Salaft: An RFID-Based Item-Level Localization Algorithm With Fluctuation Textures

Received signal strength indicator (RSSI), the most accessible physical metric for commercial ultrahigh frequency radio frequency identification (UHF RFID) systems, is highly affected by multipath propagation, noise, and other factors, making it difficult to achieve high ranging accuracy for localization. Thus, similarity-based scene analysis algorithms become critical by the advantage of nonranging. Nevertheless, as the number of reference tags used rises, tag antenna interference emerges as a new crucial factor in further reducing localization errors. In this article, we present Salaft, a scene analysis localization algorithm with Fluctuation Textures which uses fluctuations instead of similarity as the criterion for selecting neighboring tags in localization. Salaft was compared with two of the most classical scene analysis algorithms with similarity,
${k}$
-nearest neighbor (KNN) algorithm and revised KNN algorithm (Re-KNN). Considering the effect of liquid and metal on electromagnetic (EM) wave, medicinal liquid, and metal staple are also involved as the item-level targets in testing. The experimental results show that Salaft achieves the best localization accuracy no matter in locating single target or multiple targets, indicating that the degree of fluctuation is a potentially important variable used in localization.