Contactless Respiratory Sensing Using mmWave 5G NR and FMCW Radar

Accurate detection of respiratory patterns is critical for the early diagnosis of respiratory disorders, timely medical intervention, and long-term health monitoring. Conventional respiration monitoring techniques typically rely on specialized medical equipment and trained personnel, which constrains their applicability for real-time use in home care and self-monitoring. Recent advancements in contactless vital sign sensing have enabled the monitoring of key physiological indicators, such as respiration, without the need for invasive instrumentation. Among these, millimeter-wave (mmWave) technologies have shown promise for non-intrusive respiratory monitoring. However, existing mmWave-based approaches often require dedicated radar hardware and exclusive spectrum resources, thereby limiting their adaptability across diverse and resource-constrained environments.

This thesis presents a contactless respiration sensing framework that leverages millimeter-wave (mmWave) waveforms—namely Frequency-Modulated Continuous Wave (FMCW) radar and 5G New Radio (NR) communication signals to detect and classify human respiratory patterns without physical contact or specialized medical hardware, without sacrificing the communication capabilities all within a unified Integrated Sensing and Communication (ISAC) framework. The system employs a narrowband 2 MHz FMCW sweep alongside a 40 MHz 5G communication channel, enabling efficient respiratory sensing while maintaining spectral compatibility with communication requirements. Preliminary evaluations using USRP hardware demonstrate that a Convolutional Neural Network (CNN) trained on extracted signal features achieves a 98% classification accuracy, validating the effectiveness and feasibility of the proposed design for deployment in home and ambient health monitoring scenarios.