6 August 2021 in IEEE Sensors Journal
Target detection in the presence of main-lobe deception jamming has received intensive attention. Generally, traditional target detection with additive jamming can be regarded as an adjacent process of jamming suppression and target detection. Since target information can be lost in jamming suppression, the detection performance can degrade severely. Aiming at this issue, an end-to-end anti-jamming target detector is proposed based on convolutional neural network (CNN) in the presence of interrupted-sampling repeater jamming. The CNN method first converts the single-pulse compression filter into a series of feature extraction filters, accumulating the energy of various features in various dimensions and on multiple scales. Then, a feature fusion filter added after the feature extraction stage is utilized to suppress jamming features while simultaneously retaining the actual target features. The two stages can be achieved in the CNN structure, and all the filters in this method are obtained by training the CNN. Numerical simulations show the efficiency of the proposed method.
25 April 2024 in IEEE Transactions on Aerospace and Electronic Systems
Long time coherent integration for high-speed maneuvering targets in radar signal processing poses substantial difficulties due to the occurrence of Range Migration (RM) and Doppler Frequency Migration (DFM). The Generalized Radon-Fourier Transform (GRFT), an effective method for accumulating target energy, is hampered by its high computational cost. To address this, the Full-Dimensional Partial-Search Generalized Radon-Fourier Transform (FSGRFT), a fast implementation of the GRFT algorithm, is proposed in this paper. Differing from the exhaustive full-dimensional search of the GRFT algorithm, the FSGRFT algorithm first employs a pre-trained residual network to coarse estimate range cells where targets might exist, along with their corresponding motion parameter sub-spaces. Guided by this initial coarse estimate, the FSGRFT algorithm narrows its search to a subset of range cells and the designated sub-space of motion parameters, thereby considerably decreasing the number of searches. Meanwhile, the introduction of a lightweight network structure with robust representation capabilities ensures the accuracy of the coarse estimation, without bringing excessive computational cost. Therefore, the FSGRFT algorithm obtains a good trade-off between the computational cost and integration performance, in comparison with other long-time coherent integration algorithms. Finally, simulation results demonstrate that the FSGRFT algorithm can achieve detection performance comparable to the GRFT algorithm, but with considerably lower computational cost.
28 June 2024 in IEEE Transactions on Aerospace and Electronic Systems
Interrupted-sampling repeater jamming with convolution kernel modulation based on digital radio frequency memory technology usually enters from the main lobe of the radar antenna, exhibiting high levels of coherence and deception in both the time and frequency domains. Therefore, it is difficult for traditional radars to identify and suppress jamming while retaining the target signal, potentially leading to problems with target detection. In recent years, since the introduction of deep learning and the implementation of end-to-end target detection for anti-jamming, the field of radar signal processing has been revolutionized. However, existing end-to-end detectors encounters the limitations in information utilization and handling hard samples, leading to performance degradation in target detection. Therefore, an attention mechanism-based anti-jamming detector with generative adversarial training is proposed, which works directly with radar echo information in both the time and timefrequency domains. Firstly, the conception of generative adversarial is leveraged to model the framework to mitigate the impact of hard samples on network performance by confrontation interaction. The sample equalization factor is then introduced into the cross-entropy loss function to alleviate the influence of target and jamming overlap rate on detection performance. Finally, by drawing on the attention mechanism and the idea of multi-scale feature extraction, the network structure is modified to introduce time-frequency domain information and enhance the feature extraction ability of the detector, which is insensitive to variations in the jamming-to-signal ratio. Experimental results demonstrate that the proposed detector outperforms the state-of-the-art methods in terms of detection performance and robustness.
1 January 2026 in IEEE Transactions on Aerospace and Electronic Systems
1 January 2026 in IEEE Sensors Journal