Zengping Chen

Zengping Chen

Auto-generated profile, based on publicly available information

Zengping Chen

School of Electronics and Communications Engineering, Sun Yat-sen University, Shenzhen, China

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 December 2019 in IEEE Access
Verified
Journal Article
Open access
HSHanning Su
Hanning Su
QBQinglong Bao
Qinglong Bao
ZCZengping Chen
Zengping Chen
This paper addresses the problem of estimating the parameters of constant-amplitude chirp signals that have single or multiple components and are embedded in noise. Chirp signals are widely employed in applications such as radar and telecommunications, and it is a key task in countermeasure techniques to estimate their parameters without prior information. Hence, a parameter estimation processor based on a complex-valued deep neural network (CV DNN) is proposed to perform this task efficiently. The CV DNN, which is designed for regression, consists of a function fitter and a predictor. The function fitter acts like a eigenfunction mapping: it maps the one-dimensional input into a two-dimensional feature map suitable for subsequent network learning. As a special feature extraction tool, the predictor extracts local features from the feature map and estimates parameters. Simulation results indicate that the CV DNN outperforms conventional processors. Moreover, it is more accurate than the Wigner–Hough transform while being several orders of magnitude faster, which will enable real-time signal processing with fewer computational resources. Furthermore, we demonstrate that the CV DNN shows strong robustness to changes in modulation parameters and the number of components of a chirp signal. This study shows the advantages of deep learning systems for signal parameter estimation.

9 June 2025 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
YHYongbing Huang
Yongbing Huang
RGRui Guo
Rui Guo
FYFengning Yu
Fengning Yu
YZYue Zhang
Yue Zhang
SXShiyou Xu
Shiyou Xu
ZCZengping Chen
Zengping Chen
To enhance the adaptive capabilities of radar, this work focuses on the problem of joint waveform adaptation and data-assisted jamming suppression in multiple target tracking scenarios with high-power sidelobe jamming. Utilizing deep reinforcement learning (DRL) methods, we develop agents capable of decision-making based on environmental perception information, thus enabling beamforming for jamming suppression and enhancing tracking performance. Jamming suppression faces challenges due to the computational complexity of beamforming and the large sample size of jamming. Furthermore, the multidimensionality of waveform parameters and the fluctuating states of targets pose additional challenges. To address these challenges, we employ DRL for jamming number perception, leverage convolutional neural networks to extract features, and incorporate the self-attention mechanism for jamming number decision-making. For waveform adaptation, we segment each parameter into a distinct decision space, with distinct agents making decisions using value-decomposition networks. These agents are augmented with gated recurrent units to manage diverse target states and maintain historical data. The focus on tracking robustness and accuracy can be tailored by designing the reward function weights. A separate training approach is developed to obtain effective joint policies. To avoid signal-to-noise ratio (SNR) constraints and enhance applicability to real radar systems, we develop an intelligent radar tracking simulation at the signal level and evaluate the performance of the proposed method. Simulation results demonstrate that the receiver agent can balance the perception of jamming numbers and real-time capabilities to assist in jamming suppression. Furthermore, compared to criterion-based methods and an entropy reward-based method, the transmitter agent can enhance tracking robustness and accuracy by adapting waveform parameters. Moreover, field experiments are conducted to verify that the proposed DRL method can improve the actual multi-target tracking performance through waveform adaptation.

1 March 2021
Verified
Conference Paper
Closed access
JLJingtang Li
Jingtang Li
Hongtu XieHongtu Xie
Hongtu Xie
GWGuoqian Wang
Guoqian Wang
ZCZengping Chen
Zengping Chen
Synthetic aperture radar (SAR) imagery is widely applied to recognize target ships in marine surveillance since it can operate in all-weather day-and-night conditions. Deep learning and neural networks have also been used more frequently in recent years with SAR data and have achieved significant progress. However, it takes a lot of energy for deep neural networks to learn and hinders the further development of SAR technology. Compared with the traditional deep neural networks, Spiking Neural Networks (SNNs) have temporal information processing ability, low power consumption, and high biological plausibility, which are suitable for future radar intelligent sensing. In this paper, we take inspiration from SNNs and propose an energy-efficient method, namely SpikingSAR. The experimental results show that the proposed method outperforms the state-of-the-art deep learning methods as well as the conventional methods, while it also has fewer model parameters.