Radio and Sensor Interfaces for Energy-autonomous Wireless Sensing

University dissertation from KTH Royal Institute of Technology

Abstract: Along with rapid development of sensing and communication technology, Internet of Things (IoTs) has enabled a tremendous number of applications in health care, agriculture, and industry. As the fundamental element, the wireless sensing node, such as radio tags need to be operating under micro power level for energy autonomy. The evolution of electronics towards highly energy-efficient systems requires joint efforts in developing innovative architectures and circuit techniques. In this dissertation, we explore ultra-low power circuits and systems for micropower wireless sensing in the context of IoTs, with a special focus on radio interfaces and sensor interfaces. The system architecture of UHF/UWB asymmetric radio is introduced firstly. The active UWB radio is employed for the tag-to-reader communication while the conventional UHF radio is used to power up and inventory the tag. On the tag side, an ultra-low power, high pulse swing, and power scalable UWB transmitter is studied. On the reader side, an asymmetric UHF/UWB reader is designed. Secondly, to eliminate power-hungry frequency synthesis circuitry, an energy-efficient UWB transmitter with wireless clock harvesting is presented. The transmitter is powered by an UHF signal wirelessly and respond UWB pulses by locking-gating-amplifying the sub-harmonic of the UHF signal. 21% locking range can be achieved to prevent PVT variations with -15 dBm injected power. Finally, radio-sensing interface co-design is explored. Taking the advantage of RC readout circuit and UWB pulse generator, the sensing information is directly extracted and transmitted in the time domain, exploiting high time-domain resolution UWB pulses. It eliminates the need of ADC of the sensor interface, meanwhile, reduces the number of bits to be transmitted for energy saving. The measurement results show that the proposed system exhibits 7.7 bits ENOB with an average relative error of 0.42%.