Abstract:

Abstract: The exponential growth of Internet of Things (IoT) networks has amplified the threat of Distributed Denial of Service (DDoS) attacks, which jeopardize the stability and functionality of these systems. Traditional centralized DDoS detection methods often fall short in addressing the scale and complexity of IoT networks, highlighting the need for decentralized approaches. This study explores the use of Federated Learning (FL) for decentralized DDoS detection in IoT environments, utilizing the CIC-IDS 2017 dataset. Several deep learning models, including Convolutional Neural Networks (CNN), CNN with Gated Recurrent Unit (GRU), and CNN with Long Short-Term Memory (LSTM), were employed to enhance detection capabilities. Additionally, an ensemble approach was implemented to combine the predictions of individual models, producing a more resilient and accurate final detection outcome. Results indicate that the CNN + LSTM ensemble model achieved the highest performance, showcasing its effectiveness in identifying DDoS attacks across decentralized IoT networks. This work demonstrates the potential of FL and ensemble learning to improve the robustness and scalability of DDoS detection in IoT.

Description:

.