+ All Categories
Home > Documents > By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless...

By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless...

Date post: 11-Jan-2016
Category:
Upload: cory-hampton
View: 212 times
Download: 0 times
Share this document with a friend
Popular Tags:
20
Experimental Analysis of Channel Impairments on the Performance of RF Fingerprinting using Low-end Receivers By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA
Transcript
Page 1: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

Experimental Analysis of Channel Impairments onthe Performance of RF Fingerprinting using

Low-end ReceiversBy

Kevin Sowerby Co authors:

Saeed Ur RehmanColin Coghill

23rd Virginia Tech Symposium on Wireless Personal Communication, USA

Page 2: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

2

Outline

Radio Frequency (RF) Fingerprinting Problem definition Objective Experiment setup Results Conclusion

Page 3: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

3

RF Fingerprinting

Radio Frequency (RF) fingerprinting is the process of identifying a radio transmitter by the unique features present in its analog waveform.

DACDSP PA

RF Front End of Transmitter

Page 4: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

4

RF Fingerprinting Transient based

Steady state (modulation based)

Page 5: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

5

Research gap in RF Fingerprinting

Page 6: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

6

Research gap in RF Fingerprinting

High End Receiver setup

2. Controlled environment

1. High-end receiver with sampling rates in Giga’s and high quality analogue components

Page 7: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

7

Objective of this research work

To analyse the effect of channel impairments and interference on the classification accuracy of RF fingerprinting using low-end (i.e. low specification) receivers.

Page 8: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

8

Experimental SetupUniversal Software Radio Peripheral (USRP) is used as a low-end transceivers for measurements in a screened (anechoic chamber) and an operational (laboratory) environment.

Page 9: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

9

Data collection IEEE 802.16a preamble signal is transmitted through USRP daughter

board and is captured by low-end receivers

Page 10: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

10

Data collection

Page 11: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

11

Data collection A total of 10,000 signals from each transmitter were captured and

stored at each of the receivers, giving a total data set of 420,000 received signals.

Page 12: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

12

Classification Process The k-fold cross-validation method is used for performance

evaluation in order to enhance the certainty of the results

Page 13: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

13

Classification Process

Page 14: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

14

Experiment Results

Page 15: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

15

Experiment Results

Page 16: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

16

Experiment Results

Page 17: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

17

Conclusion and Future work

Results show that RF fingerprinting accuracy varies across the receivers for the same experimental setup in different environments.

The maximum accuracy achieved in an anechoic chamber was always less than in the operational (laboratory) environment.

Further experiment will be carried out with high-end receivers to further validate our results.

Page 18: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

18

Thank you&

Questions

Page 19: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

19

Feature Extraction

The RF fingerprint of a specific transmitter consists of normalized Power Spectral Density (PSD) coefficient values and is given by [1]

Where X(k) is the coefficient of discrete Fourier transform for the input signal x(m) given by

[1]. W. Suski, M. Temple, M. Mendenhall, and R. Mills, “Using spectral fingerprints to improve wireless network security,” in IEEE GLOBECOM 2008

Page 20: By Kevin Sowerby Co authors: Saeed Ur Rehman Colin Coghill 23 rd Virginia Tech Symposium on Wireless Personal Communication, USA.

20

Preamble Extraction from the captured signals

To extract the preamble from each acquired signal, the signal is first normalized and then the preamble is extracted from each acquired signal using the Amplitude-based variance detection technique


Recommended