Countering DNS Vulnerability to Attacks Using Ensemble Learning
Love Allen Chijioke Ahakonye, Cosmas Ifeanyi Nwakanma, Simeon Okechukwu Ajakwe, Jae Min Lee, Dong‐Seong Kim · 2022
The Domain Name System (DNS) is the hub of the cyberspace and communications services which also plays enabling role in the Industrial Internet of Things (IIoT) and transmission at large. DNS enciphering in HyperText Transfer Protocol Secure (HTTPS) as DoH did not eliminate vulnerability and intrusion into critical systems. This study proposed a time-efficient Ensemble Learning (EL) model for countering DNS vulnerability to attack. The proposed EL candidate incorporates feature selection capability in extracting relevant features for enhanced model optimization. The simulation results showed that the proposed EL candidate effectively mitigates vulnera-bility, classifying DNS traffic into Non-DoH, Malicious DoH and Benign-DoH. The proposed model outperforms other compared state-of-art EL techniques with a combined advantage of accuracy and training time of 99.5% and 13.96s.