Adaptive Linearization for the Sub-Nyquist Photonic Receiver Based on Deep Learning
Liyuan Zhao,
Jianghua Zhang,
Lei Huang,
Yuanxi Peng,
Ke Yin,
Xin Zheng,
Zhuohang Zhang,
Meili Shen,
Denghui Song,
Hongxiao Niu
Affiliations
Liyuan Zhao
State Key Laboratory of High Performance Computing, College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, China
Jianghua Zhang
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Lei Huang
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Yuanxi Peng
State Key Laboratory of High Performance Computing, College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, China
Ke Yin
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Xin Zheng
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Zhuohang Zhang
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Meili Shen
National Innovation Institute of Defense Technology, Academy of Military Sciences PLA China, Beijing 100071, China
Denghui Song
College of Advanced Interdisciplinary Studies, National University of Defense Technology, Changsha 410073, China
Hongxiao Niu
School of Optoelectronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China
Due to the nonlinear and aliasing effects, the sub-Nyquist photonic receiver for radio frequency (RF) signals with large instantaneous bandwidth suffers limited dynamic range and noise performance. We designated a deep residual network (Resnet) to realize adaptive linearization across 40 GHz bandwidth. In contrast to conventional linearization methods, the deep learning method achieves the suppression of multifactorial spurious distortions and the noise floor simultaneously. It does not require an accurate calculation of the nonlinear transfer function or prior signal information. The experiments demonstrated that the proposed Resnet could improve the spur-free dynamic range (SFDR) and the signal-to-noise ratio (SNR) significantly by testing with single-tone signals, dual-tone signals, wireless communication signals, and modulated radar signals.
