Motivation

The rapid development of cellular networks has transformed wireless signals from pure communication carriers into powerful sensing modalities, enabling a new class of ubiquitous and contactless human sensing applications. Among these applications, non-contact respiration monitoring has attracted increasing attention due to its importance in healthcare, sleep monitoring, and long term physiological assessment. Conventional respiration monitoring methods typically rely on wearable devices or contact based sensors, which may cause discomfort, limit natural movement, or hinder long term deployment. In contrast, wireless sensing based on cellular signals provides a passive and unobtrusive alternative, making it particularly suitable for continuous monitoring in everyday environments.

Main Idea

In this work, we present a real time respiration monitoring system based on 5G Channel State Information (CSI). The key observation behind this approach is that human breathing induces subtle periodic chest movements, which modulate the wireless propagation environment and lead to measurable variations in the amplitude and phase of CSI. By continuously capturing CSI measurements from a commercial 5G system and analyzing these fine-grained signal fluctuations, our system is able to extract respiratory patterns without requiring the subject to carry any device or wear any sensor. The proposed system adopts an end-to-end signal processing pipeline designed for real time operation. Raw CSI measurements are first preprocessed and denoised to suppress hardware impairments and environmental interference. Subsequently, amplitude and phase processing are applied to enhance sensitivity to respiration-induced motion. To further improve robustness, principal component analysis and temporal filtering are employed to isolate the dominant breathing related component while suppressing unrelated movements and noise. The extracted respiration signal is then continuously updated and visualized in real time, enabling instant observation of breathing dynamics.

Result

To demonstrate the practicality of our approach, we implemented the real time demo system that performs online CSI processing and respiration signal extraction with low latency. The video illustrates the system’s ability to track respiratory activity continuously under realistic indoor conditions. The results highlight that accurate and stable respiration monitoring can be achieved using existing 5G infrastructure, without any modification to the communication protocol or additional sensing hardware. This demo shows the potential of integrating health sensing capabilities into future cellular networks, verifies the feasibility of a scalable, device-free, and unobtrusive health monitoring solution.