The Identification of Communication Signals Based on Fractal Box Dimension and Index Entropy

Abstract

The current recognition algorithms of communication signals often have high complexity and poor anti-noise performance. To this problem, the paper proposed a communication signal modulation identification algorithm, which is based on fractal box dimension and index entropy. This paper first extracts box counting dimension characteristics of signals which characterize signals’ complexity to classify 7 kinds of signals roughly. Then it extracts index entropy characteristics of signals which characterize the distributions of signals to subdivide them. Finally it uses the extracted features to recognize the modulation of signals by setting thresholds, simulates and calculates the error rate of the identification. Simulation results show that, the features extracted by this method can effectively classify the modulations of signals, and it can identify signals' types more effectively than the signal recognition method based on the information entropy. Due to the non-sensitive to noise quality of fractal box dimension and index entropy, it can get higher recognition rate in lower SNR.