Design and Identification of Wireless Transmitters for a Low-power and Secure Internet of Things

Abstract: Wireless communication is a key enabler for connecting billions of Internet of Things devices. For networked embedded devices operating on limited energy resources, wireless communication dominates the power consumption. Moreover, as networked devices increasingly handle sensitive data, security concerns in wireless communication are continuously expanding. This dissertation develops novel solutions for low-power and secure wireless communication. Wireless transmitters consist of a series of steps, involving both analog and digital components, each playing a distinct role in the transmit chain. Conventional transmitters employ power-hungry analog components, leading to power consumption on the order of milliwatt. Backscatter transmitters significantly reduce communication power consumption to levels well below one milliwatt. This remarkable power efficiency is achieved by offloading power-hungry components to an external carrier emitter. However, backscatter transmitters encounter challenges in applications that demand medium to long communication range, because they rely heavily on powerful emitters in their proximity for an effective communication range. Instead of removing power-hungry components, our solution integrates the functions of these components into a low-power design. While still requiring an emitter, our transmitter does not reflect the carrier signal. Instead, we utilize a weak carrier signal to stabilize the transmitter, allowing a communication range of over one hundred meters even when the emitter is far away. This contribution takes a step forward in moving low-power communication beyond backscatter.Passive radiometric fingerprinting leverages imperfections of hardware components to identify and authenticate transmitters. Its passive nature fits well to secure low-power transmitters operating within constrained resources. To enhance the viability of radiometric fingerprinting, we make three contributions in this dissertation to facilitate its widespread deployment. First, compared to conventional radios, low-power backscatter communication has a fundamentally different composition of hardware components in its transmit chain. In our work, we decompose fingerprints in a backscatter system for dual identification of tags and emitters. Beyond security purposes, recognizing the emitter embeds a notion of locality, enabling fingerprinting usage in backscatter network management tasks such as emitter coordination. Second, the dynamic nature of real-world wireless channels significantly impacts the robustness of fingerprinting. We decompose channel impacts and develop a hybrid system. This system employs pertinent strategies for different channel factors, ensuring reliable performance across complex wireless conditions. Lastly, based on the understanding of components' contributions to the transmit chain, we design a lightweight fingerprinting system. We demonstrate a complete implementation seamlessly integrated within the constraints of a single low-cost off-the-shelf chip. This contribution simplifies the conventionally bulky setup using sophisticated signal acquisition equipment and dedicated computer processing resources, which facilitates the practical deployment of fingerprinting on low-cost embedded devices.