Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A V2X-based Privacy Preserving Federated Measuring and Learning System

Levente Alekszejenkó, Tadeusz Dobrowiecki

TL;DR

The paper addresses privacy-preserving data sharing and predictive modeling in V2X-enabled autonomous vehicle networks under non-IID data. It proposes a system combining trusted V2V data exchange with untrusted V2N FL aggregation (FedAvg) to balance datasets and improve privacy. The MNIST-based simulations show that data sharing speeds convergence toward the IID baseline and reduces susceptibility to passive eavesdropping compared with standard non-IID FL. The results demonstrate practical potential for real-time transportation forecasting and privacy-preserving federated learning in urban vehicular networks.

Abstract

Future autonomous vehicles (AVs) will use a variety of sensors that generate a vast amount of data. Naturally, this data not only serves self-driving algorithms; but can also assist other vehicles or the infrastructure in real-time decision-making. Consequently, vehicles shall exchange their measurement data over Vehicle-to-Everything (V2X) technologies. Moreover, predicting the state of the road network might be beneficial too. With such a prediction, we might mitigate road congestion, balance parking lot usage, or optimize the traffic flow. That would decrease transportation costs as well as reduce its environmental impact. In this paper, we propose a federated measurement and learning system that provides real-time data to fellow vehicles over Vehicle-to-Vehicle (V2V) communication while also operating a federated learning (FL) scheme over the Vehicle-to-Network (V2N) link to create a predictive model of the transportation network. As we are yet to have real-world AV data, we model it with a non-IID (independent and identically distributed) dataset to evaluate the capabilities of the proposed system in terms of performance and privacy. Results indicate that the proposed FL scheme improves learning performance and prevents eavesdropping at the aggregator server side.

A V2X-based Privacy Preserving Federated Measuring and Learning System

TL;DR

The paper addresses privacy-preserving data sharing and predictive modeling in V2X-enabled autonomous vehicle networks under non-IID data. It proposes a system combining trusted V2V data exchange with untrusted V2N FL aggregation (FedAvg) to balance datasets and improve privacy. The MNIST-based simulations show that data sharing speeds convergence toward the IID baseline and reduces susceptibility to passive eavesdropping compared with standard non-IID FL. The results demonstrate practical potential for real-time transportation forecasting and privacy-preserving federated learning in urban vehicular networks.

Abstract

Future autonomous vehicles (AVs) will use a variety of sensors that generate a vast amount of data. Naturally, this data not only serves self-driving algorithms; but can also assist other vehicles or the infrastructure in real-time decision-making. Consequently, vehicles shall exchange their measurement data over Vehicle-to-Everything (V2X) technologies. Moreover, predicting the state of the road network might be beneficial too. With such a prediction, we might mitigate road congestion, balance parking lot usage, or optimize the traffic flow. That would decrease transportation costs as well as reduce its environmental impact. In this paper, we propose a federated measurement and learning system that provides real-time data to fellow vehicles over Vehicle-to-Vehicle (V2V) communication while also operating a federated learning (FL) scheme over the Vehicle-to-Network (V2N) link to create a predictive model of the transportation network. As we are yet to have real-world AV data, we model it with a non-IID (independent and identically distributed) dataset to evaluate the capabilities of the proposed system in terms of performance and privacy. Results indicate that the proposed FL scheme improves learning performance and prevents eavesdropping at the aggregator server side.
Paper Structure (12 sections, 3 equations, 4 figures, 2 tables)

This paper contains 12 sections, 3 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. System Architecture
  4. Evaluation
  5. Data Modeling
  6. Data Sharing
  7. Measurement Cases
  8. Training Process
  9. Performance Evaluation
  10. Privacy Threats
  11. Conclusion
  12. Discussion

Figures (4)

  • Figure 1: Illustration of the FL setup. Participant AVs can communicate with each other and the server. Participants can share data on a trusted V2V link, while the server receives and sends the model parameters of the FL process on an untrusted V2N channel.
  • Figure 2: Modeling commuting vehicles' data by partitioning the MNIST dataset.
  • Figure 3: Test accuracy of the learning methods when reaching convergence, average of 10 measurements.
  • Figure 4: Most accurately recognized overrepresented classes by the attacker. Results corresponds to single training processes with each method.