Anomaly-Based Detection of Distributed Denial of Service Attacks in Internet of Things Using Deep Autoencoder
Author(s)
Abstract
The rapid expansion of the Internet of Things (IoT) has introduced unprecedented levels of connectivity and automation, but it has also increased susceptibility to cyber threats, particularly distributed denial of service (DDoS) attacks. Traditional detection systems often struggle to adapt to the dynamic and resource-constrained nature of IoT environments. This research investigates the application of deep learning (DL) -based unsupervised learning methods, such as various autoencoder (AE) architectures, for effective anomaly detection of DDoS attacks in IoT networks. We evaluate four autoencoder variants, such as conventional AE, denoising AE (DAE), sparse AE (SAE), and variational AE (VAE), to model and reconstruct normal traffic behaviour. Any significant deviation, measured via reconstruction error, is flagged as a potential anomaly. Experimental results on benchmark IoT datasets demonstrate that VAE achieves the highest detection performance, with improved precision and robustness in identifying diverse DDoS patterns. The finding underscores the effectiveness of deep AE for unsupervised DDoS and offers valuable insights for developing lightweight, scalable, and adaptive security frameworks for IoT systems.
Keywords