A.B. Nemov1

1 JSC “GOZ “Obukhov Plant” (St-Petersburg, Russia)

1 an.nilov2011@yandex.ru

High-quality GNSS consumer navigation equipment can use sophisticated digital signal processing algorithms based on the method of principal component correlation analysis (MPC) in application to the spatial sampling of signals received by consumer equipment. It is assumed that this sample contains an additive mixture of relatively powerful interference signals, relatively weak GNSS navigation signals and the equipment's own quasi-white noise. Capacities are taken into account relative to the equipment's own noise. MЗС is critical to the data on the number of various signals in the sample. Therefore, it is an urgent task to obtain information about the number of relatively powerful interference signals, which, unlike navigation signals, will be matched with the subspace of signals. The sample is processed in a digital GNSS antenna array.

The purpose of the article is to consider an algorithm for classifying the signals contained in a spatial sample into interference and other conditions.

Results. The classification and determination of the number of signals are based on dividing the space of the received signal mixture into subspaces of signals and noise. The algorithm uses information about the gradients of the eigenvalues of the estimated spatial covariance matrix of the received oscillations. It is shown that a linear eigenvalue classifier can be used to separate subspaces. It is proposed to train the classifier using an artificial neural network. True or simulated data on the normalized values of eigenvalues in various situations is used for training. As a result of the training, the slope parameter p of the linear classifier is determined. The probability of a correct decision on the number of interference signals is determined, which coincides with the probability of a correct decision by the classifier.

The practical significance of the material lies in the fact that the classification algorithm can be used to increase the efficiency of spatial processing of GNSS signals.

Nemov A.B. On the application of the perceptron to classify signals in a spatial sample of GNSS signals. Radiotekhnika. 2025. V. 89. № 2. P. 73−79. DOI: https://doi.org/10.18127/j00338486-202502-10 (in Russian)

1. Min' D.L., Nemov A.V., Tjuftjakov D.Ju. Primenenie metoda glavnyh komponent dlja kompensacii pomeh pri prieme signalov global'noj navigacionnoj sistemy. Izvestija vuzov Rossii. Ser. Radiojelektronika. 2017. № 3. S. 16-23 (in Russian).
2. Nemov A.V. Cifrovye antennye reshetki: novoe kachestvo sputnikovyh radionavigacionnyh sistem. SPb: Izd-vo SPbGJeTU «LJeTI». 2014. 159 s. (in Russian).
3. Litvinov O.S., Murod'janc D.V., Boruta V.S., Vintajkin B.E. Podavlenie shumovyh pomeh v adaptivnyh antennyh reshetkah na osnove nejrosetevyh algoritmov. Antenny. 2018. Vyp. 2(246). S. 40-43 (in Russian).
4. Nemov A.V., Tjuftjakov D.V. Algoritmy ocenivanija kolichestva signalov v vyborkah dannyh, obrabatyvaemyh CAR GNSS. Sb. XXIX Sankt-Peterburgskoj konf. po integrirovannym navigacionnym sistemam. 2022. S. 332-336 (in Russian).
5. Hajkin S. Nejronnye seti: polnyj kurs. Izd. 2-e: Per. s angl. M.: OOO «I.D. Vil'jams». 2016. 1104 s. (in Russian).
6. Ermolaev V.T., Mal'cev A.A., Rodjushkin K.V. Statisticheskie harakteristiki kriteriev AIC i MDL v zadache ocenki chisla istochnikov mnogomernyh signalov v sluchae korotkoj vyborki. Izvestija vuzov. Ser. Radiofizika. 2001. № 12. S. 1062-1069 (in Russian).