A New Method of Secure Authentication Based on Electromagnetic Signatures of Chipless RFID Tags and Machine Learning...

Area of Science:

• Electromagnetics
• Machine Learning
• Cybersecurity

Background:

• Counterfeiting poses a significant threat to manufacturers, necessitating secure object authentication methods.
• Traditional RFID tags can be vulnerable to spoofing, driving the need for advanced identification techniques.
• Chipless RFID technology offers a unique, unclonable fingerprint based on electromagnetic characteristics.

Purpose of the Study:

• To implement a neural network model for classifying chipless RFID tags using their electromagnetic signatures.
• To develop a robust authentication system capable of preventing counterfeiting.
• To validate the effectiveness of machine learning in characterizing and classifying unique RFID tag fingerprints.

Main Methods:

• Design and fabrication of 18 chipless RFID tags operating in the V band (65-72 GHz).
• Measurement of electromagnetic (EM) signatures for each tag.
• Application of machine learning (ML) algorithms, including supervised learning and random search, for classification and network configuration optimization.

Main Results:

• The neural network model achieved a maximum recognition rate of 100% for classifying RFID tag EM signatures.
• The V band's sensitivity to dimensional variations effectively highlighted differences between tag signatures.
• The proposed method demonstrated superior accuracy compared to existing RFID fingerprinting techniques.

Conclusions:

• The developed neural network model provides a highly accurate and effective method for authenticating chipless RFID tags.
• Electromagnetic signatures, when analyzed with ML, offer a secure and unclonable fingerprint for anti-counterfeiting applications.
• This approach significantly enhances object authentication security in supply chains and product verification.