Analysis of the impact of the finite step size in the autocorrelated measurement noise model on the performance of the discrete filtering algorithm

The article presents the results of a study analyzing the impact of the finite step size l in a model of autocorrelated measurement noise on the performance of the discrete filtering algorithm in the class of linear discrete stochastic systems. Using computer simulation in MATLAB, a comparison was made of the accuracy of state vector estimates in a discrete linear stochastic system for various values of the finite step of autocorrelated measurement noise, calculated using the standard Kalman filtering algorithm and its modified version that takes autocorrelated noise into account in the measurement model. The results of the computer simulations showed that as the value of l increases, the quality of state vector estimates obtained through the modified algorithm deteriorates only slightly compared to the standard Kalman algorithm, which loses effectiveness as l increases. Thus, the modified algorithm, despite its greater computational complexity, can be recommended for use when measurements contain autocorrelated noise.