Blind separation algorithm without nonlinear functions - Electronics Letters

Introduction: In a large number of cases the signal received by a sensor is the sum of elementary contributions that we can call sources. In these cases, without any knowledge of the sources (except an independence assumption), we would like to separate these independent sources. This problem is called 'blind separation of sources'. Ever since the remarkable results of the blind separation of sources by Herault and Jutten [ l , 21 were given for the first time, a few novel learning algorithms for blind separation have been reported in the literature [3, 41. The basic problem of blind separation of sources can be formulated as follows: let a given set of sensor signals e,(t) be described bY