Development of a Class-Specific Module for Hyperbolic, Frequency-Modulated Signals

Abstract : The class-specific method (CSM) is a novel technique for signal classification. CSM operates by computing low-dimensional feature sets, each tailored to a given signal class. In this manner, CSM avoids the difficulties inherent with classical Bayesian signal classification methods that attempt to estimate probability distributions in high-dimensional feature space. The fundamental building block of a CSM classifier is the feature extraction module. Each module is designed for a specific signal class. This report presents the development of a module for hyperbolic, frequency-modulated signals. This development also serves to acquaint the CSM novice with the general considerations in module design.