This paper describes an algorithm that reduces computational requirement of the Kalman filter and estimates the signal efficiently. The reference signals are mapped onto the orthogonal wavelet transform domain so that the eigenvalue spread of its autocorrelation matrix could be smaller than that in the time domain. In the wavelet transform domain the autocorrelation matrix is nearly diagonal. Therefore, the transformed signal can be decomposed each orthogonal elements. The Kalman filter can be applied to each orthogonal elements and computational requirement is reduced. The possibility of applying the parallel Kalman filter was verified through the theory and simulation. The eigenvalue spread in the wavelet transform domain is smaller 8.35 times than that in the time domain and the computational requirement is reduced from 1.4 times to 2. 93 times than that of the conventional Kalman filter.