Artificial Intelligence-based Hidden Communications Detection in Wireless Networks

Ubiquitous wireless networks are becoming a reality in our everyday lives. In this scenario, vulnerabilities and cyber-attacks represent issues that must be efficiently fulfilled, to avoid corruption of the system ‘s security policies as well as theft of sensible data. This paper investigates about the use of artificial intelligence (AI) techniques to overcome security threats. We develop a hidden wireless communication by modulating the inter-arrival packet delays (i.e., a covert timing channel) and propose 4 deep learning (DL) frameworks to discriminate between legitimate and illegitimate traffic. Our results, after comparisons with existing conventional machine learning (ML) methods, demonstrates the flexibility and robustness of such approaches for detecting hidden communications in wireless networks.