Anomaly Detection for Roadside Devices Based on Scene Consistency

Résumé fourni par la source

This paper presents a novel framework for anomaly detection in roadside devices by utilizing scene consistency to enhance the reliability of intelligent transportation systems. As roadside devices become more sophisticated, they face increased security risks, particularly in the accuracy of data displayed by intelligent message signs, which can lead to traffic disruptions. Our approach involves two key components: detecting anomalies in sensor data through comparative analysis across multiple devices and verifying actuator behavior against received commands. We achieve effective real-time anomaly detection without modifying existing hardware or software by employing a dynamic correlation modeling technique on an association graph of device interactions. Experimental results indicate that our method significantly improves detection accuracy and reduces false alarms, contributing to safer and more efficient urban transportation management.