PC-CAN69: A 69-Hour High-Fidelity Physical Consistency Dataset for In-Vehicle IDS

This dataset provides a large-scale, process-oriented benchmark for automotive cybersecurity research, specifically designed for physical consistency-based Intrusion Detection Systems (IDS). Collected from a 2019 Renault Clio (0.9 TCe, Euro 6) over a duration of 69 hours, the PC-CAN69 dataset contains approximately 15 million data points, representing a significant 50-fold increase in scale compared to early physical-consistency studies. Key Features: High-Dimensional Data: Includes 30+ synchronized physical parameters (e.g., fuel flow, manifold pressure, engine torque) decoded from raw CAN traffic. Manual Transmission Support: Provides estimated gear positions derived via a novel RANSAC-based regression and K-Nearest Neighbors (KNN) classification methodology. Context-Aware Benchmarking: Designed to support the detection of sophisticated "semantic" attacks (e.g., masquerade, replay) by verifying the mechanical relationship between cyber-data and physical vehicle states. Technical Details: Vehicle: 2019 Renault Clio IV (0.9L TCe engine). Sampling Rate: Systematic polling via OBD-II with high temporal resolution. Format: Comma-Separated Values (CSV). This dataset is released in conjunction with the paper submitted to CETA 2026. Please cite the original paper when using this data in your research.