Overview

• The research, dubbed Pulse‑Fi, is published in the 2025 DCOSS‑IoT conference proceedings.
• In tests with 118 participants, the method reached about 0.5 beats per minute error after five seconds, with longer monitoring improving accuracy.
• Performance held across 17 body positions and typical room setups with reliable readings at distances up to roughly three meters.
• Experiments used inexpensive ESP32 and Raspberry Pi hardware, and the team says commercial routers could further enhance results.
• Models were trained on an ESP32 dataset paired with oximeter ground truth and validated against a Brazilian Raspberry Pi dataset, with ongoing work targeting breathing rate and sleep‑apnea detection based on early but unpublished findings.