You are on page 1of 3

INFORMATIONMAXIMIZATIONAPPROACHFORCOMMUNICATIONSIGNAL

SEPARATION

Independent Component Analysis (ICA) is a method of finding unknown source signals


fromsignalmixtures,anditisjustoneofmanysolutionstotheBSSproblem.JustasICAis
one of many methods of resolving signal mixtures into their original component source
signals,manyapproacheshavebeendevelopedtoperformICA.Thisprojectfocusesonthe
information maximization i.e. Infomax algorithm, which finds a number of independent
sourcesignalsfromthesamenumberofsignalmixturesbymaximizingtheentropyofthe
signals.TheInfomaxalgorithmachievesthemaximumentropyofafunctionusinggradient
ascent,aniterativeprocessoftakingastepinthedirectionofmaximumgradientuntila
localmaximumisreached.Ifthisprocessisrepeatedsufficiently,theglobalmaximumwill
eventuallybefound.Whentheglobalmaximumofentropyisfoundusinggradientascent,
entropyhasbeenmaximized,andtheresultingsignalsarethesourcesignals.
Infomax is a method of ICA grounded in information theory which aims to find
independent source signals by maximizing entropy. The details, calculations, and
assumptionsinvolvedwiththeInfomaxmethodarediscussedlaterinthischapter,butthe
general strategy of Infomax begins with Equation (1), where the extracted signals y are
obtainedfromsignalmixturesxbyoptimizinganunmixingmatrixW.Infomaxholdsthat
the extracted signals are source signals if they are mutually independent. While
independence of the signals cannot be measured, entropy can. Entropy is related to
independence in that maximum entropy implies independent signals. Therefore, the
objective of ICA is to find the unmixing matrix W that maximizes the entropy in the
extractedsignalsy.
In this Project, we will implement information maximization approach to a simple
communications signal, the polar nonreturn to zero (NRZ) waveform separation. Figure
showsalgorithmusingthisapproachforunmixingofthecommunicationsignal.

Page1

Figure:Algorithmusinginformationmaximizationapproach

Page2

REFERENCES:

[1]A.J.BellandT.J.Sejnowski,Aninformationmaximizationapproachtoblindseparation
andblinddeconvolution,NeuralComputation,vol.7,pp.11291159,1995.
[2] Richard E. Blahut, Principles and Practice of Information Theory, pp. 5960, Addison
Wesley,Massachusetts,1987.
[3]SimonHaykin,AnIntroductiontoAnalogandDigitalCommunications,pp.197200,John
Wiley&Sons,NewYork,1989.
[4]AapoHyvrinen,JuhaKarhunen,ErkkiOja,IndependentComponentAnalysis,JohnWiley
&Sons,NewYork,2001.
[5]PeytonZ.Peebles,Jr.Probability,RandomVariables,andRandomSignalPrinciples,4th
Edition,pp.189193,TataMcGrawHill,NewDelhi,2002.
[6]BernardSklar,DigitalCommunications:FundamentalsandApplications,2ndEdition,pp.
10351045,PrenticeHallPTR,NewJersey,2001.
[7]JamesV.Stone,IndependentComponentAnalysis:ATutorialIntroduction,Massachusetts
InstituteofTechnology,Massachusetts,2004.

Page3

You might also like