Towards the efficient use of LoRa for wireless sensor networks

Bor, Martin and Roedig, Utz (2020) Towards the efficient use of LoRa for wireless sensor networks. PhD thesis, Lancaster University.

Text (2020borphd)
2020borphd.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (4MB)

Abstract

Since their inception in 1998 with the Smart Dust Project from University of Berkeley, Wireless Sensor Networks (WSNs) had a tremendous impact on both science and society, influencing many (new) research fields, like Cyber-physical System (CPS), Machine to Machine (M2M), and Internet of Things (IoT). In over two decades, WSN researchers have delivered a wide-range of hardware, communication protocols, operating systems, and applications, to deal with the now classic problems of resourceconstrained devices, limited energy sources, and harsh communication environments. However, WSN research happened mostly on the same kind of hardware. With wireless communication and embedded hardware evolving, there are new opportunities to resolve the long standing issues of scaling, deploying, and maintaining a WSN. To this end, we explore in this work the most recent advances in low-power, longrange wireless communication, and the new challenges these new wireless communication techniques introduce. Specifically, we focus on the most promising such technology: LoRa. LoRa is a novel low-power, long-range communication technology, which promises a single-hop network with millions of sensor nodes. Using practical experiments, we evaluate the unique properties of LoRa, like orthogonal spreading factors, nondestructive concurrent transmissions, and carrier activity detection. Utilising these unique properties, we build a novel TDMA-style multi-hop Medium Access Control (MAC) protocol called LoRaBlink. Based on empirical results, we develop a communication model and simulator called LoRaSim to explore the scalability of a LoRa network. We conclude that, in its current deployment, LoRa cannot support the scale it is envisioned to operate at. One way to improve this scalability issue is Adaptive Data Rate (ADR). We develop two ADR protocols, Probing and Optimistic Probing, and compare them with the de facto standard ADR protocol used in the crowdsourced TTN LoRaWAN network. We demonstrate that our algorithms are much more responsive, energy efficient, and able to reach a more efficient configuration quicker, though reaching a suboptimal configuration for poor links, which is offset by the savings caused by the convergence speed. Overall, this work provides theoretical and empirical proofs that LoRa can tackle some of the long standing problems within WSN. We envision that future work, in particular on ADR and MAC protocols for LoRa and other low-power, long-range communication technologies, will help push these new communication technologies to main-stream status in WSNs.