Respiratory pattern tracking proved to be critical for many applications ranging from well-being monitoring and stress management to dealing with chronic breathing abnormalities. Specific breathing and meditation exercises have been designed to improve well-being of users based on monitoring the complete breathing waveform. While wearable systems had leveraged a wealth of information available from respiration stream in a variety of applications, contact-less sensing systems are lagging behind when it comes to capturing detailed breathing metrics. In this work we propose WiRelax; the first non-contact respiratory biofeedback system that relies solely on WiFi availability. We propose algorithms that map the changes in the Channel State Information (CSI) to the instantaneous breathing state. The key contribution is a model that relates relative phase of the received signal and the micro-motion of the chest during breathing. A novel processing pipeline is developed to extract a single breathing waveform from CSI data captured across noisy multiple sub-carriers in real-time. Our evaluation in a real-world setup shows that WiRelax can estimate real-time breath-by-breath cycle time with median error less than 0.25 s ( < 13% relative to the cycle length) thus enabling accurate device-free respiratory biofeedback.

事实证明，呼吸模式跟踪对许多应用至关重要，从健康监测和压力管理到应对慢性呼吸异常。经过专门设计的呼吸和冥想练习，旨在通过监测完整的呼吸波形来改善用户的健康状况。尽管可穿戴系统在各种应用中利用了来自呼吸流的大量信息，但非接触式传感系统在捕获详细的呼吸指标方面却落后于其他人。在这项工作中，我们提出了WiRelax。第一个完全依靠WiFi可用性的非接触式呼吸生物反馈系统。我们提出了将通道状态信息（CSI）中的变化映射到瞬时呼吸状态的算法。关键的贡献是一个模型，该模型将接收信号的相对相位与呼吸过程中胸部的微动作联系起来。开发了一种新颖的处理流水线，以实时从在嘈杂的多个子载波之间捕获的CSI数据中提取单个呼吸波形。