Context-Aware Intrusion Detection on the Vehicle High-Performance Computer: Securing the Software-Defined Vehicle

The shift to software-defined vehicles collapses dozens of distributed ECUs onto a handful of high-performance computers, consolidating functionality – and attack surface – onto centralized, update-driven platforms. Traditional signature- and specification-based intrusion detection struggles here: it inspects messages in isolation and misses multi-stage attacks that only reveal themselves in context. This talk presents a context-aware IDS that runs on the vehicle HPC and correlates signals across domains – in-vehicle networks, service-oriented communication, connectivity, and vehicle state – to distinguish genuine threats from benign anomalies and cut false positives. We examine how the HPC’s compute headroom enables richer contextual and ML-driven analysis than legacy ECUs allow, how detection integrates with the AUTOSAR IDS framework and a backend V-SOC, and how to keep such analysis advisory and freedom-from-interference-safe alongside safety-critical workloads. We close with deployment, evaluation, and lifecycle considerations under ISO/SAE 21434 and UNECE R155.

Biography

Automotive technology leader with 12+ years in connected cars and cybersecurity. Formerly held senior roles at ClearMotion UK, EDAG Engineering, and TRL, delivering OEM security programmes from concept to launch. Holds an MSc in Embedded and Wireless Systems from Coventry University.