Statistical Characterization of RFID Tag Performance for Agile Nuclear Material Operations in Cluttered Glovebox Environments

The complex propagation of radio frequency (RF) waves in metallic environments such as nuclear material gloveboxes complicates the evaluation of radio frequency identification (RFID) performance. While many RFID applications perform highly controlled testing in anechoic RF chambers and translate these results to real-world environments, this approach is challenged by RF reflective environments, necessitating testing in realistic environments that approximate the RF characteristics of the application location. We previously constructed an RFID test bed comprising mock gloveboxes with dimensions and layouts based on gloveboxes and rooms used for nuclear material processing and reported on RFID performance as a function of position within a glovebox and proximity to a nearby metal container. In this paper we demonstrate an experimental design and analysis approach to comprehensively evaluate RFID performance in a realistic cluttered glovebox environment containing variable quantities and sizes of RFID tagged metal containers of a type used in nuclear material processing. We fit second-order logistic response surface models, ultimately identifying a reduced model that accurately matches the collected data. We will discuss the selected statistical model and the relative performance between three candidate commercial-off-the-shelf RFID tags in the context of tag match probability over the entire useful range of RF power for this application. We will also present novel approaches for automating the RFID data collection and improving experimental efficiency, as well as the subsequent benefits of deploying these approaches in nuclear facilities.