Blind channel estimation based TLS algorithm in fully connected network with two signal sources
Zohre Foroushi, Vahid Tabatabavakili · 2014
Blind channel estimation for dereverberation is noticed in most applications such as, blind source separation, communication and etc. The important challenge in this application is convolutive mixtures in a reverberant acoustic environment. In this paper, a MIMO system that has 2 input signals will be converted into 2 SIMO systems and their spatial interference will be separated. A tree structure is chosen from fully connected network and one virtual node is added to this network at signal source 1. The equations of 2 SIMO systems are merged and a linear equation is obtained from them. This equation is solved by TLS estimation and channel coefficients are obtained iteratively. We have provided MC simulation that demonstrates the effectiveness of the algorithm.