BLIND SOURCE SEPARATION OF CONVOLUTIVE MIXTURE

Guided By Er. Manorama Swain Presented By Indu sekhar samanta

Outlines
Motivation for work Introduction Blind source separation Independent component analysis Future work Conclusion References

Motivation For Work

Blind source separation is receiving a growing interest due to its applications in diverse fields such as array processing,multi user communications, etc. While most of the work has been developed in the context of instantaneous mixtures, the more difficult problem of convolutive mixtures has received less attention.

INTRODUCTION
It is a well-established statistical signal processing technique that aims at decomposing a set of multivariate signals into a base of signals as statistically independent/ uncorrelated as possible, with the minimal loss of information content. The BSS problem arises in many fields of studies, including speech processing, data communication, biomedical signal, and the mixing and filtering processes of the unknown input sources. An extensive part of the literature has been directed toward the simple case of linear instantaneous mixing; when the observed signals are a linear combination of the sources and no time-delays are involved in the mixing model. A more challenging case is when the mixing system is linear convolutive; i.e., when the sources are mixed through a linear filtering operation and the observed signals are linear combinations of the sources and their corresponding delayed versions.

BLIND SOURCE SEPARATION

Blind source separation (BSS) refers to the problem of recovering signals from several observed Linear mixtures. The strength of the BSS model is that only mutual statistical independence between the source signals is assumed and not a priori information about, e.g., the characteristics of the source signals, the mixing matrix or the arrangement of the sensors is needed. Therefore BSS can be applied to a variety of situations such as, e.g., the separation of simultaneous speakers, analysis of biomedical signals obtained by EEG or in wireless telecommunications to separate several received .

Cont

Cont..
With respect to audio signals, a linear mixture of sources is commonly referred to as "cocktail Party problem". How can humans select the voice of a particular speaker from an ensemble of different voices corrupted by music and noise in the background? One approach to solve this problem is to record the mixed audio signals with microphone arrays and subsequently apply blind source separation methods . Several simultaneously active signal sources at different spatial locations can then be separated by exploiting mutual independence of the sources. In the field of audio processing BSS is applicable, e.g., to the realization of noise robust speech, recognition, high-quality hands-free telecommunication systems or speech enhancement in hearing aids. Because temporal redundancies (statistical regularities in the time domain) are "clumped "in this way into the resulting signals , the resulting signals can be more effectively deconvolved than the original signals.

An illustration of BSS using four source signals

Linear mixtures of the source signals due to some external circumstances..

Estimates of the source signals..

INDEPENDENT COMPONENT ANALYSIS

Independent component analysis (ICA) belongs to a class of blind source separation method for separating data into underlying components , where such data can take the form of images, sounds, telecommunication channels or stock market prices . It is a computational method for separating a multivariate signal into additive subcomponents supposing the mutual statistical independence of the non-Gaussian source signals . It is a special case of blind source separation.

Cont..
ICA defines a generative model for the observed multivariate data, which is typically given as a Large data base of samples. In the model, the data variables are assumed to be linear mixtures of some unknown latent variables, and the mixing system is also unknown. The latent variables are assumed non gaussian and mutually independent , and they are called the independent components of the observed data. These independent components, also called sources or factors, can be found by ICA

Cont
ICA is superficially related to principal component analysis and factor analysis.ICA is a much more powerful technique , however , capable of finding the underlying factors or sources when these classic methods fail completely. The data analyzed by ICA could originate from many different kinds of application fields , including digital images ,document data bases , economic indicators and psychometric measurements . In many cases ,the measurements are given as a set of parallel signals or time series ; the term blind source separation is used to characterize this problem. Typical examples are mixtures of simultaneous speech signals that have been picked up by several microphones, brain waves recorded by multiple sensors , interfering radio signals arriving at a mobile.

Cont.
When the independence assumption is correct, blind ICA separation of a mixed signal gives very good results. It is also used for signals that are not supposed to be generated by a mixing for analysis purposes. A simple application of ICA is the cocktail party problem, where the underlying speech signals are separated from a sample data consisting of people talking simultaneously in a room. Usually the problem is simplified by assuming no time delays and echoes. An important note to consider is that if N sources are present, at least N observations 13 (e.g. microphones) are needed to get the original signals. This constitutes the square (J = D, where D is the input dimension of the data and J is the dimension of the model). Other cases of underdetermined (J < D) and overdetermined (J > D) have been investigated.

Cont.

Cont
Non-Gaussianity, motivated by the central limit theorem, is one method for measuring the independence of the components. NonGaussianity can be measured, for instance, by kurtosis or approximations of entropy. Mutual information is another popular criterion for measuring statistical independence of signals. x(t)= As(t) .(1) y(t)= Wx(t) ...(2) ICA goal is finding a linear transform given by matrix W so that the Output y(t) is the copy or estimate of source signal s(t): In which s(t)=[s1, s2, ...sn] is a 1xn vector composed by n source signals, x(t)=[ x1, x2,.xn]T is a (nx1) vector composed of n measuring signals and the (nxn) matrix A is called as mixture matrix.

Cont

Cont.

Cont..

Cont

Cont.

Cont.