Article:
A team of researchers has unveiled Pulse-Fi, a groundbreaking AI-powered heart rate monitoring system that uses everyday Wi-Fi signals to measure heartbeats without physical contact. This innovation could redefine remote health monitoring, offering an inexpensive and accessible alternative to traditional wearable devices.
Developed by Pranay Kochetla, Nayan Bhatia, and Katya Obraczka from the University of California, Santa Cruz (UC Santa Cruz), the Pulse-Fi prototype demonstrates that standard Wi-Fi infrastructure can be repurposed to detect subtle physiological signals — including a person’s heart rate — through intelligent signal interpretation.
1. How Pulse-Fi Works
The system relies on channel state information (CSI) — a dataset that represents how Wi-Fi signals travel between a transmitter and a receiver. As these radio waves pass through a person’s chest, they experience slight distortions corresponding to the movement caused by heartbeats. By analyzing these fluctuations, the system can accurately extract the heart rate signal without any physical sensors.
To test the concept, the team positioned two ESP32 microcontrollers, one as a transmitter and one as a receiver, around volunteers at varying distances — 1, 2, and 3 meters apart. A pulse oximeter was used simultaneously to record accurate heart rate readings for comparison. Over several five-minute sessions, Pulse-Fi consistently produced results with a margin of error under 0.5 beats per minute, proving that the technology maintains high accuracy even as distance increases.
2. Machine Learning and Signal Processing
At the core of Pulse-Fi lies a recurrent neural network (RNN) with long short-term memory (LSTM) capabilities — an advanced AI architecture designed for sequential data analysis. This neural network identifies heart rate patterns over time while filtering out noise from environmental interference. The researchers also implemented frequency filters to isolate heart rate frequencies between 0.8 and 2.17 Hz, corresponding to 48–130 beats per minute.
This machine learning pipeline allows the system to interpret the minute variations in Wi-Fi signal amplitude as real-time heart activity, turning ordinary home routers into powerful biosensors.
3. Real-World Validation and Results
To further validate the model, researchers tested Pulse-Fi using open-source health data collected from 118 adults in Brazil. Participants performed 17 different body postures and light activities — from sitting to sweeping floors — while being monitored by Wi-Fi signals and a Raspberry Pi 3B+ data collection system. Even under these varied conditions, Pulse-Fi’s performance closely matched smartwatch readings, with an average error of just 0.2 beats per minute.
The results confirm that Pulse-Fi’s accuracy remains stable regardless of a user’s body position or activity level, showcasing its robustness and potential for integration into smart homes, elder care, and medical IoT ecosystems.
4. Why It Matters
Most commercial health trackers rely on direct skin contact — such as chest straps and wrist-based sensors — making them costly and inconvenient for continuous monitoring. Pulse-Fi eliminates this barrier, leveraging the ubiquity of Wi-Fi networks to deliver non-invasive, affordable health analytics. Its scalability and low cost could help bring reliable heart rate monitoring to underserved regions, hospital settings, and telemedicine platforms.
Conclusion
The Pulse-Fi system represents a major step toward contactless, AI-enhanced healthcare, using Wi-Fi as a new frontier for biomedical sensing. As machine learning models become more sophisticated, everyday environments may soon double as health monitoring systems — providing seamless, invisible, and accurate insights into human physiology.
