Self-Powered Internet-of-Things Nonvolatile Processor and System Exploration and Optimizatio

Every shift in the way our devices are connected or powered brings with it a potential for revolution in the usage and capabilities of the systems built around them. Just as the transition from wired to wireless telephones led to unprecedented changes in our communications and the shift from wall-power to battery-power transformed our expectations for computational systems, the shift from battery-powered systems to self-powered systems promises to fuel the next revolution in the Internet of Things (IoT). The ability to utilize ambient, scavenged energy, such as solar energy, radio-frequency (RF) radiation, piezoelectric effect, thermal gradients, etc., can liberate IoT devices from the lifetime, deployment, and service limitations of a fixed battery. However, the power supplied by energy harvesting sources is highly unreliable and dependent upon ambient environment factors. Hence, it is necessary to develop specialized IoT architectures and systems that are tolerant to this power variation, and also capable of making forward progress on their computation tasks.;This dissertation provides a holistic exploration in applying nonvolatility into the processor and system design in energy harvesting application scenarios. Various techniques are proposed for optimization in nodes level and system level. To begin with, architectural design space is explored in a nonvolatile processor, in which nonvolatility feature is designed within a processor to overcome the unstable power supply through distributed energy and time efficient backup and recovery operations. This dissertation focuses on design space of different architectures, different input power sources, and policies for maximizing forward progress. It is presented that different complexity levels of nonvolatile microarchitectures provide best fit for different power sources and even different trails within same power source. To further overcome this challenge of architectural dependency on application scenario, various techniques are proposed in this dissertation, including frequency scaling and resource allocation to dynamically adjust the microarchitecture to achieve the maximum forward progress. Furthermore, noticing that such nodes usually perform similar operations across each new input record,;I observe optimization opportunities for mining the potential information in buffered historical data, at the potentially lower effort, while processing new data rather than abandoning old inputs due to limited computational energy. This approach is proposed as incidental computing, and synergies between this approach and approximation techniques are explored. After a progress of nonvolatilizing pieces within an energy harvesting node, IoT fog computing in Wireless Sensor Networks (WSN) is re-optimized in this dissertation as one application example to perform system-level optimization. I discuss how nonvolatility features including nonvolatile processors and nonvolatile RF can benefit the system, and how other optimizations like load balance under unstable power, as well as increasing nodes density for the quality of service can be applied to the fog computing system.;This dissertation provides a holistic exploration in designing a battery-less energy harvesting system, from individual parts within a node to an applicable energy harvesting wireless sensor network system optimization. With the increasing demands of maintains-free IoTs with green energy, this dissertation foresees the vision of self-powered IoTs in the near future.