Data and analysis scripts for "Current Affairs: A Security Measurement Study of CCS EV Charging Deployments"

Project organization

Device IDs

Tested chargers are divided into four layers:

• Countries
• Charging Parks: (A set of multiple chargers deployed in the same location by the same operator)
• Chargers: Boxes inside of a park.
• Plugs: The individual plugs. (Usually 1 or 2 per charger)

Each device is given a unique ID for dataset management in the format: <country code>.<park ID>.<charger ID>.<plug ID>. The code heavily assumes that the below file/folder naming conventions are used, and that none of the ID components contain a . or _.

Data format

Data is stored in the data/ folder, the contents of this are provided as a separate Zip file on Zenodo, and must be extract to this location without additional nested data directories.

High level information about the parks, charger and plugs (location, manufacturer, serial number) is stored in three metadata CSV files in data/metadata/.

Experimental data for each plug is stored in data/chargers/<country code>/<park ID>/<charger ID>_<plug ID>/.

Each experiment has a folder indexed by a UTC timestamp YYYY_MM_DD_hh_mm_ss inside the plug folder. For each experiment there is a result.json file from the measurement tool, as well as a backup.bak.txt that is flushed to disk line by line in case the data collector crashes. The analysis scripts can automatically re-generate a damaged result file from the backup. Both of these files contain a full record of every action and command the test tool executed against the charger. For each charger a pcap.pcap file is also recorded of the test.

During processing, the user is prompted to inspect NMKs for visible patterns. The results of this are cached in nmk_review.json for each plug.

Processed data of each experiment is saved to an overview.json file for each plug, listing the final conclusions. In some cases where an old version of the test tool was used, we only have these overview files.

Photos are stored in data/photos, named based on UTC timestamp, and split into folders by day. The metadata files specify which photos belong to which devices.

According to our dataset anonymity discussed in the paper, metadata, photos, and overview.json files are provided, but the raw experiments are not.

Code files

The code was written in Python 3.9 on Windows 10. It has also been tested in Python 3.12 on Kubuntu 24.04.1 LTS, the remaining instructions are for this version. Only standard packages are used, the code should be portable to most platforms and future versions.

To set up, download the repository, and execute

./install.sh
source venv/bin/activate

This installs the necessary system packages, downloads our dataset, creates a Python venv in the folder venv, and installs the necessary packages.

The proc_code/ folder contains a small python library for loading, managing and analysing data in accordance with the above format.

Inside this, the proc_code/webserver/ folder contains a small webserver that allows interactive management of the data. To launch the server, in the folder of this md file run python3 -m proc_code.webserver Open http://localhost:8000 for a static view or http://localhost:8000?edit to manage the data.

Warning: Do not edit any data file or use any other script while the webserver is running. On exit (SIGINT) it will save the data back to disk, overwriting any external changes.

To build a safe to view static website from the data, run python3 -m proc_code.build_static. This will start the webserver, wget each page, export a static HTML website into the data/ folder. This can then be hosted using any static webserver, such as python3 -m http.server. For simplicity, .sh and .bat files are provided in data/ to help with this.

Adding data

To add new experiments, the following steps are necessary:

• Create metadata entries for each new park, charger and plug. This can be done by editing the CSV or using the webserver. Do not do both at the same time.
• Add the experimental folders from the data collector to the correct folders for each plug
• Run python3 -m proc_code.process_data in the root folder. This will interactively prompt about the NMKs for the new devices, for a human to review whether they appear to have a pattern.

Running analysis

The plots.ipynb file can be used to generate statistics and plots from the data. This can be ran using only the dataset we publish. To run this in a non-interactive way, run python3 -m jupyter execute plots.ipynb.