The term Software-Defined Vehicle (SDV) describes the vision of software-driven
automotive development, where new features, such as improved autonomous driving,
are added through software updates. Groups like SOAFEE advocate cloud-native
approaches – i.e., service-oriented architectures and distributed workloads – in
vehicles. However, monitoring and diagnosing such vehicle architectures remain
largely unaddressed. ASAM’s SOVD API (ISO 17978) fills this gap by providing a
foundation for diagnosing vehicles with service-oriented architectures and
connected vehicles based on high-performance computing units (HPCs).
For service-oriented architectures, aspects like the execution environment,
service orchestration, functionalities, dependencies, and execution times must
be diagnosable. Since SDVs depend on cloud services, diagnostic functionality
must extend beyond the vehicle to include the cloud for identifying the root
cause of a malfunction. Due to SDVs’ dynamic nature, vehicle systems must be
monitored as service degradation is more likely than a complete failure.
Established monitoring and error analysis approaches for cloud environments
cannot easily be transferred to vehicles. Monitored values must be aggregated
and correlated to error events before cloud transmission, or suspects must be
created in the vehicle for thorough analysis, reducing the data exchanged with
the backend.
The SOVD API provides a good foundation to diagnose service-oriented
architectures and HPCs. While SOVD offers a wide range of diagnostic and
monitoring features, it currently lacks solutions for diagnosing certain aspects
and especially monitoring of a service-oriented architecture. This paper
addresses these gaps, showcasing approaches and techniques to enhance monitoring
and diagnostics.