A.P. Pavlov1, I.E. Kashchenko2, A.V. Bakhmutskaya3

1-3 Omsk Scientific Center SB RAS (Institute of Radiophysics and Physical Electronics) (Omsk, Russia)

Currently, one of the key factors determining the energy efficiency of a radio transmitting system is the degree of matching between the radio transmitting device and the antenna. For this reason, HF radio transmission systems often include automatic antenna matching devices that give the best degree of matching over the entire frequency band. However, these devices, due to the slow convergence of algorithms based on classical search methods used to determine the optimal state of the matching network, require significant tuning time when changing the frequency This is contrary to the latest trends in the development of radio transmitting systems, which involve the use of fast frequency hopping to ensure the noise immunity of radio links. Thus, research on the development of matching methods and algorithms that will significantly reduce the matching time of the radio transmitting path is a relevant task.

The aim of the article is to study of the possibility of reducing the time of matching the antenna-feeder path with a HF radio transmitter using an automatic switching antenna-matching device by using machine learning and neural networks as an algorithm or part of an algorithm for determining the corresponding state of the matching network. The article presents the implementation of the algorithm for matching the antenna-feeder path of the HF band of radio waves based on neural networks with the distribution of responsibility, as well as their architecture and methods for their training. The use of the presented algorithm makes it possible to significantly reduce the time for determining the state of the matching network of the antenna-matching device in order to achieve the best matching. The architecture of neural networks used in the proposed algorithm is adapted for implementation on low-power microcontrollers.

Pavlov A.P., Kashchenko I.E., Bakhmutskaya A.V. Algorithm of matching of the antenna-feeder path of HF radio waves based on neural networks with the distribution of responsibility zones. Electromagnetic waves and electronic systems. 2023. V. 28. № 1. P. 37−46. DOI: https://doi.org/10.18127/j15604128-202301-05 (in Russian)

1. Kovalevich D.A., Listopad N.I., Batura M.P. Optimizatsiya raschetnogo algoritma raboty dlya avtomaticheskikh soglasuyushchikh ustroistv korotkovolnovogo diapazona. SVCh-tekhnika i telekommunikatsionnye tekhnologii. 2020. № 1−2. S. 215−216. (in Russian)
2. Ermolaev V.V., Zhukov V.M. Sokrashchenie vremeni nastroiki tsifrovogo avtomaticheskogo soglasuyushchego ustroistva s pomoshchyu vychislitelnogo metoda. Vestnik Tambovskogo gosudarstvennogo tekhnicheskogo universiteta. 2013. T. 19. № 3. S. 544−552. (in Russian)
3. Rodriguez J.L., Garca-Tunon I., Taboada J.M., Obelleiro F., Basteiro. Broadband HF Antenna Matching Network Design Using a Real-Coded Genetic Algorithm. IEEE Transactions on Antennas and Propagation. 2007. V. 55. № 3. P. 611−618.
4. Alibakhshikenari M., Virdee B.S., Shukla P., See C.H., Abd-Alhameed R.A., Falcone F., Limiti E. Improved adaptive impedance matching for RF front-end systems of wireless transceivers. Sci. Rep. 2020. 10. 14065.
5. Cybenko G. Approximation by superpositions of a sigmoidal function. Math. Control Signal Systems 2. 1989. P. 303−314.
6. Nicholas J.C., Bogdan M.W. Compensation of Nonlinearities Using Neural Networks Implemented on Inexpensive Microcontrollers. IEEE Transactionm on industrial electronics. 2011. V. 58. P. 733−740.
7. Babkov V.Yu., Muravev Yu.K. Osnovy postroeniya ustroistv soglasovaniya antenn. L.: VAS. 1980. 240 s. (in Russian)
8. Fainitski D. Avtomaticheskii antennyi tyuner s avtonomnym pitaniem (prototip). URL: http://www.sdr-deluxe.com/publ/ avtomaticheskij_antennyj_tjuner_s_avtonomnym_pitaniem_prototip/1-1-0-55 (data obrashcheniya 02.08.2022). (in Russian)
9. Pavlov A.P., Bakhmutskaya A.V., Kashchenko I.E. Soglasovanie antenno-fidernogo trakta korotkovolnovogo diapazona radiovoln s primeneniem kommutatsionnogo antenno-soglasuyushchego ustroistva na baze neironnoi seti. Elektromagnitnye volny i elektronnye sistemy. 2021. T. 26. № 5. S. 67−74. DOI: https://doi.org/10.18127/j15604128-202105-08 (in Russian)
10. Pavlov A.P., Kashchenko I.E. Soglasovanie antenn-fidernogo trakta korotkovolnovogo diapazona radiovoln s primeneniem razlichnykh metodov optimizatsii. Tekhnika radiosvyazi. 2020. № 4(47). S. 77−86. DOI 10.33286/2075-8693-2020-47-77-86. (in Russian)
11. Hirose A., Yoshida S. Generalization Characteristics of Complex-Valued Feedforward Neural Networks in Relation to Signal Coherence. IEEE Transactions on Neural Networks and Learning Systems. 2012. V. 23. № 4. P. 541−551. doi: 10.1109/TNNLS.2012.2183613.
12. Marieme Ngom, Oana Marin. Fourier neural networks as function approximators and differential equation solvers. Statistical Analysis and Data Mining (The ASA Data Science Journal). 2021. V. 14. P. 647−661.
13. Khaikin S. Preimushchestva i ogranicheniya obucheniya metodom obratnogo rasprostraneniya. Neironnye seti. M.: Vilyams. 2006. S. 304−314. (in Russian)
14. Transtrum M.K., Sethna J.P. Improvements to the Levenberg-Marquardt algorithm for nonlinear least-squares minimization. URL: https://arxiv.org/pdf/1201.5885.pdf (data obrashcheniya 02.08.2022).