Deep Learning based Wireless Radio Identification Architecture for Edge Computing

Abstract

With the popularity of Near-field Communication(NFC), attacks and defenses of access control systems have also attracted much attention. However, the identity verification of many commercial electronic access control systems at present is still based on the Unique Identifier (UID) of the access control card for identity identification. In recent years, the literature shows the difference between hardware components can effectively identify different device fingerprinting, thereby indirectly judging the legitimacy of the user, to improve the information security of radio communication equipment. To solve the risk of identification of access control devices, the software-defined radio is used to collect NFC band signals to extract I/Q signal samples as features in the study. The extracted radio device features are handed to learn and build a model, and further combined with the Edge Computing architecture to verify and identify NFC card replication.