Long-Term Stable Communication in Centrally Scheduled Low-Power Wireless Networks

Sammanfattning: With the emergence of the Internet of Things (IoT), more devices are connected than ever before. Most of these communicate wirelessly, forming Wireless Sensor Networks. In recent years, there has been a shift from personal networks, like Smart Home, to industrial networks. Industrial networks monitor pipelines or handle the communication between robots in factories. These new applications form the Industrial Internet of Things (IIoT). Many industrial applications have high requirements for communication, higher than the requirements of common IoT networks. Communications must stick to hard deadlines to avoid harm, and they must be highly reliable as skipping information is not a viable option when communicating critical information. Moreover, communication has to remain reliable over longer periods of time. As many sensor locations do not offer a power source, the devices have to run on battery and thus have to be power efficient. Current systems offer solutions for some of these requirements. However, they especially lack long-term stable communication that can dynamically adapt to changes in the wireless medium. In this thesis, we study the problem of stable and reliable communication in centrally scheduled low-power wireless networks. This communication ought to be stable when it can dynamically adapt to changes in the wireless medium while keeping latency at a minimum. We design and investigate approaches to solve the problem of low to high degrees of interference in the wireless medium. We propose three solutions to overcome interference: MASTER with Sliding Windows brings dynamic numbers of retransmissions to centrally scheduled low-power wireless networks, OVERTAKE allows to skip nodes affected by interference along the path, and AUTOBAHN combines opportunistic routing and synchronous transmissions with the Time-Slotted Channel Hopping (TSCH) MAC protocol to overcome local wide-band interference with the lowest possible latency. We evaluate our approaches in detail on testbed deployments and provide open-source implementations of the protocols to enable others to build their work upon them.