What runs here
Platform services support applications across the high-performance computer (HPC). They commonly run as privileged processes or services within selected virtual machines, with controlled interfaces for applications and maintenance tools.
| Component | Role |
|---|---|
| Over-the-air (OTA) update client | Receives update instructions and packages, checks authorization and compatibility, stages installation, and reports progress and results. |
| Service-Oriented Vehicle Diagnostics (SOVD) server | Exposes diagnostic capabilities, faults, software information and supported operations through a REST/HTTP API. |
| Host intrusion detection system (IDS) | Observes selected host activity, detects suspicious behavior and produces security events. |
| Logging and tracing service | Collects application and platform records with timestamps, source identifiers and bounded storage. |
An AUTOSAR Adaptive implementation can use Update and Configuration Management, Vehicle Update and Configuration Management, Intrusion Detection System Manager, and Log and Trace. These are logical functional clusters; the grid is a teaching model, not a mandatory AUTOSAR layering scheme. An IDS manager processes security events; detection also requires event sources and detection logic. AUTOSAR Adaptive Platform
SOVD provides diagnostic access to HPC software and can also expose connected ECU diagnostics through adapters. ISO 17978-3:2026 specifies its API. It does not imply that every connected ECU implements SOVD itself. ISO 17978-3
Why it looks like this
An application processor provides the memory, storage and operating-system facilities needed for package handling, diagnostic servers and substantial event collection. Centralizing these capabilities reduces duplicated application code, although integration, validation and persistent storage remain significant costs. GPU and NPU availability does not make these services accelerator workloads.
Applications and security rules may change independently of the base operating system. Shared services therefore need versioned interfaces, explicit dependencies and controlled deployment. Update engineering covers the vehicle, infrastructure and organizational responsibilities, rather than ending at successful package delivery. ISO 24089
Most maintenance work can tolerate variable latency, but unrestricted logging, scanning or installation can disrupt workloads with deadlines. The node’s QM–ASIL B description does not assign an ASIL to every service. Safety relevance depends on allocated requirements and failure effects. Shared CPU time, memory bandwidth, storage and network access must be considered when establishing freedom from interference.
How it connects
Above, applications use APIs to publish logs and security events, expose diagnostic data and participate in update lifecycle transitions. Define ownership of faults, operation cancellation and behavior when a service is unavailable.
Below, services use middleware and operating-system facilities for communication, process control, files, persistent state and cryptographic operations. Package verification must connect to provisioned trust anchors; boot firmware must enforce the intended boot policy. An authenticated download alone does not establish a trusted executable image.
Across nodes, the HPC can coordinate updates with other controllers, access legacy diagnostics through a gateway, and forward selected security events through the vehicle’s backend connection. It need not own the entire vehicle update campaign. Remote access requires authentication, authorization and network policy; SOVD capability discovery does not grant permission to execute an operation.
Compared with other node classes
- HPC safety island: Watchdog management, end-to-end communication protection and fault memory emphasize bounded supervision and fault response; bulk maintenance traffic needs controlled interaction with these functions.
- Zone controller: Network Management, Dcm/Dem, IdsM and eFuse management emphasize local communication, diagnostic events, security-event handling and power control.
- MCU-based ECU: Dcm/Dem, NvM, SecOC and the crypto stack typically operate within a statically configured platform with tighter memory budgets.
- Smart node / LIN slave: Configuration and basic diagnostics expose a smaller maintenance surface, usually reached through a supervising controller.
Best practices
- Budget CPU, memory, storage and network use; bound queues and define which records to discard under overload.
- Verify package authenticity, target compatibility and dependencies before activation. Test interrupted installation, failed activation and recovery; reconcile rollback with anti-rollback policy.
- Gate intrusive diagnostics and installation on explicit vehicle and application states, with clear ownership of authorization.
- Apply least privilege to maintenance services and protect credentials through the platform’s key-management facilities.
- Correlate faults, security events and logs using software versions, source identities and timestamps with known clock quality. Minimize sensitive data and define retention.
- Validate service failure and overload effects on safety-related workloads, including shared storage and communication paths.
Common pitfalls
- Treating ordinary logs as a complete IDS, or assuming an event manager detects attacks without sensors and rules.
- Reporting an update as successful before activation, health checks and version reconciliation finish.
- Allowing diagnostic requests or log export to exhaust resources needed by driving functions.
- Assuming one application-processor service can observe every guest VM or the safety island without explicit instrumentation and access.