Enhancing DDoS Detection in SDNs: Integrating AFSOA with BiLSTM for Real-Time Threat Management

Software-defined networking (SDN) has emerged as a popular solution to the inherent difficulties of traditional dispersed networks. The primary focus of this work is the early identification of anomalous assaults. We investigate the efficacy of combining a Bidirectional Long Short-Term Memory (BiLSTM) network with an Artificial Fish Swarm Optimization Algorithm (AFSOA) to detect Distributed Denial of Service (DDoS) attacks in an SDN environment. For dimensionality reduction, we combine information gain (IG) with principal component analysis (PCA), allowing for a more effective and economic analysis of complex, high-dimensional datasets. The AFSOA optimizes the BiLSTM model's parameters to improve its ability to detect patterns in network traffic indicative of DDoS attacks. The proposed solution leverages the programmable and dynamic nature of SDNs to build and test the detection system, enabling adaptive threat management in accuracy of 95.20%, precision of 97.64%, recall of 98.20%, and an F-measure of 98.67%, significantly outperforming existing methods. This approach opens up new possibilities for enhancing network security in the future.