"Ethernet card discrimination using unintentional cable emissions and c" by Timothy J. Carbino, Michael A. Temple et al. 10.1109/ICCNC.2015.7069371">
 

Ethernet card discrimination using unintentional cable emissions and constellation-based fingerprinting

Document Type

Conference Proceeding

Publication Date

2-16-2015

Abstract

Improved network security is addressed using device dependent physical-layer (PHY) based fingerprints from Ethernet cards to augment traditional MAC-based ID verification. The investigation uses unintentional Ethernet cable emissions and device fingerprints comprised of Constellation-Based, Distinct Native Attribute (CB-DNA) features. Near-field collection probe derivative effects dictated the need for developing a two-dimensional (2D) binary constellation for demodulation and CB-DNA extraction. Results show that the 2D constellation provides reliable demodulation (bit estimation) and device discrimination using symbol cluster statistics for CB-DNA. Bit Error Rate (BER) and Cross-Manufacturer Discrimination (CMD) results are provided for 16 devices from 4 different manufactures. Device discrimination is assessed using both Nearest Neighbor (NN) and Multiple Discriminant Analysis, Maximum Likelihood (MDA/ML) classifiers. Overall results are promising and include CMD average classification accuracy of %C = 76.73% (NN) and %C = 91.38% (MDA/ML).

Comments

Copyright © 2015, IEEE.

This conference paper is available through subscription or purchase from the publisher, IEEE, using the DOI link below.

Author notes:

  • Carbino was an AFIT PhD candidate at the time of this conference. (AFIT-ENG-DS-15-S-008, September 2015)
  • Trevor Bihl was an AFIT PhD candidate at the time of this conference. (AFIT-ENG-DS-15-D-003, December 2015)

Source Publication

2015 International Conference on Computing, Networking and Communications (ICNC)

Share

COinS