Predição do Consumo Energético de Dispositivos LoRa usando Aprendizagem de Máquina

Abstract

The Internet of Things (IoT) is a constantly evolving concept that has gained prominence in both the academic community and industry. Within it, energy consumption is a fundamental factor in determining the operating time of devices and the frequency necessary to carry out their maintenance. This paper investigates the application of machine learning algorithms to predict the energy consumption of IoT-LoRa devices, allowing estimation of the devices’ battery life and autonomy. The methodology considered the creation of a data set based on experiments with Event stream processing (ESP32) development boards, capturing metrics such as sleep time, connection type and energy consumption. Artificial Intelligence (AI) techniques are then applied to predict energy consumption based on these variables. According to the results obtained, the best technique for predicting energy consumption is Decision Tree, with a coefficient of determination greater than 96%. The study contributes to decision-making processes that aim to select IoT devices considering the autonomy designed for the batteries of such devices.