Source-linked AI summary
Vehicular Communications: Survey and Challenges of Channel and Propagation Models
Wantanee Viriyasitavat, Mate Boban, Hsin-Mu Tsai, Athanasios Vasilakos
TL;DR
Vehicular channel modeling must address dynamic environments, high mobility, low antenna heights, and diverse obstacles that make propagation difficult to represent. The paper surveys and classifies existing models by propagation mechanisms, implementation, and channel properties, emphasizing usability for protocol and application evaluation. It identifies model-selection guidance and underresearched vehicle types and environments requiring further measurement and modeling.
Problem
Dynamic environments, high mobility, low antenna heights, and diverse obstacles make vehicular propagation and channel modeling particularly challenging.
Method
The paper surveys and classifies vehicular channel models by propagation mechanisms, implementation approach, and channel properties, emphasizing usability for large-scale evaluation.
Results
The comparison finds that geographically informed GBD models are preferable when geographic information is available, while measurement-fitted models perform better for a specific location.
Takeaways & Limitations
The survey provides guidelines for choosing channel models according to protocol scale, available geographic information, and deployment needs.
Abstract
from arXiv · showhide
Vehicular communication is characterized by a dynamic environment, high mobility, and comparatively low antenna heights on the communicating entities (vehicles and roadside units). These characteristics make vehicular propagation and channel modeling particularly challenging. In this article, we classify and describe the most relevant vehicular propagation and channel models, with a particular focus on the usability of the models for the evaluation of protocols and applications. We first classify the models based on the propagation mechanisms they employ and their implementation approach. We also classify the models based on the channel properties they implement and pay special attention to the usability of the models, including the complexity of implementation, scalability, and the input requirements (e.g., geographical data input). We also discuss the less-explored aspects in vehicular channel modeling, including modeling specific environments (e.g., tunnels, overpasses, and parking lots) and types of communicating vehicles (e.g., scooters and public transportation vehicles). We conclude by identifying the underresearched aspects of vehicular propagation and channel modeling that require further modeling and measurement studies.
I. INTRODUCTION
Vehicular channel modeling is difficult because communication combines diverse environments, link types, mobile and static obstacles, high mobility, and low antenna heights. The paper surveys models with emphasis on their usability for large-scale protocol and application evaluation.
- Diverse environments, communication types, and static or mobile objects create complex vehicular propagation conditions.
- Low antenna heights, high vehicle mobility, and dynamic surroundings cause rapid variation in small- and large-scale signal statistics.
- Vehicular channels differ from cellular channels in antenna geometry, operating frequency, and communication distance.
- The survey classifies models by propagation mechanisms, implementation approach, and channel properties, while assessing scalability and geographic-data requirements.
- The models are evaluated for usability in large-scale simulations supporting protocol and application assessment before deployment.
II. SPECIFIC CONSIDERATIONS FOR VEHICULAR CHANNELS
Vehicular propagation depends strongly on environment, link type, vehicle type, and the objects surrounding communicating entities. These factors determine which propagation obstacles and channel assumptions a model must represent.
- Buildings, vehicles, and vegetation strongly influence propagation through their type, number, size, density, and combination.
- Vehicular environments require qualitative classification because similar object types can create different propagation settings.
- V2V, V2I, and V2P links differ in mobility, antenna height, and shadowing conditions.
- Personal vehicles, vans, trucks, scooters, and public transportation vehicles have distinct dimensions and mobility dynamics.
- A large truck blocking line of sight can add more than 20 dB of attenuation beyond the attenuation caused by personal vehicles.
D. Objects
Vehicular channels vary across space and time and exhibit Doppler shift and frequency-selective fading from mobile and static objects. Models therefore commonly divide the problem into manageable propagation components and distinguish link conditions.
- Non-LOS V2V channels have noticeably larger RMS delay spreads than LOS channels across highway and urban environments.Stronger attenuation and multipath from additional reflections and diffractions account for the difference.
- The survey selects models according to usability, including scalability, geographic database requirements, extensibility, and environmental realism.
- Vehicular channels exhibit spatially and temporally varying path loss, potentially high Doppler shift, and frequency-selective fading.
- Log-distance path loss is the most commonly used large-scale model, with its path loss exponent estimated from empirical measurements.
- GEMV2 uses separate LOS and non-LOS path loss models, while Maurer et al. use ray tracing for large-scale propagation.
2) Small-scale fading:
Vehicular channel models represent small-scale fading and link conditions using statistical, ray-tracing, and geometry-based approaches. Their implementation choices determine how mobility, obstacles, geographic information, and scalability are handled.
- Small-scale fading: Small-scale variations arise from multipath propagation and Doppler effects caused by vehicle and surrounding-object mobility.
- Small-scale fading: Weibull, Nakagami, and Gaussian distributions are commonly used to model small-scale fading.
- Link conditions: Recent models distinguish LOS, vehicle-blocked non-LOS, and building- or foliage-blocked non-LOS links.
- Geometry-based models: Ray-tracing models require detailed environmental descriptions to calculate channel statistics for a specific environment.
- Geometry-based models: RADII improves scalability by preprocessing regional attenuation and using interpolation and lookup tables during simulation.
- Geometry-based models: Simplified geometry-based models estimate channel parameters from measurements or simulations and can model LOS and different non-LOS conditions.
2) Non-geometry-based (NG) models:
Non-geometry-based models derive channel parameters from measurements in specific environments, emphasizing usability through properties such as scalability, correlation, non-stationarity, extensibility, applicability, and antenna support.
- NG models typically measure channel characteristics in a specific environment and adjust path loss, shadowing, and small-scale fading parameters.
- Model usability is assessed through properties including scalability, spatial and temporal dependency, non-stationarity, environmental extensibility, applicability, and antenna-configuration support.
- Spatial and temporal dependency: Measurements show that vehicular-channel attenuation is correlated across time and space, reflecting shared effects from static and dynamic physical features.
- Temporal variance and non-stationarity: Non-stationary models represent changing channel statistics and sudden LOS appearance or disappearance caused by mobile and static objects.
- Extensibility to different environments: Measurement-calibrated models provide no accuracy guarantees for locations with considerably different characteristics, whereas geometry-informed models can generalize beyond measured environments.
- Applicability: Models may support application-specific scenarios and antenna configurations, including MIMO, to exploit beneficial and mitigate adverse small-scale-fading effects.
- Scalability: Scalability depends largely on the complexity of the mechanisms used to calculate channel statistics, because large-scale simulations require efficient models.
D. Comparison of Selected Channel Models
A Porto V2V measurement comparison shows that model suitability depends on available environmental measurements and geographic information: NG models can be inconsistent without local calibration, while GBD models can perform better when geometry is available.
- Without measurements for a specific environment, NG models produce inconsistent received-power estimates, including unrealistically low values from the dual-slope Cheng model beyond 100 m.
- When parameters are extracted from local measurement data, the log-distance model gives better estimates than the uncalibrated alternatives in the Porto comparison.
- When geographic information is available, GBD models such as GEMV2 are identified as the better choice for the comparison.
E. Guidelines for Choosing a Suitable Channel Model
Choosing a channel model requires balancing accuracy, complexity, scalability, processing power, and available geographic or measurement data, with application requirements determining the suitable model category.
- Channel models offer different accuracy–complexity and scalability trade-offs, ranging from simple stochastic models to computationally demanding ray-tracing models.
- Comparison of selected models: 6.7 dB mean absolute error was reported for GEMV2, compared with 11.1 dB for Cheng single slope, 14.4 dB for Cheng dual slope, and 7.7 dB for log-distance.
- The model choice should depend on the application or protocol being evaluated, constrained by processing power and the availability of geographic or measurement data.
- Ray-tracing-based GBD models are suggested when complete geographic information is available and processing speed is not a concern, prioritizing maximum accuracy.
- Simplified GB models are suggested when environmental information is limited and processing speed matters, while other GB models can use qualitative environment descriptions.
IV. TOWARD REALISTIC AND EFFICIENT VEHICULAR CHANNEL MODELING
Realistic and efficient vehicular channel modeling is needed for large-scale evaluation before deployment because commonly used simulator models do not capture diverse vehicular propagation conditions.
- Large-scale simulations require realistic channel models to evaluate vehicular applications efficiently before real-world deployment.
- Common VANET simulators use simple statistical models indiscriminately across environments, limiting their ability to capture differing LOS conditions and environments.
- Realistic small-scale channel models should provide environment-specific delay and Doppler statistics and be implemented in large-scale network simulators.
- GEMV2 combines computational efficiency with city-wide simulation of thousands of vehicles across highway, rural, urban, and complex-intersection environments.
B. Vehicular Channel Emulations
Channel emulation combines simulated propagation with parts of real communication systems to retain repeatability and configurability while increasing testbed realism. The broader modeling landscape also exposes gaps in vehicle types and environments that require further study.
- B. Vehicular Channel Emulations: Channel emulation combines real communication-system components with simulated propagation to balance repeatability, configurability, and realism.The CMU Wireless Emulator takes signals from real devices and subjects them to simulated realistic propagation.
- B. Vehicular Channel Emulations: Real-hardware experiments are realistic but costly and difficult to repeat at scales involving tens or hundreds of vehicles.
- 1) Channel models for different vehicle types:: Vehicular channel measurements have primarily focused on personal cars, leaving commercial vans, trucks, scooters, and public transportation vehicles comparatively understudied.
- 1) Channel models for different vehicle types:: Distinct vehicle dimensions and mobility dynamics can produce different propagation characteristics, reliable communication ranges, and packet error rates.The passage specifically contrasts scooters, motorcycles, vans, trucks, and personal cars.
- 1) Channel models for different vehicle types:: Measurements in multi-level highways, tunnels, parking garages, bridges, and roundabouts remain rare despite their distinct vehicular communication use cases.The paper calls for further measurements and modeling in these environments.
2) Under-explored environments:
Vehicular channel research has concentrated more heavily on V2V than on V2I and V2P, even though their propagation conditions differ. V2P channel models and measurements remain especially incomplete across enabling technologies.
- 2) Under-explored environments:: V2I and V2P propagation characteristics are less researched than V2V, and the models surveyed in Table I focus on V2V communications.
- 2) Under-explored environments:: Only a few dedicated V2I channel models exist despite V2I measurement campaigns in urban, suburban, and highway environments.
- 2) Under-explored environments:: V2I differs from cellular links because roadside units are positioned close to roads at heights considerably lower than cellular base stations.
- 3) Vehicle-to-X channels:: V2P channel modeling remains incomplete across technologies such as DSRC, WiFi, and cellular-based systems.Recent studies have explored basic V2I link channel properties in relation to V2P communications.
- 3) Vehicle-to-X channels:: The paper anticipates gradual convergence between V2X and 5G channel-modeling efforts as 5G research addresses ITS applications and stricter delay requirements.
V. CONCLUSIONS
The paper surveys vehicular propagation and channel models with emphasis on their mechanisms, implementation approaches, and usability for evaluating protocols and applications. It also identifies underexplored settings and vehicle types requiring further measurement and modeling.
- V. CONCLUSIONS: The paper surveys recent vehicular propagation and channel-modeling developments, classifying models by propagation mechanisms and implementation approach.
- V. CONCLUSIONS: It evaluates model usability for large-scale protocol and application studies, including geographic-information requirements and available simulation processing power.
- V. CONCLUSIONS: The paper provides guidelines for choosing channel models according to the protocol or application under investigation and available geographical information.
- V. CONCLUSIONS: It identifies less-explored aspects of vehicular channel modeling and areas requiring further research efforts.