"QF-LOG: Quantum Forensic Dataset for QKD Networks"

"This dataset is a set of logs from a quantum key distribution network, made to help detect attacks using normal system records. It has 50,000 labeled entries, and each entry includes four values like error rate, photon count, delay, and whether the system stopped the session. The data is split into four equal parts with 12,500 samples each: normal behavior, partial intercept resend, detector blinding, and fiber-related issues. Some values are meant to overlap so the task feels real and not too easy. The delay numbers describe processing and buffering inside the controller, not the time light takes to travel through the fiber. It also comes with Python scripts that can recreate the data, train models, test them with 5-fold validation, and remake charts like PCA and feature importance."