Variable Periodic Pseudonym-based Location Privacy Preserving for ICV

Xinling Zhou, Lei Peng, Huiyun Li, Mingsong Chen, Hongwei Zhu, Shi Dong2017 · DOI

Abstract

Intelligent connected vehicles (ICV) share a lot of traffic information involving location through V2X communication to improve the efficiency of transportation. Hence, location-aware privacy protection is the key issue in ICV. In this paper, a scheme called variable periodic pseudonym (VPP) is proposed to improve the privacy security without loss of location precision via breaking the link between vehicle's identity and real-time location. The main idea of VPP is that vehicle can change pseudonym at appropriate time with the impact of the privacy leak, overcoming the weakness of constant period of changing pseudonym. And the impact of the privacy leak is modeled as gamma distribution in this paper. Then it is convoluted with the original constant pseudonym period, to achieve the effect of adaptive changing. Finally, the simulation shows the effectiveness of the VPP with different privacy leak parameters.

Introduction

Intelligent Connected Vehicles (ICVs) rely on V2X (Vehicle-to-Everything) communication to share real-time location, speed, and heading. While this data is essential for preventing accidents and managing traffic, it creates a massive privacy hole. An adversary collecting these "heartbeat" messages can reconstruct a vehicle’s entire trajectory, eventually uncovering sensitive information like a driver’s home address or workplace.

Existing solutions generally fall into two camps, both with flaws:

  1. Location Hiding: Blurring or falsifying coordinates. This protects privacy but makes the data useless for safety-critical applications that require sub-meter precision.
  2. Fixed-Period Pseudonyms: Changing the vehicle's ID every $X$ minutes. The problem here is that "data leakage" isn't just a function of time; it’s a function of how much information is shared. A "chatty" vehicle might leak its identity long before the timer expires.

The Variable Periodic Pseudonym (VPP) Scheme

The authors propose the Variable Periodic Pseudonym (VPP) scheme. Instead of waiting for a fixed timer ($T$), the vehicle monitors its own "privacy leakage" and swaps its pseudonym as soon as a risk threshold ($P$) is reached. This creates a dynamic, sawtooth-like pattern of privacy management where the identity is refreshed exactly when the risk becomes too high.

Figure 1: The VPP mechanism resets the vehicle's identity once the accumulated privacy leakage reaches threshold P.

Modeling Privacy Leakage

The core of the VPP scheme is the mathematical modeling of the privacy leak ($g(x)$) using a Gamma distribution. This distribution is chosen because it effectively represents the cumulative risk associated with two critical factors: $\alpha$ (the number of messages sent) and $\beta$ (the time interval between messages).

To determine the optimal change point, the authors convolve the standard periodic change formula with the Gamma-based privacy leak function. This creates a dynamic trigger ($y(x)$) that responds to the density of information being shared. The system relies on a Trusted Institution (TA) to provide sets of pseudonyms and certificates to vehicles before they begin their journey, ensuring the vehicle can switch IDs locally without needing a constant, laggy connection to a central authority.

Simulation and Evaluation

The researchers evaluated their model using Vissim 9, a professional-grade traffic flow simulation tool. The scenario involved a simulated version of the Shenzhen Science and Technology Park with 500 vehicles traveling within a specific coordinate area. The simulation used a baseline pseudonym period ($T$) of 5 minutes and a privacy threshold ($P$) of 1.

Key Results

The simulation demonstrated that the VPP scheme is much more responsive than traditional methods. The primary insight is that privacy is dynamic, not linear: a vehicle stuck in a driveway leaks almost zero identity information compared to a vehicle navigating a complex intersection and broadcasting high-frequency safety updates.

Figure 2: Cumulative privacy leak g(x) for varying message interval parameters (beta).

As shown in the results:

  • Adaptive Timing: When the interval between messages was small ($\beta = 0.5$, meaning frequent updates), the privacy threshold was reached much faster, at approximately $t = 3.3$ minutes.
  • The "Safety Gap": Under a traditional 5-minute fixed-period scheme, that same vehicle would have been vulnerable for an extra 1.7 minutes after its identity was effectively compromised.
  • Efficiency: Conversely, if a vehicle sends information infrequently ($\beta = 2$), the system is smart enough to keep the current pseudonym until the full 5-minute period expires, reducing the overhead of constantly requesting and changing certificates.

Conclusion

VPP allows the car to broadcast its exact location, preserving the precision needed for safety, while changing who the car claims to be before an observer can figure out its true identity. Future work will focus on real-world validation with hardware-in-the-loop and exploring how VPP holds up against advanced machine learning-based inference attacks that look for unique driving "fingerprints" like braking or acceleration patterns.

Summary updated

Abstract

Intelligent connected vehicles (ICV) share a lot of traffic information involving location through V2X communication to improve the efficiency of transportation. Hence, location-aware privacy protection is the key issue in ICV. In this paper, a scheme called variable periodic pseudonym (VPP) is proposed to improve the privacy security without loss of location precision via breaking the link between vehicle's identity and real-time location. The main idea of VPP is that vehicle can change pseudonym at appropriate time with the impact of the privacy leak, overcoming the weakness of constant period of changing pseudonym. And the impact of the privacy leak is modeled as gamma distribution in this paper. Then it is convoluted with the original constant pseudonym period, to achieve the effect of adaptive changing. Finally, the simulation shows the effectiveness of the VPP with different privacy leak parameters.
BibTeX
@misc{zhou2017variable,
  author = {Xinling Zhou and Lei Peng and Huiyun Li and Mingsong Chen and Hongwei Zhu and Shi Dong},
  title = {{Variable Periodic Pseudonym-based Location Privacy Preserving for ICV}},
  year = {2017},
  doi = {10.2991/itim-17.2017.19},
}
Figure