Ultra-low Power/Energy Harvesting CMOS Sensor Designs in Wireless Sensing Platforms.

摘要

The recent rapid development in wireless communication technologies has opened up the market for wireless sensing systems, which is attracting more and more attentions both in the industry and academia. In order to perform accurate sensing using multiple sensors, more calibration requirement and higher power consumption are expected. This inevitably increases the unit cost of wireless sensing nodes. One of the possibilities is that instead of using discrete sensors which will increase system volume and power consumption, integrated sensors can be used to achieve miniaturized size and reduced power requirement, which are of utmost importance in wireless sensing platforms deployment. Another hurdle for massive deployment of wireless sensing platforms is the limited energy capacity of batteries, which determines the system lifetime and hence increases the maintenance cost. Energy harvesting techniques are therefore generally introduced to solve this problem.;In this research, various aspects of sensor design requirements in wireless sensing platform, especially passive ones where the system energy is obtained from the environment, is presented. Based on the choice of CMOS compatibility for low-cost implementations, two different case studies for implementing sensors in power/energy limited wireless sensing systems, namely CMOS temperature sensor designs in passive RFID tag and CMOS image sensor design in passive wireless image sensor network, are identified and studied.;Various aspects of sensor designs in passively powered wireless sensing platforms are discussed and analyzed. First, power limited/energy limited sensing system is classified. As a result, the choice between low power or low energy sensor architectures can be duly made to achieve optimal system performance. The possibility of harvesting energy using reconfigurable sensor array can also be exploited for improved system power/energy efficiency.;For CMOS temperature sensor design in passive RFID tag, the temperature dependence of delay generated using MOSFET operating in all operating regions (i.e. saturation, linear and subthreshold) are studied, and optimization to achieve high linearity for reduced calibration consideration is presented. The use of sensor gain compensation scheme in BJT-based temperature sensor to suppress the inaccuracy introduced due to process variation is proposed. A time-domain differential readout scheme, that can achieve ultra-low power operation and improve SNR, is also proposed. We have successfully implemented and experimentally characterized all the designs to validate our ideas, both in the sensor block level and the RFID tag system level. For all the designs, a required power budget of as low as hundreds of nW and an acceptable sensing inaccuracy of less than +/-1°C are achieved.;For CMOS image sensor design in wireless sensing platform, a systematic approach to generate a high output voltage using integrated photodiodes in standard CMOS technology is presented. This approach has the benefit that it can be readily implemented in bulk-CMOS, without the use of extra mask or special process. As a consequence, it can be useful in realizing low-cost single-chip energy harvesting systems. We have also explored the use of reconfigurable sensor array in CMOS image sensor to harvest energy during the idle period, so that energy can be first stored and then later released for system use. Apart from that, reconfigurable full/half resolution readout and two-level quantization schemes are also proposed to reduce the readout power/energy requirement. Both the sensing and energy harvesting capabilities of the fabricated image sensor are characterized, and an estimated duty cycle of 0.2% can be achieved if the power generated by the sensor array is to compensate for the power consumed. This can be useful for developing a hybrid battery-assisted system for extended system lifetime, or even a truly self-powered CMOS image sensor system for theoretically unlimited lifetime.;To conclude, this research studies the role of different sensor designs in wireless sensing platforms to address the important issues of ultra-low power consumption, calibration consideration and energy harvesting capability in such systems.