Characterisation of capacitive front-ends for indoor person localization

Sensors for locating and identifying people indoors are currently extensively designed to monitor and automate systems. In the field of home automation, location sensors have become a key factor. There are different types of sensors that differ in their distinctive features: data processing, ease of installation, ease of use, privacy management: the latter is really important due to scalpers of the data of users in modern society. For these reasons, a localization system should have some particular features such as being tagless, secure the user's privacy and passive. In fact some localization systems require people to wear or carry a tag to be monitored. The most commonly used location sensors are radio frequency, infrared, pressure, ultrasound, video and capacitive sensors. Capacitive sensors are here the subject of study because they have many advantages including good sensitivity compared to their size, are tagless and therefore suitable for any user, and are safe from the point of view of privacy. The project involves several areas of research, from the optimization of front-ends to the use of neural networks for location and identification algorithms. My project focuses on the characterization of three front-ends: a period modulator based on an RC astable oscillator (RC-FE) circuit, another one based on constant current charge-discharge (IC-FE) and a last one based on the ramp slope modulation. All front-ends are based on the capacitive coupling between a metal plate and the human body: the resulting capacitance depends on the distance. Therefore, this capacitance is used to swing the RC-FE in astable mode with fixed resistance values. For the front-end based on the IC-FE, oscillation is due to a Schmitt trigger that compares the capacitance voltage with the thresholds and always returns a square waveform at the output. In conclusion, the output period (frequency) of the circuit is measured. In the circuit based on slope modulation, the output value returns a triangular waveform, the slope of the ramp is measured at the output instead of the oscillation period. The main objective of my thesis was to evaluate the measurement sensitivity to various noise source, such as the oscillation period or the slope of the ramp and the capacitance of the plate, compared to other quantities of the circuit itself or external (such as environmental noise properly modelled). During this thesis work on capacitive sensors, I worked with three different front-ends. For the front-ends I did simulations to evaluate different sensitivities. In detail, I analyzed the drift of the output value and of the plate capacitance calculation. After data evaluations I found a very significant noise rejection of the dual ramp based circuit compared to the RC-FE based circuit and quite significant compared to the constant current based circuit. The results are significant since the very slow ambient noise turns out to invalidate the results obtained by the capacitive sensor: these capacitive sensors could work for several hours. In the future we could investigate the use of a new front-end to lower the sensitivity value even more compared to the noise component.