Fighting Fraud in the GNSS World: How onocoy Plans to Keep Data Honest
Aug 18, 2025
TL;DR: Fraudsters might try to fake it, but with smart checks and AI-powered defenses, onocoy is making sure they don’t make it.
Imagine you’re building a huge community project where thousands of people share valuable data every second of the day. Everyone who contributes gets rewarded, and the data powers critical applications like navigation, robotics, and geolocation services. Sounds great, right?
But here’s the catch: whenever there’s money or rewards involved, someone will eventually try to game the system. In their article, a16z describes this general issue amongst DePIN projects.
That’s exactly the challenge we as onocoy are preparing for in our mission to transform the GNSS (Global Navigation Satellite System) industry. By creating a dense, community-driven global network of GNSS reference stations, onocoy enables precise positioning for everyone. But with thousands of contributors feeding data into the system, there’s always the risk that bad actors might try to cheat by injecting fake GNSS data to earn rewards they don’t deserve.
This kind of fraudulent activity doesn’t help anyone, it adds no value to the community and only serves personal gain. So, the big question is: how do you spot fake signals in a sea of real data?
The Problem: Fraudulent GNSS Data Injection
Think of onocoy’s GNSS network like a massive orchestra. Each station (a “musician”) contributes its own notes (satellite signals) that, together, create a symphony of accurate global positioning data.
Fraudulent actors are like someone trying to fake playing an instrument while still collecting the paycheck. They inject synthetic or manipulated GNSS signals into the network with the sole aim of being rewarded. If undetected, this could:
Corrupt the system
Undermine user trust
Waste rewards on bad data
Clearly, that’s not acceptable for a system built on community value to serve high demands.
Traditional Defense: Spotting the Fakes
The good news? Fake signals often leave fingerprints. Here are some of the clever checks that catch fraud:
1. Geographic & GNSS Consistency Checks
Location sanity tests: Is the station really where it says it is? Satellite measurements (pseudoranges) should match the claimed coordinates.
Stability monitoring: A real station doesn’t wander — its position should only vary by a few centimeters over time. Big jumps are suspicious.
Sky plot checks: From any given spot on Earth, only certain satellites are visible at certain angles. If a station reports seeing satellites that shouldn’t be visible, that’s a red flag.
Neighbor comparisons: By cross-checking estimated parameters with trusted nearby stations, inconsistencies can be quickly spotted.
2. Signal Quality & Raw Data Validation
Noise patterns: Real-world signals always contain some noise, and the noise might be irregular. Synthetic streams often look too perfect.
Multipath signatures: Every antenna has a unique “echo” profile, like a fingerprint, shaped by reflections from its surroundings. Fake data lacks this.
Timing quirks: Real stations have tiny irregularities in timing. Fakes can be unnaturally precise. The pattern of the receiver clock is an important parameter here.
Slip statistics: Genuine signals occasionally “slip” due to atmospheric or environmental effects. These temporary losses of lock are called cycle slips. If everything looks flawless, it might actually be fake.
These methods are powerful, but they often look at individual pieces of data in isolation. And fraudsters are clever; over time, they may learn to mimic these patterns.
A Smarter Solution: Teaching Machines to Detect Fraud
On top of chasing fraud one variable at a time, onocoy is taking an additional holistic approach, together with the University of Zurich (ETH): machine learning.
Here’s why it makes sense:
Every GNSS receiver provides four core types of data for every satellite it tracks:
Code pseudo-range (distance to the satellite)
Carrier phase (cycle counts, a super-precise distance measure)
Doppler shift (frequency changes due to relative motion)
Signal-to-noise ratio (signal quality)
With a typical view of the sky, a single station can see 100+ signals at once — that’s 400 variables per second! Multiply that by tens of thousands of stations in the onocoy network, and you’ve got a treasure chest of data.
This massive dataset can be used to train deep learning models to recognize what “real” GNSS looks like, with all its natural imperfections, quirks, and fingerprints. Then, if a stream of data doesn’t match these patterns, the AI can flag it as potentially fraudulent. And as fraud methods might get more sophisticated, AI can learn on the go to keep up with the new methods.
Backed by ESA: Taking Fraud Prevention to the Next Level
This approach isn’t just theory. We have already presented a proposal to the European Space Agency’s Navigation Innovation and Support Program (ESA NAVISP), which was accepted. The project’s mission is to investigate and develop the most effective AI techniques for fraud detection in the GNSS ecosystem.
In this project, we are joining forces with top experts in the field:
Benedikt Soja, professor of Space Geodesy at ETH Zurich, published numerous frequently cited papers about GNSS and AI in the last years
Rodrigo Leandro, technology executive at u-blox, long experience in GNSS technology and PhD in Spatial Geodesy
Alejandro Pérez Conesa, Associate Professor at Universitat Autònoma de Barcelona
By combining traditional consistency checks with advanced AI, this team is building a robust shield against fraud. This ensures that rewards go to honest contributors and that the GNSS data users rely on remains accurate and trustworthy.
Why This Matters
For the everyday person, this might sound like a technical arms race. But the stakes are high: GNSS data underpins everything from smartphone maps to autonomous vehicles to precision agriculture. Keeping this data reliable isn’t just about fairness — it’s about ensuring that critical services we all depend on remain accurate.
onocoy’s proactive stance shows that community-driven innovation doesn’t mean lowering standards. In fact, it’s proof that with the right tools, transparency and trust can be scaled globally.