Highlighting the influence of artifact signals on the equilibrium state of the feedback structure

In this paper, we address the problem of the determination of the equilibrium point of the feedback structure initially used for two-channel adaptive noise cancellation in the presence of crosstalk. We concentrate on an important characteristic of an adaptive filter, i.e. the steady-state mean-square error observed once the algorithm has converged. Our approach relies on the analysis of the relationships between the useful signals and their artifacts (distortion, residual noise, crosstalk) at the system outputs. The obtained equilibrium point essentially comes from experimental observations, nevertheless, we propose in the first part a theoretical framework that enables to highlight the links between the different signals in the feedback structure and also used to define the equilibrium point. Then, we give experimental results that show that the equilibrium state is obtained when the energy of the artifacts on the output signals is the same on each channel.