Vital signs such as heart rate (HR) and respiration rate (RR) are critical for the clinical assessment of an individual's health and have high predictive value in identifying undesired events like cardiac arrest, critical dizziness, arrhythmias, and synchronization with respiration rate. Traditional contact-based methods are limited by their reliance on accessories attached to the body, making continuous monitoring challenging in both clinical and home environments. Moreover, wearable devices may not be suitable for patients with skin allergies, burns, or infants. Microwave radar sensing, which offers superior penetration through materials and clothing, and is less affected by environmental conditions, is posited as a promising alternative for continuous, non-contact monitoring.

However, extracting physiological information from radar signals presents a significant challenge, primarily because the phase changes in the received signal are highly susceptible to environmental noise and interference, particularly when measuring HR in realistic circumstances. Even the individual's body movement can substantially impact HR readings. Skin displacement from heart activity is much smaller than that caused by respiration, leading to a weaker reflected signal from heartbeats compared to respiration. Furthermore, the HR spectrum's entire frequency range contains significant noise from the second and third-order harmonics of the respiration and intermodulation products. Environmental clutter and random body movements also add to the noise in the received reflected signal, posing a considerable challenge in developing an efficient system for HR estimation.

The research focuses on developing an efficient, accurate, and privacy-aware non-contact vital sign (NCVS) monitoring method using mm-wave radar technology and devising a signal processing algorithm to improve the accuracy of heart rate and respiration rate measurements. The research adopts a multi-pronged approach to achieve these goals. This includes analytical modelling of the chest wall motion due to cardiovascular activity, facilitating a nuanced understanding of trade-offs between various radar parameters. Furthermore, the study proposes using a non-linear signal analysis technique, resonance sparse spectrum decomposition (RSSD), to better capture and analyze the complex dynamics of non-stationary signals. RSSD decomposes the signal into time-varying frequency components using wavelet decomposition and sparse approximation, identifies and isolates resonant frequencies, and constructs a sparse representation of the signal. This approach offers a highly accurate and efficient method for analyzing non-stationary signals with time-varying spectral characteristics. A harmonic-based algorithm is formulated to improve the accuracy of HR measurement.

Additionally, target localization, crucial for the practical deployment of radar-based NCVS systems, becomes even more challenging due to the inevitable positional changes in real-world scenarios. We introduce an automatic, real-time beam steering and beam forming algorithm for identifying target locations, which augments the signal-to-noise ratio (SNR) and enhances vital sign estimation accuracy. The effectiveness of the proposed method is evaluated through a series of experiments carried out in various realistic settings, including artificial clutter and body movements such as reading a book, drinking water, and forward and backward body movements. To mitigate the noise and interference due to these additional attributes, we optimize the Q factor selection for each dataset by modifying the RSSD algorithm parameter selection by leveraging the sub-band energy distribution leading to a more precise extraction of HR. The findings demonstrate that the proposed method effectively mitigates issues caused by unwanted clutter, manages random body motion and harmonic interference, and significantly improves HR estimation accuracy by reducing noise in the phase signal.