S.A. Kuznetzov, A.S. Chumakov

This article describes the algorithm of the interpolation of Gaussian stochastic process (GSP) realizations using polynomial splines. The algorithm is based on correlations between different observations of the process. The survey of literature and the comparison of known signal reconstruction methods with developed algorithm were made. The algorithm is developed for two cases: quadratic spline and cubic spline. The interpolation quality was evaluated by the calculation of the deviation mean square between original and interpolated realizations. Was found that the interpolation quality obtained using quadratic and cubic splines differs less than 10% (when sampling step to GSP correlation time ratio is 0.5). The upper bound of the mean-square error of the GSP realization interpolation using this algorithm is 30% less than analogous values obtained using Kotelnikov's series