The Extend-Infomax Blind Signal Separation Algorithm with Parameter Adjustable Hyperbolic Tangent Function

In this paper, by analyzing the probability density function of sub Gaussian source signal in Extend-Infomax algorithm, the hyperbolic tangent function in the algorithm is changed to hyperbolic tangent function with adjustable parameter. This change is based on the fact that when the expectation of the source signal is constant, the non Gaussian property of the output signal of the separation system increases with the increase of the parameters, which will improve the performance evaluations of the Extend-Infomax algorithm. Matlab simulation results show the effectiveness of this improvement.