Deep learning-based anomaly detection method for enhancing IoT security

The Internet of Things (IoT) is developing very rapidly. It makes communication between devices much easier, greatly improving many fields, such as smart homes, healthcare, and industrial automation. IoT networks are becoming increasingly popular, but security risks are also growing. This is because IoT devices do not have much computing power and are connected to unsecured networks. Signature-based detection and other older security methods do not work well in IoT environments where attack strategies constantly change and become increasingly complex. This paper proposes a more advanced anomaly-finding method using a hybrid deep learning model that combines a temporal convolutional network (TCN) and a gated recurrent unit (GRU). The model uses TCN to extract features that show long-range temporal dependencies from network traffic data. These features are fed into the GRU for sequential pattern learning, improving detection accuracy and speed. The proposed TCN-GRU model does a better job of discovering network attacks than traditional machine learning models and other deep learning methods, with high accuracy (99.23%), recall (99.46%), and F1 score (99.18%). The results show that the TCN-GRU model works well in the real-time discovery of intrusions in resource-limited IoT situations providing a robust and scalable security solution.