Radar Sleep Heart Rate Variability Monitoring

ISEF Category: Biomedical Engineering

Subcategory: Biomedical Sensors and Imaging  ·  Difficulty: Advanced  ·  Setup: School Lab  ·  Time: Full Year

The Hook

Your chest moves less than a millimeter when your heart beats. A radar sensor can still pick that up. That means you can try to measure sleep signals without stickers, wires, or a camera. If you can separate breathing from tiny cardiac motion, you have a real biomedical engineering problem.

What Is It?

This project uses radar to sense motion without touching the body. The sensor sends out radio waves, then measures the tiny changes that bounce back from a person. Breathing makes a bigger, slower motion. Heartbeats make a much smaller, faster motion. Your job is to pull those two signals apart.

Think of it like listening to two people talking in the same room. One voice is loud and slow. The other is quiet and quick. A good model can separate them if you give it the right features and enough clean data. In this project, those features come from micro-Doppler signals, which are tiny motion patterns caused by chest movement. A state-space model is a math model that tracks how a signal changes over time, so it can help estimate respiration and cardiac activity from the same radar stream.

Why This Is a Good Topic

This is a strong science fair topic because you can test real signal processing choices, compare algorithms, and measure how well each method works on noisy human data. It connects to sleep tracking, patient monitoring, and noncontact health sensing. You can learn sensor calibration, filtering, model evaluation, and statistical comparison. You do not need a hospital lab, but you do need careful testing and good controls.

Research Questions

• How does mattress thickness affect radar signal quality for breathing and heart-rate-variability estimates?
• What is the effect of subject position, such as back, side, or slight angle, on separation of respiration and cardiac micro-Doppler signatures?
• Does a state-space model outperform a band-pass filter approach for estimating heart-rate-variability from radar data?
• To what extent does movement during sleep-like conditions reduce accuracy in radar-based cardiac signal extraction?
• Which feature set best improves classification of respiration versus cardiac motion in radar returns?
• How does sensor placement distance change the signal-to-noise ratio for through-mattress monitoring?

Basic Materials

• HLK-LD6001 radar module or similar low-cost radar sensor.
• Microcontroller or single-board computer with serial input support.
• Laptop for data logging and analysis.
• Stable bed frame or foam testing platform.
• Different mattress materials or thicknesses for comparison.
• Tripod, clamp, or fixed mount for the sensor.
• Tape measure or ruler.
• Notebook for labeling trials and conditions.

Advanced Materials

• Radar module with raw I/Q output access.
• Oscilloscope or logic analyzer for signal debugging.
• Reference respiratory belt or chest sensor for ground truth.
• ECG or PPG system for comparison data if available in a school or university lab.
• Vibration-isolated mounting setup.
• Phantom target or motion stage for calibration trials.
• Acoustic or electromagnetic shielding materials for interference tests.
• Medical-grade data collection computer with synchronized timestamping.

Software & Tools

• Python: Processes raw radar data, filters signals, and runs model comparisons.
• NumPy: Handles arrays and signal calculations for radar time series.
• SciPy: Supports filtering, spectral analysis, and peak detection.
• Pandas: Organizes trial metadata, labels, and results tables.
• MATLAB: Helps if your school already licenses it for signal processing and model testing.

Experiment Steps

1. Define the exact output you will measure, such as breathing rate, heart rate, or heart-rate-variability estimate.
2. Design a comparison plan that tests at least one simple signal method against one model-based method.
3. Choose your ground-truth reference so you can score accuracy, not just show pretty waveforms.
4. Plan a calibration set that covers different mattress types, body positions, and distances.
5. Build controls that separate true physiological motion from bed movement, clothing noise, and ambient vibration.
6. Decide how you will score performance with error metrics, signal-to-noise ratio, and statistical tests.

Common Pitfalls

• Trying to measure heart-rate-variability without a reliable reference signal, which makes validation weak.
• Using a mattress or bed setup that moves too much, which hides the tiny cardiac motion you want.
• Collecting data from only one position, which makes the model fail when the subject shifts.
• Mixing respiration and heartbeat in the same frequency band without a clear separation plan, which blurs the result.
• Skipping repeated trials, which leaves you with no way to tell noise from a real effect.

What Makes This Competitive

A competitive version of this project would compare multiple separation methods and prove one works better under harder conditions, not just ideal ones. You could test different mattress materials, body positions, or movement levels and report how each one changes performance. Strong statistics matter here, especially if you track error, sensitivity, and failure cases across many trials. A great entry would also explain why the model works, not only that it works.

Project Variations

• Compare radar performance across memory foam, spring, and latex mattresses to see which one preserves cardiac micro-Doppler best.
• Swap the state-space model for a classical spectral filter pipeline and compare accuracy under the same sleep-like conditions.
• Test whether adding a second radar angle improves separation of breathing and heartbeat signals through the mattress.

Learn More

• PubMed: Search review articles on contactless vital-sign monitoring, radar sensing, and micro-Doppler biomedical applications.
• NASA: Search open technical reports on Doppler sensing and signal processing methods used in remote motion detection.
• NIH RePORTER: Find funded projects on noncontact physiological monitoring and related biosignal analysis.
• MIT OpenCourseWare: Look for signals and systems or biomedical instrumentation lecture materials to build your filtering and modeling background.
• Biomedical Signal Processing and Control: Search recent papers on radar-based respiration, heartbeat extraction, and state-space modeling in this journal.