Hybrid Ensemble based Feature Engineering for Detecting Direct DDOS Flooding Attack in Cloud

In recent years, both the general public and commercial enterprises have grown more and more interested in using cloud services. The majority of businesses use cloud computing technologies for production operations, which draws hackers. Nowadays, one of the most popular methods used to obstruct the availability of essential internet services is Distributed Denial of Service (DDoS) floods. These attacks either totally destroy the victim or overload it with a massive amount of traffic that prevents it from carrying out normal communication. The cloud's services are fully suspended if there are any delays in identifying flooding attacks. A preprocessing stage in cloud DDoS attack defence known as feature engineering has been recognized as having the potential to improve classification accuracy and lower computing complexity. In this article, we suggest a DDoS intrusion detection system for use in cloud environments. The application of filter and wrapper-based feature selection techniques along with machine learning is proposed as a hybrid ensemble-based feature engineering strategy. The benchmark dataset, which fills in the gaps in the current datasets and comprises a wide range of direct DDoS flooding attacks, is used to assess the model. To prevent data overfitting issues, area-user-curve analysis is also assessed. When compared to previous benchmarking techniques, the evaluation and findings revealed a considerable improvement in attack detection. This framework offers a high detection rate and classification accuracy when contrasted with the existing framework. Hence, it is more suitable for protecting the cloud.