The datasets below were collected from CAN bus of a real car.
Some of the datasets seem very similar, but they have differences in collected time, car status (idling or driving), injected attack types, and car models. For example, Car Hacking: Attack & Defense Challenge 2020 and Car-Hacking Dataset have the following differences:
The former includes replay attacks which are not in the latter.
Rest of the attack types (flooding=DoS, spoofing, fuzzing) are same, but they have different injection frequencies and also different CAN IDs of spoofing messages.
They were collected at different time from different cars. The former was collected from Hyundai Avante CN7 while the latter was collected from Hyundai YF Sonata. This means their normal traffic differs (e.g., the number of unique CAN IDs, payload values) due to the cars' different functionalities.
Please refer to the dataset descriptions and the relevant publications for more details.
On the other hand, the datasets below were collected from our Automotive Ethernet testbed.