Contactless breathing waveform detection through RF sensing: radar vs. Wi-Fi techniques

Human breathing detection plays a vital role in healthcare, safety, and various other applications. This research paper explores the use of radio-frequency (RF) sensing technologies, specifically radar and Wi-Fi, for detecting human breathing patterns. Abnormal breathing patterns can indicate respiratory or cardiovascular diseases, and early detection is crucial for timely diagnosis and treatment. Radar-based systems utilize low-power RF pulses to capture subtle chest movements associated with breathing, while software-defined radio (SDR)-based systems analyze Wi-Fi signals to detect variations caused by human chest motion. Deep learning algorithms, namely residual neural network (ResNet) and deep multilayer perceptron (DMLP), are employed to classify breathing patterns based on the collected data. ResNet attained classification accuracy up to 90% on radar-based spectrogram images data, while DMLP attained classification accuracy up to 99% on SDR-based channel state information data. The proposed approaches offer non-intrusive, remote-operable, and cost-effective solutions for breathing detection. The research demonstrates the potential of RF sensing technologies in healthcare, eldercare, sleep monitoring, and emergency response systems, paving the way for enhanced well-being and safety.