Reliable communication for outdoor sensing: improving link quality in long range wide area networks

Wireless sensor networks are a major driver of the increasing digitalization in various domains. They consist of embedded systems, so-called nodes, which combine energy efficient operation, sensing of physical parameters, and radio communication. As a result, nodes can be deployed even in hard-to-reach areas and report relevant parameters remotely. Common application scenarios are widely spread and include environmental monitoring, smart cities, digital twins, and many more. Over the years, Long Range Wide Area Network (LoRaWAN) has emerged as the de-facto standard communication protocol for numerous applications. It offers high energy efficiency, allowing for years of battery life, and long range connectivity up to several kilometers. However, many real-world use cases require nodes to be deployed outdoors, exposing them to the environment and prevailing weather conditions. Both can have a negative impact on the link quality and ultimately result in communication failures and data loss. Such an unavailability of data can impair the functionality of applications and lead to severe consequences in safety-critical contexts, e.g., in early warning systems. This thesis thus provides several approaches and methods to tackle different real-world challenges and increase the reliability in LoRaWANs. The first contribution is an advanced path loss model which was developed with empirical data and enables link quality and coverage estimations based on the topographical settings in an area of interest. It allows for the identification of suitable locations for nodes and base stations already ahead of deployment. Subsequently, the second contribution is an in-depth investigation of the link quality deterioration caused by individual weather parameters. It includes a statistical analysis and the development of regression models to estimate the link quality based on the prevailing weather conditions. Building on that, as the third contribution, three methods for increasing the reliability of the LoRaWAN protocol itself are proposed. The first approach, PreCo, is based on the pre-computation of all plausible data packets a node could send and correlating them with the received signal on the physical layer, which improves demodulation performance for weak signals. CEC-LoRa shifts this approach to the data link layer and allows for platform-independent correction of corrupted packets. Lastly, Backpack-LoRa extends the LoRaWAN protocol with multi-hop functionality on the network layer and enables packet forwarding for lossy links. The practical feasibility of the proposed methods is shown through prototypical implementations and evaluations with actual hardware.