Researchers found that by listening to the signals of eight LEO satellites for about 10 minutes, their algorithm could achieve unprecedented accuracy in locating a stationary receiver on the ground and was able to converge on it with an error of only about 5.8 meters.

The research, led by Zak Kassas, a professor of electrical and computer engineering at The Ohio State University and director of the Department of Transportation Center for Automated Vehicles Research with Multimodal AssurEd Navigation (CARMEN), was presented last week at the IEEE/ION Position Location and Navigation Symposium (PLANS) 2023 conference in Monterey, California. Along with Ohio State PhD students Sharbel Kozhaya and Haitham Kanj, the paper, which demonstrated the first ever exploitation of unknown OneWeb LEO satellite signals, won the conference’s Best Student Paper award. 

The researchers did not need assistance from the satellite operators to use the signals, and they emphasized that they had no access to the actual data being sent through the satellites – only to publicly available information related to the satellites’ downlink transmission frequency and a rough estimate of the satellites’ location.

From transportation to communication systems to the power grid and emergency services, nearly every aspect of modern society relies on positioning, navigation and timing data from global navigation satellite systems (GNSS), or GPS, that orbit the Earth. Despite this, because GPS system signals are weak and susceptible to interference, they can often become unreliable in certain places such as indoor environments or in deep urban canyons. In addition, GNSS signals are spoofable, which poses serious security risks in safety-critical applications, such as aviation.

In the long term, such complications could lead to a number of navigational and cybersecurity issues, especially as virtually all of our current systems rely heavily on GPS, Kassas said. Technologies on the rise, such as autonomous vehicles, he noted, are beginning to amplify the limitations of our current GNSS systems.

“It’s becoming more pressing to find civilian and military alternatives to GPS, whether as a backup or in the case when GPS isn’t there whatsoever,” said Kassas. 

This study builds on previous research by Kassas’ lab that solely used six SpaceX satellite signals to pinpoint a location within 10 meters of accuracy, which was recently reduced to 6.5 meters.

“The Starlink study scratched the surface of what is possible,” said Kassas.

His work suggests utilizing signals from LEO satellites as an alternative for humans’ positioning, navigation and timing needs, as they reside about 20 times closer to Earth compared to GNSS satellites, which reside in medium Earth orbit – a little more than 20,000 kilometers above the planet. According to Kassas, the technology could potentially usher in a new era of positioning, navigation and timing. 

“We are witnessing a space renaissance. Tens of thousands of LEO satellites will be launched into space over the next decade, leading to what is referred to as mega-constellations,” he said. “Signals transmitted by these satellites will revolutionize numerous technologies and benefit scientific inquiry in fields such as remote sensing.” 

What also makes the study so different from all other attempts at creating an alternative to GPS is, unlike previous studies, this algorithm doesn’t reverse engineer the signal, said Kassas.  

“Our algorithm is agnostic to the LEO constellation,” said Kassas. “Our receiver can listen to virtually any satellite signal, trains on the data it’s receiving on-the-fly, then deciphers certain features of the signal in a way where we can reconstruct what they are transmitting into location data.” To demonstrate the team’s new approach, the team applied the algorithm to four different LEO satellite constellations: Starlink, OneWeb, Orbcomm and Iridium. The algorithm cracked all these signals, with virtually no prior knowledge about what is being transmitted.

Additionally, their algorithm is so sophisticated that the researchers were also able to estimate where the satellites are in space. In order to use the satellite to position ourselves, we need to know where the satellite is located. “That’s a very challenging problem because LEO satellites don’t normally broadcast their location, and our publicly available estimates of where they are is off by a few kilometers,” said Kassas.

During a stationary experiment to test how the signals worked as an accurate positioning system, researchers set a ground receiver's initial position estimate to the roof of an engineering parking structure at the University of California, Irvine, a spot more than 2,000 miles away from its actual position: the roof of Ohio State’s Electroscience Laboratory (ESL) in Columbus, Ohio. Using the satellite constellations to guess where exactly in the country the receiver actually was, the algorithm was only off by about 5 meters. 

In another experiment, the researchers tested how the algorithm would fare on a moving vehicle, and mounted the receiver onto the top of a car. First, they used today’s navigation technology, which relies on a GPS receiver coupled with an inertial navigation system (INS). They navigated for about 100 meters before cutting GPS off, after which they drove for nearly a kilometer. They found that by relying on today’s GPS-INS system, they were said to be located about 500 meters away from their true location, but with their algorithm, they were found about 4.4 meters away. “Our result showed that our system is getting close to what you can do with GPS today,” said Kassas.

Although a patent has been filed on the algorithm, the team does plan to continue evolving all of the algorithm’s technical abilities, said Kassas.

“GPS is a very mature system that we trust with our lives,” he said. “To be able to trust new types of signals with our lives, there will have to be more studies on their accuracy, integrity and continuity.”

Other Ohio State co-authors were Joe Saroufim, Samer Hayek and Mohammad Neinavaie. The work was supported by the Office of Naval Research, the Air Force Office of Scientific Research, the Department of Transportation and the National Science Foundation.

Over the last few decades, rising global temperatures have caused many natural phenomena like hurricanes, snowstorms, floods and wildfires to grow in intensity and frequency. 

While humans can’t prevent these disasters from occurring, the rapidly increasing number of satellites that orbit the Earth from space offers a greater opportunity to monitor their evolution, said C.K Shum, co-author of the study and a professor at the Byrd Polar Research Center and in earth sciences at The Ohio State University. He said that potentially allowing people in the area to make informed decisions could improve the effectiveness of local disaster response and management. 

“Predicting the future is a pretty difficult task, but by using remote sensing and machine learning, our research aims to help create a system that will be able to monitor these climate-induced hazards in a manner that enables a timely and informed disaster response,” said Shum. 

Shum’s research uses geodesy — the science of measuring the planet’s size, shape and orientation in space — to study phenomena related to global climate change. 

Using geodetic data gathered from various space agency satellites, researchers conducted several case studies to test whether a mix of remote sensing and deep machine learning analytics could accurately monitor abrupt weather episodes, including floods, droughts and storm surges in some areas of the world. 

In one experiment, the team used these methods to determine if radar signals from Earth’s Global Navigation Satellite System (GNSS), which were reflected over the ocean and received by GNSS receivers located at towns offshore in the Gulf of Mexico, could be used to track hurricane evolution by measuring rising sea levels after landfall. Between 2020 and 2021, the team studied how seven storms, such as Hurricane Hana and Hurricane Delta, affected coastal sea levels before they made landfall in the Gulf of Mexico. By monitoring these complex changes, they found a positive correlation between higher sea levels and how intense the storm surges were.

The data they used was collected by NASA and the German Aerospace Center’s Gravity Recovery And Climate Experiment (GRACE) mission, and its successor, GRACE Follow-On. Both satellites have been used to monitor changes in Earth’s mass over the past two decades, but so far, have only been able to view the planet from a little more than 400 miles up. But using deep machine learning analytics, Shum’s team was able to reduce this resolution to about 15 miles, effectively improving society’s ability to monitor natural hazards. 

“Taking advantage of deep machine learning means having to condition the algorithm to continuously learn from various data inputs to achieve the goal you want to accomplish,” Shum said. In this instance, satellites allowed researchers to quantify the path and evolution of two Category 4 Atlantic hurricane-induced storm surges during their landfalls over Texas and Louisiana, Hurricane Harvey in August 2017 and Hurricane Laura in August 2020, respectively.

Accurate measurements of these natural hazards could one day help improve hurricane forecasting, said Shum. But in the short term, Shum would like to see countries and organizations make their satellite data more readily available to scientists, as projects that rely on deep machine learning often need large amounts of wide-ranging data to help make accurate forecasts.   

“Many of these novel satellite techniques require time and effort to process massive amounts of accurate data,” said Shum. “If researchers have access to more resources, we’ll be able to potentially develop technologies to better prepare people to adapt, as well as allow disaster management agencies to improve their response to intense and frequent climate-induced natural hazards.”

Co-authors of the project were Yu Zhang, Yuanyuan Jia, Yihang Ding and Junyi Guo of Ohio State; Orhan Akyilmaz and Metehan Uz of Istanbul Technical University; and Kazim Atman of Queen Mary University of London. This work was supported by the United States Agency for International Development (USAID), the National Science Foundation (NSF), the National Aeronautics and Space Administration and the Scientific and Technological Research Council of Türkiye (TÜBİTAK).

The tool could be used to detect significant natural disasters in remote regions of the globe soon after they happen, giving first responders accurate information about the needs of the region affected.

The Planetscope satellite constellation, operated by the satellite data company Planet, collects weekly and sometimes even daily images of the entire globe. On average, its fleet of Cubesats, or miniature satellites, has about 1,700 images of every location on Earth. The data they capture has been used to monitor the spread of wildfires, detect changes in crop health and survey areas of deforestation.

That kind of global coverage is unprecedented, said Rongjun Qin, co-author of the study and an associate professor of civil, environmental and geodetic engineering at The Ohio State University.

“There are a lot of great benefits in terms of having satellites cover the globe very quickly,” said Qin, who is also a core faculty member of Ohio State’s Translational Data Analytics Institute. “We’re focused on informing the community about changes to our cities, forests and ecosystems.”

The study, published in the Journal GIScience and Remote Sensing, found that Planetscope’s vast datasets could be used to create 3D reconstructions, or digital surface models, of any given area.

“Remote sensing could help us estimate the area impacted by a natural disaster,” Qin said. “We could figure out how many people to send over for rescue operations, and observe the level of damage these events actually create.”

Previous remote-sensing-based disaster studies have been limited by their lack of available data and coverage and their resolution, or how frequently images are collected or updated.

For instance, many people are familiar with Google Earth, a computer program that renders a 3D representation of the globe using satellite images and aerial photography. But the popular program is one of the reasons why Qin’s team wanted to create a model capable of a much higher resolution, or update rate.

“Some places on the site have very nice 3D reconstructions,” Qin said. “But there are a lot of places where those images attached to the Earth are distorted.”

Purely flat images overlaid on a globe can make objects or locations on the map appear out of scale with each other, and negatively influence the entire program’s accuracy.

However Qin’s 3D reconstructions, which take into account different elevation levels and landscapes, are accurate down to about 6 meters from the ground. In terms of mapping data, he said it’s akin to achieving “almost approximately one pixel accuracy.”

And because Planetscope’s data is open access to educators, other scientists can use the same datasets the study used to create their own simulations. According to Qin, for an area as big as Ohio State’s Columbus campus (1,600 acres), it would take less than an hour to turn satellite images into an accurate 3D reconstruction of the region.

But to put their method to the test, Qin’s team devised three different case studies, or experiments using thousands of Planetscope images collected between 2016 and 2021.

One test case showed that they could use the satellite images to make a 3D reconstruction of an urban and a rural area in Spain. A second test case showed that they could detect 3D changes over time in an urban and a forested area near Allentown, Pennsylvania.

To determine how good their model was at post-disaster assessment, one experiment investigated a glacial area in Chamoli, India.

Last year, the area around the region experienced a devastating flood that killed hundreds of people and destroyed two nearby power plants. Advances in satellite technology later revealed that the flood was caused by a rock and ice avalanche.

Their results showed that their model could not only recreate the changed topography that led to the disaster, but account for the volume of the rocks and ice in the avalanche. “We verified that Planetscope’s digital surface model can be used to evaluate mass changes for similar global natural disasters to the avalanche event,” said Qin.

Qin’s findings will help engineer better ways to utilize satellite data, especially as the number of satellites and their various applications grow.

“This is still in its incubation stage and will still require some engineering efforts,” he said, “but I think it’s going to be a big deal in the industry and for scientists interested in combating climate change.”

The research was supported by the Office of Naval Research. Co-authors include Debao Huang and Yang Tang of Ohio State.