EUVIMEDEUVIMED entdecken
Crossref

C-PIWDnet: A Conditional PINN for NLFM Waveform Design in Target Tracking

Qidi Pan, Jing Liang, Zinan Wang

IEEE Transactions on Radar Systems · 2026 · 4 · 1334-1348

DOI: 10.1109/trs.2026.3714490 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Radar Systems, 2026. Verfasst von Qidi Pan und 2 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben. Gefördert von: National Natural Science Foundation of China, 111 Project.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The adaptive waveform design is essential for cognitive radar to achieve high-precision target tracking in dynamic environments. However, conventional optimization-based methods typically require reoptimization when the scenario changes, while existing deep learning approaches often fail to enforce physical constraints required for radar implementation. To address these limitations, we propose a conditional physics-informed neural network (C-PIWDnet) for nonlinear frequency modulation (NLFM) waveform design. Specifically, this framework transforms the problem from a discrete parameter search into a continuous functional optimization. The theoretical foundation relies on a variational analysis of the Fisher-information criterion, which yields a parameterized ordinary differential equation (ODE) that links the target state to the waveform phase. C-PIWDnet embeds this equation directly into its loss function as a physical constraint, learning a continuous state-to-phase mapping conditioned on target states. Furthermore, to resolve phase-dependent coefficients, a nested training strategy is developed to guarantee stable convergence and hardware compliance. This approach enables direct waveform generation for multiple targets without scenario-specific retraining. Numerical simulations show that C-PIWDnet simultaneously improves tracking accuracy and reduces computational latency compared with conventional methods. Hardware experiments on an IF transceiver confirm the physical realizability of the proposed method.

Förderung

  • National Natural Science Foundation of China
  • 111 Project

Zitieren

APA

Pan, Q., Liang, J., & Wang, Z. (2026). C-PIWDnet: A Conditional PINN for NLFM Waveform Design in Target Tracking. IEEE Transactions on Radar Systems, 4, 1334-1348. https://doi.org/10.1109/trs.2026.3714490

Vancouver

Pan Q, Liang J, Wang Z. C-PIWDnet: A Conditional PINN for NLFM Waveform Design in Target Tracking. IEEE Transactions on Radar Systems. 2026;4:1334-1348. doi:10.1109/trs.2026.3714490

BibTeX
@article{Pan2026,
  author = {Pan, Qidi and Liang, Jing and Wang, Zinan},
  title = {C-PIWDnet: A Conditional PINN for NLFM Waveform Design in Target Tracking},
  journal = {IEEE Transactions on Radar Systems},
  year = {2026},
  volume = {4},
  pages = {1334--1348},
  doi = {10.1109/trs.2026.3714490}
}
Kostenlos starten →