Blind source separation and identification of nonlinear multiuser channels using second order statistics and modulation codes

In this paper, a Blind Source Separation (BSS) and channel identification method using second order statistics is proposed for nonlinear multiuser communication channels. It is based on the joint diagonalization of a set of covariance matrices. Modulation codes (constrained codes) are used to ensure the orthogonality of nonlinear combinations of the transmitted signals, allowing the application of a joint diagonalization based estimation algorithm. This constitutes a new application of modulation codes, used to introduce temporal redundancy and to ensure some statistical constraints. Identifiability conditions for the problem under consideration are addressed and some simulation results illustrate the performance of the proposed method.