Abstract:

Current spectrum sensing methods are sensitive to noise uncertain, inaccurate in low signal to noise ratio (SNR) situations and low in calculation speed. To solve these issues, a fast blind spectrum sensing method based on Katz fractal dimension in frequency domain is proposed. It senses the presence of primary users according to different characteristics of Katz fractal dimension in frequency domain between signals and noise. Monte Carlo simulations show that the proposed method has some excellent advantages, such as insensitivity to noise uncertain, less influence from modulation parameters, high accuracy in low SNR situations, and high calculation speed. Furthermore, it can overcome the limitation of box dimension method which cannot detect the signals of some modulation parameters, so it can be used for fast blind spectrum sensing in situations of low SNR and uncertain noise.