Afternoon Tutorial 3

The purposes of this tutorial are to instruct the attendees on a methodology for developing automatic signal classifiers and recognizers, and to provide them with relevant information and references to facilitate such development. The tutorial deals with both the theoretical and practical aspects of signal classification, in particular on the derivation of appropriate signal "features" - numerical quantities that capture some distinctive characteristic(s) of one or more of the signal types under consideration - and on various alternatives for the decision logic for making classification decisions.

1. Feature selection, including robustness testing and determining the most powerful features; graphical and statistical tools.
2. The Likelihood Theory Approach, including determining theoretically optimal detection and classification features;
3. Exploitable properties of various modulation types, including cyclostationarity and other characteristics;
4. Decision Logic Alternatives, including Hierarchical Decision Trees, Bayesian Likelihood Ratios, Neural Networks.
5. Figures of Merit; The relationship between Signal Quality Estimation and Signal Classification;
6. Effects of interference; The importance of background whitening; Detecting the presence of interference; interference mitigation methods.
7. Logical Integration of Classification Decisions;

A set of viewgraphs, an extensive list of references, and reprints of selected papers will be provided.

Dr. Bart Rice is a Senior Consulting Engineer for Lockheed Martin, Inc. He has worked in communications, signal processing, and related areas since 1972, and he has been involved with signal classification/recognition/identification, signal quality estimation, interference mitigation, and signal detection and analysis since 1980. He holds a BA (mathematics) from Rice University, Masters Degrees in Computer Science and Electrical Engineering from The Johns Hopkins University, and a PhD (mathematics) from Louisiana State University. Until recently, he was an adjunct lecturer at Santa Clara University, and before that at Virginia Polytechnic Institute, the University of Maryland, and George Washington University. He was the founder and past chairman of the Santa Clara Valley of the IEEE Signal Processing Society. He is a registered Professional Engineer in the state of California, with six patents and more than twenty-five papers in various journals, conference proceedings, and a volume of selected papers.

< Return to overview of Tutorials.

Last Update:  February 4, 1999         Ingo Höntsch