Spectrum-based Fingerprint Extraction and Identification Method of 100M Ethernet Card

Abstract

In the local area network (LAN) system, most terminals are connected to edge switches through fast or gigabit Ethernet connections. The terminal access security problem has always been a key concern. This paper proposes a method of Ethernet card fingerprint extraction and identification based on spectrum characteristics, which solves the problem of illegal terminal access with counterfeit media access control (MAC) addresses. The extracted Ethernet card fingerprint is used as the identity of the terminal, which is unique and difficult to be counterfeited. The frequency-domain features of the signals can be extracted by analyzing the Ethernet card signals of wired terminals received by the switch. The dimension of these features is reduced to obtain their Ethernet card fingerprints, which can be effectively classified and identified. In the classification and recognition experiments on 7 Ethernet cards of 100M produced by the same manufacturer, 26 Ethernet cards by different manufacturers, and 65 Ethernet cards by mixed manufacturers, all Ethernet cards can achieve an accuracy of 100%. This method can be widely used for identity authentication during the access and connection of terminals and provides a secure access control scheme.