An Efficient DDoS Attack Detection Method Using Optimized Long Short-Term Memory based on an Enhanced Salp Swarm Optimization Approach

P. Sakthivel, Dr. A. Manikandan

Abstract: DDoS attacks are currently rapidly evolving and represent a significant challenge to the stability and security of contemporary network infrastructures. Higher-dimensional traffic data, optimization inefficiencies, class imbalance, and slow convergence are the primary causes of performance degradation of existing deep learning-based Intrusion Detection Systems (IDS), which leads to high false alarm rates and low generalization. In order to address these shortcomings, this study offers a new Long Short-Term Memory model that is optimized by a Centroid Opposition-Based Learning and an Enhanced Simplex Search Salp Swarm Algorithm (CSSSA-LSTM). The integrated CSSSA improves search capability on a global scale, converges faster, and avoids premature stagnation through the combination of the exploratory opposition learning with the exploitative simplex refinement. Four standard IDS datasets, such as NSL-KDD, CICIDS2017, CICIDS2019, and UNSW-NB15, are experimentally validated following powerful preprocessing procedures, including normalization, feature encoding, imbalance management, and dimensionality reduction. The proposed model has been compared with LSTM, GA-LSTM, PSO-LSTM, SSA-LSTM, and COSSA-LSTM; the model has a superior performance with a maximum accuracy of 98.76, better F-score, sensitivity, specificity, and AUC, and faster convergence in terms of minimizing the MSE. As the results confirm, CSSSA-LSTM is an effective tool in terms of improving the ability of detecting attacks and minimizing the number of false positives as well as providing excellent stability in both binary and multi-class DDoS situations.

Keywords: DDoS Detection, Intrusion Detection System, Deep Learning, Long Short-Term Memory, Salp Swarm Algorithm, Centroid Opposition-Based Learning, Simplex Search, Swarm Intelligence Optimization, Network Security, Cyber-attack Detection

How to Cite?: P. Sakthivel, Dr. A. Manikandan, "An Efficient DDoS Attack Detection Method Using Optimized Long Short-Term Memory based on an Enhanced Salp Swarm Optimization Approach", Volume 15 Issue 3, March 2026, International Journal of Science and Research (IJSR), Pages: 125-134, https://www.ijsr.net/getabstract.php?paperid=SR26225144453, DOI: https://dx.dx.doi.org/10.21275/SR26225144453

Download Citation: APA | MLA | BibTeX | EndNote | RefMan