Design of a Novel Network Intrusion Detection Technique for SDN-based IoT Network Using Machine Learning

Abstract— The exponential expansion of Internet-connected smart environment devices, particularly in the IoT domain, has made software-defined network (SDN) security a major concern. This paper introduces a novel intrusion detection system (IDS) that combines machine learning (ML) and deep learning (DL) techniques, optimized for SDN-based IoT networks. The recommended strategy emphasizes comprehensive data preprocessing—including feature transformation, normalization, and correlation-based feature selection—to improve detection precision and efficacy. Three deep learning approaches (LSTM, RNN, and GRU) and four machine learning classifiers (logistic regression, naive Bayes, decision tree, and XGBoost) are assessed using two standard datasets, NSL-KDD and CIC-IDS-2017. Experimental findings indicate that XGBoost attains superior results compared to other ML classifiers on NSL-KDD, achieving an F1-score of 0.9909, while LSTM outperforms other models on CIC-IDS-2017 with an F1-score of 0.9991 and on NSL-KDD with an F1-score of 0.9927. Additionally, a comparative analysis demonstrates that the preprocessing pipeline improves accuracy by up to 15\(\%\). These findings highlight the potential of combining DL methods with SDN architectures to develop scalable and dependable intrusion detection solutions for next-generation networks.