Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey on Differential Privacy for SpatioTemporal Data in Transportation Research

Rahul Bhadani

TL;DR

This survey addresses the privacy risks inherent in releasing spatiotemporal transportation data and surveys differential privacy methods, algorithms, and software applicable to such data. It covers foundational DP concepts ($\eplison$-DP, $(\eplison,\\delta)$-DP), central and local models, and core algorithms (Randomized Response, Exponential Mechanism, $oldsymbol{\\alpha}$-net) along with DP software ecosystems (OpenDP, Google DP, TensorFlow Privacy, SecretFlow, Opacus). It then surveys DP applications to real-time traffic data, trajectory publication, and the use of spatiotemporal correlated noise, highlighting innovations like the $w$-event model, dynamic budget updates, time-generalization with clustering, and least-squares-based noise alignment. The paper concludes with open challenges, including addressing high correlation and high dimensionality in multimodal data and the feasibility of DP in autonomous vehicle deployment, while suggesting synthetic data generation as a promising interim solution.

Abstract

With low-cost computing devices, improved sensor technology, and the proliferation of data-driven algorithms, we have more data than we know what to do with. In transportation, we are seeing a surge in spatiotemporal data collection. At the same time, concerns over user privacy have led to research on differential privacy in applied settings. In this paper, we look at some recent developments in differential privacy in the context of spatiotemporal data. Spatiotemporal data contain not only features about users but also the geographical locations of their frequent visits. Hence, the public release of such data carries extreme risks. To address the need for such data in research and inference without exposing private information, significant work has been proposed. This survey paper aims to summarize these efforts and provide a review of differential privacy mechanisms and related software. We also discuss related work in transportation where such mechanisms have been applied. Furthermore, we address the challenges in the deployment and mass adoption of differential privacy in transportation spatiotemporal data for downstream analyses.

A Survey on Differential Privacy for SpatioTemporal Data in Transportation Research

TL;DR

This survey addresses the privacy risks inherent in releasing spatiotemporal transportation data and surveys differential privacy methods, algorithms, and software applicable to such data. It covers foundational DP concepts (-DP, -DP), central and local models, and core algorithms (Randomized Response, Exponential Mechanism, -net) along with DP software ecosystems (OpenDP, Google DP, TensorFlow Privacy, SecretFlow, Opacus). It then surveys DP applications to real-time traffic data, trajectory publication, and the use of spatiotemporal correlated noise, highlighting innovations like the -event model, dynamic budget updates, time-generalization with clustering, and least-squares-based noise alignment. The paper concludes with open challenges, including addressing high correlation and high dimensionality in multimodal data and the feasibility of DP in autonomous vehicle deployment, while suggesting synthetic data generation as a promising interim solution.

Abstract

With low-cost computing devices, improved sensor technology, and the proliferation of data-driven algorithms, we have more data than we know what to do with. In transportation, we are seeing a surge in spatiotemporal data collection. At the same time, concerns over user privacy have led to research on differential privacy in applied settings. In this paper, we look at some recent developments in differential privacy in the context of spatiotemporal data. Spatiotemporal data contain not only features about users but also the geographical locations of their frequent visits. Hence, the public release of such data carries extreme risks. To address the need for such data in research and inference without exposing private information, significant work has been proposed. This survey paper aims to summarize these efforts and provide a review of differential privacy mechanisms and related software. We also discuss related work in transportation where such mechanisms have been applied. Furthermore, we address the challenges in the deployment and mass adoption of differential privacy in transportation spatiotemporal data for downstream analyses.
Paper Structure (19 sections, 10 equations, 3 figures)

This paper contains 19 sections, 10 equations, 3 figures.

Table of Contents

  1. Introduction
  2. SpatioTemporal Data
  3. Analysis using SpatioTemporal Transportation Data
  4. A Primer on Differential Privacy
  5. Algorithms for Differential Privacy
  6. Randomized Response
  7. Exponential Mechanism
  8. $\alpha$-net Mechanism
  9. Software and Tools for Differential Privacy
  10. OpenDP
  11. Differential Privacy from Google
  12. TensorFlow Privacy
  13. SecretFlow
  14. Opacus
  15. Differential Privacy Methods on SpatioTemporal Data in Transportation
  16. ...and 4 more sections

Figures (3)

  • Figure 1: Probability Density Function (PDF) of the Laplace distribution with central parameter $\mu = 0$ and width parameters $b = 0.5$, $b = 1.0$, and $b = 3.0$.
  • Figure 2: The central model and the local model for differentially private mechanism. A trusted curator collects data from its source, applies some transformation ${\mathcal{M}}$, and disseminates it for downstream analyses. In the local model, the transformation is applied at the source as there may not be a trusted curator.
  • Figure 3: $w$-event model where two spatiotemporal data in a window of size $w$ are not identical.

Theorems & Definitions (2)

  • Definition 1
  • Definition 2