Rate Adaptation Mechanism for LoRaWAN Network Applied to Semi-Intensive Livestock Farming using Machine Learning
LoRaWAN; Smart Livestock; ADR
The objective of this work is to explore techniques for adapting the transmission rate (ADR) in LoRaWAN networks intended for monitoring livestock in semi-intensive livestock farming, evaluating the performance of the network depending on the number of animals monitored and the extent of the area of pasture. The evaluated scenario considers the seasonal movement of animals, which depends on their breeding regime. In this study, a semi-intensive livestock farming regime is presented, which consists of a daily period of confinement in quarters or corrals (for feed feeding) and a period of pasture for food supplementation. An investigation of classic ADR techniques applied to this scenario proves that different algorithms have better performance for the different scenarios addressed, varying in terms of capacity and coverage. With the evidence that there is a need to dynamically adjust ADR parameters and algorithms to meet the nature of the proposed scenario due to its dynamic characteristics, this work proposes using machine learning for this purpose. This document presents the preliminary results produced by simulations with the ns-3 network simulator, characterizes the problem, and presents the proposed solution of this dissertation, ending with its execution schedule.