A Novel Segment-Based Model for Non-Stationary Vehicle-to-Vehicle Channels With Velocity Variations

Weidong Li, Xiaomin Chen, Qiuming Zhu, Weizhi Zhong, Dazhuan Xu, Fei BaiIEEE Access 2019 · DOI

Abstract

Accurate channel modeling is a prerequisite for the development of reliable 5G and autonomous vehicle-to-vehicle (V2V) communication systems. Traditional models often rely on the assumption of Wide-Sense Stationarity (WSS) and constant velocity, which fails to capture the dynamics of real-world driving environments involving acceleration, deceleration, and complex trajectories. This paper presents a novel segment-based geometry-based stochastic model (GBSM) that accounts for the varying velocities of both vehicles and surrounding scatterers. By partitioning trajectories into small segments and utilizing Taylor series expansions, the model provides closed-form statistical properties that align closely with real-world measurements.

Introduction

In the context of autonomous driving and intelligent transportation systems, V2V communication must maintain high reliability under rapidly changing conditions. Modeling the physical radio channel is challenging because V2V environments are inherently non-stationary. Vehicles frequently change speed and direction, and the surrounding environment is populated by other moving objects (scatterers) that reflect signals.

Existing modeling approaches often simplify these dynamics by assuming constant speeds or stationary scatterers. Such simplifications lead to significant inaccuracies when vehicles navigate intersections or engage in sudden maneuvers. This research bridges the gap by introducing a mathematical framework capable of handling arbitrary trajectories and velocity variations for all entities in the environment.

The Segment-Based GBSM Framework

The core of the proposed approach is a Geometry-Based Stochastic Model (GBSM) enhanced by a segmentation strategy. Instead of modeling a long, complex trajectory with a single set of equations, the time domain is divided into small segments of duration $T_0$. Within each segment, the velocity and direction of the mobile transmitter (MT), mobile receiver (MR), and scatterer clusters are approximated as linear functions of time, effectively incorporating acceleration.

Geometry-based stochastic model (GBSM) illustrating the interaction between mobile transmitters (MT), mobile receivers (MR), and moving scatterers under varying velocities.

To maintain mathematical tractability, the researchers employed Taylor series expansions for time-evolving parameters such as Doppler frequencies, path delays, and angles of arrival/departure. This allows for the explicit observation of how acceleration and angular changes impact signal degradation. The model specifically accounts for the relative movement between the MT, MR, and the clusters of scatterers, ensuring the dynamic nature of the environment is fully captured.

Statistical Characterization

To validate the model's performance, the authors derived four key statistical metrics:

  1. Temporal Correlation Function (TCF): Describes how the signal relates to itself over time.
  2. Doppler Power Spectrum Density (DPSD): Analyzes the spread of frequencies caused by motion.
  3. Level-Crossing Rate (LCR): Measures how often the signal envelope crosses a specific threshold.
  4. Average Fading Duration (AFD): Determines the average length of time the signal remains below a threshold.

Results and Validation

The model was evaluated using simulated data across four realistic V2V scenarios: vehicles moving in the same direction, vehicles in opposite directions, a single vehicle turning a corner, and both vehicles turning corners simultaneously.

Key findings include:

  • Superior Accuracy: Compared to traditional constant-velocity models, the proposed segment-based model maintained high accuracy over a 5-second window, whereas older models diverged significantly after a fraction of a second.
  • Theoretical Consistency: The derived closed-form formulas for TCF, DPSD, LCR, and AFD matched simulation results almost perfectly across all driving scenarios.
  • Empirical Validation: The model’s TCF showed strong agreement with real-world measured data from previous V2V experiment campaigns (e.g., Makhoul et al., 2017), confirming its practical utility for system design.

Conclusion

This research demonstrates that velocity variation is not a negligible factor but a fundamental driver of V2V channel behavior. By employing a segment-based approach and Taylor series simplifications, the authors have created a flexible and accurate tool for simulating arbitrary driving trajectories. While currently focused on Single-Input Single-Output (SISO) systems, the framework provides a robust foundation for future extensions into Multiple-Input Multiple-Output (MIMO) and dense urban environment simulations.

Summary updated

BibTeX
@article{li2019novel,
  author = {Weidong Li and Xiaomin Chen and Qiuming Zhu and Weizhi Zhong and Dazhuan Xu and Fei Bai},
  title = {{A Novel Segment-Based Model for Non-Stationary Vehicle-to-Vehicle Channels With Velocity Variations}},
  journal = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2941042},
}
Figure