The 1/f family of fractal signals constitutes an important calss in signal processing. In many applications a precise estimation of the fractial parameter is needed, which is not easy in many environmets. This paper is a study on estimation the fractal parameter, i.e. Hurst index, under the wavelet domain and using genetic algorithms. The results show that this method is robust and coverges quickly. Due to the intrinsically parallel nature of genetic algorithms, this method is especially useful in real time applications, e.g. estimation of parameter in self-similar traffic in ATM networks.
展开▼