Wednesday, December 15th, 2021
Optimization and experimental characterization of Low Power Wide Area Networks
LPWAN are getting increased interest from academia and industry. They utterly address the IoT needs for low power and long-range networks, providing efficient connectivity over wide areas and enabling innovative applications.

This thesis focuses on the LoRa/LoRaWAN technology, a promising LPWAN candidate, which offers many advantages but presents some challenges in terms of scalability and reliability.

In the first part of this thesis, we study, evaluate, and characterize the LoRaWAN network link quality; then, we consider one of the major limitations of the technology, namely the channel access, by proposing a technique to improve the capacity of the network.
In the first contribution, we explore the actual LoRaWAN network state by monitoring all traffic on a LoRaWAN gateway and, subsequently, conducting a thorough analysis of the current practice for setting the different parameters by the ambient traffic.  

In the second contribution, we evaluate and characterize the transmission qualityof LoRa links by measuring the Packet Reception Rate (PRR) as a function of the payload length. We conducted extensive experiments on a test-bed in The Things Network (TTN) and investigated the resulting analysis, which shows only a slight impact on the payload length on PRR.

In the third contribution, we characterize the wireless channel experimentally, determine its behavior, and examine what factors depend on it. For both an indoor and outdoor sender, we have identified different patterns, considering the time variability of the channel at different gateways.

Finally, we address one of the major limiting factors in LoRaWAN networks, the Aloha-like access method. We study how the \gls{mim} technique for LoRaWAN would have the potential to improve the network capacity. We show that while LoRaWAN with capture effect only allows reaching 23\% of the channel utilization, MIM enables us to increase this utilization rate up to 35\% in a single LoRaWAN cell scenario.
Mis à jour le 14 December 2021