Deep Learning Utilization for DDoS Attack Detection with Federated Learning: A Case Study on the CICDDoS2019 Dataset

Modern systems are highly susceptible to Distributed Denial of Service (DDoS) attacks, which often cause major service interruptions and financial losses. These attacks have become more frequent and sophisticated due to the unexpected explosion in the use of wireless technologies, making them more challenging to detect with traditional methods. In this paper we investigate the effective detection of DDoS attacks using deep learning, specifically within a federated learning architecture. The CICDDoS2019 dataset was used to train and evaluate the proposed intrusion detection scheme. Federated learning trains models on several dispersed servers or devices while maintaining local data, thereby preserving user privacy and enabling collaborative learning across clients. The results demonstrate that Federated Deep Learning (FDL) outperforms conventional machine learning algorithms such as Random Forest (RF), Support Vector Machines (SVM), and Recurrent Neural Networks (RNNs), achieving 98% detection accuracy. In addition, the proposed method improves recall and reduces false positives, making it ideal for real-time intrusion detection in dynamic environments.