Desarrollo de un modelo de radio propagación para tecnología Lora en la región interandina usando Machine Learning.

Abstract

Abstract: This paper details the entire process involved in developing a path loss prediction model for LoRa technology under the specific conditions of the Inter-Andean region. To this end, an exhaustive measurement campaign was carried out within the city of Loja, which has a varied geography and urban density, collecting data (41,901 samples) that represent the signal behavior within this scenario. The processing and analysis of these data allowed the training of two propagation models, one for the Uplink and one for the Downlink. These models, through preprocessing, normalization, and evaluation techniques, were able to fully exploit the implicit information of the data collected during the learning process. Consequently, two models were obtained that were capable of accurately predicting path loss, overcoming the limitations of traditional empirical methods and thus allowing for the optimization of the deployment of IoT networks in mountainous environments.