A Low-Latency and High-Performance Microwave Photonic AOA and IFM System Based on Deep Learning and FPGA

• Published in: IEEE sensors journal Vol. 25; no. 6; pp. 9934 - 9945
• Main Authors: Zhang, Longlong, Li, Yin, Liao, Xuan, Hu, Xiang, Peng, Yuanxi, Zhou, Tong
• Format: Journal Article
• Language: English
• Published: New York IEEE 15.03.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Acceleration system; Accuracy; Algorithms; Angle of arrival; angle of arrival (AOA) estimation; Deep learning; deep learning (DL); Electronic warfare; Field programmable gate arrays; field-programmable gate arrays (FPGAs); Frequency measurement; Hardware; instantaneous frequency measurement (IFM); Machine learning; Microwave measurement; microwave photonic (MWP); Microwave photonics; Microwave theory and techniques; Optical sensors; Parameters; Phase measurement; Photonics; Real time; Signal measurement; Signal processing; Signal processing algorithms
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2025.3535152

The angle of arrival (AOA) estimation and instantaneous frequency measurement (IFM) are the important aspects of radar, communication, and electronic warfare, and there is an urgent need to find intelligent solutions that are faster and more accurate than the traditional methods. Currently, deep learning (DL) has increasingly strong learning and feature extraction capabilities, which brings new opportunities to the field of electromagnetic signal processing. However, the models in DL have a large number of parameters, high storage requirements, and long computational latency and not practical for deploying applications on resource-limited devices. In this article, we build a complete AOA and IFM system for the first time, taking advantage of microwave photonics (MWPs) technology, DL, and features of field-programmable gate array (FPGA). Then, a novel hardware-friendly pruning-quantization-iteration compression (PQIC) algorithm is proposed for the post-processing of MWP signal measurement applications. By reducing the data width from 32 bits to four bits, the compressed algorithm reduces parameter storage requirements and hardware implementation complexity with negligible performance loss. Finally, we deploy the acceleration system and test the actual collected signal data on FPGA while running 200 MHz. The results show that with a <inline-formula> <tex-math notation="LaTeX">14\times </tex-math></inline-formula> model compression and a <inline-formula> <tex-math notation="LaTeX">3.99\times </tex-math></inline-formula> reduction in operations, the accuracy of AOA reaches 98.82%, with the mean absolute error (MAE) of 0.27°. More importantly, the running latency is only <inline-formula> <tex-math notation="LaTeX">20.78~\mu </tex-math></inline-formula>s, meeting the real-time processing requirements of MWP signal processing. It is also applicable to real-world signal intelligence processing and demonstrates superior performance compared to other existing algorithms in this field.