Hypervisor-based Double Extortion Ransomware Detection Method Using Kitsune Network Features

Abstract

Double extortion ransomware attacks have become mainstream since many organizations adopt more robust and resilient data backup strategies against conventional crypto-ransomware. This paper presents detailed attack stages, tactics, procedures, and tools used in the double extortion ransomware attacks. We then present a novel detection method using low-level storage and memory behavioral features and network traffic features obtained from a thin hypervisor to establish a defense-in-depth strategy for when attackers compromise OS-level protection. We employed the lightweight \emph{Kitsune} Network Intrusion Detection System (NIDS)'s network feature to detect the data exfiltration phase in double extortion ransomware attacks. Our experimental results showed that the presented method improved by 0.166 in the macro F score of the data exfiltration phase detection rate. Lastly, we discuss the limitations of the presented method and future work.