Distributed Attack Detection Framework in Cloud Environment using Hybrid Activation-based BiLSTM with Virus Colony Optimization

One of the key advantages of cloud computing is its ability to swiftly roll out services and applications. Cyberattacks within cloud storage have dramatically increased since smart devices are becoming more and more popular and the many security flaws in networks. Modern society is becoming increasingly dependent on information and communication technology, which is increased its susceptibility to many types of cyberattacks. Distributed denial-of-service (DDoS) assaults are difficult to halt by blocking a single source since they overwhelm victims with traffic from several sources. Web servers, network equipment, and application servers are the targets of these attacks. In this research, a Hybrid Activation-based Bidirectional Long Short-Term Memory (HABiLSTM) model improved by Virus Colony Optimization (VCO) is employed to create a framework for distributed attack detection in cloud environments. This system improves the efficacy and precision of cyber threat detection, hence mitigating the security threats associated with cloud computing. It uses K-Nearest Neighbors (KNN) to handle missing values and Generative Adversarial Networks (GANs) to balance data, increasing the accuracy of threat identification and reducing cloud security issues. The results indicate that the HA-BiLSTMVCO model performs better than current approaches across a range of criteria. The HABiLSTM- VCO model effectively uses the BoT-IoT and UNSW datasets to show enhanced performance metrics. Regarding the BoT-IoT data, it attains 97.35% accuracy, 97.52% specificity, and 97.19% sensitivity. On the UNSW dataset, it also demonstrates 97.08% sensitivity, 97.80% specificity, and 97.44% accuracy. Overall, by offering a strong defense against dispersed attacks and advancing the continuous development of cloud security frameworks, this proposed technique greatly improves security measures in cloud environments.