Contactless human vital signals monitoring based on a FMCW Radar

Breath and heart rate, highly used during the recent Covid emergency, are among the first vital parameters of clinical interest. They can provide lots of information about a person's state of health. Their continuous monitoring will be a sector of great interest in the next decades by companies operating in various fields, mostly in healthcare (home monitoring, elderly patients, or for sleep analysis). According to the latest report by the WHO (World Health Organization), the world population is aging, thanks to the advancements in medicine, leading to an increase in geriatric diseases related to breathing and the cardiovascular system. Nowadays, monitoring breath and heart rate technologies are mainly wearable, and based on electrodes or contact sensors. However, these technologies have the disadvantage of being uncomfortable for the subject, sometimes irritating to the skin, and highly sensitive to the level of skin-sensor contact. New remote technologies are therefore being studied, and the thesis project is part of this new research field. The aim is to design an automatic system based on Infineon's BGT60ATR24C FM (Frequency Modulated) radar, developed in an ECSEL-JU project called AI4CSM (Automotive Intelligence for Connected Shared Mobility) with Polytechnic of Turin as partner, chosen because of its reduced size. Being a prototype, the radar required a careful characterization and investigation of the performance, aiming at monitoring the millimeter displacement of the chest wall from a distance. Then, an algorithm was designed with the aim of locating the human and extract its vital parameters by processing the data acquired from the radar. Signal frequency analysis based on a high-resolution PSD (Power Spectral Density) was used to obtain the necessary spatial resolution. The first validation phase was carried out using manually generated signals on MATLAB, by superimposing variable noise on the breathing signals. The second validation phase was performed using the filament 3D printer, REPRAP X400, reprogramming it in its native language (gcode) to simulate vital signs. The modelling of breathing was in these steps of the sinusoidal type with different amplitude levels, in the literature between 4 and 12 mm. After fine tuning the algorithm to improve its robustness, a third validation phase was carried out on volunteer subjects. Standing one meter far from the radar, the subject was asked to breathe in a relaxed condition. The vital parameters extracted were compared with those obtained by respiratory belt and photoplethysmography, taken as the gold standard.