An Improved Fractal Box Dimension Extraction Method of Signals in Cognitive Radio under Low Signal-to-noise Ratio ⋆

Fractal box dimension as a feature is widely used in many fields such as spectrum sensing in cognitive radio, image processing, modulation recognition of signals, parameter estimation of signals, and so on. In order to improve the estimation accuracy of it, an improved feature extraction method of modulation signal is proposed under low signal-to-noise ratio(SNR). In the absence of a priori knowledge for the signal, the received signal is filtered with segments in frequency domain at first, and then the box dimension feature is calculated from the reconstructed signals after filtering. Simulation results show that the estimation accuracy of the box dimension of the modulation signals can dramatically be improved under low signal-to-noise ratio by using the proposed method.