Abstract
The Internet-of-Things (IoT) has the potential to revolutionize our understanding and real-time awareness about the world around us in order to make our lives simpler, safer, and more efficient. However, IoT growth depends strongly on how cost effective the solutions are. Battery-based systems may suffer from limited lifetimes, which incur regular maintenance, replacement, and downtime costs. Energy-harvesting based systems can suffer from poor dependability when environmental sources supply insufficient power, which leads to poor application quality. Generally, lifetime and dependability both increase when system power consumption is lowered. Thus, to achieve global IoT scale beyond the 10s of billions of devices, dominating factors of system power consumption must be addressed.
Many edge IoT devices perform events with relatively low-activity factors, such that their dominant source of power consumption stems from components performing essential functions. Given that these systems communicate wirelessly, they must keep their receivers active in order to hear wireless requests even in these low-activity factor scenarios. This energy inefficient idle listening can easily account for a significant portion of the floor power. Wake-up receivers (WuRXs) are a candidate solution to this challenge by acting as energy-efficient wireless notifiers that wake up a system upon reception of a specific wireless code. However, many of the reported WuRXs require excessive power consumption for desired levels of performance. Further, the real power improvement from wake-up receivers, and other components, in an integrated system is difficult to predict without the context of system architecture, design options, and operational constraints. This limits the ability of designers to realize energy-efficient IoT systems.
This dissertation addresses WuRX design challenges in three phases. An analysis and comparison are completed on two conventional WuRX duty-cycling techniques as well as the proposed within-packet duty-cycling mechanism, which was developed to further reduce WuRX power consumption. A variety of critical WuRX digital baseband techniques and architectures are presented with accompanying modeled and measured results. Lastly, the discussed concepts are applied in commercial CMOS on four different wake-up receiver designs achieving state-of-the-art power at near and sub-µW levels, due to the presented duty-cycling methods, along with high-performance in terms of RF sensitivity, wake-up latency, and interference rejection.
This dissertation then addresses low-power IoT system design challenges by proposing a system-level modeling framework based on system architecture, design, and operation information to enable fast virtual prototyping and an understanding of trade-offs across a variety of scenarios. Several experiments concerning the integration of wake-up radios into the system context are conducted with this framework to demonstrate examples of the component's impact.
