Packet Reception Probabilities in Vehicular Communications Close to Intersections

Erik Steinmetz, Matthias Wildemeersch, Tony Q. S. Quek, Henk Wymeersch

TL;DR

This paper presents a general analytical framework using stochastic geometry to model packet reception probabilities and throughput in vehicular communication networks, specifically focusing on the challenging scenario of intersections. The authors provide a flexible methodology that accounts for various propagation conditions, road geometries, and MAC protocols, offering a valuable tool for designing efficient vehicular communication systems.

What Problem Does This Paper Address?

Vehicular communication systems are crucial for future intelligent transportation systems (ITS), enabling vehicles to share information for enhanced safety and efficiency. However, designing these systems is challenging due to high mobility, rapidly changing propagation conditions, and the complex traffic dynamics at intersections. Existing methods like simulations and measurements are time-consuming and scenario-specific. This paper addresses the need for analytical tools to predict the performance of vehicular networks, particularly at intersections, which are critical and accident-prone areas. The authors aim to provide a general procedure to analytically determine key performance metrics like packet reception probability and throughput, considering the unique spatial distribution of vehicles on roads.

Methods

The core of this paper lies in its application of stochastic geometry to model vehicular communication networks. The authors abstract the intersection scenario into two perpendicular roads, each populated by vehicles modeled as one-dimensional homogeneous Poisson point processes (PPPs). They then develop a general procedure to calculate:

  • Packet Reception Probability (PRP): This involves deriving expressions for the probability that a receiver can successfully decode a transmission from a transmitter, considering signal propagation (path loss and fading), noise, and interference from other vehicles. They leverage the Laplace Transform (LT) of the interference distribution and the complementary cumulative distribution function (CCDF) of the fading distribution of the useful link.
  • Throughput: This metric is calculated by combining the PRP with the probability of gaining access to the channel, which is dictated by the Medium Access Control (MAC) protocol.

The methodology is designed to be flexible, allowing for the incorporation of:

  • Different Path Loss Models: Euclidean distance for Line-of-Sight (LOS) propagation and Manhattan distance for Non-Line-of-Sight (NLOS) propagation.
  • Various Fading Distributions: Exponential fading for LOS and log-normal (approximated by Erlang) for NLOS.
  • Common MAC Protocols: Aloha and CSMA are analyzed, with CSMA's complex interference patterns approximated using thinned PPPs.

Datasets

Not discussed in the paper. The paper focuses on developing an analytical model and validating it through numerical results and comparisons with simulations, rather than using specific real-world datasets for training or evaluation.

Key Insights and Ideas

  • Intersections are a critical bottleneck: The spatial clustering of vehicles at intersections significantly impacts network performance, making it distinct from traditional linear road scenarios.
  • Stochastic geometry is powerful for V2X: The paper demonstrates the effectiveness of stochastic geometry in providing tractable analytical expressions for complex vehicular network scenarios, which are difficult to capture with simpler models.
  • MAC protocol choice matters significantly: The choice between Aloha and CSMA, and their specific parameters (like contention radius in CSMA), has a substantial impact on both packet reception probability and throughput, with CSMA showing significant throughput gains in certain scenarios.
  • Fading and path loss models are crucial: The paper highlights how different propagation environments (rural vs. urban) and their associated path loss and fading models directly influence network performance.
  • Analytical tractability vs. accuracy trade-off: While complex models offer higher accuracy, simpler approximations (like Erlang for log-normal fading) are often necessary for analytical tractability, and the paper shows that these approximations can be quite effective.

Key Results

The paper presents three case studies to illustrate the application of their methodology:

  • Case I (Rural Intersection with Aloha): Demonstrates that outage probability increases with distance and transmit probability. Interference significantly reduces the communication range.
  • Case II (Urban Intersection with Aloha): Shows that using Manhattan path loss and Erlang fading for NLOS links (approximating log-normal) provides a good approximation to simulations. The Erlang approximation is shown to be tractable and reasonably accurate.
  • Case III (Aloha vs. CSMA): Compares Aloha and CSMA in a rural intersection scenario. CSMA significantly outperforms Aloha in terms of throughput, achieving more than a tenfold increase for the same communication range while maintaining acceptable outage probabilities.

Limitations and Future Work

  • Latency and Mobility: The current model does not explicitly account for latency and the dynamic mobility of vehicles, which are important aspects of vehicular communication.
  • Generalization of CSMA: While CSMA is approximated using thinned PPPs, the paper acknowledges that the exact hard-core process induced by CSMA is complex.
  • Validation with Real-World Measurements: The authors suggest future work to validate the model against actual measurements.
  • Advanced MAC Schemes and 5G D2D: The paper proposes exploring advanced MAC schemes and incorporating 5G D2D features into the model.
  • Non-homogeneous PPPs: For scenarios with vehicle clustering (e.g., traffic congestion), numerical integration might be required, though closed-form solutions can be found for specific intensity functions.

Summary updated

Abstract

Vehicular networks allow vehicles to share information and are expected to be an integral part of future intelligent transportation systems (ITS). To guide and validate the design process, analytical expressions of key performance metrics such as packet reception probabilities and throughput are necessary, in particular for accident-prone scenarios such as intersections. In this paper, we present a procedure to analytically determine the packet reception probability and throughput of a selected link, taking into account the relative increase in the number of vehicles (i.e., possible interferers) close to an intersection. We consider both slotted Aloha and CSMA/CA MAC protocols, and show how the procedure can be used to model different propagation environments of practical relevance. The procedure is validated for a selected set of case studies at low traffic densities.
BibTeX
@article{steinmetz2021packet,
  author = {Erik Steinmetz and Matthias Wildemeersch and Tony Q. S. Quek and Henk Wymeersch},
  title = {{Packet Reception Probabilities in Vehicular Communications Close to Intersections}},
  journal = {IEEE Trans. Intell. Transp. Syst.},
  year = {2021},
  doi = {10.1109/tits.2020.2976564},
}
Figure