Abstract

In recent years, the use of Internet of Things (IoT) devices has increased rapidly, which brings challenges related to security, computational complexity and power consumption. Traditional intrusion detection systems in cyber physical systems often fail to provide efficient and scalable solutions for resource-constrained environments. This paper proposes an intrusion detection system (IDS) based on a hybrid model (Autoencoder and Random Forest), and this hybrid model is energy efficient. The Autoencoder is used for feature reduction to minimize computational complexity, while the Random Forest classifier is employed for accurate threat detection. The hybrid model is trained using IoT datasets and achieves a high detection accuracy of 97.73% with an inference time of 0.1458 seconds for 20,000 records. Experimental results show that our hybrid model reduces computational cost and improves energy efficiency while maintaining detection performance. Therefore, the proposed model is suitable for intrusion detection in resource-constrained IoT environments.