Machine-Learning Anomaly Detection for Early Identification of DDoS in Smart Home IoT Devices

作者：Roland Lamptey, Mohammad Saedi, Vladimir Stanković · 年份：2025 · DOI：10.1109/csr64739.2025.11130003 · 被引用次数：4 · 研究领域：Network Security and Intrusion Detection

The rapid adoption of Internet of Things (IoT) devices in smart homes has introduced security vulnerabilities, with Distributed Denial of Service (DDoS) emerging as a critical threat. Exploiting the often-unsecured nature of these interconnected devices, such attacks overwhelm network resources, causing severe disruptions and privacy breaches. We present a novel anomaly detection system for early-stage DDoS attack identification in smart home IoT environments. Using NS-3 simulator, a realistic IoT network dataset was generated, capturing normal and malicious traffic. Key traffic features, e.g., packet size and inter-arrival times, were extracted to train two lightweight Machine Learning (ML) models: One-Class Support Vector Machine (OCSVM) and Isolation Forest (IF). OCSVM model achieved superior performance with accuracies from 96% to $99 \%$ for various attacks, while the IF model performed marginally worse. We offer a lightweight and scalable solution for real-time deployment in resource-constrained IoT environments, a significant step to enhance smart home security.