Electric Vehicle Profiling Leveraging Early Charging Voltage Patterns
Paper in progress »
Anonymous Authors
Electric Vehicles (EVs) are rapidly gaining adoption as a sustainable alternative to traditional fuel-powered vehicles. Securing their charging infrastructure becomes essential as their presence on the roads increases. Despite adopting traditional authentication algorithms at the protocol level, recent results showed that an attacker may be able to steal energy through tailored relay attacks. A solution to counter such attacks may be to leverage the EV's fingerprint on the current exchanged during the charging process. However, existing approaches focus on the final stages of the charging process, leaving a window of opportunity for malicious actors to consume substantial amounts of energy before being detected and repudiated. This gap underscores the need for earlier and more effective authentication methods to enhance security and prevent unauthorized charging. At the same time, profiling may jeopardize users' privacy, as the possibility of uniquely identifying EVs based on their charging patterns potentially enables unauthorized tracking and profiling. In this paper, we propose a framework to uniquely identify EVs using physical measurements over the early stages of their charging process. Our intuition is that the voltage behavior during the initial stages of the charging process exhibits similar characteristics to the current behavior during the final stages. Thus, by extracting features from the early voltage measurements, we prove the feasibility of profiling EVs. Our approach improves existing works enabling faster and more reliable vehicle identification. We test our solution on a dataset containing 7408 usable charges from 49 EVs, achieving an accuracy of up to 0.86. Our feature importance analysis shows that our models achieve near-optimal performance by considering only the 10 most important features, further enhancing efficiency on top of the already lightweight architectures used. This research poses the basis for a novel authentication factor while highlighting the possible privacy leakages that may arise from an attacker who can get charging information about a victim's vehicle.
To train the models and reproduce the results, start by cloning the repository.
git clone https://github.com/Mhackiori/EV-Volt-Auth.git
cd EV-Volt-AuthThen, install the required Python packages by running the following command. We reccomend setting up a dedicated environment to run the experiments.
pip install -r requirements.txt