Is Your ECU Really Yours?
summary
Claims a physical fingerprinting system that distinguishes identical ECUs by capturing analog switching transients with a Red Pitaya data acquisition board and extracting 211 features across rise/fall/steady transients. A hybrid classifier uses a cycle‑based Random Forest and an OVA stacked ensemble to reach 99.5% real-time accuracy under controlled mounting (93.6% without) and 100% device‑type accuracy across eight devices spanning four hardware types including same‑model clones. Prototype runs on embedded development platforms as a proof of concept.
AI-written summary · about
chapters (11)
- 0:00ECU authentication challenges
- 1:14ECU spoofing vulnerabilities
- 2:17Prior ECU fingerprinting limits
- 2:59CAN bus sender authentication failure
- 3:20Chip imperfections as fingerprints
- 4:45Signal characteristics extraction
- 6:16Data acquisition pipeline
- 7:43Hybrid classifier for identification
- 8:29Real-time ECU test bed
- 9:10Study results and takeaways
- 9:51Future work on ECUs
related talks
30mFingerprinting Electronic Control Units for Vehicle Intrusion Detection
37mAutomatic Automotive Hacking
07mNot Quite Yourself Today: Behaviour-Based Continuous Authentication in IoT Environments
55mReliable and Secure Embedded Systems
50mSensor and Process Fingerprinting in Industrial Control Systems
44mMy Car, My Keys: Obtaining CAN Bus SecOC Signing Keys
14mStateful Behavior Inference and Runtime Enforcement for Vehicle Network Security
49mEstablishing and Maintaining Device Trust: A Case Study