In this paper, we consider the problem of detecting performance related advanced persistent threats on a system of connected automated vehicles (CAVs) possibly mixed with connected, but manually driven vehicles (CMVs), operating over an urban area or highway and its feeder roads. We show how the emerging edge computing infrastructure deployed on the roadside for traffic monitoring can be exploited for this task. We briefly propose an approach that estimates an "abnormality" measure of each vehicle and performance anomalies for each type of attack in order to better track the vehicles that are affected early on in a performance attack and thereby facilitate root cause analysis. We also note several challenges in using such an approach.
Paper
Full text
Advanced Persistent Threats in Autonomous Driving
Semantic Scholar · Engineering · 2020
Abstract
In this paper, we consider the problem of detecting performance related advanced persistent threats on a system of connected automated vehicles (CAVs) possibly mixed with connected, but manually driven vehicles (CMVs), operating over an urban area or highway and its feeder roads. We show how the emerging edge computing infrastructure deployed on the roadside for traffic monitoring can be exploited for this task. We briefly propose an approach that estimates an "abnormality" measure of each vehicle and performance anomalies for each type of attack in order to better track the vehicles that are affected early on in a performance attack and thereby facilitate root cause analysis. We also note several challenges in using such an approach.