Abstract:
The virtualization and software-definition of network functions, controls, and applications enhance performance and operation costs while bringing new value to the infras...Show MoreMetadata
Abstract:
The virtualization and software-definition of network functions, controls, and applications enhance performance and operation costs while bringing new value to the infrastructures. However, due to the stateless, gossipy, centralized, periodic, and tardy nature, the network control and management approaches in Software-Defined Networking (SDN), such as the current OpenFlow Discovery Protocol (OFDP), pose scalability, latency, and reliability challenges. This paper designs a novel Centrality-Aware Multitemporal (CAMEL) discovery protocol for SDN to enhance the centralized discovery mechanism’s scalability and latency issues. We facilitate multiple discovery timers for each target according to the significance instead of using a single timer for the entire network. CAMEL generalizes the significance measurement for various network topologies by using the centrality models. We combine the normalized degree and betweenness centrality values to find an unbiased impact factor of each node. Applying the identified significance to a multitemporal discovery mechanism, CAMEL reduces network impact by decreasing the discovery delay to the significant nodes and enhances control message efficiency by lowering the discovery frequency to the less significant targets. We have implemented CAMEL on the RYU controller. The experimental results validate that CAMEL improves discovery message efficiency, makes the control traffic less bursty, and enhances the network service quality by reducing discovery delay to the significant nodes.
Date of Conference: 19-22 July 2021
Date Added to IEEE Xplore: 31 August 2021
ISBN Information: