THE RANDOM PROCESS REPRESENTATION WITH THE HELP OF VECTORIAL RECURRENT
MODEL OF THE SECOND ORDER
V. P. Volchkov, N. E.
Poborchaya, A. M. Shloma
The paper is received
on December 12, 2013
Abstract. One of à
foreground task of digital signal processing is a decrease of computational
complexity of appropriate algorithms, but without substantial loss accuracy. Good
adequate signal models are necessary for synthesis of these algorithms. In the
article, the vectorial recurrent circulant model of the second dynamic order is
proposed for random signals representation. For given correlation function of
signal the computational procedures of parameters this model and expression of
approximating correlation function are derived. The comparative analysis of
approximation property of the proposal model and other known models are
executed. Optimal recurrent algorithms of filtration and smoothing based on the
proposal model are produced. Accuracy behavior and efficiency of these
algorithms are confirmed in experimental research.
Key words: random process, random signal, optimal estimate, Kalman
filter, Winner filter, circulant model, autoregressive model, stochastic
difference equalization, correlation function, Fourier transform.