Scale transform approach for impulse responses identification

It is well known that noise is very difficult to deal with in system identification. Based on multiresolution analysis and wavelets, a scale transform approach for impulse response identification is proposed. The impulse response function is expressed with the expansion of an orthonormal scale function, an orthonormal scale transform is exerted on the signal, and least squares are used for parameter identification. Simulation results give a description of the advantages of this approach.