AFL-IDS: Reproducibility Software and Experimental Artifacts for Resource-Aware Federated Intrusion Detection

This record contains version 1.0.0 of AFL-IDS, the reproducibility software and experimental artifacts supporting the manuscript “AFL-IDS: Resource-Aware Federated Intrusion Detection for IoT Edge Computing.” The package includes the Python source code, software environment specification, preprocessing and training scripts, seed-specific train/test indices, non-IID client partitions, simulated resource traces, per-round logs, trial-level performance metrics, saved test labels and prediction probabilities, and scripts used to generate the manuscript tables and figures. AFL-IDS evaluates a threshold-based CPU–memory–battery client-selection strategy for federated binary intrusion detection using the TON_IoT network dataset. The original TON_IoT dataset is not redistributed in this record. Users must obtain the source dataset from UNSW Canberra and cite the applicable TON_IoT publications. Version 1.0.0 corresponds to the software and experimental configuration used for the PeerJ Computer Science submission.