A.A. Naumenko1, K.A. Kivva2

1, JSC “CNIRTI named after academician A.I. Berg” (Moscow, Russia)

2 Bauman Moscow State Technical University (Moscow, Russia)

1 post@cnirti.ru

Problem statement. In modern radio engineering systems, processing and analysis of data on received radar signals is of great importance. One of the main tasks is to determine the number of radiation sources, as well as grouping the received radar signals by their belonging to different devices. However, measurement errors, as well as reflections of radio signals from different objects, lead to the fact that the measured characteristics differ significantly from the original ones, which leads to the formation of overlapping point clouds. Therefore, to solve the problem of correct grouping of radar signals by radiation sources, including both directly received and reflected ones, it is necessary to develop an appropriate clustering method.

Goal. To conduct a comparative analysis of existing methods for clustering reflected radar signals, which can form the basis for a new method.

Results. The clustering problem is formulated as applied to the received radar signals reflected from objects. The parameters of radar pulses, on the basis of which signal clustering can be carried out, are considered, and the frequency, duration and period of the pulse are determined as optimal. It is noted that a change in the ambient temperature leads to the occurrence of errors in the measurements of signal characteristics. The following groups of clustering methods are described: partitioning, hierarchical, probabilistic, methods based on density, grids, fuzzy logic, neural networks. The criteria for comparing the considered clustering methods, essential for solving the problem, are selected: the shape of the detected clusters, the need to pre-set their number, resistance to emissions, applicability for processing big data. A comparative analysis of the described groups of methods according to the selected criteria is carried out, as a result of which it is established that the optimal methods for solving the problem of clustering reflected radio signals, taking into account the limitations and requirements imposed by the problem, are density-based methods and methods using neural networks.

Practical significance. The presented results allow us to reasonably select or develop the most effective method for clustering reflected radar signals, which helps to increase the accuracy and reliability of radar data analysis.

Naumenko A.A., Kivva K.A. Comparative analysis of clustering methods of reflected radar signals. Radiotekhnika. 2025. V. 89. № 5. P. 137−146. DOI: https://doi.org/10.18127/j00338486-202505-15 (In Russian)

1. Ji Z., Bu Y. A signal sorting algorithm based on LOF de-noised clustering. Springer. Singapore. 2020.
2. J. Han, et al. A new method for sorting radar signal based on entropy features. Atlantis Press. 2018.
3. Ryzhov A.O. i dr. Vlijanie termostabilizacii opornogo generatora na tochnost' izmerenija chastoty radiosignalov i rekomendacii po minimizacii ee negativnyh pobochnyh jeffektov. Bulletin of VSTU. 2018 (in Russian).
4. Kugaevskih A.V., Muromcev D.I., Kirsanova O.V. Klassicheskie metody mashinnogo obuchenija. SPb: Universitet ITMO. 2022. 53 s. (in Russian).
5. Omran M., A.Salman A., Engelbrecht A. An overview of clustering methods. IOS Press. Intelligent Data Analysis 11. 2007.
6. Kang K., et al. Key radar signal sorting and recognition method based on clustering combined with PRI transform algorithm. Journal of Artificial Intelligence and Technology. 2022.
7. Wang S.Q., et al. Analysis of radar emitter signal sorting and recognition model structure. Procedia Computer Science. 2019.
8. Homenko I.V., Kosyh A.V. Kvarcevye rezonatory i generatory. Omsk: OmGTU. 2018. 160 s.
9. Aggarwal C.C., Reddy K.C. Data clustering. Minneapolis: CRC Press. 2011. 321 р.
10. Lia M., et al. A new clustering and sorting algorithm for radar emitter signals. Journal of Physics: Conference Series. 2020.
11. Everitt B.S., et al. Cluster Analysis. London: Wiley. 2022. 53 р.
12. Gao T., et al. Adaptive density peaks clustering: Towards exploratory EEG analysis. 2022.
13. Kovalev S.P. Ispol'zovanie algoritma klasterizacii DBSCAN dlja fil'tracii vybrosov v dannyh. BGUIR. 2019 (in Russian).
14. Khalid U., Karim N., Rahnavard N. RF signal transformation and classification using deep neural networks. University of Central Florida. Orlando. USA. 2022.
15. Zhilov R.A. Primenenie nejronyh setej pri klasterizacii dannyh. Izvestiya KBSC RAS. 2021 (in Russian).