ABSTRACT Predictive full‐wave modelling of additively manufactured (AM) antennas is limited by uncertainty in printed‐conductor loss, metallization thickness, surface morphology and nonuniform dielectric properties. This letter presents a process‐constrained modelling workflow for inkjet‐metallized antennas fabricated on additively manufactured dielectrics. Material properties extracted from coplanar waveguide (CPWG) structures, resonators and test coupons provide useful electrical‐property inputs, but do not by themselves verify transfer to the printed antenna geometry. This work addresses that geometry‐transfer problem by combining extracted material‐property information with antenna‐specific process metrology and validating the resulting model against measured electromagnetic response. The approach is demonstrated using a magneto‐electric dipole. Four‐point‐probe sheet resistance, profilometry‐derived thickness and roughness, cross‐sectional scanning electron microscopy (SEM) and dielectric characterization provide quantitative HFSS inputs, while surface SEM observations identify cracking, granular morphology and apparent porosity that support the reduced‐conductivity representation. The process‐constrained model reproduces measured input‐match behaviour, radiation‐pattern trends and transverse near‐field distribution without post‐hoc fitting. Relative to the idealized HFSS model, near‐field correlation improves by 46.8%. The principal novelty is a physically traceable AM characterization framework linking geometry‐specific conductor metrology to spatial electromagnetic validation, overcoming geometry‐transfer ambiguity inherent when coupon, CPWG or resonator‐derived effective properties are used alone.
ABSTRACT Recent deep learning‐based reconfigurable intelligent surfaces (RIS) phase control models achieve competitive performance, but their scalability is constrained by rapidly increasing computational complexity as the number of reflective elements grows. We propose RISnet‐LiteMix, a lightweight MLP‐Mixer‐based RIS phase control model that captures inter‐user interactions with significantly lower computational complexity. In particular, the LiteMixer block in RISnet‐LiteMix applies dimensional compression or expansion across different perspectives, thereby reducing computational burden while enabling effective learning of channel representations. Consequently, RISnet‐LiteMix reduces the number of parameters by 70% and computational complexity by 47% relative to existing MLP‐Mixer‐based models, while maintaining comparable transmission performance.
ABSTRACT Floating photovoltaic (FPV) systems improve photovoltaic performance through water‐induced cooling; however, the influence of cooling‐model parameters on energy production remains insufficiently quantified. This letter proposes a computationally efficient FPV assessment framework by combining Seasonal Typical Meteorological Days with a semi‐empirical cooling‐model driven by wind speed, humidity and a constant cooling term. Unlike conventional studies that employ fixed cooling coefficients, the proposed formulation treats these coefficients as engineering design parameters and systematically evaluates their influence. Simulations using NASA POWER meteorological data show that the FPV achieves a 3.7% annual energy gain compared with a conventional PV, with the largest improvement occurring during summer. The wind‐speed cooling coefficient is the dominant factor governing FPV performance.
ABSTRACT The unmanned aerial vehicle (UAV) cluster can accomplish complex tasks that a single platform cannot handle. Precise time synchronisation among the nodes in a formation flight of a UAV cluster is an essential foundation for resource scheduling, cooperative positioning and data fusion. The connectivity of the time comparison link exhibits time‐varying characteristics, which pose a challenge to the realisation of continuous high‐precision time synchronisation. This letter proposes a precise time synchronisation scheme based on time‐varying topology aggregation for a UAV cluster and an adaptive clock skew exchange protocol was designed to enable each node to achieve leader‐following consistency time synchronisation in the optimal path through edge computing. This method can achieve synchronisation accuracy better than 10ns and can provide support for collaborative applications of a UAV cluster.
ABSTRACT Rainfall estimation plays a pivotal role in meteorological early warning systems, urban flood mitigation and hydrological scheduling. To tackle the challenges of heavy rainfall sample scarcity, long‐tailed distribution imbalance and noise sensitivity in Doppler radar sensor (DRS) spectral‐moment block‐level features, a physics‐guided evidence routing mixture network (PERM‐Net) framework for adaptive rainfall estimation is proposed. By integrating evidence‐driven dynamic routing (EDR), heterogeneous expert modelling (HEM) and fusion with uncertainty estimation (FUE) into a unified framework, PERM‐Net enhances feature representation capability and improves the robustness and reliability of rainfall estimation under complex precipitation scenarios. Experimental results demonstrate that the proposed framework consistently outperforms some learning methods in terms of overall estimation accuracy and adaptability to complex raining condition. These superior performance further validates the effectiveness and generalization capability of PERM‐Net for DRS rainfall estimation.
ABSTRACT Overhead ground wires (OGWs) of transmission lines are prone to broken‐strand faults under the combined effects of strong winds, ice, lightning and corrosion, which may further trigger ground faults and unplanned power outages. Traditional maintenance methods suffer from high operational risks, heavy labour intensity or mandatory power cuts. To address these issues, this paper proposes a dual‐robot cooperative system for live‐line repair of OGW broken‐strand defects. The system consists of a strand‐repositioning robot and an armour‐winding robot. The mechanical structure of each robot is elaborately designed with an outer‐fixation and inner‐rotation architecture to realise strand resetting, tape fixation and preformed armour rod winding. Meanwhile, the overall hardware framework and operational workflow of the control system are established. Laboratory prototype tests achieved an average repair speed of 1.48 m/min for a 3 m‐long local splaying defect. The average repair speed of the robot is 68.2% faster than manual operation, and the repaired section exhibits favourable structural integrity. This system effectively avoids power outages and reduces maintenance risks, providing a feasible technical solution for intelligent live‐line maintenance of transmission lines.
ABSTRACT Unknown custom protocols in space communication often complicate security analysis. A key challenge in protocol reverse engineering (PRE) is identifying the original protocol that a new variant is based on—a critical step that is often manual and time‐consuming. This paper presents SpaceProtoNet , a novel framework that employs a convolutional neural network (CNN) to classify protocol types from image representations of raw packet data. Experimental results demonstrate that SpaceProtoNet effectively generalises from known protocols to classify unseen variants into their correct base families, showing an F1‐score of 0.96 even when only four packets are available. The framework also maintains robustness in adverse conditions, sustaining an F1‐score of 0.91 even with a 10% bit error rate (BER). By automating this crucial identification step, SpaceProtoNet provides a systematic foundation for PRE, where the predicted base protocol family can help narrow the analysis scope and reduce the complexity of security analysis for space communication systems.
ABSTRACT In high‐resolution radars, targets span multiple range‐Doppler (R‐D) cells, forming 2‐D range‐spread targets. Post‐CFAR detection, scattering fluctuations often fragment these targets. While standard connected component labelling efficiently processes sparse grids, its restrictive 1‐pixel adjacency fails to bridge these fractures. Grid‐based gap‐bridging incurs heavy computational burdens, and point‐cloud density algorithms suffer from an isotropic dilemma causing false adhesion or fragmentation. To address these bottlenecks, we propose a sparse vectorized clustering algorithm. By reducing the detection matrix to a sparse coordinate set and formulating an anisotropic Chebyshev tolerance, it employs a vectorized breadth‐first search to seamlessly integrate fragmented targets. Experiments on simulated and measured data demonstrate the algorithm effectively resolves the topological limitations of symmetric radii, achieving robust assignments while minimizing heuristic parameter tuning. By circumventing floating‐point operations and spatial tree searches, it exhibits strictly linear scaling, delivering computational efficiency substantially higher than conventional paradigms for real‐time radar applications.
ABSTRACT This letter proposes A‐CQFOA, an adaptive chaotic quantum‐inspired fruit fly optimisation framework for multi‐objective feature selection coupled with CNN, LSTM and Bi‐LSTM sentiment classifiers. It unifies a logistic chaotic map and quantum rotation operator within a single FOA‐based selector, with a classifier‐coupled fitness function optimising accuracy, feature compactness and prediction consistency, and an adaptive stagnation‐aware chaos‐decay mechanism. Over 30 independent 10‐fold CV runs on Sentiment140 (1.6 M tweets), SST‐2 (67k reviews) and Telecom Twitter (22k opinions), A‐CQFOA obtains mean accuracies of 0.9267, 0.9412 and 0.9328—gains of +3.17–3.28% over PSO while reducing features by 35%–45%. Open‐set AUC = 0.997 and >94% few‐shot accuracy (5 samples/class), all at p < 0.01.
ABSTRACT To overcome the degradation of conventional direction‐finding accuracy for VLF and LW signals caused by sky‑wave‑induced polarisation deflection, this paper employs a tri‑axial magnetic antenna to capture the three‑dimensional magnetic field components and proposes an analytic closed‑form joint estimation algorithm. The direction‑of‑arrival angles and polarisation parameters are solved through eigen‑decomposition of the covariance matrix, vector cross product and polarisation projection. Simulations verify that the algorithm effectively circumvents the systematic errors introduced by elliptical polarisation, yields direction‑finding performance markedly superior to that of the conventional amplitude‑comparison method, exhibits good low‑SNR adaptability and can satisfy the engineering requirements of ship navigation and radio sounding.
ABSTRACT This paper proposes MixBinNet, a neural network‐based model that generates binary sequences with low autocorrelation sidelobes and low cross‐correlation while maintaining a balanced bit distribution. Conventional sequence design methods lack flexibility in sequence length and set size, while often failing to jointly optimize correlation and balance properties. To overcome these limitations, MixBinNet employs an embedding block for diverse initialization, intra‐ and inter‐mixing blocks that learn local and global correlation patterns and a binarization block for converting real‐valued outputs into binary sequences. In addition, we design custom loss functions that enable the joint optimization of autocorrelation sidelobes, cross‐correlation and balanced bit distributions during training. Experimental results show that MixBinNet consistently outperforms existing chaotic map‐based methods in terms of autocorrelation and cross‐correlation, while maintaining a bit distribution close to the ideal ratio of 0.5. These results demonstrate the effectiveness of MixBinNet in generating high‐quality binary sequence sets.
ABSTRACT Apart from the radar or/and communication performance improvement, power‐saving is also an important metric in multicarrier dual‐functional radar‐communication (DFRC) system. To this end, this letter explores the power allocation problem in multicarrier DFRC system by minimizing the total radiated power with the constraints of communication data rate, Cramér–Rao lower bound (CRLB) of target time delay estimation and subcarrier power. We present two power allocation strategies that are suitable for different scenarios, one considering the total data rate constraint while the other considering the data rate constraint on each subcarrier. Two resulting optimization problems are solved by converting them into convex problem and linear programming problem, respectively. Via numerical simulations, the proposed two strategies are compared with other power allocation strategies under different communication data rates and CRLB thresholds. The results indicate that our power allocation strategies are power‐saving and can achieve a great trade‐off between the total radiated power and the performance of radar and communication.
ABSTRACT A low‐power, high‐linearity input buffer for multi‐GS/s analogue‐to‐digital converters is presented. A third‐order Volterra‐series model is utilised to characterise a source follower biased near the saturation–triode boundary, revealing the interaction between static resistive and dynamic capacitive nonlinearities. Based on this analysis, a dual‐band compensation topology employing parallel auxiliary differential pairs is designed to mitigate these distortions, enabling a lower core supply to save power. As a pre‐silicon design study, the proposed buffer is validated through transistor‐level simulations using a 5‐nm FinFET PDK, achieving an SFDR of 81 dBc at 0.95 GHz and 65 dBc at 10.2 GHz for a 0.8‐V pp differential swing, while consuming 48 mW from 1.2‐V and 1.5‐V supplies.
ABSTRACT Target detection and recognition in synthetic aperture images remain challenging. Given the advantages of the YOLO series, four enhancement schemes (Marine, SDS, DINOv3, MoonNet) are evaluated comprehensively. These modules are integrated into YOLOv8, YOLOv11, YOLOv12 and YOLOv13 in this letter. Sixteen comparative models are constructed. Three main conclusions are obtained. First, MoonNet is the most robust method. Performance improvements are achieved for YOLOv8 and YOLOv11. Slight drops in performance are observed for YOLOv12 and YOLOv13. Second, high sensitivity to the base architecture is shown by SDS and marine. A 0.0172 drop is observed in marine‐based YOLOv11. This is the largest drop. Third, the performance of all DINOv3‐based methods is poor. No positive improvements are achieved at all. These conclusions provide practical guidance for users in model selection.
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.
ABSTRACT This letter presents a two‐stage RC‐network‐based differential ring voltage‐controlled oscillator (VCO) featuring common‐mode false‐lock suppression and fast settling. A cross‐coupled latch is incorporated into the delay cell to mitigate the common‐mode false‐lock state that tends to occur in even‐stage ring VCOs. In addition, a bypass transistor is introduced in the bias path to shorten the settling time during turn‐on and is disabled in the steady state to preserve phase‐noise performance. Simulation results demonstrate a reduction in settling time from 40 µs to 2 µs while ensuring lock‐free startup. Fabricated in a 55 nm CMOS process, the prototype consumes 2.6 mW from a 1.2 V supply and operates over a frequency range of 1.31–1.89 GHz. The measured phase noise is −99.33 dBc/Hz at a 1 MHz offset from a 1.6 GHz carrier.
ABSTRACT Small traffic targets in unmanned aerial vehicle (UAV) imagery are easily missed because of overhead viewpoints, dense distributions, occlusion and complex backgrounds. RGB images provide texture and colour cues but degrade under low‐light or shadow conditions, whereas infrared images are more illumination‐robust but often have blurred boundaries. This letter proposes ERDF‐Anchor, a lightweight reliability‐guided decision fusion method for UAV RGB‐infrared traffic detection. A four‐channel early‐fusion detector supplies primary anchors, while RGB‐only and infrared‐only detectors serve only as auxiliary experts for consistency verification, confidence reweighting, unsupported‐anchor filtering and weighted box refinement. Object‐level expert statistics on DroneVehicle show that 50.6% of objects are more reliable in RGB and 46.8% in infrared, confirming local complementarity. Under a unified post‐processing protocol, ERDF‐Anchor improves recall from 0.9141 to 0.9321 and mAP50 from 0.8197 to 0.8252 over RGBIR4‐Early, with less than 1 ms post‐processing overhead per validation image on saved predictions.
ABSTRACT For planar arrays, adaptive sum and difference beamforming (ADBF‐SD) forms ‘canyon‐shaped’ null regions (CSNRs), which cause severe estimation errors and echo attenuation for off‐interference targets in the CSNRs, especially under main lobe interference. A monopulse angle estimation method based on adaptive dual sum‐and‐difference beamforming is proposed for planar monopulse arrays to alleviate the impacts of CSNRs. By setting main lobe constraint points to generate two sum beams with mutually perpendicular CSNRs and their corresponding difference beams, two monopulse ratio surfaces and estimation results are obtained, with the final target angle estimation derived from the fusion of these results. Compared with ADBF‐SD methods, simulation results demonstrate that the proposed method significantly improves the accuracy of target angle estimation
To address system biases and non-stationary noise in distributed radar networks, this paper proposes a robust track fusion algorithm based on a hybrid weighting mechanism. Drawing on the Interactive Multiple Model (IMM) framework, individual radar nodes are treated as parallel sub-filters. The algorithm integrates statistical maximum likelihood estimation with spatial geometric support to dynamically update fusion weights. A weight transition matrix is employed to describe the evolution of node credibility, while a Gaussian kernel-based similarity model evaluates geometric consistency to effectively suppress measurement outliers. By incorporating a mixing factor, the system adaptively balances statistical reliability and spatial support. Results indicate that this approach significantly enhances tracking accuracy and environmental adaptability, providing superior resilience against complex electromagnetic interference compared to traditional likelihood-based methods.
