Detecting framerate-oriented cyber attacks on user experience in virtual reality

Virtual Reality (VR) is expected to become an enabling technology for training in realistic conditions, data visualisation, education and many other applications. However, there is still limited research on cyber threats to VR environments and even less on technical protections against them. We are currently developing a VR testbed specifically designed for assessing different cyber threats, their impact to user experience and corresponding defences. In this work in progress, we demonstrate two novel approaches by which a cyber attack can potentially cause VR sickness on demand based on frame rate manipulation by taking advantage of GPU and network vulnerabilities. We further show that a simple unsupervised machine learning method using Isolation Forest can provide early warning of such attacks likely before they have significant impact on the VR system and its user