The Research Of Blind Source Separation Based On Unsteady Signal

Blind signal processing has become a research issue in signal processing fields during recent ten years.Blind source separation has been used in such as earthquake detection, mobile communication, image procession, speech signal processing, biological medical engineering and so on.Blind source separation (BSS) is the recovery unknown sources from the observed outputs of an unknown linear systems with static parameters.The research of unsteady signal blind source separation is issue because it is very important in theory and useful in value with the technology advancing.The major contribution of this paper is summarized as follow(1) The research tendency of BSS algorithms is analyzed systematacially in is paper. The foundational principles of BSS are expatiates.Three mathematical models of BSS are discussed and the typical performance evaluating the effect of its algorithms are indexed.(2)This paper analysizes not only the sepration nature, premise and structure of unsteady signals but also the typical algorithms of unsteady signals.(3) A loss function based on maxmimum function estimation signal separation is deduced in this paper.Adaptive blind source separation algorithm of unsteady signal is deduced on natural gradient algorithm aimed at unsteady nature.It is sepreaded to the convolution mixture state which simplifies compution procession and reduces computionAt the same time, Infomax algorithm of BSS is deduced on the feature of unsteady signal which is steady in a short time.These algorithms are simulated in computer and analyzed.Simulation results show then can better separate mixed signals.(4) Blind source separation algorithm based on immunity algorithm is sugested in this paper to what natural gradient algorithm depends on nonlinear function choice and it is easy to be immersed in part best.The advantage of this algorithm is that it searches in all aspects.the disadvantage of this algorithm is running slowly.(5) This paper discusses several applications of BSS algorithms based on unsteady signal in digital communication. Computer simulations illustrate that the algorithms can well separate mixed these signals.