ISO 15118-2 EV Charging Traffic Dataset with 734 Labeled Benign and Attack Sessions

This dataset contains 734 labeled ISO 15118-2 network traffic sessions generated in a controlled RISE-V2G v1.2.6 testbed with EVCC, SECC, and attacker virtual machines. It covers benign charging communication and six attack classes: replay, connection-oriented denial of service, NDP-based man-in-the-middle, EXI fuzzing, certificate spoofing, and false data injection. The package includes one anonymized PCAP file per session, a CSV file with 19 session-level traffic features, per-file metadata identifying source-session groups and variants, feature-extraction and model-evaluation scripts, software requirements, and a SHA-256 checksum manifest. The evaluation script supports group-aware five-fold cross-validation and a group-preserving holdout based on session generation order. The data were generated on an isolated laboratory network without physical vehicles or charging equipment. They are intended for reproducible research on intrusion detection in EV charging communication.