PErformance Evaluation of Detection Mechanisms Towards the Slow Ddos Attack of Edge Computing
X. Lee, Yu‐Beng Leau · 2023
Edge Computing is a network topology made up of three layers: Cloud Server Layer (CSL), Edge Server Layer (ESL), and Edge Device Layer (EDL). Therefore, it is vulnerable since the processing capacity of loT devices or mobile devices is insufficient to withstand and they are unable to implement high- level security measures such as using robust security protocols like HTTP/HTTPS, FTP, or SMTP at this layer. Hence, it poses considerable security risks, particularly the Slow Distributed Denial of Service (DDoS) attacks. Various detection methods have been investigated in this study, and three machine learning algorithms that are classified as anomaly-based detection techniques have been chosen, such as CDAAE (Conditional Denoising Adversarial AutoEncoder), CNN (Convolutional Neural Network), and deep learning using ‘relu’ and ‘softmax’ in detecting slow DDoS attacks. These models are evaluated with the CICIDS2017 dataset and show that CNN has achieved better overall performance with shorter training times and high detection accuracy.