High-resolution range profile (HRRP) sequences have great potential for space target classification because they can provide both scattering information and micro-motion information. However, many factors cause an obtained HRRP sequence for a space target to be corrupted in real cases due to noise interference, limited radar resources, and the requirement of multi-target observations. Many space target classification methods cease to be effective when HRRP sequences are corrupted, so classifying space targets with corrupted HRRP sequences is still a challenging problem. To solve this problem, a novel space target classification method based on a temporal-spatial feature aggregation network (TSFA-Net) is proposed by using the corrupted HRRP sequences directly. First, a sequence-to-token module (S2T-module) is designed to extract low-level and fine-grained features from the raw inputs. Second, to effectively model the long-range dependencies among corrupted HRRP sequences and capture global representations without losing target local features, we propose a parallel and dual-branch block, i.e., a temporal-spatial feature aggregation block (TSFA-block), by combining a Transformer network and a convolutional neural network (CNN). Then, via progressively hierarchically stacking TSFA blocks, a hierarchical temporal-spatial feature aggregation subnetwork (H-TSFA-subnetwork) is constructed to obtain the final temporal-spatial features. Finally, a token-to-label module (T2L-module) is adopted to obtain the classification results. Extensive experiments demonstrate that the proposed method achieves state-of-the-art classification accuracy for space target classification with HRRP sequences, especially under the conditions of a low signal-to-noise ratio and a high missing rate.

The challenge of effectively suppressing interference in radar systems, particularly the complex and unknown mainlobe interference, is a significant concern in radar signal processing. Traditional anti-jamming methods in single radar often fail to address this issue. The paper proposes a novel approach for suppressing mainlobe interference in radar networks, capitalizing on the low-rank representation of interferences and the sparse representation of echoes. Interference signals can be extracted by minimizing their rank with a regularization constraint after performing range and Doppler equalization on the received signals. Target echoes can be recovered through joint sparse reconstruction, exploiting their unique motion states across multiple observation points. To solve the underlying optimization problem, which involves the simultaneous reconstruction of low-rank and sparse matrices, we propose two algorithms based on the augmented Lagrangian method (ALM), with one algorithm focusing on precision and another emphasizing efficiency. This method leverages the robust spatial correlation of the interference signal and the sparsity of the target spatial distribution, allowing for effective interference suppression and accurate target echo recovery without prior knowledge of the interference type. Numerical experiments validate the effectiveness of this proposed approach and its superiority compared with other methods.

This study investigates the target detection and discrimination method in the presence of deceptive jamming using a full-polarization radar network. Within the target detection framework by multiple radars with full polarization capabilities, we formulate the problem of detecting false targets as a multivariate hypothesis testing task. Based on the generalized likelihood ratio test, we propose a two-stage detection scheme comprising a pre-detection stage at each radar system and a discrimination stage at the information fusion center. The pre-detection step is crucial when interference signals cannot deceive all radars. The discrimination stage exploits spatial distinctions in polarization scattering properties between physical and false targets. By leveraging the capabilities of full polarization radar for acquiring and processing target polarization information, along with the multi-view observation potential of radar networks, our proposed approach can obtain more comprehensive scattering information about physical and false targets as well as their differences, thereby achieving enhanced accuracy in target discrimination. We validate the effectiveness of our algorithm through Monte Carlo experiments.

Multi-rotor aircraft have the advantages of a simple structure, low cost, and flexible operation in the unmanned aerial vehicle (UAV) family, and have developed rapidly in recent years. Radar surveillance and classification of the growing number of multi-rotor aircraft has become a challenging problem due to their low-slow-small (LSS) characteristics. Estimation of the blade number is an important step in distinguishing LSS targets. However, most of the current research on micro-motion parameters estimation has focused on the analysis of rotational frequency, length, and the initial phase of blades with a prior of blade number, affecting its ability to identify LSS targets. In this article, a micro-motion parameters estimation method for multi-rotor targets without a prior is proposed. On the basis of estimating the flashing frequency of the blades, a validation function is constructed through spectral analysis to judge the number of blades, and then the rotational frequency is estimated. The blade length is calculated by estimating the maximum Doppler shift. Moreover, the variational mode decomposition (VMD)-based atomic scaling orthogonal matching pursuit (AS-OMP) method is jointly applied to estimate the blade length when suffering from the low PRF and insufficient SNR conditions. Extensive experiments on the simulated and measured data demonstrate that the proposed method outperforms robust micro-motion parameter estimation capability in low PRF and insufficient SNR conditions compared to the traditional time-frequency analysis methods.

When echoes of micromotion targets are overlapping in the time-frequency (TF) domain and sampling data are missing, the decomposition of the multicomponent micro-Doppler (m-D) signals is challenging. To address this issue, this letter proposes a method for multicomponent m-D signal decomposition by iterations of the instantaneous frequencies (IFs), individual components, and complex envelopes. To initialize the IFs, the well-focused time-frequency representation (TFR) is obtained by sparse reconstruction of the incomplete data, and then the IFs of the TFR can be estimated by the short-time variational mode decomposition (STVMD) algorithm. After initialization, the IFs, individual components, and complex envelopes are updated by the intrinsic chirp component decomposition (ICCD), alternating direction method (ADMM) of multipliers, and least-square-error criterion (LSEC), respectively. Finally, the proposed method is verified by simulation and application to real data.

In micromotion feature extraction, the incomplete and phase-corrupted radar echo may cause bad time-frequency representation (TFR) and prevent micromotion feature extraction. To solve the problem, we establish a sparse regularization model to reconstruct well-focused TF distributions. The regularization model is solved by the iterative soft-thresholding algorithm (ISTA). In each iteration, the hard thresholding function and least-square-error criterion are developed to estimate the phase errors. For micro-Doppler signal real-time processing to save radar time resources, the received signal can be directly sparse recovered in real time rather than waiting for the complete signal. Finally, the effectiveness of the proposed method is validated by the simulation results.