Training Dataset for Industrial TSN Traffic Regulation

Training dataset used to learn IEEE 802.1Qcr Asynchronous Traffic Shaper (ATS) parameters Committed Information Rate (CIR) and Committed Burst Size (CBS) for cloud-edge industrial Time-Sensitive Networking (TSN). The dataset contains 600 labeled samples, each describing one synthetic traffic scenario (stream-level frame size, frame count, inter-arrival interval, burstiness, and maximum residence time) paired with the optimal (CIR, CBS) pair found for it via exhaustive grid search under a composite delay/drop/resource-usage objective. It is used to train ensemble regression models (Random Forest, Histogram-based Gradient Boosting, and XGBoost) that predict ATS shaping parameters directly from traffic descriptors, avoiding costly online grid search at deployment time. Full field-by-field schema documentation and the dataset construction procedure are included in the accompanying data/README.md file within the uploaded archive. This dataset accompanies the paper "Machine Learning-Driven Adaptive Deterministic Traffic Regulation Strategy in Cloud-Edge Industrial TSN" (A. Boulahdour, M. Bagaa, F. Ahmed Ouameur, S. Bertrand, A. Ksentini, D. Massicotte), submitted to IEEE Transactions on Cognitive Communications and Networking (TCCN), 2026. A DOI/reference to the published paper will be added once available.