A proposed model for urban links between terrestrial networks and non-terrestrial networks (NTN) produced higher simulated coupling gains than the current 3GPP model across the carrier frequencies and terminal-height distributions tested. In this analysis, coupling gain is the simulated link measure used to compare the two models. The authors associate the result with lower clutter loss and a higher probability of line of sight, or LOS, for terminals placed higher above the ground. In specific situations, the clutter-loss reduction can exceed 20 dB. The finding comes from a model comparison under stated simulation assumptions, not a field test.
A model built around height and angle
The work targets large-scale channel models, using terminal height and elevation angle as inputs for above-ground devices such as drones. It covers S-Band and Ka-Band frequencies and is intended to represent urban terrestrial-to-NTN communication.
Inside the synthetic city
To build the underlying data, the researchers used an open-source 3D scene generator with OpenStreetMap maps. The system randomly assigned building materials as 80% brick-stone, 10% concrete and 10% marble, limited building heights to 22.5 metres and produced 20 Urban scenes.
Those scenes were examined with Sionna ray tracing. The path-gain samples represented outdoor terminals at heights up to 300 metres and elevation angles up to 90 degrees. The analysis treated total path gain as free-space gain minus a clutter-loss term plus random shadowing, separating the free-space component from the effects represented by clutter loss and shadowing.
To measure line of sight, the analysis classified a terminal as LOS when the ray was unobstructed. It calculated the LOS share for each location at each height and elevation, then averaged those ratios across locations. Shadow fading was summarized by the standard deviation of measured path gains, while clutter loss was the median gap between free-space and measured path gains.
Instead of applying one expression everywhere, the clutter-loss and shadow-fading models split the data by line-of-sight and non-line-of-sight conditions and by whether the terminal was below or above the rooftops.
A close fit, with a measured caveat
The resulting analytical functions were described as showing excellent agreement with the ray-tracing data for LOS probability, shadow fading and clutter loss. That description is qualitative. The report does not provide a numerical fit error, confidence interval or independent validation result to quantify the agreement.
One specific feature appeared in the LOS case. The paper says LOS clutter loss remained small across elevation and terminal height and suggests setting it to 0, matching the value used in the current 3GPP model.
What the comparison assumed
For the broader coupling-gain comparison, the authors used carrier frequencies of 2 GHz and 28 GHz. The simulation covered a 1000-metre radius, used uniformly distributed terminal heights, assigned a 100% outdoor-terminal probability and fixed terminal elevation at 40 degrees. It did not model fast fading and reported 100,000 terminal drops.
Under those conditions, the proposed model consistently yielded higher coupling gains than the current 3GPP model across the tested frequencies and terminal-height distributions. The authors identify two associated features: lower clutter loss, with reductions that can exceed 20 dB in specific situations, and higher LOS probability for elevated terminals.
The gap between simulation and deployment
The evidence comes from generated Urban scenes and Sionna ray-tracing outputs rather than field measurements. The paper describes 20 Urban scenes, limits building heights to 22.5 metres and models terminals as 100% outdoor. The supplied analysis therefore does not establish how well the parameterization generalizes beyond the described Urban scenarios, terminal conditions and frequency settings.
The coupling-gain comparison also leaves out fast-fading effects, so it does not show how rapid signal variation would change the result. The report gives no independent validation set, numerical fit errors or uncertainty intervals. Whether the proposed expressions hold beyond the selected scenes and assumptions remains open.
The next test is outside the city model
The authors say the aerial-terminal model may also apply to in-building terminals and plan to extend it to Dense Urban, Suburban and Rural scenarios, using more refined scene generation and ray-tracing accuracy. Those applications are proposed next steps, not results demonstrated by the present analysis.
Taken together, the preprint describes a height- and elevation-based way to represent urban terrestrial-to-NTN channel parameters, with higher simulated gains than the current 3GPP model under the tested conditions. That is a modeling result, not a demonstration of real-world network performance. Independent field measurements and tests in additional environments remain open questions.
The manuscript is a preprint on arXiv and says it was submitted to IEEE for possible publication. The acknowledgment reports support from Germany's BMFTR under grant 16KIS2424, identified as 6G-Coverage.
Paper data and sources
Original title: Enhancing 3GPP Urban Channel Models For Terrestrial-to-Non-Terrestrial Communication
Authors: Gerhard Schreiber, Chenrui Sun, Joerg Schaepperle et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-28
DOI: Not available
Original paper · Full text