Journal of Radio Electronics. eISSN 1684-1719. 2026. ¹5
Full text in Russian (pdf)
DOI: https://doi.org/10.30898/1684-1719.2026.5.15
ANALYSIS OF VARIOUS NEURAL NETWORK ALGORITHMS
FOR ESTIMATING THE AMPLITUDE OF A RADIO SIGNAL
WITH UNKNOWN INITIAL PHASE, DURATION, AND FREQUENCY
Yu.E. Korchagin, Nguyen Van Thuy
Voronezh State University,
394006, Russia, Voronezh, Universitetskaya Ploshchad, 1
The paper was received April 30, 2026.
Abstract. This paper presents a study of various neural network–based algorithms for estimating the amplitude of a radio signal with an arbitrarily shaped envelope under conditions of a priori uncertainty in the signal duration, initial phase, and carrier frequency. Using the Fisher information matrix, the lower bound on the variance of the amplitude estimator is derived. A comparative evaluation of estimation accuracy across different neural network architectures is then carried out, together with benchmarking against the Cramer–Rao lower bound. The performance of these architectures is analyzed for two types of training data representation: time-domain signals and time–frequency spectra. The results show that the accuracy of neural network–based estimation depends not only on the network architecture but also on the representation of the training data. Convolutional neural networks demonstrate high performance in both scenarios, with the best results achieved when the training data are represented in discrete form in the time–frequency domain. In contrast, fully connected neural networks exhibit low performance for both types of data representation. The gated recurrent unit performs effectively when the input data are represented as discrete time-domain samples; however, a substantial degradation in performance is observed when time–frequency spectra are used as the training data representation.
Key words: radio signal, neural network algorithm, amplitude estimation, convolutional neural network, multilayer perceptron, gated recurrent unit, unknown frequency, duration, initial phase.
Financing: The study was supported by a grant from the Russian Science Foundation Project No. 24-19-00891, https://rscf.ru/project/24-19-00891/.
Corresponding author: Yury Eduardovich Korchagin, korchagin@phys.vsu.ru
References
1. Zheng S. et al. Estimation of echo amplitude and time delay for OFDM-based ground-penetrating radar // IEEE Geoscience and Remote Sensing Letters. – 2015. – Vol. 12. – No. 12. – P. 2384-2388. https://doi.org/10.1109/LGRS.2015.2478279
2. Zanella A. Best practice in RSS measurements and ranging // IEEE Communications Surveys and Tutorials. – 2017. – Vol. 18. – No. 4. – P. 2662-2686. https://doi.org/10.1109/COMST.2016.2553452
3. Kulikov E.I., Trifonov A.P. Estimation of signal parameters against a background of interference. – M.: Sov. radio, 1978. – 296 p.
4. Trifonov A.P. et al. Estimation of the amplitude of a rectangular narrowband radio pulse with unknown duration and initial phase // Bulletin of higher educational institutions Radio Electronics. – 2017. – Vol. 60. – No. 12. – P. 675-684. https://doi.org/10.20535/S0021347017120019
5. Trifonov A.P., Shinakov Yu.S. Joint discrimination of signals and assessment of their parameters against a background of interference. – M.: Radio and Communications, 1986. – 268 p.
6. Jwo D.J., Biswal A., Mir I.A. Artificial neural networks for navigation systems: A review of recent research // Applied Sciences. – 2023. – Vol. 13. – No. 7. – P. 4475. https://doi.org/10.3390/app13074475
7. Jiao J. et al. An overview of wireless communication technology using deep learning // China Communications. – 2021. – Vol. 18. – No. 12. – P. 1-36. https://doi.org/10.23919/JCC.2021.12.001
8. Andreev V.V., Slavutskiy L.A., Slavutskaya E.V. Signal processing by a feedforward neural network: approximation and decision making // Bulletin of the Chuvash University. – 2022. – No. 1. – P. 14-22. https://doi.org/10.47026/1810-1909-2022-1-14-22
9. Trifonov A.P., Buteyko V.K. Characteristics of joint estimates of signal parameters with partial violation of regularity conditions // Radio Engineering and Electronics. – 1991. – Vol. 36. – No. 2. – P. 319-327.
10. Melnik E.V. et al. Design of MLP and CNN neural network modules on FPGA for image classification problems // Bulletin of the Southern Federal University. Technical sciences. – 2025. – No. 5 (247). – P. 214-229.
11. Li Y., Xie T., Mei D. Application of convolutional neural networks for parallel multi-scale feature extraction in noise image denoising // IEEE Access. – 2024. – Vol. 12. – P. 98599-98610. https://doi.org/10.1109/ACCESS.2024.3427143
12. Zhanga J., Zhangb X., Jic X. Application of Convolutional Neural Networks in High Score Remote Sensing Image Classification // Academic Journal of Computing & Information Science. – 2024. – Vol. 7. – No. 2. – P. 79-84. https://doi.org/10.25236/AJCIS.2024.070211
13. Schneider N., Lermer M., Reich C. Investigating the performance of CNN feature extractor-based LSTM and GRU variants for time series classification // Procedia Computer Science. – 2024. – Vol. 246. – P. 1070-1079. https://doi.org/10.1016/j.procs.2024.09.526
14. Yang T.Y. et al. GNSS–VTEC prediction based on CNN–GRU neural network model during high solar activities // Scientific reports. – 2025. – Vol. 15. – No. 1. – P. 9109. https://doi.org/10.1038/s41598-025-93628-8
For citation:
Korchagin Yu.E, Nguyen Van Thuy. Analysis of various neural network algorithms for estimating the amplitude of a radio signal with unknown initial phase, duration, and frequency // Journal of Radio Electronics. – 2026. – ¹. 5. https://doi.org/10.30898/1684-1719.2026.5.15 (In Russian)