On synthesizing blind pulse signal extraction algorithm from the mixture with white Gaussian noise based on maximization of the third moment

 

A. E. Manokhin

Ural Federal University

 

The paper is received on March 14, 2016,  after correction - on May 9, 2016

 

Abstract. In the work a problem of synthesizing the blind pulse signal extraction algorithm from the mixture with white Gaussian noise based on maximization of neuron network output signal third moment is decided. In contrast to widespread blind algorithms based on maximizing kurtosis, the third moment maximization algorithm possesses higher convergence rate and doesn’t depend on initial conditions and the pulse process intensity.

Key words: blind sources extraction, kurtosis, third moment, artificial neural network, convergence rate.