Research On Correctness Detection And Fault Location In Software Defined Network’s Flow Table

Software-Defined Network (SDN) provides a high degree of scalability for network management. With the expansion of network size, network rules, frequent updates, SDN switch requires frequent communication between the control plane and the data plane, which leaves the correctness of SDN switch vulnerable, especially the flow table inside SDN switch.The key to the correctness of SDN switch is the correctness of flow table. After study and analyze the testing methods to the validity of the existing SDN switches, we found one kind of switch fault has not yet been resolved. This fault is caused by the priority of forwarding rules. The existing attestation methods are unable to accurately detect such fault. In this paper, we design new algorithms to solve this problem, by expanding the coverage of the attestation.Secondly, the sub-problem of correction (generating test packages) is proved as a NP-complete problem. On this basis, we propose an efficient detection algorithm to detection and fault diagnosis algorithms to infer the entire flow table in switch. Practice shows that these algorithms can effectively reduce the complexity of the packet generation process.This paper presents an attestation system for enforce the detection algorithms. The system includes an OpenFlow switches, controllers and virtual hosts. By the experimental system, the paper well validated the correctness and efficiency of detection algorithms and fault diagnosis algorithms.