Abstract:
Software Defined Networking (SDN) is an emerging architecture that decouples the network control and forwarding functions. In SDN, the functionality of static configuration and routing table in a traditional network has been replaced by forwarding table entries (FTEs). Thus a systematic research on FTE to better monitor traffic and manage networking resources becomes crucial in SDN. There are already some initial works on FTE representation from mathematical/logical perspective. However, they usually concentrate on the abstraction and expression of FTE rather than the applications in real network. Based on existing research, a controller is unable to monitor networking traffic and manage networking resources from a network-wide perspective. To address these challenges, Boolean algebra is chosen and extended in this thesis to examine the relations and manipulations among FTEs together with traffic statistics. Specifically, three SDN applications: i) equivalence evaluation during FTE deployment, ii) non-invasive traffic estimation and iii) anomaly detection, have been proposed and verified with the help of Boolean algebra. All of these applications rely on the mining of the FTEs and their associated statistics, thus no overhead will be introduced to the switch's original packet forwarding functionalities. They can be easily deployed in production networks due to the non-invasive strategy as well as the feasibility and flexibility in real networking scenarios.
