Open Reproducible Ray-Tracing Pipeline for 5G NR Multicell Digital Twins: From Public 3D City Data to an Interactive Artefact

Ray-traced radio digital twins are usually built on proprietary city scenes and paid compute, which makes them hard for students and independent researchers to reproduce. We describe an end-to-end pipeline that runs a 5G NR multicell ray trace (Sionna RT 2.2.0) on free resources: a city scene built from open municipal data, a free-tier Colab T4 GPU, and an interactive web artefact that reads the trace output directly. The scene, 1,471 buildings extruded from the City of San Francisco's published footprints at their LiDAR-measured heights, over public terrain data, is itself redistributable, so every input to the study can be downloaded. For a three-site, nine-sector layout at 3.6 GHz over a 1 km × 1 km tile at Telegraph Hill / North Beach, 80.8% of street grid points obtain a path but only 24.3% reach an RSRP of −90 dBm (44.2% at −110 dBm), and the median best-server SINR is 4.8 dB. The deficit is concentrated spatially: within 150 m of the one site with open sightlines (B), 91% of points are covered, against 13% and 28% around the two sites whose antennas sit within 2 m of the local roofline (A, C). Varying tilt, the antenna height reference and solver depth moves tile-wide coverage by at most 1.6 points, whereas coverage differs by 78 points between sites of identical antenna height: the scene, and the sites' relation to it, dominate the result. We do not validate against field measurements; absolute levels are model outputs, and the package is offered as a reproducible instrument for teaching and relative comparison, not as a coverage prediction. The notebook, site file, scene and its build script, results and artefact are openly licensed, and the site file defines tower-shift experiments for classroom use. We present the study as a demonstration: the coverage figures are the pipeline's outputs on this scene, shown so that a reader can see what the pipeline produces end to end.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24210759
Primary Topic
Millimeter-Wave Propagation and Modeling
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Open Reproducible Ray-Tracing Pipeline for 5G NR Multicell Digital Twins: From Public 3D City Data to an Interactive Artefact

Mubanga Nsofu, Ram Asokan, Rao S. Yenamandra, Sameer Lalwani
Zenodo (CERN European Organization for Nuclear Research)
Millimeter-Wave Propagation and Modeling
preprint

Open Reproducible Ray-Tracing Pipeline for 5G NR Multicell Digital Twins: From Public 3D City Data to an Interactive Artefact

Mubanga Nsofu, Ram Asokan, Rao S. Yenamandra, Sameer Lalwani
preprint en

Abstract

Ray-traced radio digital twins are usually built on proprietary city scenes and paid compute, which makes them hard for students and independent researchers to reproduce. We describe an end-to-end pipeline that runs a 5G NR multicell ray trace (Sionna RT 2.2.0) on free resources: a city scene built from open municipal data, a free-tier Colab T4 GPU, and an interactive web artefact that reads the trace output directly. The scene, 1,471 buildings extruded from the City of San Francisco's published footprints at their LiDAR-measured heights, over public terrain data, is itself redistributable, so every input to the study can be downloaded. For a three-site, nine-sector layout at 3.6 GHz over a 1 km × 1 km tile at Telegraph Hill / North Beach, 80.8% of street grid points obtain a path but only 24.3% reach an RSRP of −90 dBm (44.2% at −110 dBm), and the median best-server SINR is 4.8 dB. The deficit is concentrated spatially: within 150 m of the one site with open sightlines (B), 91% of points are covered, against 13% and 28% around the two sites whose antennas sit within 2 m of the local roofline (A, C). Varying tilt, the antenna height reference and solver depth moves tile-wide coverage by at most 1.6 points, whereas coverage differs by 78 points between sites of identical antenna height: the scene, and the sites' relation to it, dominate the result. We do not validate against field measurements; absolute levels are model outputs, and the package is offered as a reproducible instrument for teaching and relative comparison, not as a coverage prediction. The notebook, site file, scene and its build script, results and artefact are openly licensed, and the site file defines tower-shift experiments for classroom use. We present the study as a demonstration: the coverage figures are the pipeline's outputs on this scene, shown so that a reader can see what the pipeline produces end to end.

Zenodo (CERN European Organization for Nuclear Research)
Wake Forest University (US)
Millimeter-Wave Propagation and Modeling
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon
Open Reproducible Ray-Tracing Pipeline for 5G NR Multicell Digital Twins: From Public 3D City Data to an Interactive Artefact — Mubanga Nsofu, Ram Asokan, et al. · Zenodo (CERN European Organization for Nuclear Research) (2026) | TGRS Research Map | TGRS