Grey Wolf Optimizer-Enhanced CNN for DDoS Detection in ۵G D۲D Networks

The proliferation of ۵G networks with Device-to-Device (D۲D) communication introduces new security challenges, as resource-constrained devices without centralized oversight become targets for Distributed Denial-of-Service (DDoS) attacks. This paper proposes a novel DDoS detection approach that combines the Grey Wolf Optimizer (GWO) algorithm with Convolutional Neural Networks (CNN) to protect ۵G D۲D communications. GWO is employed to select an optimal subset of features from network traffic data, improving detection accuracy and reducing false alarms while minimizing computational overhead. Using the CICDDoS۲۰۱۹ benchmark dataset of DDoS traffic, we train a CNN on GWO-selected features to distinguish benign and attack flows. Experimental results demonstrate that the proposed GWO-CNN model achieves a DDoS detection accuracy of ۹۹.۰۷%, outperforming traditional machine learning models (Random Forest, XGBoost, AdaBoost) on the same dataset. The hybrid approach effectively identifies complex and emerging DDoS attacks in ۵G D۲D scenarios and highlights the potential of integrating meta-heuristic optimization with deep learning for robust, scalable network defense.

کلیدواژه ها:

DDoS attacks ، ۵G networks ، attacker signature detection ، machine learning ، deep learning ، Grey Wolf Optimizer (GWO) algorithm ، Convolutional Neural Networks (CNNs)