Improving DDoS detection in cloud computing through deep learning and feature fusion with TWBO optimization

Abstract A Distributed Denial of Service (DDoS) attack is considered a subclass of Denial of Service (DoS) that causes the severity of the attack in the cloud environment, and the actual service of the network is disturbed by several malicious attempts. One of the general forms of attack is a DDoS attack, which leads to substantial harm and negatively impacts the efficiency of cloud services. Hence, the hybrid Sequential Convolutional Neural Network with Taylor’s Wolf Bird Optimization (SCNN-TWBO) is introduced. The cloud is simulated initially, where the attacker stops the valid users from accessing a specific network resource through a DDoS attack. Here, recorded log files are sourced and normalized by the Max-Min Normalization algorithm. Then, the features are fused by Spearman’s Rank Correlation Coefficient (SRCC) and Deep Neural Network (DNN). Next, fused features are augmented, and the DDoS attack detection is done by SCNN-TWBO. SCNN is trained by the TWBO, which is the integration of Taylor Series and Wolf Bird Optimization (WBO). The superior analytical findings are recorded by the BoT-IoT dataset, while the newly developed SCNN-TWBO acquired 92.88% of accuracy, 94.88% of True Positive Rate (TPR), and 90.88% of True Negative Rate (TNR) for 90% of the learning data.