GPS Spoofing Detection With Particle Filters

ISEF Category: Embedded Systems

Subcategory: Signal Processing  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A fake GPS signal can make a device think it is somewhere else. That can break drones, phones, ships, and timing systems. You can test whether a receiver notices when the numbers stop agreeing. Your project becomes a detective story for signals.

What Is It?

GPS spoofing means sending false satellite-like signals so a receiver trusts the wrong position or time. Think of it like a bad actor reading you fake street signs while your own steps say you are walking in a different direction. Your detector looks for that mismatch.

The project pairs two sources of location clues. One comes from GPS navigation messages and signal timing. The other comes from an IMU, which is a sensor pack that tracks motion from acceleration and rotation. A particle filter then compares many possible paths and gives a consistency score. If GPS and dead-reckoning stop lining up, the score drops and your system can flag a possible spoofing event.

Why This Is a Good Topic

This is a strong science fair topic because you can test a clear yes or no claim: does your detector spot bad GPS signals better than simple threshold checks? It connects to real problems in navigation, robotics, aviation, and infrastructure timing. You can learn signal analysis, sensor fusion, and basic state estimation, all with a project that has a real security angle.

Research Questions

• How does the particle-filter consistency score change when GPS timing disagrees with IMU dead-reckoning?
• What is the effect of different spoofing patterns on detection accuracy?
• Does adding more IMU-derived motion data improve false-alarm rate?
• To what extent can a low-cost RTL-SDR capture enough signal detail for spoofing flags?
• Which consistency threshold best separates normal GPS drift from spoofing-like behavior?
• How does receiver motion, such as walking versus driving, affect detector performance?
• What is the effect of sub-Nyquist sampling on the quality of the detection signal?

Basic Materials

• Raspberry Pi with power supply and microSD card
• RTL-SDR dongle
• GPS antenna
• Low-cost IMU module, such as an MPU-6050 or similar breakout
• Breadboard and jumper wires
• Laptop for coding and analysis
• Smartphone or handheld GPS receiver for comparison logs
• Spreadsheet software for organizing results.

Advanced Materials

• Raspberry Pi or other single-board computer with GPIO access
• RTL-SDR with calibrated antenna setup
• Higher-grade IMU or inertial measurement unit
• GPS module with raw NMEA and, if available, raw observation output
• Shielded test enclosure or RF test setup for controlled signal comparisons
• Signal generator or authorized GPS test source, if your lab has one
• Time-synchronization tools for logging sensor streams
• RF analysis accessories and attenuators, if needed by your lab supervisor.

Software & Tools

• GNU Radio: Processes SDR data and helps you inspect signal chains and timing behavior.
• Python: Runs the particle filter, data cleaning, and statistical tests.
• ImageJ: Not needed for this topic, so skip it unless you make visual plots from screenshots.
• QGroundControl: Helps if you compare your detector with drone or robot navigation logs.
• LibreOffice Calc: Organizes trial data and calculates detection metrics.

Experiment Steps

1. Define the specific spoofing pattern or inconsistency you want to detect.
2. Choose the sensor signals you will compare, then map each one to a measurable feature.
3. Build a baseline model of normal GPS and IMU agreement before testing any anomaly.
4. Design the particle-filter logic that turns sensor mismatch into a single consistency score.
5. Plan controls that separate real motion error from spoofing-like signal tampering.
6. Decide how you will judge success using detection rate, false alarms, and timing lag.

Common Pitfalls

• Treating ordinary GPS drift as spoofing, which inflates false alarms.
• Logging the IMU and GPS streams with mismatched timestamps, which breaks the consistency score.
• Using too little motion data, which makes dead-reckoning too noisy to compare against GPS.
• Testing only one spoofing pattern, which makes the detector look better than it really is.
• Changing antenna position or signal strength between trials, which adds RF noise that hides the real effect.

What Makes This Competitive

A strong version of this project does more than spot a mismatch. It tests several motion conditions, several spoofing styles, and several threshold rules. You get a stronger result if you report detection accuracy, false positives, and how the system behaves when the sensor data gets noisy. A well-designed comparison against a simple baseline can make the work feel much more like real engineering research.

Project Variations

• Test the detector on walking, biking, and vehicle motion to see how dynamics change the consistency score.
• Compare a particle filter against a simpler Kalman-style threshold method for the same GPS and IMU data.
• Swap in different low-cost IMUs to see which sensor quality gives the cleanest spoofing flag.
• Focus on timing-only inconsistencies instead of position errors, then see whether the detector still works.

Learn More

• NASA CDDIS GNSS resources: Search NASA's Earth data and GNSS documentation for background on satellite navigation signals and timing.
• NIH PubMed: Search review articles on GPS spoofing, sensor fusion, and inertial navigation.
• NOAA GPS reference materials: Look for GPS and satellite timing explanations in NOAA educational and technical pages.
• MIT OpenCourseWare: Search for signals, systems, estimation, and robotics courses that cover sensor fusion and filtering.
• GNU Radio documentation and tutorials: Find free SDR signal-processing guides and flowgraph examples on the GNU Radio site.
• USGS GPS and geodesy resources: Search USGS pages for satellite positioning concepts, error sources, and practical applications.