Optimal deployment of mobile gateways in LoRaWAN environments

• Autor(a) principal: Mendes, Bruno Eusébio
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.1/19422
• topic: IoT; LoRaWAN; LPWAN; Chirpstack; OMNet++; Domínio/Área Científica::Engenharia e Tecnologia::Outras Engenharias e Tecnologias

The recent growth of the Internet of Things (IoT) has given rise to new applications and technologies. Of these technologies, LoRa is the one that has stood out recently due to its ability to transmit packets over long distances at low energy costs. In addition to this, this technology also uses unlicensed frequency bands, and all these factors make it possible to build low energy cost networks with large coverage areas at low monetary cost. This makes LoRa very appealing for environments where multiple square kilometers need to be covered for monitoring, such as agriculture. This thesis focuses primarily on positioning gateways in a Lo- RaWAN in order to achieve energy fairness in the network.The target in question is an environmental sensor network that monitors conditions inside tree canopies in an orange orchard in the Algarve, south of Portugal.The peculiar characteristics of these orange trees, with heights up to 3.5 m and very dense foliage, makes it a very challenging environment for radio waves propagation and causes a rapid drop in signal quality. The power consumption of the end-nodes of the network is defined by 7 combinations of spreading factor and bandwidth (0 to 6) where 0 represents the slowest and most reliable transmission at the cost of higher power consumption while 6 represents the opposite. The combination of bandwidth and spreading factor is denominated data rate. Environmental factors can negatively impact the quality of LoRa packets and the necessary power adjustments of the end-node to overcome this, and increase signal reliability, can easily define whether a device is able to transmit for 1 year or 10! The main factors that can affect signal quality are obstruction, distance and meteorology. In the case study, of these 3 factors, obstruction affects transmission quality the most. Most of the literature suggests solutions within the framework of optimizing the datarate optimization algorithm (ADR). ADR aims to minimize energy consumption while ensuring the best possible packet transmission rate and achieves this by changing the data rate based on the quality of the last 20 packets received.However, this optimization is done directly to individual end-nodes and does not solve the problem of energy fairness over the whole network because, regardless of how optimized this algorithm is, the algorithm cannot transcend the physical constraints imposed by the devices and the technology itself. Distance and obstruction will always be obstacles to signal quality. Since these physical constraints will always be present in a network and the solutions proposed by the literature only improve performance at the level of individual devices, this ends up creating a large lifetime discrepancy between devices depending on their placement. In the case of LHT65s, the discrepancy in device life expectancy is high. For example the difference between using a data rate of 0 or 5 is about 10 years. The solution proposed in this thesis to overcome this problem is to precompute the optimal position for the gateways in order to guarantee the highest life expectancy for the network. Given a number of available positions for the gateways and having a certain number of gateways less than the number of positions, the goal is to compute the optimal positioning of the gateways in order to maximize the overall network life expectancy by ensuring a fair energy consumption among different end-nodes. The first step in this process was to collect information about signal quality from a real case LoRaWAN deployment. This allowed to better understand the constraints and problems associated with its implementation. This was done using 25 LTH65 devices, 1 RAK 7244 gateway and Chirpstack as the framework to manage the network. Regarding the study of the algorithm before applying it to the practical case, a simulator was used to collect data. The simulator chosen for the development of the application was OMNet++, which besides being easier to use is also better documented than the other options considered. This simulator also offers a graphical interface with great detail that allows you to easily observe the behavior of the network. Using the Flora module it was simulated a LoRaWAN network with the structure suggested by the LoRa Alliance® with 25 devices using Oulu’s path loss model. The information obtained from this simulation was used as input and test for the algorithm that was compiled by CPLEX. In each simulation about 10,000 packets were sent per device and each experiment was repeated 30 times. The results show that the optimization model has the ability to identify the best placement for the gateway given a predefined locations and network geometry. This is due to the fact that the algorithm identifies the lowest value in the highest energy consumption per packet, and minimizing this value creates a balance of consumption among the devices and consequently extends the life expectancy of the network. It can then be concluded that this methodology is indeed efficient for deployments where changing network devices cannot be done frequently. Although it is not easy to relocate gateways in already implemented networks, but in new environments where monitoring and optimization are requirements, and these new environments are built considering the network structure, we can use this methodology since it has proven to be able to improve network life expectancy.