Hybrid Bio‐Inspired Combined Deep Learning Model for DDoS Attack Detection in Cloud: A Big Data Perspective

One of the most prevalent attacks that cause significant harm and impair cloud performance is Distributed Denial of Service (DDoS). DDoS attacks pose a significant threat to cloud environments, degrading performance and disrupting services. To address this issue, we propose a hybrid bio‐inspired deep learning model for DDoS attack detection that leverages big data analytics in the cloud. The proposed model incorporates a MapReduce framework to efficiently process large‐scale network traffic data, extracting crucial features such as raw features, packet‐based features, improved correlations, and statistical features. These extracted features are further refined using an improved recursive feature elimination (RFE) method, which selects the most relevant attributes for attack detection. The attack detection phase employs a hybrid classifier (HC) that integrates Long Short‐Term Memory (LSTM) and Deep MaxOut (DMO) models. To ensure optimal performance, the weights of LSTM and DMO are fine‐tuned using the White Shark Updated Remora Optimization (WSU‐ROA), enhancing classification accuracy. The proposed HC + WSU‐ROA model outperforms other methods, achieving the highest accuracy of 93.98%, compared to the other existing methods, demonstrating its superior effectiveness in DDoS attack detection.