D. L. Myasnikov1
1 Povolzhskiy State University of Telecommunications and Informatics (Samara, Russia)

1 myasnikov.danil@internet.ru

Modern neural networks require significant computing resources, which makes it difficult to integrate them into real-time systems such as RFID technologies, where high data processing speed is important.

The objective of the article is to optimize the neural network architecture by reducing the number of fully connected layers and changing a number of important parameters, while maintaining the recognition quality and functionality of the model.

It has been shown that it is possible to reduce the number of fully connected layers and optimize the key parameters of the neural network, providing a significant increase in its performance without compromising the accuracy of classifications.

The optimized architecture obtained makes it possible to successfully integrate a neural network into real RFID systems, increasing the efficiency of object identification processes and reducing computing power costs.

Myasnikov D.L. Optimization of the convolutional neural network architecture to increase the noise immunity of the RFID signal relay system // Antennas. 2026. № 3. P. 86–93. DOI: https://doi.org/10.18127/j03209601-202603-08

1. Zakaria R., Abdelmajid H., Dya Z. et al. CXAI: Explaining convolutional neural networks for medical imaging diagnostic. Electronic Devices and Systems for Biomedical Applications. 2022 [Elektronnyj resurs]. URL: https://doi.org/10.3390/electronics11111775.
2. Kim D.-Y., Yook J.-G., Yoon H.-G. et al. Interference analysis of UHF RFID systems. Progress In Electromagnetics Research B. 2008. V. 4. P. 115–126.
3. Myasnikov D.L., Lozhkin L.D., Sokolova Yu.V. Opredelenie vliyaniya pomekh na izobrazheniya raznykh tsvetov posredstvom PO. Fizika volnovykh protsessov i radiotekhnicheskie sistemy. 2023. T. 26. № 1. S. 116–122. (in Russian)
4. Python [Elektronnyj resurs]. URL: https://www.python.org.
5. Tensorflow [Elektronnyj resurs]. URL: https://www.tensorflow.org.
6. Klyuev D.S., Myasnikov D.L. Sozdanie nejronnoj seti dlya klassifikatsii zashumlennykh signalov radiochastotnoj identifikatsii. Sb. trudov XXXI Mezhdunar. nauch.-tekhnich. konf. «Radiolokatsiya, navigatsiya, svyaz'». 2025. T. 5. S. 292–297. (in Russian)
7. Novomejskij D.N., Piganov M.N., Luptsov A.A. i dr. Metodika sinteza topologii platy uzla radiochastotnogo trakta. Fizika volnovykh protsessov i radiotekhnicheskie sistemy. 2025. T. 28. № 3. S. 49–55. (in Russian)
8. Sharifrazi D., Khozeimeha F., Izadic N.H. et al. CNN AE: Convolution neural network combined with autoencoder approach to detect survival chance of COVID 19 patients. 2021.
9. Mishin D.V., Gladkikh A.A., Kutuzov V.I. i dr. Issledovanie protsessov kognitivnoj obrabotki dannykh v sistemakh radiosvyazi s perestanovochnym dekodirovaniem. Fizika volnovykh protsessov i radiotekhnicheskie sistemy. 2024. T. 27. № 1. S. 103–112. (in Russian)