Detecting EDoS Attacks in Cloud Environments Using Machine Learning and Metaheuristic Algorithms

Cloud computing has emerged as the most cost-effective method for delivering public and private IT services. Virtualization and Network Function Virtualization are key enabling technologies that enhance the efficiency and cost-effectiveness of cloud services. However, these advancements also bring forth new challenges, including the threat of Economic Denial of Sustainability (EDoS) attacks. EDoS uses pay-per-use services to gradually increase resource utilization over time without the cloud customer’s knowledge, ultimately imposing steep fees to keep the service operational. By analyzing network traffic patterns, Machine learning (ML) algorithms, such as Random Forest (RF), have been employed for detecting EDoS attacks and enabling proactive defense measures. However, the process of finding the optimal hyperparameters values can be challenging and computationally demanding. In this work, we investigate the use of metaheuristics from Nature-inspired algorithms to find the most suitable hyperparameters for the RF algorithm. The proposed approach has shown promising results in enhancing the performance, reducing computational time and improving accuracy of the RF algorithm in the detection of EDoS attacks.