Capturing Customer Browsing Insights through RFID Tag Motion Detection in High Tag Density Environments

Retail industry is moving into the era of "responsive retail" where real-time detection and data analytics of customeritem interactions and preferences are becoming critical differentiators for brick-and-mortar retail stores to compete with increasing online alternatives. As RFID tag deployments in retail stores continue to grow, RFID technology is positioned to be one of the best means to capture customer-item interactions. This can generate insights retailers use to optimize store layout, shop floor item arrangements and even the placement of associated items in close proximity to increase the chances of sales. Generation of such key insights in a retail environment starts with the capability of detecting motion of a few tags among thousands of stationary tags, which is still an open research problem. We have created a fully FCC compliant UHF RFID testbed with individually tagged 1000 clothing items. By collecting data for customer-item interaction use cases relevant to realistic retail environments, we show why existing algorithms that work for low tag density environments (i.e. < 100 tags) will start to fail in denser environments with a few hundred tags or more. To address this gap, we developed a motion detection algorithm that utilizes the RSSI and phase information from tag reads and is capable of accurately discriminating a few moving tags from hundreds of co-located stationary tags (up to 1000 tags) in near-real time. The algorithm can run on standard RFID systems with no hardware modifications, and simply adds acceptable processing overhead. Our results, captured in two different RF environments, show it is possible to achieve > 90% accuracy in real-world retail environments.