Detecting Iot Botnet Attacks Using Deep Learning Approach

Résumé: Recently, Network Intrusion Detection (NIDS) has emerged as an essential part of cyber security, with a focus on protection Internet of Things (IoT) systems from unauthorized use and attacks. Nevertheless, NIDS techniques are facing several challenges due to the emergence of new numerous sophisticated attacks. Using Deep Learning approaches, the main objective of this research is to automatically learn useful feature representations, extracted from large amounts of unlabeled raw network traffic data. Therefore, we propose an effective Network Intrusion Detection System STL-NIDS based on the self-taught learning (STL) framework. The proposed model is build using the deep AutoEncoder neural network, which is an effective learning technique for reconstructing a new feature representation in an unsupervised manner.After the pre-training stage, the new features are fed into the SVM algorithm to improve its detection capability for intrusion and classification accuracy. The proposed method is validated experimentally and a comparison between competitive methods using IoT Bot dataset is conducted. Hence, results of the evaluation using several appropriate metrics show that the proposed method outperforms the related works’ results. Moreover, it is proven that our model is effective for large-scale and real-world network environments.

Publié dans la revue: المجلة الجزائرية للعلوم