image: UAlbany's Mustafa Aksoy is leading the development of machine-learning algorithms that can detect and remove radio frequency interference that impairs the accuracy of U.S. satellite measurements.
Credit: UAlbany
ALBANY, N.Y. (Nov. 20, 2025) — As the National Oceanic and Atmospheric Administration prepares to launch the next generation of satellites responsible for U.S. weather forecasting and climate monitoring, it’s asking researchers for help ensuring accurate measurements in the face of growing radio frequency interference from wireless technologies.
Mustafa Aksoy, an associate professor in the University at Albany’s College of Nanotechnology, Science, and Engineering, is part of a large research team that was awarded a two-year, $1.1 million grant to develop radio frequency interference detection and mitigation strategies for future NOAA satellites. Other collaborators on the project include NASA Jet Propulsion Laboratory, NASA Goddard Space Flight Center, the Ohio State University and Noctua Technologies.
Aksoy, who joined UAlbany’s Department of Electrical & Computer Engineering in 2017 and directs the Microwave Remote Sensing Laboratory, is leading the development of machine-learning algorithms for NOAA that can detect and remove radio frequency interference from its satellite measurements, improving the accuracy of weather forecasts and climate monitoring.
“When you think of big weather events like hurricanes, drought, flooding, when you think of climate change, these events cannot be understood if you don’t have reliable atmospheric measurements,” Aksoy said. “And one of the biggest threats to reliable measurements right now is radio frequency interference. Even the smallest interference can bias your measurements, giving you a completely different estimation that may cost billions of dollars.”
Radio Frequency Interference
Weather and climate modelers rely on the radio frequency spectrum to make and communicate sensitive observations about Earth’s atmosphere, oceans and surface. Satellites and other wireless communication technologies transmit this data over radio waves, an invisible portion of the electromagnetic spectrum with the longest wavelengths and lowest frequencies, ranging from 3 kilohertz to 300 gigahertz.
Because this portion of the spectrum is finite, only so many users can operate on a specific frequency at a specific time. And because radio waves are foundational to wireless communications, which have experienced exponential growth in recent decades, a growing number of users are now competing for a limited number of radio frequencies.
This has led to a rise in radio frequency interference, or RFI, which occurs when unwanted radio signals disrupt the normal operation of other technologies using the spectrum. Unwanted signals may come from cell phones, power lines, GPS and vehicle radar, as well as natural sources like lightning, solar flares and auroras. This interference can lead to disruptions in service, and can degrade or even block data from being collected by weather satellites and radars.
“There are plenty of interference sources, and they change from country to country and region to region because different countries may give different licenses to different users at different frequencies,” Aksoy said.
Governments and regulatory bodies tasked with managing the radio frequency spectrum attempt to prevent RFI by licensing who can use the spectrum and when, but unwanted signals continue to occur as the spectrum becomes increasingly crowded and complex.
Commercial users who can profit from their technologies are also crowding out scientific use of the spectrum, which is not immediately profitable but critical for delivering accurate and life-saving weather forecasts and climate tracking.
Detecting and Removing RFI
Aksoy was tapped by NOAA to develop advanced strategies for detecting and mitigating RFI using sophisticated machine-learning algorithms.
“We are applying machine learning algorithms because we realized traditional algorithms we used in previous studies are not efficient enough in this very crowded radio spectrum,” he said. “It’s getting more crowded and more complicated. So traditional algorithms are not enough anymore.”
Aksoy’s work in this area dates back to his time as a PhD student at The Ohio State University, where he worked on research to assist NASA satellites with accurately assessing moisture levels in Earth’s soil, a critical measurement for understanding Earth’s climate.
Under this new project, NASA’s Jet Propulsion Laboratory is developing atmospheric models that can simulate what atmospheric radiation would look like without interference. Aksoy, in turn, is training algorithms on these interference-free models so that they are able to detect and correct for when an anomaly occurs.
“So our algorithm will know what radiation should be without interference, and when there is interference, it will deviate from that expectation and the algorithm will detect it as an anomaly,” he said.
Grant funding for the first phase of the research lasts through February 2027. The next phase will likely involve hardware development, Aksoy said.
“Right now we are building the algorithms,” he said. “But our next step is the hardware, so that eventually these algorithms are built right into future satellites.”