Seagull-optimized Deep Convolutional Neural Network for Detecting Distributed Denial of Service Attack in Cloud Computing

Due to the enormous benefits and applications of cloud computing, there is a lot of interest from researchers on cloud computing. The most significant security risk associated with cloud computing is insider distributed denial of service (DDoS), which affects the entire deactivation of services due to its dispersed architecture and utter dependency on the internet for service supply. Machine learning and Deep learning have appeared as a potent tool in cyber security recently, and numerous attack detection methods have been put out in recent studies. Nevertheless, existing studies on attack detection encounter several challenges, including data imbalance, interpretability issues, inaccurate detection, overfitting, and time complexity issues. Therefore, in order to overcome these difficulties, a deep learning model, Seagull optimized Synthetic Minority Over-sampling Technique enabled Deep Convolutional Neural Network (SgO-SCNN) is developed in this research to detect DDoS attacks. The optimized SMOTE employed in the model addresses the imbalanced data issue and improves the model’s performance by effectively rebalancing the data. The DCNN classifier can detect the attacks more efficiently and reduces the computation of the model. The seagull optimization tunes the DCNN classifier’s parameters and improves the model’s performance in attack detection with fast convergence, which reduces the model’s time complexity. The experimental outcomes indicate that the developed $\mathrm{SgO}-\mathrm{SCNN}$ model outperforms the existing approaches by attaining enhanced values of accuracy 96.05%, F1-score 94.66%, precision 94.84%, and recall 94.48%.