⚡ electrical engineering

Space-Based GNSS Radio Frequency Interference Detection Evaluation Through Multi-Satellite Data Integration

This study evaluates the impact of constellation size on space-based GNSS radio frequency interference detection using CYGNSS data, demonstrating that a minimum of three satellites significantly reduces detection latency and improves spatial coherence compared to single-satellite deployments.

Original authors: Anouar Boumeftah, Peter Klimas, Gunes Karabulut Kurt

Published 2026-06-15

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Anouar Boumeftah, Peter Klimas, Gunes Karabulut Kurt

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Earth is a giant, quiet library where everyone is trying to listen to a very faint whisper (the GPS signal) from satellites high above. Unfortunately, some people in the library are shouting (Radio Frequency Interference, or RFI), drowning out the whispers. This paper is about building a team of "noise detectives" in space to find out who is shouting and where.

Here is the story of how the researchers tested whether having one detective is enough, or if they need a whole squad to do the job effectively.

The Setup: The CYGNSS "Squad"

The researchers used data from a real team of satellites called CYGNSS. Think of these satellites as a fleet of seven tiny drones flying in a specific ring around the Earth (mostly over the tropics and mid-latitudes). Each drone has a special ear that listens for the "forbidden" parts of the GPS signal—places where noise shouldn't exist. If it hears a loud static hiss there, it knows someone is jamming the signal on the ground.

The big question the paper asks is: Does it matter if we use all seven drones, or would just one work just as well?

The Four Ways They Tested the Team

The researchers compared a "One-Detective" scenario against a "Full-Squad" scenario (and some middle-ground teams of 3 or 5) using four different tests:

1. Speed: The "Fire Alarm" Test

The Analogy: Imagine a fire starts in a building. If you have only one security guard walking a single loop, they might not see the smoke for hours. If you have seven guards covering different angles, someone will see it almost immediately.
The Result: The full team of seven satellites found the interference 4.7 times faster than a single satellite.
- One satellite: It took a median of nearly 4 hours to find the noise.
- Seven satellites: It took less than an hour.
- Why it matters: For short, bursty shouts (like a 5-minute jamming signal), a single satellite only catches them 2% of the time. The full squad catches them 11.5% of the time. It's the difference between missing a flash of lightning entirely and actually seeing it.

2. Coverage: The "Mowing the Lawn" Test

The Analogy: Imagine mowing a huge lawn. One person with a mower takes a long time and leaves big gaps between the strips of grass. Seven people mowing at once cover the whole lawn much faster and leave almost no grass uncut.
The Result:
- One satellite: It takes about 6 hours to cover 38% of the target area. It leaves "blind spots" where no one is looking for a long time.
- Seven satellites: They cover 60% of the area in just 6 hours and nearly the whole area (99.8%) in two days.
- Revisit Time: This is how often they check the same spot. One satellite checks a spot every 5.8 hours on average. The full squad checks it every 1.8 hours. This is crucial because if the "shouter" stops and starts, you need to be looking at the right time to catch them.

3. Clarity: The "Puzzle Piece" Test

The Analogy: Imagine trying to figure out the shape of a hidden object by looking at it through a keyhole. If you only look from one angle (one satellite), you might see a blurry line and think it's a stick. If you look from seven different angles, you can see the whole picture and realize it's actually a chair.
The Result: A single satellite leaves up to 72% of the "shape" of the noise unresolved. It's like trying to solve a puzzle with only a few scattered pieces. The full team fills in the gaps, allowing them to see exactly where the interference is coming from and confirm it's a real source, not just a glitch in the satellite's ear.
- Key Finding: The biggest jump in clarity happens when you go from 1 to 3 satellites. Adding more than three helps, but the biggest "aha!" moment happens once you have a small team.

4. Persistence: The "Stalking" Test

The Analogy: Imagine trying to prove a neighbor is throwing parties every night. If you only have one person watching for 10 minutes a day, they might miss the party entirely. If you have a team watching in shifts, you can prove the party is happening every night.
The Result: The researchers looked at a real jamming event in Venezuela.
- Seven satellites: They confirmed the jamming started on October 29.
- One satellite: It didn't confirm the jamming until December 8.
- The Gap: The full team was 39 days faster at confirming the event.
- The Reliability Ceiling: For a small, specific target (like a small country), a single satellite has a "ceiling." Even if you wait forever, one specific satellite might never catch two signals in a row to prove the jamming is real. It gets stuck at an 86% success rate. The full squad removes this ceiling, guaranteeing detection.

The Main Takeaway

The paper concludes that you cannot rely on just one satellite for this job.

The Magic Number: The biggest improvements happen when you go from 1 satellite to 3 satellites. Three is the "minimum effective team" to get reliable results.
Why? One satellite is too slow, misses too many short bursts, and leaves too many blind spots. It's like trying to catch a fly with a single net in a dark room.
The Verdict: To effectively monitor the Earth for GPS interference, you need a constellation (a team) of satellites. The full seven-satellite team is the gold standard, but even a small team of three is vastly superior to a lone satellite.

In short: One detective is better than none, but a squad is what you need to actually solve the crime.

Technical Summary: Space-Based GNSS Radio Frequency Interference Detection Evaluation Through Multi-Satellite Data Integration

Problem Statement
Global Navigation Satellite Systems (GNSS) are critical infrastructure for positioning, navigation, and timing (PNT), yet their low-power signals are inherently vulnerable to Radio Frequency Interference (RFI). While space-based monitoring using Low Earth Orbit (LEO) satellites equipped with GNSS Reflectometry (GNSS-R) payloads has demonstrated the ability to detect terrestrial interference via elevated noise floors in Delay-Doppler Maps (DDMs), the specific performance gains of integrating observations from multiple satellites versus a single satellite have not been systematically quantified. Operational monitoring architectures require precise metrics regarding detection latency, spatial-temporal coverage, and reliability to justify constellation sizing. This study addresses the gap in understanding how constellation size impacts the capability to detect, localize, and monitor both persistent and transient RFI sources.

Methodology
The authors developed a quantitative evaluation framework using real Level 1 data from the seven operational CYGNSS spacecraft (excluding FM06 due to an anomaly) collected over a three-month period in 2025 from the NASA PO.DAAC archive. The study employed an automated data ingestion pipeline to process DDM data, applying geometric corrections for the 28° off-nadir antenna pointing and projecting data onto a geographic grid.

The evaluation focused on four key performance dimensions:

Temporal Coverage: Analysis of detection latency for persistent sources, probability of intercept (POI) for transient emissions (via Monte Carlo simulation), and revisit intervals for high-intensity hotspots.
Spatial Coverage: Assessment of the fraction of the ±40° latitude zone covered over time and the mean revisit time for grid cells.
Geometric Coherence: Use of Normalised Residual Variance (NRV) to quantify how well the spatial distribution of noise floor elevations conforms to a physically consistent point-source model (Gaussian template) across different constellation sizes ( $N=1, 3, 5, 7$ ).
Persistence Monitoring: Application of a detection algorithm using a rolling baseline of quiet ocean regions to calculate z-scores. Detections required a "streak" of consecutive anomalous windows (using a nanmean aggregation strategy) to confirm activity, while a nanmax strategy was used to analyze sensitivity limits.

Key Contributions
The paper makes four primary contributions:

Quantitative Framework: It establishes a method to compare single-satellite and multi-satellite GNSS-R RFI detection performance using real-world CYGNSS observations.
Constellation Characterization: It characterizes performance improvements across operational dimensions, demonstrating that the value of a constellation extends beyond simple data availability to include significant gains in latency, coverage, and reliability.
Data Pipeline: It implements an automated pipeline for retrieving, filtering, and processing CYGNSS Level 1 data, generating analysis-ready datasets for large-scale RFI studies.
Visualization Environment: It provides an interactive geospatial environment for exploring detected RFI activity and constellation coverage patterns.

Results
The analysis of the seven-satellite constellation versus single-satellite baselines yielded the following results:

Detection Latency: The full constellation ( $N=7$ ) reduced median detection latency by a factor of 4.7 compared to a single satellite ( $N=1$ ), from 3 hours 46 minutes to 48 minutes. For 5-minute transient emissions, the probability of intercept increased from 2.0% ( $N=1$ ) to 11.5% ( $N=7$ ).
Revisit Time: Median footprint revisit time improved from 5.8 hours ( $N=1$ ) to under 2.0 hours ( $N=7$ ). The 90th-percentile revisit time decreased from 10.5 hours to 1.5 hours.
Spatial Coverage: While a single satellite covers 98.4% of the target zone within 48 hours, the full constellation achieves 99.8%. More critically, the full constellation reduces the number of unvisited grid cells from 1.6% to 0.2% and significantly increases the density of passes per cell (from ~8 to ~49).
Spatial Coherence: A single satellite left up to 72% of a source's spatial structure unresolved (high NRV). The NRV decreased monotonically with constellation size, with the steepest improvement occurring between $N=1$ and $N=3$ . This indicates that a single satellite often cannot distinguish a genuine point source from sampling artifacts due to anisotropic ground track coverage.
Persistence Monitoring: For a verified interference event in Region H1 (Venezuela), the full constellation confirmed the onset 39 days earlier than the single-satellite deployment. For a compact target (Region H5), a single-satellite deployment had a hard reliability ceiling of 86% (one of seven satellites never achieved the required consecutive passes), whereas $N=3$ and $N=7$ achieved 100% reliability.

Significance and Claims
The paper claims that the most significant performance gains occur between one and three satellites, establishing three satellites as the minimum effective constellation size for operational RFI monitoring. Below this threshold, the system operates in an "information-limited regime" where orbital geometry, rather than signal physics, governs detection failure.

The authors assert that while single-satellite systems are valuable for mapping and historical analysis, they are insufficient for real-time threat response due to sparse temporal sampling and the inability to guarantee detection for compact targets. The study concludes that effective operational GNSS interference monitoring requires multi-satellite architectures to ensure both rapid detection latency and high reliability, particularly for transient events and geographically constrained sources. The findings provide empirical evidence to inform future mission design and constellation sizing for space-based RFI monitoring systems.