Reflected electro-material signatures for self-sensing passive RFID sensors

In this paper, we evaluate realizations for implementing an RFID reflected electro-material signature (REMS) sensor. REMS sensors allow passive measurement, recording, and reading of environmental data such as temperature in a small, low cost device. This paper presents results from two configurations: a three-section lossless microstrip transmission line and a monopole probe inserted into a lossy medium. A neural network is used to recover the permittivity profile in either case, based on the reflection coefficient of the wave backscattered from an RF tag. The neural network incorporating the Levenberg Marquardt back-propagation algorithm is evaluated in terms of average error, regression analysis and computational efficiency in the presence of realistic noise. A unique contribution of this paper is the exploration of REMS using a dissipative electro-material medium. In the lossy case, two real-valued neural networks are integrated together to reconstruct the complex permittivity from the measured reflection coefficient. The approach is verified over the frequency range 4.0-5.0 GHz and less than 4% error was observed in presence of white Gaussian noise with 10dB SNR.