An attractive approach to distributed multiresolutional estimation

An optimal algorithm for multiresolutional estimation is derived using wavelet transforms. Distributed filtering with feedback is used. Wavelet transforms are utilized as a bridge linking different resolution levels, and to convert dense error covariance matrices into sparse ones. Therefore, computationally efficient algorithms can be obtained. The algorithm can be employed for dynamic multiresolutional sensor/data fusion. The discrete wavelet transformation is reviewed, with particular emphasis on wavelet transformations on random sequences