed framework by evaluating two key knowledge-based service provisioning use cases: proactive inter-domain traffic-prediction- based autonomous traffic engineering and quality-of-transmission-aware inter-domain lightpath provisioning. Numerical results demonstrate remarkable benefits of the proposed framework over conventional solutions.