Higher-Order Blind Signal Feature Separation: An Enabling Technology for Battlefield Awareness; Conference paper

摘要

Higher-order transform blind signal feature classification is discussed for separating bar-shaped, circular, squared, circular-squared, and offset-diamonded constellation patterns of digital linear signals. This technique is used for automatic modulation classification of unknown signals. The higher-order transform technique is powerful in distinguishing the geometrical patterns of the signal constellations and can be used to sort a large amount of modulation schemes into several pattern groups with smaller amount of modulation schemes so a fine modulation classification can be performed more effectively and reliably on each pattern group. Robustness of higher-order cumulants is discussed to show that the constellation pattern separation is less sensitive to variations in signal-to-noise ratio.