Fast and efficient estimation of the symmetric alpha-stable impulsive signal or noise parameters

Some methods to estimate parameters of symmetric alpha-stable impulsive signals or noise have been presented, such as Fama and Roll's method, Tsihrintzis's method and Brorsen's method, all with their advantages and disadvantages respectively. Synthesizing all these methods, we present a new method, which solves the problem of the large amount of computation resulting from the highly nonlinear optimization problem and no initialization when seeking the maximum value of the maximum likelihood function. It is shown by simulation that the estimation performance is better than Fama and Roll's method and at the same time, faster than Brorsen's method without loss of precision.