Timing Analysis of the TSN Time-aware Shaper Based on Compositional Performance Analysis

Introduction: Ethernet-based automotive networks require additional mechanisms to ensure low latency for time-sensitive streams. The Time-Sensitive Networking (TSN) task group for real-time systems has proposed several mechanisms, including the time-aware shaper (TAS), to provide real-time performance. TAS enables the setting of separate transmission time windows at output ports for different queues, thereby preventing interference from other priorities. The primary objective of this study is to determine the worst-case end-to-end delay of class-based TAS streams in vehicular networks with less conservative error, which can then be used to guide the reverse design of TAS parameters. Method: This study employs Compositional Performance Analysis (CPA) to compute the worst-- case end-to-end (E2E) delay of class-based TAS streams at the data link layer. The optimized CPA results are compared with pre-optimization calculations and simulation outcomes. Result: The optimized CPA calculation reduces the end-to-end delay by an average of 42.6% for two-hop TAS streams and 57.8% for three-hop TAS streams compared to the pre-optimization results. Discussion: Building on the existing CPA method, this study enhances E2E worst-case delay analysis and reduces conservative error. Additionally, synchronization errors and extreme operational scenarios are discussed. However, the methodology proposed in this study is only applicable to scenarios with aligned gating windows. Conclusion: All simulation results for TAS streams fall within the bounds predicted by the CPA calculation. The optimized CPA analysis of the worst-case end-to-end delay greatly reduces conservative error. It can be used to validate the appropriateness of the class-based gate control parameter configurations.