Abstract
Conference Title: 2017 IEEE International Symposium on Technologies for Homeland Security (HST) Conference Start Date: 2017, April 25 Conference End Date: 2017, April 26 Conference Location: Waltham, MA, USA This work applies artificial intelligence (AI) to secure wireless communications of Connected Vehicles. Vehicular Ad-hoc Network (VANET) facilitates exchange of safety messages for collision avoidance, leading to self-driving cars. An AI system continuously learns to augment its ability in discerning and recognizing its surroundings. Such ability plays a vital role in evaluating the authenticity and integrity of safety messages for cars driven by computers. Falsification of meter readings, disablement of brake function, and other unauthorized controls by spoofed messages injected into VANET emerge as security threats. Countermeasures must be considered at design stage, as opposed to afterthought patches, effectively against cyber-attacks. However, current standards oversubscribe security measures by validating every message circulating among Connected Vehicles, making VANET subject to denial-of-service (DoS) Attacks. This interdisciplinary research shows promising results by searching the pivot point to balance between message authentication and DoS prevention, making security measures practical for the real-world deployment of Connected Vehicles. Message authentication adopts Context-Adaptive Signature Verification strategy, applying AI filters to reduce both communication and computation overhead. Combining OMNET++, a data network simulator, and SUMO, a road traffic simulator, with Veins, an open source framework for VANET simulation, the study evaluates AI filters comparatively under various attacking scenarios. The results lead to an effective design choice of securing wireless communications for Connected Vehicles.