A Deep Learning Based Approach To Detect Covert Channels Attacks and Anomaly In New Generation Internet Protocol IPv6

Abstract

The increased dependence of internet-based technologies in all facets of life challenges the government and policymakers with the need for effective shield mechanism against passive and active violations. Following up with the Qatar national vision 2030 activities and its goals for “Achieving Security, stability and maintaining public safety” objectives, the present paper aims to propose a model for safeguarding the information and monitor internet communications effectively. The current study utilizes a deep learning based approach for detecting malicious communications in the network traffic. Considering the efficiency of deep learning in data analysis and classification, a convolutional neural network model was proposed. The suggested model is equipped for detecting attacks in IPv6. The performance of the proposed detection algorithm was validated using a number of datasets, including a newly created dataset. The performance of the model was evaluated for covert channel, DDoS attacks detection in IPv6 and for anomaly detection. The performance assessment produced an accuracy of 100%, 85% and 98% for covert channel detection, DDoS detection and anomaly detection respectively. The project put forward a novel approach for detecting suspicious communications in the network traffic.