A Boolean Network Tomography based Method for Deterministic Multi-point Fault Detection

Time-sensitive networking (TSN) is a deterministic network, where failures will seriously affect the quality of the real-time data transmission service. Hence, fault localization is critical in TSN. In the application domain of TSN, an ideal link failure detection method needs to have properties including execution time deterministic, low detection traffic overhead, multi-point fault detection capability for arbitrary typologies. However, current state-of-the-art fault detection algorithms can not achieve these properties simultaneously. In this paper, we propose a Boolean network tomography-based multi-point fault detection method that leverages the deterministic transmission mechanism of TSN to guarantee that the detection time is upper bounded. The method includes an offline preparation phase and an online operation phase. In the preparation phase, we use a detection flow generation algorithm to get a set of detection flows that can identify up to $K$ faults with as few paths as possible. In the operation phase, detection packets are realized as time-sensitive flows routing along the detection paths. They are sent periodically to the controller which performs a topology discovery algorithm to infer the faulty links based on the arrival status of each detection packet. Comprehensive experiments are conducted and the results show that compared with existing fault detection methods, the proposed algorithm can accurately identify multiple faulty links in deterministic time, and the generated detection paths set needs fewer paths than the prior methods.