Impact analysis and detection of time-delay attacks in time-sensitive networking

Abstract

Time-sensitive networking (TSN) will be widely used in automotive industry and industrial automation because it can provide deterministic transmission. Most of the TSN traffic shaping mechanisms rely on clock synchronization between different devices in the network. However, Time-Delay Attacks (TDAs) can interfere with synchronization, which further reduces traffic transmission quality. Although some studies have proposed detection strategies against TDAs, they are oriented to traditional ethernet and restricted by network architecture and devices. Therefore, this paper first analyzes the impact of TDAs on traffic transmission quality under different combinations of Time-Aware Shaper (TAS) or Cyclic Queue Forwarding (CQF) with link redundancy. Then, this paper provides a distributed monitoring parameter using Per-Stream Filtering and Policing (PSFP) for the detection and localization of TDAs, in which a modified token bucket mechanism is applied. Finally, the paper models the application of TSN in automotive industry for verification and further evaluates the effectiveness of this parameter through simulation. The results show that TDAs can cause undesirable variations in traffic latency under TAS and CQF. While proper parameter configuration and link redundancy can mitigate the impact of TDAs, they cannot prevent them. Besides, the proposed parameter helps detect and locate TDAs in TSN effectively.