In this work, we propose a novel approach for the detection of chipless radio frequency identification (RFID) signals. The method is founded upon the application of transformations to the measurement, in conjunction with the utilization of Artificial Intelligence (AI) algorithms. In the initial stage of the process, frequency-related features are measured. Subsequently, a time-frequency representation of these measurements is generated through the application of the Inverse Fourier Transform (IFT), a time-gating strategy, and the Continuous Wavelet Transform (CWT). The resulting representation is then employed as input to a shallow convolutional neural network (CNN), which is able to learn complex patterns while being able to generalize to new measurements. Furthermore, the proposed scheme incorporates a filtering process, based on the probabilities derived from the model, to filter out low-confidence predictions. To assess the performance of the proposed method, we considered a population of 16 tags. We collected 4,800 measurements for the training phase and 2,400 measurements for the testing phase in a real environment, collected at different days, and within a distance range of 50-140 cm from the tag to the antenna. The proposed method exhibited accuracies of 94% in the 110-140 cm range, 99% in the 80-110 cm range, and 100% in the 50-80 cm range, showcasing its suitability for chipless RFID detection.

In this work, we describe step by step a complete workflow to apply machine learning (ML) classification for chipless RFID tag identification, covering: i) the tag implementation criteria for circular ring resonator (CRR) arrays for ML interoperability; ii) the data collection procedure to get a sufficiently representative dataset of real measurements; iii) the ML techniques to visualize the data and reduce its dimensionality; iv) the evaluation of the ML classifier to ensure high accuracy predictions on new measurements; and v) a thresholding scheme to increase the certainty of the predictions. The differences of the tags’ frequency responses are maximized by optimizing the Hamming distance between the tag identifiers and by controlling the radar cross section (RCS) level of each CRR array. We show on two scenarios, fixed and flexible range (up to 160 cm), that the proposed workflow can achieve perfect accuracy in most cases for the identification of four tags.