Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures

Harun Sevinç; Levent Seyfi

doi:10.54287/gujsa.1901011

Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures

Abstract

Frequency Modulated Continuous Wave (FMCW) radars present a robust and privacy-friendly alternative for human activity recognition (HAR) systems. In the classification of micro-Doppler signatures obtained from these radars via deep learning algorithms, the structural quality of the time-frequency (TF) transformation used as input directly affects the model's performance. Therefore, the explicit objective of this study is to design and optimize lightweight Convolutional Neural Network (CNN) architectures capable of efficiently extracting micro-Doppler features from various TF representations under the constraints of limited radar data. To achieve this, signals belonging to 6 different human activities collected by an FMCW radar were converted into spectrograms using Short-Time Fourier Transform (STFT), Continuous Wavelet Transform (CWT), and Wigner-Ville Distribution (WVD). Three novel convolutional neural network architectures (CNN–Base, CNN–Wide, CNN–LSTM), capable of providing different responses to the spatial features of these transformations, were designed and analyzed using a 6-fold cross-validation strategy. Experimental results demonstrated that the logarithmic scaling of CWT misled CNN filters, while the standard STFT provided a strong baseline with 80.00% accuracy. The highest performance of the study was achieved at 80.42% with the CNN–Wide architecture, which successfully processed the high-resolution texture of WVD containing cross-terms utilizing its wide receptive field. The WVD-based model identified life-critical falling activities with 100% precision, offering a promising approach for elderly care and autonomous health monitoring systems.

Keywords

Supporting Institution

Scientific and Technological Research Council of Türkiye (TUBITAK)

Project Number

125E391

Ethical Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Konya Technical University (Protocol Code: 2025/2-3, Date of Approval: 28/04/2025). Informed consent was obtained from all subjects involved in the study.

Thanks

This study was supported by Scientific and Technological Research Council of Türkiye (TUBITAK) under Grant Number 125E391. The authors thank TUBITAK for their support.

References

Abdu, F. J., Zhang, Y., & Deng, Z. (2022). Activity classification based on feature fusion of FMCW radar human motion micro-Doppler signatures. IEEE Sensors Journal, 22(9), 8648-8662. https://doi.org/10.1109/JSEN.2022.3156762
Avinash, B. S., Aryan, R., Yadav, A., Kumar Gupta, V., & Kumar, B. (2025). Hierarchical Multiscale CNN With Frequency-Aware Attention for Enhanced HAR. IEEE Sensors Journal, 25(16), 31175-31184. https://doi.org/10.1109/JSEN.2025.3586069
Biswas, S., Manavi Alam, A., & Gurbuz, A. C. (2024). HRSpecNET: A deep learning-based high-resolution radar micro-Doppler signature reconstruction for improved HAR classification. IEEE Transactions on Radar Systems, 2, 484-497. https://doi.org/10.1109/TRS.2024.3396172
Faisal, K. N., Mir, H. S., & Sharma, R. R. (2023). Human activity recognition from FMCW radar signals utilizing cross-terms free WVD. IEEE Internet of Things Journal, 11(8), 14383-14394. https://doi.org/10.1109/JIOT.2023.3344100
Gong, S., Shi, H., Yan, X., Fang, Y., Paul, A., Wu, Z., & Long, W. (2023). Human activity recognition with FMCW radar using few-shot learning. IEEE Sensors Journal, 23(15), 17815-17824. https://doi.org/10.1109/JSEN.2023.3290565
Hakobyan, G., & Yang, B. (2019). High-performance automotive radar: A review of signal processing algorithms and modulation schemes. IEEE Signal Processing Magazine, 36(5), 32-44. https://doi.org/10.1109/MSP.2019.2911722
Hasch, J., Topak, E., Schnabel, R., Zwick, T., Weigel, R., & Waldschmidt, C. (2012). Millimeter-wave technology for automotive radar sensors in the 77 GHz frequency band. IEEE Transactions on Microwave Theory and Techniques, 60(3), 845-860. https://doi.org/10.1109/TMTT.2011.2178427
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (27-30 June 2016, pp. 770-778), Las Vegas, NV, USA. https://doi.org/10.1109/CVPR.2016.90

Dang, V. N., Hoang, N. C., Le, M. T., Nguyen, K., & Nguyen, Q. C. (2026). Deep Learning-based Human Activity Recognition with FMCW Radar: A Review. IEEE Sensors Journal, 26(4), 5302-5326. https://doi.org/10.1109/JSEN.2026.3651578
Kim, W. Y., & Seo, D. H. (2022). Radar-based human activity recognition combining range–time–Doppler maps and range-distributed-convolutional neural networks. IEEE Transactions on Geoscience and Remote Sensing, 60, 1-11. https://doi.org/10.1109/TGRS.2022.3162833
Kumawat, H. C., & Raj, A. A. B. (2022). SP-WVD with adaptive-filter-bank-supported RF sensor for low RCS targets’ nonlinear micro-Doppler signature/pattern imaging system. Sensors, 22(3), 1186. https://doi.org/10.3390/s22031186
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet Classification with Deep Convolutional Neural Networks. In: F. Pereira, C. J. Burges, L. Bottou, & K. Weinberger (Eds.), Advances in Neural Information Processing Systems 25 (NIPS 2012).
Miao, F., Huang, Y., Lu, Z., Ohtsuki, T., Gui, G., & Sari, H. (2025). Wi-Fi sensing techniques for human activity recognition: Brief survey, potential challenges, and research directions. ACM Computing Surveys, 57(5), 107. https://doi.org/10.1145/3705893
Mishra, K. K., & Pachori, R. B. (2025). FMCW radar-based human activity recognition based on higher-order synchrosqueezing transform. IEEE Sensors Journal, 25(15), 28934-28941. https://doi.org/10.1109/JSEN.2025.3579423
Patole, S. M., Torlak, M., Wang, D., & Ali, M. (2017). Automotive radars: A review of signal processing techniques. IEEE Signal Processing Magazine, 34(2), 22-35. https://doi.org/10.1109/MSP.2016.2628914
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L. C. (2018). MobileNetV2: Inverted residuals and linear bottlenecks. In: Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, (18-23 June 2018, pp. 4510-4520), Salt Lake City, UT, USA. https://doi.org/10.1109/CVPR.2018.00474
Sarna, M. L. A., Hossain, M. R., & Islam, M. A. (2024). Comparative analysis of stft and wavelet transform in time-frequency analysis of non-stationary signals. International Journal of Novel Research in Engineering and Science, 11(1), 72-78. https://doi.org/10.5281/zenodo.11230337
Ullmann, I., Guendel, R. G., Kruse, N. C., Fioranelli, F., & Yarovoy, A. (2023). A survey on radar-based continuous human activity recognition. IEEE Journal of Microwaves, 3(3), 938-950. https://doi.org/10.1109/JMW.2023.3264494
Ünal, L., Pakfiliz, A.G. (2022). LPI Radar Signal Detection Based on Autocorrelation Function and Wigner-Ville Distribution. Review of Computer Engineering Studies, 9(4), 125-135. https://doi.org/10.18280/rces.090401
Wang, C.-H., Zong, Z.-Y., Yin, X.-Y., Li, K., & Zuo, Y.-H. (2025). A novel local maximum synchrosqueezing W transform for reservoir characterization. Petroleum Science, 23(4), 1842-1859. https://doi.org/10.1016/j.petsci.2025.11.034
Wu, Y., Fioranelli, F., & Gao, C. (2025). RadMamba: Efficient Human Activity Recognition Through a Radar-Based Micro-Doppler-Oriented Mamba State-Space Model. IEEE Transactions on Radar Systems, 4, 261-272. https://doi.org/10.1109/TRS.2025.3648848
Yin, W., Shi, L.-F., & Shi, Y. (2024). Indoor human action recognition based on millimeter-wave radar micro-doppler signature. Measurement, 235, 114939. https://doi.org/10.1016/j.measurement.2024.114939
Zhang, S., Li, Y., Zhang, S., Shahabi, F., Xia, S., Deng, Y., & Alshurafa, N. (2022). Deep learning in human activity recognition with wearable sensors: A review on advances. Sensors, 22(4), 1476. https://doi.org/10.3390/s22041476

Details

Primary Language

English

Subjects

Deep Learning, Electronic Sensors, Radio Frequency Engineering

Journal Section

Research Article

Authors

Harun Sevinç ^*
0009-0006-2408-2540
Türkiye

Levent Seyfi
0000-0002-8698-5140
Türkiye

Publication Date

June 30, 2026

Submission Date

March 2, 2026

Acceptance Date

April 29, 2026

Published in Issue

Year 2026 Volume: 13 Number: 2

DOI

https://doi.org/10.54287/gujsa.1901011

IZ

https://izlik.org/JA78AW73NF

Cite

RIS / Bibtex

APA

Sevinç, H., & Seyfi, L. (2026). Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures. Gazi University Journal of Science Part A: Engineering and Innovation, 13(2), 692-709. https://doi.org/10.54287/gujsa.1901011

AMA

1.Sevinç H, Seyfi L. Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures. GU J Sci, Part A. 2026;13(2):692-709. doi:10.54287/gujsa.1901011

Chicago

Sevinç, Harun, and Levent Seyfi. 2026. “Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures”. Gazi University Journal of Science Part A: Engineering and Innovation 13 (2): 692-709. https://doi.org/10.54287/gujsa.1901011.

EndNote

Sevinç H, Seyfi L (June 1, 2026) Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures. Gazi University Journal of Science Part A: Engineering and Innovation 13 2 692–709.

IEEE

[1]H. Sevinç and L. Seyfi, “Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures”, GU J Sci, Part A, vol. 13, no. 2, pp. 692–709, June 2026, doi: 10.54287/gujsa.1901011.

ISNAD

Sevinç, Harun - Seyfi, Levent. “Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures”. Gazi University Journal of Science Part A: Engineering and Innovation 13/2 (June 1, 2026): 692-709. https://doi.org/10.54287/gujsa.1901011.

JAMA

1.Sevinç H, Seyfi L. Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures. GU J Sci, Part A. 2026;13:692–709.

MLA

Sevinç, Harun, and Levent Seyfi. “Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures”. Gazi University Journal of Science Part A: Engineering and Innovation, vol. 13, no. 2, June 2026, pp. 692-09, doi:10.54287/gujsa.1901011.

Vancouver

1.Harun Sevinç, Levent Seyfi. Comparative Analysis of Time-Frequency Transformations and Deep Learning Architectures in the Classification of FMCW Radar Micro-Doppler Signatures. GU J Sci, Part A. 2026 Jun. 1;13(2):692-709. doi:10.54287/gujsa.1901011