IoT-Botnet Detection Using Deep Learning Techniques

Botnet attacks have become a major threat to the Internet of Things (IoT) over the past few years. Malicious operations, attacks, data loss, and network compromise are all possible outcomes of botnets. This paper presents a comparative study on advanced deep learning algorithms for IoT-Botnet attacks detection and classification. We explore the performances of four classifiers including autoencoders, CNN-LSTM, artificial neural network, and multi-layer perceptron for the detection of eleven attacks associated with the Mirai and Bashlite’s malware. We intend to thoroughly analyze and evaluate the DL techniques for multiclass classification problem to successfully identify attack types from 11attacks using the N-BaIoT dataset. The experimental results indicate that for multiclass classification, the multi-layer perceptron classifier outperforms other classifiers producing an accuracy of 0.91 with a precision of 0.90. Attacks associated with the Gafgyt malware exhibit the highest detection rate with an F1-score up to 1.00.