A novel approach for blind multichannel identification is proposed on the basis of Kalman filter theory. Taking advantage of the cross relations between any pair of the multichannel outputs, the process and measurement equations are established in state space. The standard Kalman filter algorithm is simplified by exploiting the special zero observation vector which will leads to the homogenous iterative ways of the state vector and the filtered state-error correlation matrix. Simulations show that the convergence speed is significantly higher than those of LMS-like approaches.
展开▼
