Deep Learning-Based Device-Free Localization Using ZigBee

Abstract

With the rapid development of the Internet of Things (IoT), the demands for location-based services (LBS) are increasing day by day. At present, most of the indoor localization technologies require targets to carry terminal devices, which limits the practical application of indoor localization. In this paper, we propose a deep learning-based device-free localization system using ZigBee. The system employs ZigBee nodes as sensor nodes, which can locate the targets through measuring received signal strength (RSS) among these sensor nodes. In the off-line phase, we collect the RSS data of some specific locations and construct a localization model through training a deep learning convolutional neural network (CNN) model. In the on-line phase, we are able to calculate target location coordinates with the trained CNN model. The experimental result shows that the mean error of the proposed deep learning-based device-free localization system is 0.53 m, which could be a technical solution for human target localization in indoor environments.