Coherent multiple-input multiple-output (MIMO) radar could significantly improve the weak moving target detection ability by accumulating multi-channel and multi-frame echo signal. However, due to the target motion and the geometric configuration differences of each node relative to the target, the multi-channel multi-frame accumulation of target echoes by coherent MIMO radar faces several difficulties, i.e., range migration (RM) and Doppler frequency migration (DFM) issues within each channel, spatial variations of target signal envelope and phase across different channels, time variations of target signal envelope and phase across different frames. To tackle these problems, two contributions are made in this paper. Firstly, by applying the generalized Radon Fourier transform (GRFT) to correct the RM/DFM and obtain the signal integration within each channel, we propose a multi-channel accumulation algorithm to compensate the spatial variations of target signal and realize the target energy accumulation across different channels, which takes full advantage of the GRFT output characteristics and the radar system topological structure. Secondly, a multi-frame accumulation algorithm is developed to compensate the time variations of target signal and achieve the further integration of target energy distributed in different frames. We also provide explicit expressions and analysis for performance metrics, including the output response, SNR improvement performance. The proposed method's effectiveness is confirmed through simulation experiments.

In order to track multiple maneuvering extended targets accurately, an adaptive interacting multiple model algorithm based on variational inference (AIMM-VI) is proposed. An augmented state is constructed to cater for time-varying orientation angle and track realistic shape changes, resulting in better elliptical shape estimation and tracking accuracy. Multiple measurements from multiple extended targets are effectively assigned to corresponding targets through the marginal association probability distribution criterion, and the variational inference is used to accurately estimate the augmented state and shape information, which greatly improves the parameters estimation performance. The residual and likelihood functions are updated in real-time according to the results of variational inference, allowing for the updating of the model probability in real-time. The Markov probability transfer matrix is subsequently adaptively updated by the compression ratio, which makes the algorithm more adaptable to maneuvering target and significantly improves the adaptability and robustness of the algorithm. The final simulation and experiment results show that the proposed algorithm can effectively improve the tracking performance of multiple maneuvering extended targets.

Deep-learning-based radar jamming recognition focuses on identifying different types of jamming by using convolutional neural networks. Though achieving superior performance in recent years, existing deep-learning-based few-shot methods are still unable to break through the barrier of needing at least 5% labeled data for supervised learning and fail to utilize the large amounts of unlabeled data. Considering the difficulty and formidable cost of acquiring labeled data and the ease of acquiring large amounts of unlabeled data in real-world scenarios, it is meaningful to explore a semi-supervised jamming recognition method by utilizing a smaller amount of labeled data and extensive unlabeled data. To this end, a few-shot semi-supervised radar jamming recognition network based on a self-training framework is proposed in a challenging setting (1% labeled data, with only five labeled samples for each jamming class). To effectively extract recognition-related knowledge from the labeled data, a mutual learning strategy is first proposed by constraining the consistency between predictions on original jamming data and their augmented data. Using the trained mutual learning strategy, pseudo-labels of unlabeled data can be obtained by taking the unlabeled jamming data as input. Then, to select more reliable pseudo-labeled data, a pseudo-labeled sample selection mechanism is proposed by introducing confidence scores to filter the high-quality pseudo-labels. With the two aforementioned components, our framework is able to effectively exploit information contained in both labeled and unlabeled data through self-training in a semi-supervised manner. Extensive experiments on a dataset comprising a mixture of simulated and measured data demonstrate that the proposed semi-supervised jamming recognition method outperforms state-of-the-art techniques in few-shot jamming recognition and semi-supervised recognition.

Aiming at resolving the problem of low tracking accuracy for maneuvering extended targets in lidar systems, an interactive multimodel variational Bayes independent axis estimation (IMM-VB-IAE) algorithm is proposed in this article. First, the algorithm utilizes IMM to adaptively select the appropriate model to track the target according to the changes in the target's motion state, thereby improving the overall tracking performance. Second, the algorithm uses VB theory to approximate the posterior closed expression, which simplifies the solution process by transforming the original complex inference problem into a parameter optimization problem. Finally, the principle of eigendecomposition is utilized to quadratically estimate the ellipse axes' lengths, which improves the estimation accuracy of the ellipse parameters. The final simulation and experimental results demonstrate that the proposed algorithm outperforms several other algorithms in the accuracy of tracking maneuvering extended targets, with average OSPA and Gaussian-Wasserstein distances reduced by at least 55.5% and 56.1%, respectively.

This article investigates the performance of coherent multiple-input multiple-out (MIMO) radar in the presence of independent but not necessarily identically distributed random amplitude errors. We first establish the coherent MIMO radar echo model that incorporates random amplitude errors. Then, to accurately assess the multi-channel signal integration performance with amplitude errors, we develop closed-form solutions for the average power, processing gain and detection probability. In addition, we provide a simplified expression of the average power in the case of independent and identically distributed (i.i.d.) errors. Further, to bridge the gap between theory and practical applications, two representative i.i.d. error distributions (uniform distribution and truncated Gaussian distribution) are examined. The average power, processing gain and detection probability are calculated in closed-form. Finally, numerical simulations confirm the validity and accuracy of closed-form analytical results.

With flexibility in maneuverability and remarkable adaptability, airborne bistatic radar system can obtain excellent detection performance for high-speed target by employing coherent integration. However, range migration (RM) and Doppler frequency migration (DFM) could become serious issues due to the relative motion characteristics of airborne platforms and high-speed target. Meanwhile, various unpredictable factors such as atmospheric turbulence and mechanical issues, etc., resulting in additional motion errors, would have further negative impacts on motion state and flight trajectory of airborne platforms. This phenomenon would cause serious consequence on coherent integration and target detection. Thus, we make contributions to tackle these limitations and enhance coherent integration and detection performance. First, we establish signal model with high-speed target in 3-D space for airborne bistatic radar system, along with motion error model which simultaneously includes translational error and rotational error. Next, we articulate range history's mathematical expression and further derive echo signal model. We then propose an improved generalized radon Fourier transform (IGRFT) method. More specifically, the purpose of IGRFT is achieving joint search for the parameters of the target motion and the parameters of motion error, to ensure high precision parameter estimation and high gain integration. However, the computational complexity surges due to the increasing of search dimensionality. To devise computationally feasible methods for practical applications, we split the high-dimensional maximization process into two disjoint problems by sequentially searching motion parameters and then motion error parameters, and this method is named generalized Radon transform (GRT)-IGRFT. Numerical simulations show that the proposed algorithms can correctly estimate parameters and achieve signal integration and target detection. Finally, we present performance analysis of two proposed strategies considering computational complexity, detection performance, and parameter estimation.

In this paper, we propose two methods for tracking multiple extended targets or unresolved group targets with elliptical extent shape. These two methods are deduced from the famous Probability Hypothesis Density (PHD) filter and the Cardinality-PHD (CPHD) filter, respectively. In these two methods, Trajectory Set Theory (TST) is combined to establish the target trajectory estimates. Moreover, by employing a decoupled shape estimation model, the proposed methods can explicitly provide the shape estimation of the target, such as the orientation of the ellipse extension and the length of its two axes. We derived the closed Bayesian recursive of these two methods with stable trajectory generation and accurate extent estimation, resulting in the TPHD-E filter and the TCPHD-E filter. In addition, Gaussian mixture implementations of our methods are provided, which are further referred to as the GM-TPHD-E filter and the GM-TCPHD-E filters. We illustrate the ability of these methods through simulations and experiments with real data. These experiments demonstrate that the two proposed algorithms have advantages over existing algorithms in target shape estimation, as well as in the completeness and accuracy of target trajectory generation.

Most synthetic aperture radar (SAR) automatic target recognition (ATR) methods can achieve good recognition results only under the closed-set assumption. However, in practical applications, ATR models are often exposed to open environments, the general closed-set method may misclassify unknown categories as known categories, which is not reasonable. To tackle this issue, this article proposes an end-to-end hybrid model for SAR image open-set recognition (OSR), named GANFlow, which combines a generative adversarial network (GAN) with a flow-based module. The GANFlow achieves accurate classification of known categories and effective rejection of unknown categories. In this model, a classifiable convolution GAN is first designed to complete the training of the feature extraction module and classifier. Through adversarial training, the generated images enrich the training samples, which improves the ability of feature extraction and classification of the discriminator. Then, to find the difference in the probability density distribution of the extracted features, a flow-based module is adopted. Also, the features avoid interference from irrelevant background information in SAR images. Furthermore, by establishing an appropriate threshold, unknown categories can be efficiently rejected. Finally, the outputs of the classifier and the flow-based module are combined to complete the OSR of the SAR image target. The experimental results on the MSTAR and OpenSARShip public-measured datasets verify the robustness and generalization of the proposed method.

Uncrewed aerial vehicle (UAV) swarms have the characteristics of small size, high density, and agile maneuverability. These attributes have given rise to substantial difficulties for radar in achieving precise detection and resolution of UAV swarms. Meanwhile, their potential malicious use poses a significant threat to national security, making accurate identification of UAV swarms of utmost importance. This article presents a super-resolution method for UAV swarms that integrate the coherent long-time integration technique with the gridless sparse recovery method based on iterative weighted atomic norm minimization (IW-ANM). In the method, a framework for UAV swarm detection and super-resolution processing is first established. Then, based on the framework, the IW-ANM algorithm is proposed. This algorithm encodes prior information into the Toeplitz constraint matrix, adopts a well-designed weight function, and finally super-resolves UAV swarms in the spatial dimension through iterative weighting. Numerical simulations demonstrate that, compared with the reweighted atomic norm minimization, ANM, multiple signal classification, and so on, the proposed IW-ANM algorithm is more practical and robust, and has a better super-resolution performance in conditions of low signal-to-noise ratio, high-density swarms, and small angle intervals. Furthermore, a real experiment is conducted to validate the effectiveness of the proposed IW-ANM.

In this work, we propose a method for tracking multiple extended targets or unresolvable group targets in a clutter environment. First, based on the random matrix model (RMM), each target's joint kinematic-extent state is modelled as a gamma Gaussian inverse Wishart (GGIW) distribution. Considering the uncertainty of measurement origin caused by the clutters, we adopt the idea of probabilistic data association and describe the joint association event as an unknown parameter in the joint prior distribution. Then, variational Bayesian inference (VBI) is used to approximate the intractable posterior distribution. To improve practicality, we propose two lightweight schemes to reduce computational complexity. The first is clustering-based and effectively prunes joint association events. The second simplifies the variational posterior by using marginal association probabilities. Finally, we demonstrate effectiveness on simulations and real-data experiments and show that the method outperforms state-of-the-art baselines in accuracy and adaptability.

In the tracking of unresolvable group object (URGO), the common extended Kalman filter based on the multiplicative error model (MEM-EKF*) provides an effective solution to accurately estimate the extended shape. But in multiple URGOs tracking applications, the existing methods based on MEM-EKF* and joint probabilistic data association (JPDA) are not adequate for tracking the complex behavior of URGOs such as overlapping. An ideal is presented in this article that uses fuzzy clustering technology (FCT) to complete probabilistic data association between multiple URGOs and simultaneously estimate the shape of the URGOs through MEM-EKF*. A multiple URGOs tracking method called GMMEM-MEM-EKF* is proposed; it utilizes the Gaussian mixture model-expectation maximization (GMM-EM) clustering to achieve data association and estimates the states of multiple URGOs within the MEM-EKF*, which avoided the imprecision of traditional FCT such as fuzzy C-means (FCM) clustering in data association. In addition, this article gives the solution of multiple URGOs tracking under dense clutter. The better performance and accuracy of our method in dealing with complex behaviors of URGOs is demonstrated by Monte Carlo simulations in comparison to state-of-the-art methods in literature.

In this paper, the possibility to improve target detection performance in passive bistatic radar by jointly exploiting multiple signals at different carrier frequencies emitted by the same illuminator is investigated, namely multi-frequency passive bistatic radar (MFPBR) coherent integration. Since the carrier frequency of each signal is agile, the MFPBR coherent integration suffers from the problems of range phase incoherence and Doppler broadening. In order to tackle these challenges, a multi-frequency coherent integration target detection algorithm for passive bistatic radar is proposed. Specifically, this scheme corrects the Doppler broadening effect via TSP. Then, low sidelobe filtering based on convex optimization is carried out to remove the range phase incoherence and obtain the MF of the target's energy. Meanwhile, the high-range-resolution profiles (HRRP) of target can be generated. The advantage of the proposed algorithm is that it can obtain superior coherent integration and detection performance for both the single and multiple targets scenarios compared to existing methods. Finally, a series of measured and simulation results are presented to demonstrate the effectiveness of the proposed algorithm.

Effective recognition of radar jamming is of great importance in improving radar system's anti-jamming capability. Existing radar jamming recognition methods based on convolutional neural networks suffer from limitations such as low recognition accuracy and poor network stability in complex electromagnetic environments. This is due to incomplete extraction and utilization of useful information from the jamming signal. In this paper, a few-shot adaptive confidence aggregation and cross-modal refinement jamming recognition method (JR-ACAR) is designed to mine the complete information of the jamming signal for enhancing both accuracy and robustness of the method. Adaptive confidence aggregation (ACA) model is proposed to fully exploit the complementary information between the modulus, phase, real and imaginary parts of the input time-frequency spectrum of jamming signal, and adaptively aggregate these information by confidence vectors. Moreover, to further improve the recognition accuracy while enhancing the robustness of the network, a cross-modal refinement model (CMR) is proposed to mine the correlation information between the aggregated jamming recognition result and the original input time-frequency image. Experimental results on both simulated and measured mixed datasets validate that the proposed JR-ACAR is superior to existing methods for radar jamming recognition in terms of recognition accuracy and robustness.

Multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) technology is widely used in integrated radar and communication systems (IRCSs). Moreover, index modulation (IM) is a reliable OFDM transmission scheme in the field of communication, which transmits information by arranging several distinguishable constellations. In this paper, we propose a sparse reconstruction-based joint signal processing scheme for integrated MIMO-OFDM-IM systems. Combining the advantages of MIMO and OFDM-IM technologies, the integrated MIMO-OFDM-IM signal design is realized through the reasonable allocation of bits and subcarriers, resulting in better intercarrier interference (ICI) resistance and a higher transmission efficiency. Taking advantage of the sparseness of OFDM-IM, an improved target parameter estimation method based on sparse signal reconstruction is explored to eliminate the influence of empty subcarriers on the matched filtering at the receiver side. In addition, an improved sequential Monte Carlo signal detection method is introduced to realize the efficient detection of communication signals. The simulation results show that the proposed integrated system is 5 dB lower in the peak sidelobe ratio (PSLR) and 1.5 ×105 lower in the number of complex multiplications than the latest MIMO-OFDM system and can achieve almost the same parameter estimation performance. With the same spectral efficiency, it has a lower bit error rate (BER) than existing methods.

In high-resolution and wide-swath synthetic aperture radar (SAR) systems, there are azimuth Doppler ambiguity and ghost ground moving targets (GMTs). Incorporated with an improved equivalent range model, this article explores a simultaneous scene imaging and GMT indication (GMTI) scheme with multiple beamforming for hypersonic vehicle-borne dive-trajectory high-squint multichannel SAR (HSV-DTHS-MC-SAR). First, the coarse-focusing imageries are recovered. Then, the CTV estimation with the Dynamic-EASDB is explored. After that, the SAR scene processing which enjoys “customized” beamforming, minimizes the ghost GMT and reconstructs the ambiguity-free scene image. Moreover, the proposed clutter suppression method generates beamformer center in the desired GMT direction and reduces the beamformer side-lobe by using quadratic pattern constraints. Finally, the geometry and inverse projections are performed to calibrate the distortion arising from vertical velocity. Compared to the existing methods, the proposed scheme has innovations or improvements in the CTV estimation, scene recovery, and cutter suppression. To be specific, it avoids the cumbersome CTV search and thus has lower computation, enjoys more concise and robust for removing ghost GMT in scene, and frees the steering vector mismatch plus achieves higher signal-to-clutter-and-noise ratio (SCNR). The extensive simulations confirm the effectiveness of our proposed scheme.

Aiming at the problem of poor tracking performance of traditional filtering algorithms for maneuvering extended targets, an improved interactive multimodel second-order extended Kalman filter (IMM-SOEKF) algorithm is proposed in this article. First, the Markov transfer probability matrix is updated using the probabilities within the neighboring time points between each model to improve the switching speed of the models in the algorithm and the matching accuracy. In addition, according to the state parameters and shape information of the target, the second-order extended Kalman filter (SOEKF) is used for tracking the estimation of the target. By incorporating the measurement covariance equation from the filtering algorithm into the likelihood function calculation, new likelihood probability and model weight assignments are obtained using the maximum likelihood function method to improve the tracking accuracy for extended targets and the robustness of the algorithm. Finally, simulation and data processing results show that the algorithm has higher tracking accuracy compared with other state-of-the-art algorithms.

Bistatic PRI agile radar has broad application prospects in modern electronic countermeasures because of its good anti-reconnaissance and anti-interference characteristics. It is known that the bistatic PRI-agile radar's detection and estimation performance of high-speed target can be improved by signal integration. However, due to the high speed and maneuvering characteristics of the target, the integration processing is confronted with problems such as RM, DFM, and scale effect, which would lead to a reduction in integration performance. Therefore, we present a coherent integration method named NU-SCGRFT for high-speed maneuvering target detection and estimation with bistatic PRI-agile radar in this article. First, we establish the target's equivalent motion model, which considers both the intrapulse and the interpulse motion of the target, and analyze the characteristic of the received signal and the mismatch effect of the traditional pulse compression. Then, the definition of the NU-SCGRFT method is given, and the processes of intrapulse integration and interpulse integration are derived. Thereafter, the signal processing flow based on the NU-SCGRFT method is given. Finally, through detailed simulation experiments, we compare and analyze the integration, detection, and parameter estimation performance of the proposed algorithm. The result shows that NU-SCGRFT has good integration and parameter estimation performance in bistatic PRI-agile radar system.

Coherently integrating echoes of multiple-input-multiple-output (MIMO) radar could enhance detection performance and early warning capability for high-speed weak targets. Nevertheless, the echo signal integration incorporating the single channel and multichannel faces two major challenging problems: first, the across range unit (ARU) is caused by the high-speed motion of targets for single-channel returns, and second, the echo envelope and phase differences between channels make multichannel integration particularly difficult. To tackle these issues, we propose a multichannel signal integration approach for high-speed weak target detection in coherent MIMO radar system. More precisely, we establish the echo model for the high-speed target with MIMO radar, and the Radon Fourier transform (RFT) is used to accumulate the single-channel returns with ARU. Then, we analyze the output characteristics of RFT with respect to different channels. Further, aiming to achieve multichannel integration, we decompose it into two steps: intranode integration and internode integration. First, by exploiting the target kinematic constraint and spatial geometric relations between MIMO radar nodes and the target, we propose the intranode integration algorithm, wherein we derive the target's energy accumulation equations among channels. After that, based on the properties of the intranode integration, we further propose the internode integration algorithm. Ultimately, the multichannel coherent accumulation is obtained, and the theoretical integration output responses of intranode and internode are provided in detail, respectively. Through numerous simulation experiments, we demonstrate the feasibility of the proposed approach.

Radar jamming recognition aims to accurately recognize the type of jamming to provide guidance for radar countermeasures. Although previous deep learning-based methods have made promising performance, they mainly rely on convolutional neural network (CNN) based on local processing, which ignore the global information in the time-frequency domain data of the jamming signal and also require much inference time to obtain the final recognition results. In this article, a novel few-shot jamming recognition network via time-frequency self-attention and global knowledge distillation (JR-TFSAD) is proposed by jointly considering the global information in the time-frequency spectrum of the jamming signal and the real-time performance of the recognition network. A time-frequency self-attention (TFSA) model is proposed to extract the global deep features of radar jamming signals by learning the correlation between two arbitrary points in the time-frequency spectrum of the jamming signal, thus improving the recognition accuracy. Moreover, to effectively reduce the inference time of the method while preserving the recognition accuracy, a global knowledge distillation (GKD) model is further constructed to perform jamming recognition by distilling the global knowledge from the TFSA model. The experimental results on the simulated and measured mixed dataset verify that the proposed method has higher recognition accuracy and shorter inference time compared to the state-of-the-art methods.

This article explores the coherent integration and parameter estimation problem for high-speed target detection using bistatic multiple-input multiple-output (MIMO) radar. Unlike the traditional intrachannel coherent integration that could only focus the target's energy during the multipulse observation, the bistatic MIMO radar implements joint intrachannel integration and interchannel integration to improve detection performance. However, its difficulties lie in that the high-speed motion results in the range migration (RM) in the intrachannel and the phase and envelope differences among multichannel. To solve these problems, we propose a multichannel coherent integration methodology in the radon Fourier transform (RFT) domain for the bistatic MIMO radar. First, we apply the RFT to integrate the target energy in the intrachannel. Then, the multichannel coherent accumulation is performed by the design of the phase compensation function and envelope alignment function. Finally, based on the relationship between the integration peak's location and target parameters, a Broyden-like method and linear equation solving are, respectively, presented to estimate the target's position and velocity. The effectiveness of the proposed method is assessed via simulation experiments and real-measured data. It is further noted that the proposed method can detect targets at low signal-to-noise ratios (SNRs) which are not attainable using state-of-the-art methods.