L.I. Dvoiris, I.N. Kryukov
Statement of the problem. Detection/recognition of signals of various nature against the background of interference is a rather complex task, the successful solution of which depends on knowledge of the probability distribution densities of signal properties. The absence of signal models leads to the need to search for their approximations. In practice, only 15-20% of the total number of signals studied are adequately described by the normal distribution law. In other cases, there is a need for their parametric evaluation, making a choice of distribution among many alternatives. Therefore, the development of tools for approximating probability distributions of signal characteristics, ensuring the accuracy and reliability of such estimates to ensure increased noise immunity of the detection/recognition system as a whole, is an urgent task.
Goal. To investigate and show the possibility of using the method of nonparametric estimation of the densities of probability distributions of signal parameters (using the example of one-dimensional ones) in the tasks of object detection and recognition.
Results. A mathematical apparatus has been developed for nonparametric estimates of one-dimensional densities of the probability distribution of signal characteristics in the interests of detection and recognition. It is shown that the presence of an approximated probability distribution density makes it possible to find probabilistic characteristics of solutions to various problems arising during the processing of experimental data.
Practical significance. The application of the method of one-dimensional nonparametric estimation of the probability distribution densities of signal characteristics and interference makes it possible to increase the noise immunity of object detection and recognition systems in various applied fields.
Dvoiris L.I., Kryukov I.N. Nonparametric method for estimating one-dimensional probability distribution densities of experimental data. Radiotekhnika. 2022. V. 86. № 1. P. 11−15. DOI: https://doi.org/10.18127/j00338486-202201-02 (In Russian)
- Dzheffri R. Neparametricheskaja jekonometrika: vvodnyj kurs. Kvantil'. 2008. № 4. S. 7−56 (In Russian).
- Weglarczyk S. Kernel density estimation and its application. ITM Web of Conferences. XLVIII Seminar of Applied Mathematics. 2018. № 23. Р. 00037.
- Scott D.W., Sheather S.J. Kernel Density Estimation with binned Data. Computer Science and Statistics. Springer-Verlag. New York. 1985. Р. 292−294.
- Tjurin Ju.N. Neparametricheskie metody statistiki. M.: Znanie. 1978 (In Russian).