Dynamic Economic-Denial-of-Sustainability (EDoS) Detection in SDN-based Cloud

Abstract

In Cloud Computing, a new type of attack, called Economic Denial of Sustainability (EDoS) attack, exploits the pay-per-use model to scale up the resource usage over time to the extent that the cloud user has to pay for the unexpected usage charge. To prevent EDoS attacks, we propose an efficient solution in the SDN-based cloud computing environment. In this paper, we first apply an unsupervised learning approach called Long Short-Term Memory (LSTM), which is a multivariate time series anomaly detection, to detect EDoS attacks. Its key idea is to try to predict values of the resource usage of a cloud consumer (CPU load, memory usage and etc). Furthermore, unlike other existing proposals using a predefined threshold to classify the anomalies which generate high rate errors, in this work, we utilize a dynamic error threshold which delivers much better performance. Through practical experiments, the proposed EDoS attack defender is proven to outperform existing mechanisms for EDoS attack detection. Furthermore, it also outperforms some of the machine-learning-based methods, which we conducted the experiment ourselves. The comprehensive experiments conducted with various EDoS attack levels prove that the proposed mechanism is an effective, innovative approach to defense EDoS attacks in the SDN-based cloud.