Optical technologies play nowadays a crucial role in different network scenarios (e.g., data centres, metro and access networks), mainly due to their support for higher bandwidth and scalability in comparison to electronic-based opaque solutions. In this context, the deployment of the Software Defined Networking (SDN) paradigm over optical networks has gained interest from both industry and research communities, spawning several implementations over multiple well documented use cases. However, in Transparent Optical Networks (TONs), where nodes offer optical switching capabilities, neighbouring nodes adjacencies have to be manually configured at both data and control plane levels, which is a lengthy task that can potentially lead to misconfiguration. Alternative solutions rely on the use of supervisory networks or dedicated channels. Nevertheless, implementing these methods requires both effort and resources to provide adjacencies discovery, making the awareness of the correct underlying topology by the SDN controller a tedious and occasionally very complex process. In this paper, we present a novel SDN-based cost-effective topology discovery method, allowing TONs to automatically learn physical adjacencies between optical devices. In particular, this is achieved by means of a test-signal mechanism and the OpenFlow protocol. The SDN control plane and optical agent implementations, as well as the message exchange flow between the subsystems and the controller are described. Then, the proposed discovery mechanism is experimentally assessed over an emulated TON test-bed, analysing the average time required for the optical topology discovery in different network scenarios.

Peer Reviewed