Abstract

As more IoT devices such as smart cameras and sensors are now being used in hospitals and smart cities for security management and critical infrastructures, it has become difficult to manage cybersecurity because Intrusion Detection Systems (IDS) are not very scalable and their resources are limited. In this model, we use an Advanced Neural Network (ANN) that includes LSTM (Long Short-Term Memory), which remembers old information for fast and efficient processing, and an Attention Mechanism that focuses on important data to analyze and detect hidden patterns in network traffic. IoT devices are smaller in size and have low storage capacity, so the model is designed according to their size to reduce unnecessary components, which results in running the model in real time while maintaining its accuracy. The accuracy of our model is 95 percent, with precision of 97.2 percent and F1-score of 95.36 percent using the Edge-IIoTset dataset. The main purpose of this framework is to secure IoT devices.