Impacts of GPS module on energy consumption and machine-learning based battery lifetime estimation
André Teixeira de Aquino1∗, José Ailton Leão Barboza Júnior1†, Nícolas de Araújo Moreira1‡, and Paulo Peixoto Praça1§
Federal University of Ceará, Fortaleza, Brazil
DOI : https://doi.org/10.48545/advance2023-fullpapers-3_3
Abstract
The maintenance of a Wireless Sensor Network may represent a logistic challenge. Battery replacement on applications involving huge numbers of sensors spread over a wide and distant area may be expensive and difficult. So, good planning of the maintenance schedule is necessary. This work discusses the impact of the GPS module for bovine tracking on farms on current consumption and its estimation. It compares Long-Short Term Memory (LSTM) networks and Decision Trees with AdaBoost to estimate this consumption. The results show that the activation of the GPS module increases 114.36% the current consumption and Decision Trees with AdaBoost using 300 estimators and with a depth equal to 20 outperforms LTSM with Root Mean Square (RMS) error of 0.00015.
Keywords: Machine Learning, Internet of Things, Wireless Sensor Networks.