The views expressed in this thesis are those of the author and do not reflect the official policy or position of the Department
of Defense or the U.S. Government. / The main objective of this study is to find a wavelet-based, feature extracting algorithm for push-to-talk transmitter
identification. A distance-measure algorithm is introduced to classify signals belonging to one of four transmitters. The
signals are first preprocessed to put them into a form suitable for wavelet analysis. The preprocessing scheme includes
taking the envelopes and differentials. Median filtering is also applied to the outputs of the wavelet transform. The
distance algorithm uses local extrema of the wavelet coefficients, and computes the distance between the local extrema of
a template and the processed signals. A small distance implies high similarity . A signal from each transmitter is selected
as a template. A small distance measure indicates that the signal belongs to the transmitter from which the template
originated. The distance algorithm can classify correctly the four different signal sets provided for the research. Even at
lower signal-to-noise levels, good identification is achieved.
| Identifer | oai:union.ndltd.org:nps.edu/oai:calhoun.nps.edu:10945/30740 |
| Date | 12 1900 |
| Creators | Payal, Yalçin |
| Contributors | Hippenstiel, Ralph, Fargues, Monique P., Naval Postgraduate School (U.S.) |
| Publisher | Monterey, California. Naval Postgraduate School |
| Source Sets | Naval Postgraduate School |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
| Rights | Approved for public release; distribution is unlimited. |
Page generated in 0.0018 seconds