INVITED PAPERBlind Deconvolution of Dynamical Systems: A State-Space Approach

In this paper we present a general framework of the state space approach for blind deconvolution. First, we review the current state of the art of blind deconvolution using statespace models, then give a new insight into blind deconvolution in the state-space framework. The cost functions for blind deconvolution are discussed and adaptive learning algorithms for updating external parameters are developed by minimizing a certain cost function, which is derived from mutual information of output signals. The information backpropagation approach is developed for training the internal parameters. In order to compensate for the model bias and reduce the e ect of noise, we introduce the Kalman lter to the blind deconvolution setting. A new concept, called hidden innovation, is introduced so as to numerically implement the Kalman lter. Thus we propose a new method: the two-stage approach to blind deconvolution. Finally we suggest how to extend the information backpropagation approach to the nonlinear case. Computer simulations are given to show the validity and e ectiveness of the state-space approach.