Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Privacy-Preserving Data Fusion for Traffic State Estimation: A Vertical Federated Learning Approach

Qiqing Wang, Kaidi Yang

TL;DR

The paper tackles privacy concerns in traffic state estimation by proposing FedTSE, a vertical federated learning framework that fuses MA loop-detector data with MP trajectory features without sharing raw data. It introduces FedTSE-PI, a physics-informed variant that leverages traffic flow models to improve data efficiency when ground-truth labels are scarce, using secure inner-product encryption to protect MP data during gradient computations. Case studies on real-world and simulated urban networks show FedTSE nearly matches oracle performance and FedTSE-PI approaches the accuracy of privacy-unrestricted benchmarks while preserving privacy, with notable gains when MPs share higher-resolution data. The work demonstrates that privacy-preserving data fusion can meaningfully enhance TSE, and it highlights future directions in differential privacy, zero-knowledge verifications, uncertainty quantification, and reinforcement learning for traffic control. The practical impact lies in enabling cross-ownership collaboration for ITS applications without compromising sensitive operational data or customer information.

Abstract

This paper proposes a privacy-preserving data fusion method for traffic state estimation (TSE). Unlike existing works that assume all data sources to be accessible by a single trusted party, we explicitly address data privacy concerns that arise in the collaboration and data sharing between multiple data owners, such as municipal authorities (MAs) and mobility providers (MPs). To this end, we propose a novel vertical federated learning (FL) approach, FedTSE, that enables multiple data owners to collaboratively train and apply a TSE model without having to exchange their private data. To enhance the applicability of the proposed FedTSE in common TSE scenarios with limited availability of ground-truth data, we further propose a privacy-preserving physics-informed FL approach, i.e., FedTSE-PI, that integrates traffic models into FL. Real-world data validation shows that the proposed methods can protect privacy while yielding similar accuracy to the oracle method without privacy considerations.

Privacy-Preserving Data Fusion for Traffic State Estimation: A Vertical Federated Learning Approach

TL;DR

The paper tackles privacy concerns in traffic state estimation by proposing FedTSE, a vertical federated learning framework that fuses MA loop-detector data with MP trajectory features without sharing raw data. It introduces FedTSE-PI, a physics-informed variant that leverages traffic flow models to improve data efficiency when ground-truth labels are scarce, using secure inner-product encryption to protect MP data during gradient computations. Case studies on real-world and simulated urban networks show FedTSE nearly matches oracle performance and FedTSE-PI approaches the accuracy of privacy-unrestricted benchmarks while preserving privacy, with notable gains when MPs share higher-resolution data. The work demonstrates that privacy-preserving data fusion can meaningfully enhance TSE, and it highlights future directions in differential privacy, zero-knowledge verifications, uncertainty quantification, and reinforcement learning for traffic control. The practical impact lies in enabling cross-ownership collaboration for ITS applications without compromising sensitive operational data or customer information.

Abstract

This paper proposes a privacy-preserving data fusion method for traffic state estimation (TSE). Unlike existing works that assume all data sources to be accessible by a single trusted party, we explicitly address data privacy concerns that arise in the collaboration and data sharing between multiple data owners, such as municipal authorities (MAs) and mobility providers (MPs). To this end, we propose a novel vertical federated learning (FL) approach, FedTSE, that enables multiple data owners to collaboratively train and apply a TSE model without having to exchange their private data. To enhance the applicability of the proposed FedTSE in common TSE scenarios with limited availability of ground-truth data, we further propose a privacy-preserving physics-informed FL approach, i.e., FedTSE-PI, that integrates traffic models into FL. Real-world data validation shows that the proposed methods can protect privacy while yielding similar accuracy to the oracle method without privacy considerations.
Paper Structure (19 sections, 15 equations, 9 figures, 6 tables, 2 algorithms)

This paper contains 19 sections, 15 equations, 9 figures, 6 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Problem Statement
  3. FedTSE: Vertical Federated Traffic State Estimation
  4. Case study on FedTSE
  5. Settings of the case study
  6. Datasets
  7. Benchmarks for FedTSE
  8. Experiment settings
  9. Performance of FedTSE
  10. Sensitivity analysis on the number of local updates $Q$
  11. Sensitivity analysis on sample size
  12. FedTSE-PI: Physics-informed Vertical Federated Traffic State Estimation
  13. General framework of FedTSE-PI
  14. Privacy-preserving training of FedTSE-PI
  15. Case Study on FedTSE-PI
  16. ...and 4 more sections

Figures (9)

  • Figure 1: Vertically partitioned traffic data in urban transportation network
  • Figure 2: Schematic illustration of FedTSE, where MA (in blue) and multiple MPs (in orange and yellow) jointly train a neural network for TSE while keeping their data private. Only the values and gradients of private sub-models are exchanged between parties.
  • Figure 3: pNEUMA urban road corridor
  • Figure 4: Model performance comparison with different MP penetration rate.
  • Figure 5: Test RMSE curve for FedTSE
  • ...and 4 more figures

Theorems & Definitions (5)

  • Remark 1: Willingness of MPs to collaborate
  • Remark 2: Curious MA/MPs and sensitive information
  • Remark 3: Verification of MP data via zero-knowledge proof
  • Remark 4: Privacy of data owners in FedTSE
  • Remark 5: Computation requirements of FedTSE