Internet of Things Botnet Detection via Ensemble Deep Neural Networks

Abstract

The widespread use of the Internet of Things (IoT) and the rapid increase in the number of devices connected to the network bring both benefits and many problems. The most important of these problems is cyber attacks. These cyber attacks cause financial losses as well as loss of reputation and time. Intrusion detection systems (IDS) and intrusion prevention systems (IPS) are used to eliminate or minimize these losses. IDS are designed to be signature-based or anomaly-based, and are currently being developed using anomaly-based systems as machine learning methods. The aim of this study is to detect whether there is an attack on your network, with a high success rate, by considering botnet as one of the attack types. In order to develop this system, it is aimed to use Ensemble Deep Neural Networks (DNN), which is one of the machine learning methods, and to search for solution methods for the most accurate result. In the study, N-BaIoT dataset in the UCI Machine Learning library was used for scientific research. The data consists of 1 benign network stream and 9 malicious network streams carried by 2 botnets. Stacked ensemble of DNN networks has been used from the classification stage. The proposed method has achieved %99 accuracy and the results are encouraging for future studies.