Detecting misplaced RFID tags on static shelved items

Abstract

A smart shelving system can visualize stock data in real time by leveraging item-level RFID tagging so that we can minimize out-of-stock and reduce warehousing and labor costs. The key issue of smart shelving is to locate RFID tags at any time, especially after misplacing tags. The detection of misplaced tags on stationary shelved items is very challenging due to position ambiguity, phase wrapping, device diversity, and phase ambiguity. Using a combination of theoretical analysis, simulation-based prediction and experimental verification, we propose an effective way of detecting misplaced tags, called FINDS, that integrates Particle Swarm Optimization (PSO), Synthetic Minority Over-sampling TEchnique (SMOTE) and Density-based Spatial Clustering of Applications with Noise (DBSCAN) algorithms to make theoretical and measured phases consistent with each other, and observe the phase shifts caused by misplaced tags. FINDS requires neither antenna movement nor external disturbances. We have implemented a prototype of FINDS with 20 tags and evaluated its performance, demonstrating FINDS’s accuracy to be higher than 0.92 in the case of 2 stationary antennas.