Source-linked AI summary
Modeling Diffuse Scattering in Mobility Scenarios: Ray Tracing Model and Measurement Validation
Milica Djordjevic, Thomas Wilding, Anja Dakić, Benjamin Rainer, Markus Hofer, Thomas Zemen
TL;DR
Repeated ray-tracing evaluations in mobility scenarios can produce diffuse paths without preserving their correspondence, limiting consistent representation of their temporal evolution. The paper extends Sionna RT with persistent surface-associated scattering points whose contributions evolve with geometry and visibility. Against urban vehicular measurements, the approach better reproduces diffuse Doppler characteristics, while delay-domain improvements are limited.
Problem
Repeated Sionna RT evaluations for longer mobility trajectories generate diffuse paths without persistent correspondence, leaving individual diffuse components' temporal evolution unpreserved.
Source: §I · p. 1, B. Sionna RT Baseline · p. 2Method
The extension keeps surface-associated scattering points fixed across snapshots and gradually adjusts their contributions as visibility and propagation geometry change.
Source: §I · p. 1, C. Proposed Extension of Sionna RT · p. 2Results
The proposed model produces a more measurement-like continuous diffuse Doppler distribution and follows measured RMS Doppler spread more closely, while PDP and RMS delay-spread improvements are limited.
Source: C. Validation Results · p. 4Takeaways & Limitations
Preserving diffuse components' temporal evolution is important for representing time-varying channels in mobility scenarios.
Source: C. Validation Results · p. 5, §IV · p. 5The current implementation does not consider higher-order diffuse-scattering interactions.
Source: C. Validation Results · p. 5
Abstract
from arXiv · showhide
Open-source ray tracing (RT) frameworks are increasingly used for site-specific channel modeling due to their reproducibility and extensibility. In non-stationary mobility scenarios, repeated RT evaluations can generate different diffuse propagation paths without preserving their correspondence, limiting a consistent representation of their temporal evolution. To address this limitation, we extend diffuse scattering modeling in Sionna RT using persistent surface-associated scattering points. Their positions remain fixed, while their contributions evolve according to changes in the propagation geometry and visibility. The proposed approach is validated using empirical radio channel impulse response measurements in an urban vehicular scenario at 5.9 GHz. Compared with the standard Sionna RT baseline, the proposed model shows improved agreement with measurements in the delay and Doppler domains.
I. INTRODUCTION
Mobility makes channel responses time-varying, so accurate site-specific simulations must represent diffuse multipath evolution as well as dominant paths. This paper extends Sionna RT with persistent surface-associated scattering points and validates the approach against urban vehicular measurements.
I. INTRODUCTION
- Accurate channel evolution in mobile scenarios requires modeling the temporal evolution and correlation of diffuse multipath components, not only dominant paths.The channel's temporal evolution is characterized in the delay and Doppler domains.Source: §I · p. 1
- Prior approaches include road- or cluster-based stochastic models, vegetation scattering validated against measurements, and surface-based RT scattering methods.A position-invariant lattice representation was also introduced for spatially correlated diffuse scattering and time-varying DSD.Source: §I · p. 1
- The proposed Sionna RT extension preserves diffuse-component identity across snapshots using persistent surface-associated scattering points.Their contributions adjust with transmitter and receiver motion through geometry-dependent visibility and activation.Source: §I · p. 1
- The extension is validated against measured vehicular CIRs at 5.9 GHz using PDP, DSD, RMS delay spread, and RMS Doppler spread.The comparison is with the standard Sionna RT model.Source: §I · p. 1
A. Propagation Mechanisms and Channel Model
The channel model includes LOS, specular reflection, diffraction, and diffuse scattering. Mobility changes propagation geometry and paths, producing time-varying gains, delays, and Doppler shifts; the paper focuses on diffuse-scattering evolution.
A. Propagation Mechanisms and Channel Model
- In the considered channel model, diffuse scattering is non-specular propagation caused by surface roughness, structural details, and material inhomogeneities, while specular reflection follows the direction defined by geometrical optics,.Source: A. Propagation Mechanisms and Channel… · p. 1
- Mobility changes propagation geometry and paths, yielding time-varying path gains, delays, and Doppler shifts.The resulting channel behavior is non-stationary.Source: A. Propagation Mechanisms and Channel… · p. 2
- The paper focuses on the temporal evolution of the diffuse-scattering contribution Hs(t, f).The channel response comprises deterministic and diffuse-scattering contributions.Source: A. Propagation Mechanisms and Channel… · p. 2
B. Sionna RT Baseline
Sionna RT generates diffuse paths through SBR, but successive evaluations for longer mobility trajectories do not preserve correspondence between those paths. This baseline therefore does not preserve individual diffuse components' temporal evolution.
B. Sionna RT Baseline
- For longer trajectories, Sionna RT recomputes paths as transmitter and receiver move, and successive SBR evaluations generate diffuse paths without persistent correspondence.Its path-evolution approach based on Doppler shifts is intended for short trajectories, typically not exceeding a few wavelengths [11].Source: B. Sionna RT Baseline · p. 2
C. Proposed Extension of Sionna RT
The extension uses persistent surface-associated scattering points to represent diffuse propagation consistently across mobility snapshots. Their contributions evolve with visibility and geometry, while standard Sionna RT models supply the other propagation components.
C. Proposed Extension of Sionna RT
- Fixed scattering points retain their positions and surface properties across snapshots while their contributions change with mobility.Each point is characterized by its position, surface normal, effective area, and random phase.Source: C. Proposed Extension of Sionna RT · p. 2
- A regular surface lattice is generated at spacing Δs = c/(2B), and each region's points receive equal effective areas before visibility filtering.The assigned areas remain unchanged after filtering, preserving the represented surface area of the initial lattice.Source: C. Proposed Extension of Sionna RT · p. 2
- A visibility prescan retains points visible from both transmitter and receiver at least once; snapshot visibility then guides their activation weights.The prescan defines the retained point set, and instantaneous visibility is evaluated at each snapshot.Source: C. Proposed Extension of Sionna RT · p. 2
- Activation weights change gradually as visibility changes, while path lengths and interaction angles are recomputed from current geometry.Only geometrically valid interactions with weights above the selected threshold are included.Source: C. Proposed Extension of Sionna RT · p. 2
- The implementation models first-order diffuse scattering without interactions between diffuse scattering points.Source: C. Proposed Extension of Sionna RT · p. 2
- The diffuse channel frequency response coherently sums contributions from points retained after the visibility prescan; invalid or inactive points contribute zero.The built-in diffuse mechanism is disabled, while LOS, specular reflection, and diffraction use standard Sionna RT models; the band-limited CIR is obtained by inverse FFT.Source: C. Proposed Extension of Sionna RT · p. 3
A. Measurement Scenario and Site-Specific Reconstruction
The evaluation uses a 5.9 GHz urban vehicular measurement campaign at an intersection in Kaisermühlen, Vienna. The reconstructed RT scene incorporates the measured vehicle trajectories and parked vehicles.
A. Measurement Scenario and Site-Specific Reconstruction
- Measurements were recorded with two moving cars acting as transmitter and receiver using the AIT multi-node channel sounder at 5.9 GHz.The campaign used 150 MHz bandwidth, 250 kHz frequency spacing, and a 0.5 ms snapshot interval.Source: A. Measurement Scenario and… · p. 3
- The Kaisermühlen scene was reconstructed from OpenStreetMap data, with building and street geometry prepared in Blender and exported for Sionna RT.The scene also includes parked vehicles present during measurement.Source: A. Measurement Scenario and… · p. 3, Figure 2 · p. 3
- Measured transmitter and receiver positions were mapped to the reconstructed scene and spline-interpolated onto the simulation time grid.The figure identifies their trajectories and directions of motion within the site-specific scene.Source: A. Measurement Scenario and… · p. 3, Figure 2 · p. 3
B. Channel Evaluation Metrics
The channel evaluation estimates local scattering behavior within finite stationarity regions, then derives delay- and Doppler-domain distributions and summary spreads.
B. Channel Evaluation Metrics
- The sampled channel frequency response is processed within each finite stationarity region to estimate the local scattering function using a multi-taper approach.The procedure uses M time and N frequency samples per region.Source: B. Channel Evaluation Metrics · p. 3
- The two-dimensional taper is formed as a product of time- and frequency-domain DPSS tapers.Source: B. Channel Evaluation Metrics · p. 3
- Time-variant PDP and DSD are obtained as marginals of the local scattering function, characterizing the channel in delay and Doppler domains.Source: B. Channel Evaluation Metrics · p. 3
- The resulting discrete densities yield mean delay, RMS delay spread, mean Doppler frequency, and RMS Doppler spread for each stationarity index.Source: B. Channel Evaluation Metrics · p. 4
C. Validation Results
The proposed model is evaluated against the Sionna RT baseline and measurements using shared scene and channel-processing conditions. It better matches diffuse Doppler behavior, while delay-domain improvements remain limited.
C. Validation Results
- Both simulations use the same scene geometry, trajectories, and material properties, so the diffuse scattering model is the stated difference.The proposed model retains surface-associated lattice points, whereas baseline diffuse-path counts depend on ray sampling and geometry.Source: C. Validation Results · p. 4
- Measured and simulated CIRs are evaluated on a shared 0.5 ms time grid using 120 ms stationarity intervals, or 240 time samples per region.Source: C. Validation Results · p. 4
- The proposed model better reproduces the measured diffuse Doppler distribution and RMS Doppler spread, while PDP and RMS delay-spread improvements remain limited.The comparison covers time-variant PDP and DSD in Fig. 3 and RMS delay and Doppler spreads in Fig. 4.Source: C. Validation Results · p. 4, Figure 3 · p. 4, Figure 4 · p. 5
IV. CONCLUSION
The paper extends Sionna RT with persistent surface-associated scattering points and validates the approach using urban vehicular measurements at 5.9 GHz. Its strongest reported improvement is in Doppler behavior; higher-order diffuse interactions and broader scenario applicability remain for future work.
IV. CONCLUSION
- Persistent surface-associated scattering points let diffuse components evolve continuously along transmitter and receiver trajectories.Source: §IV · p. 5
- Compared with the Sionna RT baseline, the model better reproduces measured Doppler characteristics, while delay-domain improvements remain limited.Source: §IV · p. 5
- Future work will investigate higher-order diffuse interactions, other scattering-point distributions, and applicability to other propagation scenarios.Source: §IV · p. 5