An Added Value Alternative to RAIN RFID Items Characterization in Retail

Abstract

This paper deals with the difficulties of detecting UHF passive RFID tags in a RAIN RFID system implemented in a retail environment. The aim of this paper is to evolve from the notion of characterizing a tag to the approach of evaluating a tagged item in a system, in order to better understand the behavior of each item coming to life on the internet of things. To achieve this purpose, real store shelves are mounted and filled with 444 tagged products belonging to 24 different product families. Three automatic inventory systems are installed separately in order to perform two types of evaluation: first, the detectability of each product family and second the clustering of tagged items with similar performances in the same group regardless of their product family. Using inventory systems, this type of evaluation, eliminates the need to invest in characterization equipment and introduces a new way of reliably evaluating tagged items directly in the store. It enables the real-time detecting and targeting low-level tagging and arrangement problems without any prior knowledge of the tagged products present in the reading area and their initial quantities.

Keywords

• Automatic inventory
• clustering
• Internet of Things
• K-means
• machine learning
• passive