Counterfeit Chipless Radio Frequency Identification tag detection using Differential Constellation Trace Figure and machine learning

Chipless Radio Frequency Identification (RFID) tags are widely adopted due to their cost-effectiveness, lightweight design, and passive operation. However, their lack of computational capabilities makes them vulnerable to cloning and counterfeit attacks. This paper proposes a counterfeit detection framework that combines Differential Constellation Trace Figures (DCTFs) with machine learning techniques to address these challenges. Backscattered Time-Domain (TD) signals from seven identical chipless RFID tags were processed to generate DCTFs, which were enhanced using colormaps such as Turbo, Colorcube, and Prism to highlight subtle variations. Red, Green and Blue (RGB) color features were extracted across spatial (width and height), radial, and angular dimensions, forming a multi-dimensional dataset. Gaussian noise was added to simulate real-world conditions, with Signal-to-Noise Ratios (SNRs) ranging from 100 decibels (dB) to 0 dB. Machine learning models, Support Vector Machine (SVM) and Random Forest (RForest), were trained to classify authentic and counterfeit tags. RForest demonstrated superior performance, achieving an accuracy of 99.62% and Area Under the Receiver Operating Characteristic curve (ROC–AUC) of 99.18% with multi-dimensional input data at 70 dB SNR noise.