When Satellite Internet Meets Weather Forecasting: A Spectrum Coexistence Warning
When Satellite Internet Meets Weather Forecasting: A Spectrum Coexistence Warning
A research paper under review for IEEE Global Communications Conference (GLOBECOM) 2026
The satellites that beam internet down to your home and the satellites that help forecast tomorrow’s weather may soon be competing for the same sliver of the radio spectrum—and according to new research, that competition could come at a cost to weather prediction.
In this study, researchers Yankai Peng and J. Nicholas Laneman of the Department of Electrical Engineering at the University of Notre Dame examine what happens when SpaceX’s Starlink seeks to expand into a new frequency band that sits dangerously close to one used by critical Earth-observing instruments. The work was supported by SpectrumX, the National Science Foundation’s Spectrum Innovation Center, and submitted to the IEEE Global Communications Conference 2026.
The Setup: A New Land Rush for Spectrum
As mega-constellations like Starlink grow, they need more radio spectrum to connect their satellites to the ground. The “gateway” stations that link the internet backbone to the constellation are particularly hungry for bandwidth. Reflecting this, the FCC has opened a proceeding to consider letting these gateways transmit in the 51.4–52.4 GHz V-band, and an international ITU process (WRC-27) is weighing the same question.
The problem is that this band sits right inside the 50–60 GHz oxygen absorption band—a region of the spectrum that atmospheric scientists rely on to take the planet’s temperature. Instruments like the Advanced Technology Microwave Sounder (ATMS), flown by NOAA and NASA, listen for faint natural thermal radiation from the atmosphere to build the vertical temperature profiles that feed modern weather forecasts. Channel 4 of ATMS, centered at 51.76 GHz, is especially valuable for sensing near-surface conditions in cloudy, humid weather.
Here lies the vulnerability: passive sensors like ATMS only listen—they never transmit. They detect signals so weak that even a small amount of stray emission from human-made transmitters can corrupt their readings. And because a single weather satellite sweeps over an entire continent, it can pick up unwanted signals from many gateways at once.
The Innovation: A Realistic Interference Simulator
Earlier studies of this kind of interference were narrow—looking only at older geostationary satellites, or at adjacent bands rather than the band itself. The Notre Dame team built a more realistic and comprehensive simulation framework that combines three ingredients drawn from real-world data:
First, they modeled how ATMS actually scans the Earth. Rather than treating each measurement as a single frozen snapshot, they recognized that the sensor keeps moving during its roughly 18-millisecond integration window. They captured this by slicing each measurement into 19 sub-snapshots—a dynamic field-of-view model that more faithfully represents how the sensor sweeps across the ground.
Second, they modeled the Starlink gateways themselves—145 sites and over 4,600 antennas across the U.S.—using actual FCC licensing filings. Since no public data exists on how powerful V-band gateways would be, the team cleverly inferred the transmit power by assuming the new V-band links would need to match the signal quality of Starlink’s existing Ka-band links, working backward through a link-budget calculation.
Third, they used real orbital data to model where each gateway antenna would point, since gateways track satellites as they pass overhead.
Running roughly 1.4 million simulated measurement intervals over one week of real ATMS data, the framework produced a statistical picture of how often, and by how much, interference would exceed the international protection limit (ITU-R RS.2017).
The Findings: A Large Margin of Concern
The results are stark. Across the entire range of transmit power levels studied, the simulated interference exceeded the protection threshold—even at the lowest power setting, by more than 52 dB (a factor of over 100,000). At SpaceX’s filed maximum power, the exceedance grew to nearly 58 dB.
A key insight explains why: the interference is not caused by the steady hum of thousands of antennas pointing every which way. Instead, it is dominated by rare, near-direct alignments—moments when a gateway antenna happens to point almost straight at the weather satellite. These near-boresight events made up just 0.03% of cases but accounted for nearly 65% of all interference power.
That insight motivated the team to test beam avoidance: simply muting a gateway antenna whenever it points too close to the sensor. While this helped, reducing interference by up to 23 dB, it was not enough. Even when half of all in-view transmissions were silenced, the interference still sat about 29 dB above the protection threshold. The conclusion is clear: beam avoidance alone cannot solve the problem.
Why It Matters and What’s Next
This work delivers a timely, evidence-based contribution to active regulatory debates at both the FCC and the ITU. It suggests that V-band gateway operation at the power levels needed for viable internet links would require stronger measures—lower power limits or more sophisticated mitigation—to protect the weather-sensing infrastructure society depends on. The authors point to exactly this as their next direction: designing mitigation strategies and gateway hardware that can shield passive sensors while still preserving gateway performance.
In short, the study is a careful, data-grounded caution that sharing the sky between internet satellites and weather satellites will take more than good intentions and simple fixes.
Authored by Yankai Peng and J. Nicholas Laneman, University of Notre Dame. Submitted to IEEE GLOBECOM 2026 (https://globecom2026.ieee-globecom.org). Supported by NSF SpectrumX (Award AST 21-32700).