Date of Award

3-10-2010

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering

Department

Department of Electrical and Computer Engineering

First Advisor

Michael A. Temple, PhD

Abstract

The increase in availability and reduction in cost of commercial communication devices (e.g. IEEE compliant such as 802.11, WiFi, 802.16, Bluetooth etc.) has increased wireless user exposure and the need for techniques to properly identify/classify signals for increased security measures. Communication device emissions include intentional modulation that enables correct device operation. Hardware and environmental factors alter the ideal response and induce unintentional modulation effects. If these effects (features) are sufficiently unique, it becomes possible to identify a device using its fingerprint, with potential discrimination of not only the manufacturer but possibly the serial number for a given manufacturer. Many techniques in many domains have been investigated to extract features, identify a fingerprint, classify signals, and each technique has certain benefits and limitations. Previous AFIT research has demonstrated the effectiveness of RF Fingerprinting using 802.11A signals with 1) spectral correlation on Power Spectral Density (PSD) fingerprints, 2) Multiple Discriminant Analysis/Maximum Likelihood (MDA/ML) classification with fingerprints obtained from Time Domain (TD) and Wavelet Domain (WD) statistical features. Performance \gain", defined as the difference in Signal-to-Noise ratio (SNR) required to achieve comparable classification performance, has been used to demonstrate considerable improvement. Spectral Domain (SD) fingerprinting uses PSD features for device discrimination. Results presented here show some improvement over the WD approach (gain ≈ 3 dB) and significant improvement over the TD approach (gain ≈ 8 dB).

AFIT Designator

AFIT-GE-ENG-10-19

DTIC Accession Number

ADA517270

Download

Share

COinS