Modeling and Compensating Input Distortion in MIMO Systems via Blind Deconvolution

This paper proposes a method for modeling and compensating nonlinear input distortion in MIMO dynamical systems. This is conducted in three steps. First, the distorted signals are recovered from the observed output by using blind signal deconvolution based on statistical independence of the input signals. The linear part of the system is also identified in this first step. Secondly, the recovered signal, together with the input signal, gives a model of distortion as a nonlinear function. Finally the distortion is compensated by applying the inverse of the estimated function.