Experimental Data and Results for AARDF: A Knowledge-Based Method and Formal Model for Adversarial Robustness Assessment and Defense Selection in Cross-Dataset IoT Intrusion Detection Systems

Experimental data, trained-model artefacts, and figures for “A Knowledge-Based Method and Formal Model (AARDF) for Adversarial Robustness Assessment and Defense Selection in Cross-Dataset IoT Intrusion Detection Systems.” Covers a 300-configuration adversarial robustness benchmark across four IoT intrusion detection datasets (CIC-IDS 2017, UNSW-NB15, Gotham IoT 2025, CIC-YNU-IoTMal 2026), four classifier architectures, and three adversarial attack methods, together with the results of the seven-part novelty package (weighted-MCDM comparison, Pareto-frontier analysis, second defense evaluation, rule-induction comparison, decision-margin/manipulation-resistance analysis, counterfactual explanations).