The Study And Application Of Adaptive Filtering Technology

Adaptive filter is used to study a system whose structure and parameter can be changed or adjusted. It possesses of good ability of running under the unknown environment and tracking the input statistics along with time variation. It is an important research subjuct in signal and information processing.The LMS algorithm has great significance in the transverse filter. LMS seeks the cost function minimum value in the gradient descent direction, that is the erroneous smallest principle. Another kind of important transverse filter algorithm is the RLS algorithm, evolved by the least squares method in the recursion form. These two algorithms are the foundation of other kind of algorithms and their variation.This paper proposes a IIR lattice algorithm. It can be regarded as a reverse process of the FIR lattice structure, and the difference is that the proposed has one out port. One of its biggest advantages is that the system function as reflected by all-pass characteristics. Gradient adaptive lattice with zero and pole (GALZP) algorithm uses the mean square value of the difference between input signal and the M-order backward signal as the cost function, and obtains the reflection factor update in the gradient direction. So it can maintain the lattice filter of all-pass characteristics well.The signal generated by lattice structure is not a direct estimate of the input signal. Therefore, lattice structure reflects the new sequence generated by the input signal sequence filtered. The adaptive processing system uses reference signal to guide the filter adjustment. The process combined new sequence with the reference signal to make the estimate for the expect signal is the joint estimation process.Uses the backward sequence in another adaptive process is the essence in joint estimate process. Joint estimation process has two kinds of models to update their weight in filtering, one is the overall error updating, and another is the step error recursion updating.IIR lattice joint processor has lots of advantages, such as fast convergence, smaller error, lower computation and so on. Its excellent properties can have practical application to other aspects of adaptive signal processing.