22 November 2023 in IEEE Transactions on Aerospace and Electronic Systems
Recently, interrupted sampling repeater jamming (ISRJ) based on digital radio frequency memory (DRFM), which could easily form high-power false targets, has severely impacted radar detection capabilities. Conventional anti-ISRJ techniques ignore the operating characteristics of the jammer. Once the anti-ISRJ algorithm and the features of the jammer are mismatched, undesirable jamming residues or signal loss appear. In order to avoid these unwanted performance losses, an algorithm jointing the jam perception and the adaptive waveform is developed in this paper. First, a bi-directional Constant False Alarm Rate (CFAR) detector is introduced to locate and estimate the jamming component in radar echo. An ISRJ feature vector is then built to reconstruct and cancel jamming components for accurate detection under ISRJ. Second, an adaptive anti-ISRJ waveform is developed based on the vector, leading the DRFM to retransmit signals orthogonal to the real detection signal. Besides, the particle swarm optimization (PSO) algorithm efficiently works out the optimal waveform parameters to improve the suppression performance. As radar switches to the adaptive anti-ISRJ waveform, long-term suppression of ISRJ is achieved without locating, estimating, and suppressing the ISRJ components in each echo. Simulated experiments demonstrate that the suggested algorithm can accurately detect radar targets under ISRJ.
5 August 2021 in IEEE Geoscience and Remote Sensing Letters
The main difficulty of synthetic aperture radar (SAR)-optical image matching or registration lies in the significant heterogeneous characteristics introduced by the different imaging mechanisms between SAR and optical images. Instead of directly using the raw image pair, transforming the pair into a feature domain, where they have homogeneous feature representation, is believed more effective. Inspired by image segmentation, we develop an end-to-end deep learning model for the SAR-optical matching, based on a siamese U-net with a fast Fourier transform (FFT) correlation layer. First, the siamese U-net with sharing weights extracts the feature maps of the SAR and optical images and projects the heterogeneous images into a homogeneous space. Then, the two feature maps are cross-correlated or normalized cross-correlated by the FFT layer and a similarity heatmap is obtained. Finally, the heatmap is send into a softmax2d classifier to determine the best matching, and thus matching is converted into classification. The nonlinear mapping capability of deep learning can well tackle the intensity variation across the different imaging modals; the encoder–decoder architecture with skip connections in the U-net can take full advantage of the global information and simultaneously preserve the local resolution and position information and thus guarantees high accuracy and robustness; besides, the FFT correlation is helpful for the efficiency improvement and training with large image pairs. Experiments show that the proposed method can achieve a pixel-level matching error.
2 September 2022 in IEEE Transactions on Aerospace and Electronic Systems
Frequency-agile (FA) transmission strategy plays a crucial role in radar anti-jamming applications. This strategy is usually designed or trained offline, which would lose the advantage of adaptability and flexibility when facing diverse jamming patterns and non-stationary target echoes. Considering the scenarios where a radar detects a target with strong non-stationary scattering characteristics under fast-variant interference, the radar is required to immediately adjust the FA transmission strategy to react to the variation of both jamming signals and target echoes quickly. In order to enhance radar transmission strategy in both generalization and flexibility aspects, an online FA strategy, called Combinatorial Discounted Thompson Sampling (CDTS), is developed for the anti-jamming by exploiting the target's scattering change with the Multi-Armed Bandit (MAB) model. With the advantages of both optimal exploration and fast convergence, the proposed algorithm can efficiently adapt to the scattering fluctuation under dynamic frequency jamming emissions. Experimental comparisons with conventional Deep Reinforcement Learning (DRL) demonstrate the proposed algorithm's superiority for FA transmission strategy learning to boost radar detection performance under frequency response dynamics when avoiding frequency jamming.
5 June 2024 in IEEE Sensors Journal
Recognition of multifunction radar (MFR) working modes provides essential information to assess the behavior of a non-cooperative MFR. However, it typically depends on substantial previously intercepted waveforms. To enable better recognition with limited pre-observations, this paper proposes a radar-semantics-modeled LSTM recognition algorithm with embedding radar waveform functionality features and utilizing contextual recognition of an LSTM network. On the one hand, unsupervised hierarchical modeling and functional semantics embedding extract MFR behavior units and embed the artificial functional semantics of waveforms based on experts’ knowledge about waveform functionalities. On the other hand, an LSTM network captures static and dynamic features within sequences of radar semantic data, recognizing the MFR working mode. Experimental comparisons on preset datasets illustrate that the proposed algorithm adapts more effectively to parameter variations and migrates the recognition between the MFRs with similar functional behavior features.
7 August 2023 in IEEE Sensors Journal
For the preservation of low-airspace security, a comprehensive threat assessment of low-altitude slow small (LSS) flying targets is essential. Threat assessment intrinsically depends on effectively extracting the optimal attributes from tracks, which represent the target’s motion intention and specific characteristics. However, unavoidable measured noise and track jitters during observation impact assessment performance. To address this problem, long short-term memory (LSTM) is introduced due to its good performance on sequential perception. LSTM captures the permanent intent traits and reduces noise interference utilizing temporal correlations within its novel “memory’’ units. Compared to traditional algorithms, the proposed technique substantially reduces measurement noise, enhancing the accuracy of threat evaluations. Simulation experiments confirm the effectiveness and robustness of the proposed algorithms.