Evaluating TTEthernet in the Automotive Domain Using Fault Injection

Current networking technologies in many automotive applications are approaching their limitations to support the growing communication bandwidth requirements. Some of these applications include Advanced Driver Assistance Systems (ADAS) and autonomous driving. They require high data rates, and due to the increasing number of automotive applications, there is a need for more communication bandwidth. Ethernet-based network protocols have been investigated in recent times to satisfy the bandwidth requirement. Other requirements such as fault tolerance and deterministic timing have been met by extending the Ethernet to protocols such as the TTEthernet and Time- Sensitive Networking. As TTEthernet is one of the candidate networks used for safety systems with mixed-criticality levels, it is necessary to evaluate automotive applications’ dependability over TTEthernet. This work evaluates a flattened hierarchical communication architecture of automotive applications over TTEthernet. Two variants, a redundant and non-redundant setup, are evaluated. A fault injection framework is used to test the robustness of the resulting setups by deliberately inserting faults into the system. It assesses TTEtherent’s use and suitability for an in-vehicle network and observes the behaviour of the resulting system under failure. It was observed that even for a non-redundant setup, the time-triggered virtual links are always isolated from the impact of failure due to temporal partitioning provided by the TTEthernet protocol. This property makes them suitable for safety-critical automotive applications such as the braking system. However, complete fault isolation is only observed as expected in the redundant setup.