Preserving Privacy in VR Telemetry Data

Telemetry data is essential for optimizing VR systems and poses privacy risks due to re-identification vulnerabilities. Our study introduces a privacy-preserving framework using Wasserstein GANs with Gradient Penalty to generate synthetic datasets that retain the statistical integrity of real data while mitigating re-identification risks. Differentially Private Stochastic Gradient Descent is used to further enhance privacy by introducing controlled noise during model training. Our approach balances data utility with user privacy, enabling secure analysis of VR telemetry data for performance optimization, user experience improvements and potentially paving the way for safer and more ethical VR data practices.