RFID in the Wild - Analyzing Stocktake Data to Determine Detection Probabilities of Products

Matthias Wölbitsch, Thomas Hasler, Micheal Goller, Christian Gütl, Simon Walk, Denis Helic

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

Abstract

Over the course of recent years, Internet of Things (IoT) technology, and in particular Radio Frequency Identification (RFID), has seen widespread adoption across several different domains. Particularly the fashion industry has integrated RFID into their day-to-day business for accurate stock tracking and monitoring. While inventory accuracy can be increased well above 90%, perfect inventory accuracy is hard to achieve, which is often related to the inherent problems of RFID technology. Several factors can favor or adversely affect RFID reader performance, such as the materials items are made of, their placement in the store, or the location of where the RFID tag has been attached. Therefore, identifying such products, that are frequently missed during stocktakes, is crucial to reach fully accurate inventories, as they require special attention to be properly processed. In this paper, we set out to tackle this real-world problem of determining products with low detectability, based on historical stocktake data of more than 400 brick-and-mortar stores. Further, we conduct a controlled user study to evaluate and improve the detectability of frequently missed products for a total of 16 stores. Our results indicate that frequently missed products can be identified and used as a foundation to further improve stock accuracy in retail stores.
Original languageEnglish
Title of host publication 2019 Sixth International Conference on Internet of Things: Systems, Management and Security (IOTSMS)
Pages251-258
Number of pages8
Edition6
DOIs
Publication statusPublished - 2019

Fields of Expertise

  • Information, Communication & Computing

Fingerprint

Dive into the research topics of 'RFID in the Wild - Analyzing Stocktake Data to Determine Detection Probabilities of Products'. Together they form a unique fingerprint.

Cite this