Enhancing Attack Detection on IoT Devices Using Hybrid Deep Learning Model

At present, Internet of Things has become a widely used subject for research because of its applications in our modern world. With this growth, securing IoT devices against cyber threats becomes essential. It will be difficult to put security system for each and every IoT device in the network and update it to counter new threats. Machine learning (ML) can be utilized to detect the attack in the data generated from the IoT devices. However, existing security models limited in their ability to detect a wide range of attacks and often rely on outdated datasets for evaluation. This research introduces a deep learning (DL)-based security model for attack detection, combining CNN for local anomaly detection, and LSTM for global anomaly and temporal dependency analysis. The dataset comprises data from 20 malicious and three benign IoT devices (IoT-23 dataset). Results show that the CNN-LSTM model achieves 95.42% accuracy, and the LSTM-CNN model achieves 91.825% accuracy.