Adaptive and Resource-Efficient Systems for the Internet of Things

Doctoral thesis, 2024

With the growing number of Internet of Things (IoT) devices and the emergence of the Industrial Internet of Things (IIoT), there is a growing demand for adaptive and resource-efficient wireless communication protocols and systems. Industrial networks play a crucial role in monitoring pipelines and facilitating communication among collaborating devices, such as robots in a smart factory. These applications are safety-critical and necessitate long-term reliable and low-latency communication. However, the rising number of IoT communicating devices and deployments increasingly congests the wireless medium, which leads to interference and makes the latency and reliability requirements more challenging to accomplish. Current solutions and protocols are incapable of addressing these evolving demands. Therefore, there is a need for novel communication protocols and systems capable of dynamically adapting to unforeseen interference and changes in the wireless medium.

In this thesis, we design, implement, and evaluate protocols, systems, and evaluation infrastructures tailored for modern IoT solutions. To facilitate long-term stable communication within centrally scheduled IEEE 802.15.4 Time-Slotted Channel Hopping (TSCH) networks, we propose a centralized scheduler and a flow-based retransmission strategy. This strategy allocates retransmissions to be utilized at any node within a communication flow, thereby enhancing resilience against unforeseen interference. We then introduce Autobahn, a communication protocol that integrates opportunistic routing and synchronous transmissions with TSCH to mitigate local wideband interference while keeping latency to a minimum. With TBLE, we bring TSCH to Bluetooth Low Energy (BLE), further reducing latency without compromising reliability. To provide comprehensive insights into distributed wireless communication protocols on testbeds, we propose Grace, a low-cost time-synchronized General-Purpose Input/Output (GPIO) tracing system for existing testbeds. Finally, we demonstrate with BlueSeer that a device can recognize its environment—such as home, office, restaurant, or street—solely from received ambient BLE signals using an embedded machine learning model. BlueSeer enables small IoT devices like wireless headphones to adapt their behaviors to the surrounding environment.