Phase-Based Passive Temperature Sensing With Commercial UHF RFID Tags for Ambient IoT

Abstract

Ambient Internet of Things (Ambient IoT) enables low-cost and battery-free sensing through passive devices that harvest and backscatter radio-frequency signals. However, existing RFID-based temperature sensing methods often rely on customized tags or integrated temperature sensors, increasing hardware cost and limiting large-scale deployment. This paper proposes a phase-based temperature estimation method using commercial passive UHF RFID tags in a separated transmitter–receiver architecture. This architecture separates excitation and reception: an exciter energizes the tag, while a spatially separated three-channel receiver captures the backscattered signal, thereby providing flexible sensing coverage for Ambient IoT applications. A backscatter signal model is established for the separated architecture, and the multi-channel phase measurements are processed using circular-mean estimation, residual-based outlier removal, and interpolation to obtain stable phase features. A least-squares regression model is then constructed using data collected during natural cooling to map the processed phase features to temperature. Evaluation on four independent test trials that were not used for model training shows that, over the temperature range of 30°C to 75°C, averaging the temperature estimates from the three receiving channels achieves an overall MAE of 1.144°C. The MAEs of the individual test trials range from 1.088°C to 1.203°C. Under laboratory conditions, the proposed architecture supports temperature estimation at a maximum experimentally validated tag-to-receiver distance of $3.0{\,}\mathrm {m}$ . These results demonstrate the potential of commercial passive UHF RFID tags for low-cost, noncontact temperature sensing in Ambient IoT applications.