Automotive PQC

Modern vehicles are among the most complex distributed computing systems in production today. A premium vehicle contains 100–150 Electronic Control Units (ECUs) communicating over multiple bus technologies; of these, 24–30 are security-relevant — the focus of PQC migration planning. The cryptographic requirements span from real-time safety-critical signing (brake-by-wire commands in <1 ms) to bulk data protection (OTA firmware packages of several gigabytes).

V2X communication defined by IEEE 1609.2 adds an external attack surface: vehicles broadcast Basic Safety Messages (BSMs) 10 times per second, each signed with ECDSA P-256. A future CRQC could forge BSMs to cause phantom braking or mask real obstacles — a direct safety-of-life threat. Transitioning V2X to PQC signatures is complicated by the size of PQC signatures (2,420 bytes for ML-DSA-44) relative to the ~300-byte BSM payload, and the strict 100 ms end-to-end latency requirement.

The industry is simultaneously migrating from domain architecture (separate ECU clusters for powertrain, chassis, body, infotainment) to zonal architecture (zone controllers aggregating ECUs by physical location). This architectural shift creates a window of opportunity: zonal gateways become natural PQC enforcement points, concentrating cryptographic operations on fewer, more powerful processors.

In-vehicle bus protocols create hard constraints: CAN 2.0 frames are limited to 8 bytes of payload — impossibly small for any PQC signature. CAN FD extends this to 64 bytes, still insufficient for ML-DSA (3,309 bytes at security level 3). Both require application-layer fragmentation. Automotive Ethernet (1,500-byte MTU) can natively carry ML-KEM payloads (1,088 bytes) without fragmentation, but even Ethernet will require IP fragmentation for ML-DSA-44 or larger signatures.