RFID Tag Motion Detection Through Machine Learning with a Single Antenna Setup

One of the emerging technologies in Industry 4.0 is Radio Frequency Identification (RFID), due to its low cost, non-line-of-sight communication and low energy consumption. RFID can be used in a variety of applications, such as injecting RFID tags during the manufacturing process of plastic products, in the medical field to improve the logistics of medical equipment or in manufacturing and supply chain management. Studies have been carried out to solve some problems in these applications, such as determining the direction of RFID tags, but these use a set of antennas. To solve this problem, it was implemented machine learning (ML) algorithms to determine the direction of the objects using one single antenna. Of the three classifiers, Random Forest, Extra Trees, and Bagging Classifier, Extra Trees and Random Forest were the best classifiers with an accuracy of over 86.65% and 84.96% respectively for determining the direction of movement. Under optimal conditions, the Tag 2, reached an accuracy of 94.42% when 25 samples were acquired. This study analyzed the influence of tag positioning and reading on determining movement direction using a single antenna. This proposal can enable industries to monitor logistics processes more effectively, particularly in controlling the transition of materials between sections with reduced costs.