CAN Bus Intrusion Detection: Autonomous Vehicle Security

Attack

Spoofed-frame injection rate 45 Hz

Gateway Defenses

IDS timing tolerance 5 ms

Bus Telemetry

Legit delivered

False positives

Malicious blocked

Malicious slipped through

Powertrain domain (Engine) Chassis domain (Brakes, Steering) Infotainment / telematics (OTA + V2X entry point) Blocked frame

How it works

A car's CAN bus is a shared broadcast network: every ECU can transmit a frame tagged with a fixed arbitration ID, and every other ECU can read it. The CAN protocol itself has no sender authentication — any node electrically on the bus can transmit any ID. Modern vehicles defend this with a zone gateway sitting between domains (powertrain, chassis, infotainment/telematics), inspecting every frame that tries to cross a boundary.

This simulator models two independent gateway defenses used in real automotive security architectures:

Domain isolation (whitelist):
  block if  domain_owner(arbitration_id) ≠ domain(origin_ECU)

Timing-based intrusion detection:
  T_expected(id) = learned nominal period for that ID
  Δt = t_now − t_last_seen(id)
  flag if  |Δt − T_expected(id)| > tolerance τ

Domain isolation is structural: the compromised Infotainment ECU (a realistic entry point via a malicious OTA update or a spoofed V2X message) has no legitimate reason to originate the Brakes ECU's arbitration ID, so the gateway drops it on sight — regardless of timing.
Timing IDS is statistical: it only flags a frame whose arrival breaks the learned periodic cadence of its ID. An attacker who mimics the real ECU's transmission rate closely enough can slip under a loose tolerance; too tight a tolerance instead starts rejecting legitimate traffic (a false positive) whenever normal jitter exceeds τ.
Try switching domain isolation off and tuning the attack rate near the Brakes ECU's real 50 Hz cadence — the spoofed brake command can then evade the timing IDS entirely, which is exactly why real automotive gateways layer both defenses rather than relying on timing alone.

Real-world relevance: this mirrors ISO 21434 / UNECE R155 automotive cybersecurity guidance, which requires layered in-vehicle network defenses (gateway segmentation + anomaly-based IDS) precisely because CAN's arbitration protocol was never designed with authentication.

Legitimate frame Blocked at gateway Malicious frame in transit

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.