Evaluation of an Indoor Location System Using Edge Computing and Machine Learning Algorithms

Abstract

The paper aims to evaluate precise location techniques with indoor devices using edge computing technologies, which are important for services such as smart homes and health. Despite their growing importance, indoor locations lack precise and standard methods, especially in complex environments. Solving this is being attempted through technologies such as reconfigurable surfaces and deep learning models, with attention to overcoming the challenges of indoor placement. The main objective of the study is to design a low-cost indoor location system using the ESP32 module and RSSI signals, integrated with embedded machine learning algorithms. The system to be developed will allow determining the location of objects or people with a location device through SSID signals from access points. The main objective is to evaluate the performance of three machine learning algorithms—random forest (RF), decision tree (DT) and support vector machine (SVM)—in the detection of four different locations (bathroom, kitchen, bedroom, and living room), involving the definition of system characteristics, data acquisition, the development of classifiers, and their integration in the ESP32 module to transmit location data wirelessly through the MQTT protocol. As a result of the evaluation, the DT model stands out for its efficiency under limited resource conditions during real-time implementation, but it may face challenges related to overfitting and resources at the implementation stage.