Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimizing QoE-Privacy Tradeoff for Proactive VR Streaming

Xing Wei, Shengqian Han, Chenyang Yang, Chengjian Sun

TL;DR

This work tackles privacy in proactive VR streaming, where uploading predicted viewpoints and prediction errors risks leaking users' actual viewpoints. It first analyzes viewpoint leakage under existing privacy-preserving data processing and proves that leakage cannot be eliminated due to the PEA relation, yielding a positive lower bound Pr(A)min = ε/π. To overcome this, it introduces B-PEA, which breaks the PEA relation by adding noise n to the prediction errors, and derives the optimal noise strategy that minimizes QoE loss subject to Pr(A) ≤ q, even under worst-case distributions of the prediction error e. A trace-driven evaluation on real VR workloads shows that B-PEA satisfies privacy constraints for all users and substantially improves QoE compared with Gaussian or Laplace baselines, with negligible QoE loss when zero leakage is enforced. Overall, the paper provides a principled framework for zero-leakage privacy in proactive VR streaming and demonstrates practical gains in both privacy guarantees and user QoE.

Abstract

Proactive virtual reality (VR) streaming requires users to upload viewpoint-related information, raising significant privacy concerns. Existing strategies preserve privacy by introducing errors to viewpoints, which, however, compromises the quality of experience (QoE) of users. In this paper, we first delve into the analysis of the viewpoint leakage probability achieved by existing privacy-preserving approaches. We determine the optimal distribution of viewpoint errors that minimizes the viewpoint leakage probability. Our analyses show that existing approaches cannot fully eliminate viewpoint leakage. Then, we propose a novel privacy-preserving approach that introduces noise to uploaded viewpoint prediction errors, which can ensure zero viewpoint leakage probability. Given the proposed approach, the tradeoff between privacy preservation and QoE is optimized to minimize the QoE loss while satisfying the privacy requirement. Simulation results validate our analysis results and demonstrate that the proposed approach offers a promising solution for balancing privacy and QoE.

Optimizing QoE-Privacy Tradeoff for Proactive VR Streaming

TL;DR

This work tackles privacy in proactive VR streaming, where uploading predicted viewpoints and prediction errors risks leaking users' actual viewpoints. It first analyzes viewpoint leakage under existing privacy-preserving data processing and proves that leakage cannot be eliminated due to the PEA relation, yielding a positive lower bound Pr(A)min = ε/π. To overcome this, it introduces B-PEA, which breaks the PEA relation by adding noise n to the prediction errors, and derives the optimal noise strategy that minimizes QoE loss subject to Pr(A) ≤ q, even under worst-case distributions of the prediction error e. A trace-driven evaluation on real VR workloads shows that B-PEA satisfies privacy constraints for all users and substantially improves QoE compared with Gaussian or Laplace baselines, with negligible QoE loss when zero leakage is enforced. Overall, the paper provides a principled framework for zero-leakage privacy in proactive VR streaming and demonstrates practical gains in both privacy guarantees and user QoE.

Abstract

Proactive virtual reality (VR) streaming requires users to upload viewpoint-related information, raising significant privacy concerns. Existing strategies preserve privacy by introducing errors to viewpoints, which, however, compromises the quality of experience (QoE) of users. In this paper, we first delve into the analysis of the viewpoint leakage probability achieved by existing privacy-preserving approaches. We determine the optimal distribution of viewpoint errors that minimizes the viewpoint leakage probability. Our analyses show that existing approaches cannot fully eliminate viewpoint leakage. Then, we propose a novel privacy-preserving approach that introduces noise to uploaded viewpoint prediction errors, which can ensure zero viewpoint leakage probability. Given the proposed approach, the tradeoff between privacy preservation and QoE is optimized to minimize the QoE loss while satisfying the privacy requirement. Simulation results validate our analysis results and demonstrate that the proposed approach offers a promising solution for balancing privacy and QoE.

Paper Structure

This paper contains 19 sections, 1 theorem, 18 equations, 11 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System Model
  3. Viewpoint Uploading
  4. Prediction Errors Uploading
  5. Privacy Attack
  6. QoE-Privacy Requirement
  7. Analysis of Viewpoint Leakage
  8. Viewpoint Leakage Probability
  9. Distribution Optimization for Prediction Errors
  10. Optimizing QoE-Privacy Tradeoff
  11. Noisy Prediction Errors: Breaking PEA Relation
  12. Optimizing QoE-Privacy Tradeoff with B-PEA
  13. Trace-Driven Simulation Results
  14. Privacy Preservation
  15. QoE-Privacy Tradeoff
  16. ...and 4 more sections

Key Result

Proposition 2

$f^{*}_{n|e,\varepsilon}(n)$ is a Dirac Delta function. Proof: See Appendix appendix:optimal_f_e.

Figures (11)

  • Figure 1: Illustration of FoV, pFoV, actual viewpoint $V$, predicted viewpoint $\widetilde{P}$, and prediction error $E$ on a unit sphere.
  • Figure 2: Illustration of FoV, pFoV, and streamed zone in spatial-time dimensions.
  • Figure 3: Proactive VR streaming with attackers
  • Figure 4: Neighborhood of $\widehat{V}$
  • Figure 5: Optimal $\widehat{V}$, $\mathrm{Pr}(A|\widetilde{P}, e, \widehat{V}^*)=1$.
  • ...and 6 more figures

Theorems & Definitions (1)

  • Proposition 2