Imagery modeling for microwave engineering with unsupervised variational autoencoder

doi:10.23919/JSEE.2026.000140

Abstract

Abstract:

Imagery modeling for microwave engineering generally requires a large number of training samples to generate images. In this paper, a variational autoencoder (VAE) with the multilayer perceptron (MLP) is proposed for efficient imagery modeling of electromagnetic behaviors of the microwave antenna and component. The input of VAE is the randomly generated binary images of a physical structure, and VAE is an unsupervised learning scheme. The training of MLP needs the labeled samples of the latent representation of VAE and the electromagnetic responses. Compared with supervised imagery modeling, VAE generates a large number of low-cost images applied to the image processing section. The implementation of unsupervised learning eliminates the need for labeled samples from full-wave simulations, resulting in significant time saving. The proposed model is confirmed with two examples of a pixel antenna and a microstrip/coplanar waveguide ultrawideband filter, and the results show its improvement in accuracy and efficiency.

Key words: imagery modeling, latent representation, unsupervised learning, variational autoencoder (VAE)

Shuzhong Yue, Wengao Zou, Wei Shao, Xiao Ding. Imagery modeling for microwave engineering with unsupervised variational autoencoder[J]. Journal of Systems Engineering and Electronics, 2026, 37(4): 1171-1176.

Figures/Tables 12

Fig 1

Fig 2

Fig 3

Table 1

Fig 4

Table 2

Fig 5

Fig 6

Table 3

Fig 7

Table 4

Fig 8

References 33

1	Rayas-Sánchez J E EM-based optimization of microwave circuits using artificial neural networks: the state-of-the-art. IEEE Trans. on Microwave Theory and Techniques, 2004, 52 (1): 420. doi: 10.1109/TMTT.2003.820897
2	Zhang Q J, Gupta K C, Devabhaktuni V K Artificial neural networks for RF and microwave design-from theory to practice. IEEE Trans. on Microwave Theory and Techniques, 2003, 54 (4): 1339. doi: 10.1109/tmtt.2003.809179
3	Burrascano P, Fiori S, Mongiardo M A review of artificial neural networks applications in microwave computer-aided design (invited article). International Journal of RF and Microwave Computer-Aided Engineering, 1999, 9 (3): 158. doi: 10.1002/(SICI)1099-047X(199905)9:3<158::AID-MMCE3>3.0.CO;2-V
4	Feng F, Na W C, Jin J, et al Artificial neural networks for microwave computer-aided design: the state of the art. IEEE Trans. on Microwave Theory and Techniques, 2022, 70 (11): 4597. doi: 10.1109/TMTT.2022.3197751
5	Karaboga D, Guney K, Sagiroglu S, et al Neural computation of resonant frequency of electrically thin and thick rectangular microstrip antennas. IEE Proceedings-Microwaves Antennas and Propagation, 1999, 146 (2): 155. doi: 10.1049/ip-map:19990136
6	Guney K, Sagiroglu S, Erler M Generalized neural method to determine resonant frequencies of various microstrip antennas. International Journal of RF and Microwave Computer-Aided Engineering, 2002, 12, 131. doi: 10.1002/mmce.10006
7	Angiulli G, Versaci M Resonant frequency evaluation of microstrip antennas using a neural-fuzzy approach. IEEE Trans. on Magnetics, 2003, 39 (3): 1333. doi: 10.1109/TMAG.2003.810172
8	Guney K, Sarikaya N A hybrid method based on combining artificial neural network and fuzzy inference system for simultaneous computation of resonant frequencies of rectangular, circular, and triangular microstrip antennas. IEEE Trans. on Antennas and Propagation, 2007, 55 (3): 659. doi: 10.1109/TAP.2007.891566
9	Khan T, De A Estimation of radiation characteristics of different slotted microstrip antennas using a knowledge-based neural networks model. International Journal of RF and Microwave Computer-Aided Engineering, 2014, 24 (6): 673. doi: 10.1002/mmce.20811
10	Yuan L, Wang L, Yang X S, et al An efficient artificial neural network model for inverse design of metasurfaces. IEEE Antennas and Wireless Propagation Letters, 2021, 20 (6): 1013. doi: 10.1109/LAWP.2021.3069713
11	Jacobs J P Accurate modeling by convolutional neural-network regression of resonant frequencies of dual-band pixelated microstrip antenna. IEEE Antennas and Wireless Propagation Letters, 2021, 20 (12): 2417. doi: 10.1109/LAWP.2021.3113389
12	Luo H Y, Shao W, Ding X, et al Shape modeling of microstrip filters based on convolutional neural network. IEEE Microwave and Wireless Components Letters, 2022, 32 (9): 1019. doi: 10.1109/LMWC.2022.3162414
13	Luo H Y, Hong Y, Lv Y H. Parametric modeling of UWB antennas using convolutional neural networks[C]//IEEE International Symposium on Antennas and Propagation and North American Radio Science Meeting, 2020: 2055.
14	An S S, Zhang B W, Tang H, et al Multifunctional metasurface design with a generative adversarial network. Advanced Optical Materials, 2021, 9 (5): 2001433. doi: 10.1002/adom.202001433
15	Hodge J A, Mishra K V, Zaghloul A I. RF metasurface array design using deep convolutional generative adversarial networks[C]//IEEE International Symposium on Phased Array System & Technology, 2019: 1.
16	Liu P Q, Chen L, Chen Z N Prior-knowledge-guided deep-learning-enabled synthesis for broadband and large phase shift range metacells in metalens antenna. IEEE Trans. on Antennas and Propagation, 2022, 70 (7): 5024. doi: 10.1109/TAP.2021.3138517
17	Liu Z X, Shao W, Ding X, et al Equivalent circuit-guided GAN sample generation of metasurface for low-RCS scanning array. IEEE Trans. on Antennas and Propagation, 2024, 72 (9): 7201. doi: 10.1109/TAP.2024.3430078
18	Liu P Q, Chen Z N. Prior knowledge-guided deep learning algorithms for metantenna design (invited)[C]//IEEE International Workshop on Antenna Technology, 2024: 11.
19	Liu P Q, Chen Z N Full-range amplitude-phase metacells for sidelobe suppression of metalens antenna using prior-knowledge-guided deep-learning-enabled synthesis. IEEE Trans. on Antennas and Propagation, 2023, 71 (6): 5036. doi: 10.1109/TAP.2023.3263929
20	Dai M N, Jiang Y, Yang F, et al. A surrogate-assisted extended generative adversarial network for parameter optimization in free-form metasurface design[J]. Neural Networks, 2024: 106654.
21	Liu Z C, Zhu D, Rodrigues S P, et al Generative model for the inverse design of metasurfaces. Nano Letters, 2018, 18 (10): 6570. doi: 10.1021/acs.nanolett.8b03171
22	Hodge J A, Mishra K V, Zaghloul A I. Joint multi-layer GAN-based design of tensorial RF metasurfaces[C]//IEEE 29th International Workshop on Machine Learning for Signal Processing, 2019: 1.
23	Gu Z M, Li D, Wu Y L, et al A solution to the dilemma for FSS inverse design using generative models. IEEE Trans. on Antennas and Propagation, 2023, 71 (6): 5100. doi: 10.1109/TAP.2023.3266053
24	Liu M D, Zhang H, Song J M, et al Using generative model for intelligent design of dielectric resonator antennas. Microwave and Optical Technology Letters, 2024, 66 (1): e34013. doi: 10.1002/mop.34013
25	Banerjee B, Elliott R C, Krzymien W A, et al Downlink channel estimation for FDD massive MIMO using conditional generative adversarial networks. IEEE Trans. on Wireless Communications, 2023, 22 (1): 122. doi: 10.1109/TWC.2022.3191605
26	Liu M D, Zhang H, Lu M, et al. Generative adversarial network-based design of dielectric resonator antenna for mmWave 5G applications[C]//IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting, 2021: 1877.
27	Diederik P K, Max W. Auto-encoding variational bayes[EB/OL]. [2024-07-01]. https://arxiv.org/abs/1312.6114.
28	Ding M, Jin R H, Geng J P Optimal design of ultra wideband antennas using a mixed model of 2-D genetic algorithm and finite-difference time-domain. Microwave and Optical Technology Letters, 2007, 49 (12): 3177. doi: 10.1002/mop.22928
29	Sohn K, Lee H, Yan X C. Learning structured output representation using deep conditional generative models[C]//Advances in Neural Information Processing Systems, 2015: 28.
30	Mehdi M, Simon O. Conditional generative adversarial nets[J/OL]. [2024-07-01]. https://arxiv.org/abs/1411.1784.
31	Lecun Y, Bottou L, Bengio Y, et al Gradient-based learning applied to document recognition. Proceedings of the IEEE, 1998, 86 (11): 2278. doi: 10.1109/5.726791
32	Neil T, Hong J S Compact ultra-wideband microstrip/coplanar waveguide bandpass filter. IEEE Microwave and Wireless Components Letters, 2007, 17 (3): 184. doi: 10.1109/LMWC.2006.890462
33	Telzhensky N, Leviatan Y Novel method of UWB antenna optimization for specified input signal forms by means of genetic algorithm. IEEE Trans. on Antennas and Propagation, 2006, 54 (8): 2216. doi: 10.1109/TAP.2006.879201

Set	Set₁	Set₂	Set₃	Set₄	Set₅	Set₆	Set₇	Set₈
Lower frequency	3.28	3.32	3.32	3.37	3.25	3.28	3.31	3.34
Upper frequency	10.93	10.23	10.81	9.77	9.6	10.0	10.4	10.8

Model	Samples	MAPE/%
CGAN	500 (labeled)	2.74
CVAE	500 (labeled)	2.95
CNN	500 (labeled)	−
Proposed model	10000 (unlabeled) +500 (labeled)	2.01

Set	\|S₁₁\|_L = −10 dB	\|S₁₁\|_U = −10 dB	\|S₂₁\|_L = −3 dB	\|S₂₁\|_U = −3 dB
Set₁	4.30	8.96	3.88	9.18
Set₂	4.36	8.86	3.94	9.10
Set₃	4.32	9.20	3.92	9.46
Set₄	4.30	8.80	3.90	9.05
Set₅	4.35	9.00	3.80	9.25
Set₆	4.30	9.20	3.85	9.45

Model	Samples	MAPE/%
CGAN	500 (labeled)	1.46
CVAE	500 (labeled)	1.56
CNN	500 (labeled)	−
Proposed model	10000 (unlabeled) +500 (labeled)	0.93