The Functional Service Abstraction in Software-Defined Optical Networks Design

The realization and promotion of AI aided network automation (AIaNA) has resulted in a surge in research into the integration of AI with SDN in recent years. Despite AIaNA's early triumphs, its weaknesses have not been thoroughly examined; this includes the negative of the reduced human engagement obtained by the technology. Using software-defined IP over elastic optical connections, we investigate techniques to trick the AIaNA system in this work. This study proposes a practical model for allocating routes that takes into account latency, network risk, and the similarity of main and alternative paths. Subsequently, a model-solving technique based on deep reinforcement learning is implemented. The simulation findings suggest that this approach can fully use the capacities of IP + optical connections while simultaneously decreasing the network risk, providing theoretical direction for the routing strategy of MBS in the smart grid communication system.