Distributed array radar(DAR)has the benefits including flexible deployment and excellent spatial resolution.However,the independent clock and oscillator configurations of each radar unit,along with the instability introduced by trigger signals transmitted over the feeder link,introduce errors in time and phase synchronization,thereby reducing the accuracy of coherent synthesis in a DAR.This paper proposes a time and phase synchronization error estimation method based on space-time spectral entropy.The distance-angle space-time two-dimensional spectrum is first constructed by establishing a space-time covariance matrix with time and phase synchronization faults.Then,based on entropy theory,the correspondence between synchronization errors and space-time spectral shape uncertainty is established.The time and phase synchronization errors are estimated by optimizing the space-time spectral entropy to minimize its value.Simulation experiments validate the accuracy of the proposed method,particularly exhibiting good estimation performance under low signal-to-noise ratio.

In the field of synthetic aperture radar (SAR) target recognition, leveraging computer-generated labeled simulated SAR data to aid in the recognition of unlabeled real SAR data has garnered significant attention. However, the domain shift between the simulated and real domain contravenes the assumption of independent and identically distributed in deep learning (DL). Moreover, the categories of real SAR data may not align precisely with those of the simulated SAR data, and the prior knowledge of the label shift typically remains unavailable. To align the common category samples and identify target-private unknown category samples in the presence of both domain shift and potential label shift, we introduce a two-stage universal domain adaptation method for simulation-assisted SAR target recognition based on adversarial uncertainty and neighbor relation (AUNR). The proposed AUNR introduces Dirichlet distribution-based evidential DL, which not only performs classification tasks but also characterizes decision uncertainty, enabling the model to express the concept of "Unknow." In the first stage, alongside supervised learning, we leverage data augmentation-based contrastive learning to achieve robust feature representation. In the second stage, adversarial learning based on model and data uncertainty is employed to separate common and unknown class samples, and contrastive learning based on neighbor relation aligns samples in two domains according to their geometric relationships. To the best of our knowledge, this is the first study on UniDA for SAR target recognition. Experimental results demonstrate that our method outperforms previous State-of-the-Art methods in various label shift scenarios, which indicates the effectiveness of the proposed method.

Distributed array radar (DAR) achieves large-aperture performance by combining small subarrays at different locations, which reducing system burden and manufacturing complexity while meeting low-cost and highresolution requirements in modern sensing. This paper proposes a 2-dimensional (2D) subarray optimization method for regular DARs with identical subarrays, aiming to minimize beamwidth (BW) and maximum sidelobe level (MSLL) under physical layout constraints. A directional expanded beam pattern (DiEBP) is introduced to reformulate non-analytic optimization objectives into a differentiable form, which enables gradient-descent updates within an alternating direction method of multipliers (ADMM) framework for this non-convex problem, with exponential smoothing and Monte Carlo tree search (MCTS) used for stabilization and initialization. Numerical experiments show that, compared with existing algorithms, the proposed method achieves about 10-15% narrower BW and 2 dB lower MSLL on average, along with a reduced Cramer-Rao Bound (CRB). Its effectiveness is further validated through DOA estimation and near-field imaging experiments.

This letter presents an efficient algorithm for designing Doppler-tolerant radar waveforms with superior computational efficiency. The proposed method jointly minimizes the peak sidelobe level (PSL) and mainlobe loss level (MLL) under Doppler shifts by introducing a dynamic weighting strategy that adaptively balances these competing objectives. To address the resulting non-convex optimization problem, log-sumexp smoothing combined with automatic differentiation is employed, enabling efficient gradient-based optimization. Experimental results demonstrate that the proposed algorithm achieves 1 dB lower PSL than the state-of-the-art maximum block improvement (MBI) method with comparable MLL, while being up to 8 times faster computationally. Moreover, the algorithm scales efficiently to longer code lengths, making it particularly suitable for practical radar applications requiring adaptive waveform design.

Deep-learning-based few-shot synthetic aperture radar (SAR) ship detection has demonstrated considerable potential in scenarios with limited annotated samples and dynamically emerging ship categories, closely matching the practical requirements of maritime surveillance and intelligent SAR image interpretation. However, existing methods often overlook challenging negative proposals and the incomplete annotation problem inherent in few-shot learning. In complex offshore and inshore scenes, negative proposals contain either difficult background regions caused by sea clutter, port facilities, and strong scattering interference, or unlabeled ship targets resulting from missing annotations. Existing methods struggle to distinguish between these two types of proposals. Ignoring challenging negatives prevents the detector from learning precise decision boundaries between ships and complex backgrounds, whereas treating unlabeled ships as background introduces erroneous gradients during backpropagation and degrades detection performance. To address these issues, we introduce uncertainty as a measure of proposal reliability and propose two complementary components: uncertainty-guided proposal separation (UGPS) and uncertainty-aware discriminative gradient refocusing (UADGR). UGPS jointly exploits proposal uncertainty and intersection-over-union (IoU) to separate challenging negatives and confusing negatives from the negative proposal set, thereby preserving informative hard backgrounds while identifying potential unlabeled ships. Subsequently, UADGR combines proposal uncertainty with feature dissimilarity to a background prototype to adaptively regulate their training gradients. Specifically, higher weights are assigned to challenging negatives to improve discrimination between ships and complex background interference, whereas lower weights are assigned to confusing negatives to suppress erroneous supervision introduced by missing ship annotations. Extensive experiments on SRSDD-v1.0 demonstrate consistent improvements over existing few-shot detection approaches across different data splits and shot settings, while additional results on SAR-AIRcraft-1.0 further confirm the generalization of the proposed method.

High-resolution radar sensors are critical for autonomous systems but pose significant challenges to traditional tracking algorithms due to the generation of multiple measurements per object and the presence of multipath effects. Existing solutions often rely on the point target assumption or treat multipath measurements as clutter, whereas current extended target trackers often lack the capability to maintain trajectory continuity in complex multipath environments. To address these limitations, this paper proposes the multipath extended target generalized labeled multi-Bernoulli (MPET-GLMB) filter. A unified Bayesian framework based on labeled random finite set theory is derived to jointly model target existence, measurement partitioning, and the association between measurements, targets, and propagation paths. This formulation enables simultaneous trajectory estimation for both targets and reflectors without requiring heuristic post-processing. To enhance computational efficiency, a joint prediction and update implementation based on Gibbs sampling is developed. Furthermore, a measurement-driven adaptive birth model is introduced to initialize tracks without prior knowledge of target positions. Experimental results from simulated scenarios and real-world automotive radar data demonstrate that the proposed filter outperforms state-of-the-art methods, achieving superior state estimation accuracy and robust trajectory maintenance in dynamic multipath environments.

Deep learning-based Synthetic Aperture Radar (SAR) ship detection has achieved remarkable progress in recent years. However, the high cost of annotating SAR images severely limits the practical deployment of fully-supervised detection methods. Pseudo-label-based semi-supervised learning is a mainstream approach for alleviating the reliance on large-scale annotations, and it has achieved superior performance in general object detection. Nevertheless, when directly transferred to SAR ship detection, these methods face two critical challenges due to the unique imaging characteristics of SAR data: first, they struggle to balance pseudo-label precision and recall, as ship diversity and speckle noise make confidence scores unreliable; second, they cannot effectively filter out low-quality pseudo-labels during student training, which limits further performance improvement. To address the issues above, this paper proposes a novel semi-supervised object detection framework, named SPG-IAD, specifically for SAR ship detection. The framework consists of two core components: the Scattering-Point-Guided Dual-Criterion Coarse Filtering (SPG-DC) strategy and the Intersection over Union (IoU)-Aware Teacher-Student Dynamic Fine Selection (IA-TSD) strategy. Specifically, SPG-DC combines classification scores with the physical prior of ship scattering points to collaboratively filter detection outputs, effectively recalling low-confidence yet well-localized candidate predictions. IA-TSD introduces an IoU prediction branch to quantify localization quality and compares teacher-student localization quality using a scale-adaptive threshold. This strategy enables dynamic pseudo-label selection and suppresses the misleading effects of low-quality pseudo-labels. Experiments on the HRSID and SSDD datasets demonstrate that under the semi-supervised setting with 10% labeled data, the proposed method achieves a significant improvement in Average Precision (AP) over the fully-supervised baseline, and outperforms existing state-of-the-art methods.

ABSTRACT To address resolvable group target tracking (RGTT) in environments with unknown prior birth information, this paper proposes a dual‐channel virtual leader generalised labelled multi‐Bernoulli (DC‐VL‐GLMB) filter with hybrid adaptive birth. To reduce false births caused by conventional measurement‐driven adaptive birth methods that neglect group spatial compactness and coordinated motion, a dual‐channel hybrid adaptive birth method is developed. In Channel A, the robust batch group adaptive birth (RBG‐AB) algorithm extracts candidate tracks within a sliding window and converts group trajectories into Bernoulli birth components. The state deduplication step removes components that duplicate existing tracks. In Channel B, the residual measurement adaptive birth (RM‐AB) algorithm generates supplementary Bernoulli birth components from measurements that are neither used by Channel A nor associated with surviving tracks. Birth components from both channels are merged to form an adaptive labelled multi‐Bernoulli (LMB) birth prior, yielding a dual‐channel GLMB (DC‐GLMB) filter. The virtual leader (VL) model is incorporated to construct the DC‐VL‐GLMB filter, and a fast implementation strategy based on maximum a posteriori (MAP) hypothesis extraction and global label mapping is developed to reduce computational complexity. Simulations and drone‐swarm radar measurements show that the proposed filter suppresses false births, improves track continuity and achieves robust tracking.

Continuous trajectory maintenance is important for aerial vehicle monitoring, threat assessment, and warning or interception decisions in ground-to-air remote sensing systems. In the considered system, radar and radio-frequency (RF) sensing are used together to produce fused aerial tracks. In practical monitoring, the fused trajectory can still be interrupted by occlusion, maneuvering, missed detections, poor sensing geometry, or unstable measurements. These interruptions produce fragmented tracks and reduce the reliability of long-term surveillance. This paper formulates 3D trajectory restoration as a post-processing task for fused aerial tracks and proposes AeroDiff-TIR, a conditional diffusion-based restoration framework. The method represents trajectories in a local Cartesian coordinate system and treats each interrupted segment as the missing part of a time series. Given the observed points before and after a gap, AeroDiff-TIR learns the conditional distribution of the missing segment and restores it through iterative denoising. Experiments on a simulated aerial vehicle trajectory benchmark and measured unmanned aerial vehicle (UAV) trajectories collected by a ground-to-air monitoring system show that AeroDiff-TIR improves trajectory consistency over the complete restored gap and can support trajectory continuity in aerial monitoring systems.

ABSTRACT Multipath propagation in automotive radar produces ghost targets that can mislead perception systems in autonomous driving. This letter proposes a physics‐aware graph neural network (PAGNN) that exploits multipath propagation geometry for radar ghost detection. PAGNN operates in two stages: a GNN‐based affinity clustering stage that groups radar points into target‐level clusters, and a classification stage that identifies ghosts through graph convolution and physics‐informed attention modules. Experiments on real‐world automotive radar data demonstrate that PAGNN outperforms existing methods, improving mean average precision by 15.83 percentage points over the PointNet++ baseline. These results highlight the effectiveness of embedding physical models into graph‐based learning for radar perception.

Instance segmentation of ships in synthetic aperture radar (SAR) imagery is critical for applications such as maritime monitoring, environmental analysis, and national security. SAR ship images present challenges including scale variation, object density, and fuzzy target boundary, which are often overlooked in existing methods, leading to suboptimal performance. In this work, we propose O2Former, a tailored instance segmentation framework that extends Mask2Former by fully leveraging the structural characteristics of SAR imagery. We introduce two key components. The first is the Optimized Query Generator (OQG). It enables multiscale feature interaction by jointly encoding shallow positional cues and high-level semantic information. This improves query quality and convergence efficiency. The second component is the Orientation-Aware Embedding Module. It enhances directional sensitivity through direction-aware convolution and polar-coordinate encoding. This effectively addresses the challenge of uneven target orientations in SAR scenes. Together, these modules facilitate precise feature alignment from backbone to decoder and strengthen the model's capacity to capture fine-grained structural details. Extensive experiments demonstrate that O2Former outperforms state-of-the-art instance segmentation baselines, validating its effectiveness and generalization on SAR ship datasets.

Circular synthetic aperture radar (CSAR) is an advanced mechanism with the capability of three-dimensional imaging, which can continuously observe the omnidirectional scattering characteristic of the ground scene. The time-domain imaging algorithm with accurate focusing ability, such as the back-projection algorithm, is often used in large-scene CSAR imaging. Due to a large amount of echo data, time-domain imaging algorithms are typically time-consuming. The polar format algorithm (PFA) is a convenient and efficient frequency-domain imaging algorithm. However, the azimuth defocus caused by the phase error of wavefront curvature limits the depth of focus in CSAR images formed by PFA. In this article, we propose a large-scene PFA (LS-PFA) for CSAR sub-aperture imaging with space-variant post-filtering. By leveraging the phase error of wavefront curvature for arbitrary curved flight paths, a space-variant filter suitable for the large-scene subaperture CSAR image formed by PFA is designed to compensate for the azimuth defocus. LS-PFA provides an efficient and general solution to acquire well-focused large-scene subaperture CSAR images under arbitrary curved flight paths. The focusing performance of LS-PFA is verified with simulation and experimental results.

Single-frequency MIMO arc array radar is an innovative MIMO imaging mechanism capable of reconstructing the full-attitude three-dimensional (3D) scattering center. This model can offer a comprehensive description of the electromagnetic scattering characteristics of targets. However, traditional methods for reconstructing scattering centers exhibit increased computational complexity as the attitude angle and dimensionality rise, thereby limiting echo synthesis efficiency. To address this challenge, we propose a novel method for full-attitude 3D scattering center reconstruction based on the orthogonal matching pursuit (OMP) algorithm applied to singlefrequency MIMO arc array radar. By constructing a sparse dictionary matrix utilizing the 3D surface imaging results, the position and amplitude of the scattering centers are estimated iteratively according to an optimal atomic selection criterion, reconstructing the 3D scattering centers of complex targets from multiple attitude angles under the conditions of a given number of scattering centers. The proposed method can quickly reconstruct the full-attitude 3D scattering centers of complex targets. The effectiveness and efficiency of the proposed method are validated through the simulation data of electromagnetic scattering.

This letter presents an efficient algorithm for designing binary phase-coded sequences with low peak sidelobe level (PSL) for multiple-input multiple-output (MIMO) radar applications. The algorithm features a log-sum-exp function (LSEF)-based fitness function and a vectorized recursive computation method, substantially reducing computational complexity compared to conventional approaches. A two-stage search strategy effectively balances between local exploitation and global exploration. Numerical results demonstrate that the proposed algorithm achieves comparable PSL performance to state-of-the-art algorithms with enhanced computational efficiency in monostatic cases, while exhibiting superior PSL performance in MIMO scenarios.

Maneuvering target tracking remains a critical and challenging task in radar signal processing due to the inherent uncertainty and complexity of target motion. The Kalman Filter and its various extensions have been widely applied for target tracking. However, their performance often degrades when dealing with highly maneuvering targets. To address this issue, this paper proposes a novel maneuvering target tracking approach assisted by dynamic Radar Cross Section (RCS) information. The proposed method leverages the fluctuations in RCS data detected via a sliding window to adaptively adjust the filter parameters, thereby enhancing tracking performance. Simulation results demonstrate that the method significantly improves the filter's adaptability to dynamic target behavior and achieves higher tracking accuracy and robustness.

This letter investigates the low-variance broad beampattern design method in distributed phased multiple-input multiple-output (phased-MIMO) radar. The constant modulus constraint across multiple subarrays results in a low-rank and nonconvex objective function, which is traditionally addressed by reformulating it into a solvable semidefinite program through convex relaxation. In contrast, we propose a Riemannian manifold-based method to directly address the low-rank problem without relaxation. The low-variance broad beampattern design is first transformed into an unconstrained quadratic form on a complex constant modulus manifold. Then, a Riemannian conjugate gradient descent (RCGD)-based optimization is proposed to solve the nonconvex objective function by deriving the gradient descent direction and adaptive step size. Numerical simulations demonstrate the superior performance in terms of computation speed and accuracy compared to the conventional methods.

Circular Synthetic Aperture Radar (CSAR) subaperture images are important inputs for CSAR three-dimensional (3D) incoherent imaging and target location. Due to the nonlinear geometric distortion caused by the perspective difference, it is necessary to register the CSAR subaperture images. However, the computational complexity of the existing SAR image registration methods is high, limiting the efficiency of CSAR 3D imaging and target location. In this paper, we propose a novel registration method based on accelerated-KAZE features for CSAR subaperture images. The proposed method constructs the nonlinear scale space with a pyramid structure by using fast explicit diffusion to improve the efficiency of feature detection. It exploits the modified local difference binary descriptor to achieve robust feature extraction. The proposed method can realize CSAR subaperture image registration with high quality and efficiency. The effectiveness of the proposed method is verified through the large scene Gotcha data.

Near-field radar systems are widely used in military and industrial applications for the detection and high-precision imaging of complex objects. Achieving precise delineation of target features requires high angular resolution, a critical metric directly related to array aperture. Distributed radar networks obtain a large-aperture array and enhanced angular resolution by coherently combining signals from subapertures positioned at various locations. However, the spacing between subapertures often leads to sparse spatial sampling, introducing grating lobes into the reconstructed image. In this article, we propose a multiaperture fusion transformer (MAFFormer), a complex-valued (CV) dual-domain, local-global feature fusion neural network for multiaperture fusion in near-field distributed radar networks, which enables accurate recovery of missing apertures under 75% signal sparsity. By leveraging both local and global features from the antenna array and spatial spectrum domains (S-domains) of raw radar signal through the integration of convolutional neural network (CNN) and Transformer components, the proposed network achieves enhanced estimation accuracy, overcoming the limitations of current algorithms that suffer from performance degradation in complex environments. Extensive experiments, including simulations and measurements, validate that the proposed algorithm achieves high-resolution imaging of complex targets without additional physical apertures or complex hardware and shows significant superiority on imaging quality compared to traditional methods across diverse target types. Furthermore, the robustness of the proposed structure to unseen radar parameters and imaging scenarios is validated using publicly datasets, highlighting its suitability for a wide range of practical imaging applications.

Accurate track segment association plays an important role in modern sensor data processing systems to ensure the temporal and spatial consistency of target information. Traditional methods face a series of challenges in association accuracy when handling complex scenarios involving short tracks or multi-target intersections. This study proposes an intelligent association method that includes a multi-dimensional track data preprocessing algorithm and the characteristic-aware attention long short-term memory (CA-LSTM) network. The algorithm can segment and temporally align track segments containing multi-dimensional characteristics. The CA-LSTM model is built to perform track segment association and has two basic parts. One part focuses on the target characteristic dimension and utilizes the separation and importance evaluation of physical characteristics to make association decisions. The other part focuses on the time dimension, matching the application scenarios of short, medium and long tracks by obtaining the temporal characteristics of different time spans. The method is verified on a multi-source track association dataset. Experimental results show that association accuracy rate is 85.19% for short-range track segments and 96.97% for long-range track segments. Compared with the typical traditional method LSTM, this method has a 9.89% improvement in accuracy on short tracks.

To enhance radar performance in target detection, distributed CFAR detection systems are gaining increasing attention and application. Such systems effectively extend detection range, improve detection probability, and strengthen anti-interference capabilities. However, existing research has primarily focused on distributed CFAR detection under Gaussian clutter environments, with relatively limited exploration of non-Gaussian clutter environments such as log-normal clutter. To address these issues, this paper proposes a distributed OS-CFAR detection method suitable for log-normal clutter backgrounds. First, for the simple case of single-detector detection, an OS-CFAR detection method suitable for log-normal clutter backgrounds is derived. This is then extended to distributed detection systems composed of multiple detectors, yielding the corresponding distributed OS-CFAR detection method. Finally, simulation experiments are conducted to analyze the performance of the detection method, confirming that the distributed OS-CFAR detection method exhibits superior detection performance under log-normal clutter backgrounds.