A Novel Framework for DNS Exfiltration Malware—Optimizing Neural Network Hyperparameters Using Swarm Intelligent Algorithm

Domain name service (DNS) is a well-liked approach that steals confidential data from corporate organizations and keeps a hidden channel open for interactions between a hostile website and control/command servers. Given the importance of DNS services, organizations frequently configure their gateways to allow DNS requests that enable attackers to transit encrypted messages to a hacked host under their influence. This research work proposes deep learning-based neural networks in which data are trained to identify low and slow data intrusions and tunneling through DNS. Moreover, the authors used Swarm Intelligent Algorithm, the most computational intelligent technique appropriate to resolve in identification of DNS malware from overall data samples. Our experimental results show that threats were blocked with detection rate around 99.9%, and loss accuracy is very less. Our work proves unequivocally that the flow of network system will not be impacted, also no lagging in performing operation in the system.