2023  |  Vivek Nair · Wenbo Guo · James F. O’Brien · Louis Rosenberg · Dawn Song  |  doi.org/10.48550/arXiv.2311.05090

Virtual reality (VR) and "metaverse" systems have recently seen a resurgence in interest and investment as major technology companies continue to enter the space. However, recent studies have demonstrated that the motion tracking "telemetry" data used by nearly all VR applications is as uniquely identifiable as a fingerprint scan, raising significant privacy concerns surrounding metaverse technologies. Although previous attempts have been made to anonymize VR motion data, we present in this paper a state-of-the-art VR identification model that can convincingly bypass known defensive countermeasures. We then propose a new "deep motion masking" approach that scalably facilitates the real-time anonymization of VR telemetry data. Through a large-scale user study (N=182), we demonstrate that our method is significantly more usable and private than existing VR anonymity systems.

Model Architecture

Our method involves decomposing the plausible variance of human motion sequences into action-related variance and user-related variance. For this purpose, we train an "action encoder" model, which learns an embedding for the action a user is taking while ignoring the user's identity, and a "user encoder" model, which learns an embedding for the user's identity while ignoring the action they are taking. We then train an "anonymizer" model that anonymizes motion sequences by changing their user embedding without changing their action embedding.


Vivek Nair

Wenbo Guo

Louis Rosenberg

James F. O'Brien

Dawn Song

Copyright ©2022–2023 UC Regents  |  Email us at rdi@berkeley.edu.