Independent Component Analysis for the Multitag Detection of Frequency-Coded Chipless RFID

Abstract

A multitag detection technique is proposed for frequency-coded chipless radio frequency identification (RFID) to achieve minimum separations of tags, enhanced data capacity, and calibration-free features. Previous attempts for multitag detection include signal processing, comparisons with a database, and space-division multiple access (SDMA); nevertheless, these techniques rise to the challenges of large separations of tags, reduced capacity, and the complexity due to a phased array or beamforming technology, respectively. In contrast, the proposed technique overcomes these limitations by a new framework that integrates signal processing and SDMA. The capability is derived from the independent component analysis (ICA), which transforms mixing backscattering fields into an optimization model; by using the Newton method to maximize non-Gaussianity, the original resonances of each tag can be recovered. Furthermore, the proposed technique eliminates the procedure of calibration. An 8 bit system is designed and tested over 2.0–5.0 GHz. When two tags are separated by 0, 10, and 20 mm, ICA shows average reliability of 80.4%, 90.4%, and 91.5%, respectively. Thus, closely adjacent tags can be detected without calibration even for a high-density system. Real-world implementation issues, including four-tag detection, orientation mismatch, and a displacement of tags, are also analyzed to validate the proposed technique.