On the Change Detection of the Structure of the Modulated Signals

The description of signal complexity and structure is important in many fields of science, particularly in communication systems and signal analysis. The objective of the paper is to find change detection criteria based on Renyi entropy to detect changes in the modulation type of the signals. The entropy estimation is based on source identification paradigm, with a sliding window to assure stationarity of the processed data. A set of nine basic types of frequency modulated signals are used for testing, including additive noise. A quality criterion for change detection is considered, based on the distance among the average of the estimated entropies. The computer-based experiments reveal a good behavior for change detection in the modulation of the signals, especially for high order of the Renyi entropy. The effect of the signal-to-noise ratio is considered, and a working low limit of 10 dB is obtained. Based on different values of the change detection criteria, assigned to various modulation types, a classification of modulation could be considered.