A Hybrid Ensemble Machine Learning Approach (EHML) for DDOS Attack Detection in Smart City Network Traffic

The Internet has had a profound effect on society over the last couple of decades. However, various cyberattacks, such as ransomware attacks, phishing attacks, DOS/DDOS attacks, etc., are occurring at a larger rate due to the Internet's phenomenal expansion. The Internet of Things (IoT) is a key component of today's smart city infrastructure. In essence, IoT technology creates a platform for service automation by connecting disparate things to the Internet’s backbone. Smart city infrastructure is vulnerable to cyberattacks due to security concerns with IoT networks. For instance, a Distributed Denial of Service (DDoS) attack breaches the prerequisites for permission in the infrastructure of smart cities. A DDOS attack, particularly one including IP spoofing, may generate a large number of packets in a short period of time, overwhelming the target's ability to process and communicate with the outside world. A multi-pronged approach, including detection, classification, and trace-back, is necessary for a successful defense against such attacks. Training models to identify and stop the attack before it does severe harm is a major goal of machine learning (ML). The purpose of this study is to provide an ensemble hybrid machine learning (EHML) method for the detection of DDoS attacks. Our proposed detection system utilizes distance calculation and ML detection techniques independently, and it combines the findings of the two indicators to improve the system identification accuracy. For this experimental research, we used the UNSW-NB15 Dataset. Our proposed EHML method is evaluated on the UNSW-NB15 dataset using the precision, recall, F-1 score, and accuracy metrics. Our suggested approach is compared to other ML algorithms utilized in the context of DDoS attacks. In this study, we compare the results of EHML to those of conventional ML algorithms and find that EHML is superior.