Intrusion Detection Framework for IoT-Based Smart Environments Security

Abstract

Since of huge progress of Internet of things (IoT) and networking technologies, and the expanding number of devices linked to the Internet, security and privacy problems must be addressed in order to secure hardware and information networks. Real-time monitoring of network resources and information is required to guarantee security. Intrusion detection systems have been utilized to continuously monitor, detect, and notify on an intrusion incident. Indeed, intrusion detection systems (IDSs) are a powerful cybersecurity tool that can be improved with machine learning (ML) and deep learning (DP) algorithms. These are intended to improve standard quality criteria like as accuracy (ACC), recall, and precision, but they hardly take time processor performance into effect. A novel IDS for IoT-based smart environments that utilizes DL and Supervised ML, is presented in this paper. The framework typically offered an optimum anomaly detection model that combines deep extraction based on the stacked autoencoder and combining feature selection using Information gain (IG) and Genetic algorithms (GA) (DEIGASe) which employs Multi-layer Perceptron (MLP) and support vector machine (SVM), K-nearest neighbors (K-NN) for classification. The BoT-IoT dataset was used to validate the proposed model metrics when compared to previous IDS, the results show that the proposed approach produces excellent Accuracy (ACC), Recall, Roc-auc and F1-score performance metrics.