A research team at the University of California has shown that ordinary Wi‑Fi hardware can double as a contactless heart‑rate monitor, no wearables or electrodes required. By mining subtle patterns in the radio waves that already blanket a room, their prototype—aptly named Pulse‑Fi—measures beats per minute with accuracy that rivals clinical gear, all using off‑the‑shelf components like a Raspberry Pi 4B or a $5 ESP32 module.
How Wi‑Fi becomes a stethoscope
Wi‑Fi signals don’t simply travel in straight lines; they bounce off walls, furniture, and people, creating a rich tapestry of reflections. Modern routers and radios capture this complexity as Channel State Information (CSI), which describes how each subcarrier of the signal changes between transmitter and receiver. A beating heart produces minute chest motions that slightly disturb these subcarriers. Pulse‑Fi isolates those ripples from the background and turns them into a stable pulse reading.
This approach falls under “device‑free” sensing—a growing field studied in academia and industry, with IEEE publications over the past decade showing Wi‑Fi can track respiration, human presence, and even coarse gestures. Pulse‑Fi pushes the concept toward vital‑sign monitoring with clinical‑level precision while keeping the hardware simple.
The build: Raspberry Pi, ESP32, and five seconds of data
The team evaluated two low‑cost setups: a pair of ESP32 modules and a Raspberry Pi 4B. Both can expose CSI under the right firmware and driver configuration, enabling raw subcarrier data collection. The Pi’s Wi‑Fi chipset reports more subcarrier channels than the ESP32 (234 versus 64 in the experiments), giving it a denser snapshot of the radio environment—and, ultimately, higher accuracy.
Crucially, Pulse‑Fi only needs about five seconds of CSI to produce a heart‑rate estimate, enabling near real‑time updates that feel responsive rather than laggy. In tests, the subject could sit naturally without wearing any sensor, while a reference device—such as a fingertip pulse oximeter—provided ground truth for comparison.
From noisy airwaves to a clean pulse
Extracting a heartbeat from ambient Wi‑Fi isn’t trivial. The processing pipeline begins by cleaning the CSI stream to suppress environmental noise—think HVAC hum, keyboard taps, or slight body shifts. A bandpass filter targets 0.8–2.17 Hz, the frequency range that maps to roughly 48–130 beats per minute, trimming away unrelated fluctuations.
Next comes feature extraction across subcarriers and a lightweight Long Short‑Term Memory (LSTM) neural network. LSTMs are well‑suited to time‑series data, and here a compact model is enough to track periodicity without heavy compute. That’s important: the pipeline runs comfortably on edge hardware like the Raspberry Pi, so there’s no dependency on a power‑hungry server.
Accuracy that matches clinical gear
Against simultaneous measurements from standard heart‑rate devices, Pulse‑Fi performed strikingly well. The ESP32 configuration achieved 99.38% accuracy, with an average error of 0.51 BPM. The Raspberry Pi setup did even better: 99.81% accuracy and a 0.2 BPM error. Those figures are on par with bedside monitors under controlled conditions and ahead of many consumer wearables at rest, which independent evaluations often find can miss by a few beats per minute depending on skin tone, motion, and fit.
The Pi’s advantage stems from its richer subcarrier view. More subcarriers mean more chances to spot the subtle periodic micro‑movements induced by cardiac activity, improving signal‑to‑noise ratio and robustness to multipath quirks in the room.
Why this matters
Contactless vital‑sign monitoring could simplify patient observation in clinics, reduce wiring clutter in step‑down units, and enable unobtrusive wellness checks at home. There’s no need for battery‑powered wearables, adhesives, or calibration for each user—just a small device quietly reading the room. Given that cardiovascular disease remains the leading global cause of death, according to the World Health Organization, inexpensive and continuous monitoring tools carry real public‑health potential.
The cost profile is compelling: a Raspberry Pi 4B typically runs $30–$50, and ESP32 boards are widely available for under $10. That puts Pulse‑Fi’s bill of materials within reach for research labs, startups, and even advanced hobbyists.
Caveats and what comes next
As with any lab‑grade result, real‑world deployment brings challenges. Movement by the subject or others in the environment can contaminate the signal. Multi‑person scenarios require source separation. Furniture layout, distance, and antenna placement affect performance, and access to CSI can depend on chipset and driver support. Before any clinical use, the system would need rigorous validation across diverse populations and settings, and regulatory clearance from agencies like the FDA.
Privacy also deserves attention. A network that senses physiological signals must be secured and transparent about what it collects. Fortunately, the edge‑friendly design means processing can remain local, reducing the need to stream sensitive data to the cloud.
The bottom line
Pulse‑Fi shows that with clever signal processing and a modest neural network, a Raspberry Pi and everyday Wi‑Fi can quite literally pull a heartbeat out of thin air. It’s a sharp example of how connectivity infrastructure is evolving into a versatile sensing platform—and a sign that the next generation of health monitoring may arrive not as a new gadget on your wrist, but as intelligence woven into the air around you.