Adaptive Cybersecurity Framework for DDoS Mitigation Using a Dual LSTM with Secretary Bird Optimization Approach

Distributed Denial of Service (DDoS) attacks are major threats to network infrastructure, causing substantial economic losses. Traditional detection systems use static rules or conventional Machine Learning (ML) and Deep Learning (DL) techniques, which struggle to change attack patterns, causing delay and false alarm. Henceforth, this work proposes an efficient methodology for detecting DDoS attacks using advanced data preprocessing and optimized DL techniques. In this work, preprocessing steps include data cleaning and handling missing values for removing noise and inconsistencies. Feature engineering employed for extracting and creating meaningful features. Scaling and normalization used for ensuring consistency and enhance model performance by analyzing the impact of all variables. For classification, Dual Long Short-Term Memory (LSTM) based optimized a Secretory Bird Optimized Algorithm (SBOA) is implemented, for ensures accurate predictions and parameter tuning. The use of SBOA improves performance by fine tuning the classifier parameters, thus enhancing accuracy and efficient DDoS attack detection. The performance metrics, evaluated using python software, show an accuracy of 99%, precision of 99%, recall of 98% and F 1 -score of 99%, demonstrating the models high effectiveness in mitigating DDoS attack.