Deep Learning to Mitigate Economic Denial of Sustainability (EDoS) Attacks: Cloud Computing

Cloud computing, despite being a groundbreaking technology, has increased the risk of attacks that exploit cloud services, leading to the exhaustion of resource allocations. One such attack is Economic Denial of Sustainability (EDoS), which uses pay-per-use services to gradually increase resource demand, potentially leading to financial disaster for service provider. The Cyber Range Lab at the University of New South Wales (UNSW), located in Canberra, conducted nine injection attacks to test machine learning and deep learning algorithms. These experiments involved multi-classification, incorporating nine different types of attack data. Among the presented algorithms, the RNN-LSTM method achieved an accuracy of 84%. Various statistical analyses, such as mean square error (MSE), Pearson correlation coefficient (R), and root mean square error (RMSE), were used to evaluate prediction errors between input data and values generated by different machine learning and deep learning algorithms. Despite its relatively low prediction level (MSE=0.560), the Recurrent Neural Network Long Short Term Memory (RNNLSTM) algorithm achieved an R2 level of 82.09%when applied to the multi-classification dataset. The suggested system's performance was compared to existing EDoS attack detection systems, and the RNN-LSTM-based attack mitigation algorithms demonstrated superior performance. This study aims to contribute to the identification and effective protection of cloud-based resources that could be exploited by Distributed Denial-of-Service (DDoS) attacks, thereby preventing financial losses for potential victims.