Complex-Valued Neural Networks for Millimeter Wave FMCW-Radar Angle Estimations

Kaiser K, Daugalas J, Lopez-Randulfe J, Knoll A, Weigel R, Lurz F (2022)

Publication Status: Accepted

Publication Type: Conference contribution

Future Publication Type: Conference contribution

Publication year: 2022

Publisher: Institute of Electrical and Electronics Engineers Inc.

Pages Range: 145-148

Conference Proceedings Title: 2022 19th European Radar Conference, EuRAD 2022

Event location: Milan IT

ISBN: 9782874870712

DOI: 10.23919/EuRAD54643.2022.9924767

Abstract

Processing radar signals with neural networks has shown promising results in classification and regression tasks. While processed radar data is intrinsically complex-valued, most architectures using neural networks are comprised of real-values and their arithmetic. Previous work has found that keeping the complex-valued number system and extending it into the domain of neural networks can be beneficial. In this paper, we demonstrate that in two-dimensional direction-of-arrival (DoA) estimation, complex-valued neural networks (CVNNs) show better results than real-valued neural networks (RVNNs). Real-world recordings of ten different FMCW radar devices were used to train numerous models, varying in the computational complexity and varying in data properties. Over all models trained, the best CVNN surpassed the best RVNN by 14%. In terms of model complexity, CVNNs also showed better results, both per trainable parameter and per floating point operation (FLOP). Similarly, CVNNs surpass RVNNs, both when trained with decreased data quantity and decreased data quality.

Authors with CRIS profile

Kevin Kaiser Lehrstuhl für Technische Elektronik (LTE) Robert Weigel Lehrstuhl für Technische Elektronik (LTE) Fabian Lurz Lehrstuhl für Technische Elektronik (LTE)

Involved external institutions

Technische Universität München (TUM) DE Germany (DE)

How to cite

APA:

Kaiser, K., Daugalas, J., Lopez-Randulfe, J., Knoll, A., Weigel, R., & Lurz, F. (2022). Complex-Valued Neural Networks for Millimeter Wave FMCW-Radar Angle Estimations. In 2022 19th European Radar Conference, EuRAD 2022 (pp. 145-148). Milan, IT: Institute of Electrical and Electronics Engineers Inc..

MLA:

Kaiser, Kevin, et al. "Complex-Valued Neural Networks for Millimeter Wave FMCW-Radar Angle Estimations." Proceedings of the 19th European Radar Conference, EuRAD 2022, Milan Institute of Electrical and Electronics Engineers Inc., 2022. 145-148.

BibTeX: Download