Adversarial Localization Algorithms in Indirect Vehicle-to-Vehicle Communication

Abstract

Communicating autonomous vehicles (CAVs) can obtain direct measurements from their sensors or indirectly receive them via Vehicle-to-Vehicle (V2V) communication. As the CAVs are expected to share a part of their measurements, it can naturally pose a privacy threat by possibly revealing the route of the sender vehicle. Consequently, we shall assess the risks of sharing a dataset that is a mixture of direct and indirect measurements. However, a wide variety of papers focus on localization attacks for direct measurements; incorporating indirect measurements opens a new horizon for these researches. In this paper, we analyze a couple of localization algorithms for mixture datasets with applicable performance metrics. We have evaluated the algorithms in an Eclipse SUMO-based simulation. We consider these results as the baseline of future research.