Lightweight detection system for low-rate DDoS attack on software-defined-IoT

Low-rate Distributed Denial-of-Service (LR-DDoS) attack is an uninterrupted cybersecurity challenge within a Software- Defined Internet of Things (SD-IoT) network. If successful, LR-DDoS can severely impact the system availability and overall performance; therefore, this paper presents a machine learning model to detect LR-DDoS cyberattacks over SD-IoT. Three machine learning techniques, namely, J48 decision tree, Naive Bayes (NB), and Logistic regression, were implemented and evaluated based on recent network traffic with predefined attack scenarios. Our empirical results confirmed that detection-based J48 decision trees outperformed other ML-based models, scoring a high detection accuracy of 99.9%.