Abstract
Camera-channel replay attacks threaten networked vehicular cyber-physical systems by delivering syntactically valid but temporally outdated images through the in-vehicle sensing and communication architecture. This paper proposes a persistence-enhanced attack detection and event-triggered resilient control framework for 4WID autonomous vehicles under such attacks. The vehicle model is first established, and parallel camera-LiDAR channels are constructed to estimate lateral and heading errors. Based on the redundant task states, the proposed detector evaluates normalized cross-channel residuals over a sliding window, accumulates persistent temporal divergence, and updates the decision threshold using median-MAD statistics with confirmation and reset logic. After attack confirmation, LiDAR-derived task states are smoothly incorporated to reconfigure compromised camera feedback, while an integrated MPC coordinates path tracking and yaw stability. Hardware-in-the-loop results show that TDPE achieves F1-scores of 0.966, 0.947, and 0.963 under fixed-depth, random-depth, and intermittent replay attacks, while the controller maintains low lateral error and sideslip responses with mean execution times below 10 ms. Real-vehicle tests further verify reliable path tracking under different replay patterns.
| Original language | English |
|---|---|
| Journal | IEEE Internet of Things Journal |
| DOIs | |
| Publication status | Accepted/In press - 2026 |
| Externally published | Yes |
Keywords
- attack detection
- camera-channel replay attacks
- integrated control
- multi-sensor perception
- resilient control
Fingerprint
Dive into the research topics of 'Persistence-Enhanced Attack Detection and Event-Triggered Resilient Control for Autonomous Driving under Camera-Channel Replay Attacks'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver