Nonlinear Blind Source Separation for Chemical Sensor Arrays Based on a Polynomial Representation

Rafael Assato Ando (1) , Christian Jutten (1) , Bertrand Rivet (1) , Romis Attux (2) , Leonardo Duarte (3)

1 GIPSA-VIBS - GIPSA - Vision and Brain Signal Processing
2 DSPCom - Laboratory of Signal Processing for Communications
3 LAMBE - Laboratoire Analyse et Modélisation pour la Biologie et l'Environnement

—In this paper we propose an extension of a blind source separation algorithm that can be used to process the data obtained by an array of ion-selective electrodes to measure the ionic activity of different ions in an aqueous solution. While the previous algorithm used a polynomial approximation of the mixing model and used mutual information as means of estimating the mixture coefficients, but it only worked for a constrained configuration of two sources with the same ionic valence. Our proposed method is able to generalize it to any number of sources and any type of ions, and is therefore able to solve the problem for any configuration. Simulations show good results for the analyzed application.

Domaines: Traitement du signal et de l'image [eess.SP]

Rafael Assato Ando, Christian Jutten, Bertrand Rivet, Romis Attux, Leonardo Duarte. Nonlinear Blind Source Separation for Chemical Sensor Arrays Based on a Polynomial Representation. EUSIPCO 2016 - 24th European Signal Processing Conference, Aug 2016, Budapest, Hungary. ⟨hal-01400550⟩