Engineers at the University of California, Santa Cruz have developed a new way to measure heart rate in real time without the need for wearable devices, relying solely on Wi-Fi and a Raspberry Pi. According to reports, this system called Pulse-Fi only needs ordinary WiFi to send and receive signals, and no longer relies on traditional medical devices such as bracelets and smart watches. It is expected to significantly simplify health data monitoring and make health management more popular and convenient.
The project is led by Katia Obraczka, professor of computer science and engineering, and the research results were released at the 2025 IEEE International Conference on Distributed Computing for Intelligent Systems and the Internet of Things. Members of the research team include doctoral student Nayan Bhatia and visiting high school researcher Pranay Kocheta. They demonstrated that everyday WiFi networks combined with machine learning algorithms can track health signals with high accuracy.
Pulse-Fi works by utilizing the tiny changes that occur when WiFi radio frequency signals penetrate the human body and environmental objects. The system uses low-cost WiFi chips and receivers, combined with machine learning algorithms, to accurately identify signal fluctuations caused by heartbeats from complex environmental interference. Bhatia pointed out: "The signal is very sensitive, and filters must be accurately selected to remove various environmental noises."
The research team invited 118 subjects to test in 17 postures including standing, sitting, lying, and walking. It only took 5 seconds to monitor, and the average heart rate error did not exceed 0.5 beats/minute. The longer the monitoring time, the higher the accuracy. The system only requires a low-cost ESP32 chip (priced at about $5) and a Raspberry Pi motherboard (about $30). Both achieved excellent results, with the Raspberry Pi performing better. The researchers also believe that if a commercial-grade wireless router is used, the effect will be better. In practical applications, Pulse-Fi performs stably at a distance of up to 3 meters, and preliminary experiments even show potential for longer distances.
Kocheta pointed out that in the past, WiFi health monitoring systems performed unstable under changes in distance and body position, but Pulse-Fi used a machine learning model to completely solve this problem. "We found that, thanks to machine learning, distance changes have essentially no impact on performance," he said.
In order to train the algorithm, the team first used ESP32 equipment and medical oximeter to simultaneously collect data in the campus library, providing the "true value" of heart rate for neural network learning. In addition, the research incorporates the world's largest WiFi heartbeat signal data set collected by the Brazilian team based on Raspberry Pi, giving Pulse-Fi both breadth and accuracy.
Although the current results focus on heart rate measurement, the team is already trying to apply it to other health indicators such as respiratory rate and sleep apnea. Early unpublished experiments show that WiFi signals may be able to detect breathing and sleep abnormalities. If these studies are ultimately recognized by the industry, Pulse-Fi is expected to become a low-cost, non-invasive, and convenient home and clinical health monitoring tool, bringing good news to areas with limited medical resources.