In this letter, a robust autonomous timekeeping scheme for navigation constellations based on inter-satellite links is proposed. In this scheme, the clocks in the constellation are divided into two groups to generate two Ensemble Times (ET) respectively, then taking the two ETs as references to evaluate the frequency stability of each clock. Abnormal clocks can be identified and excluded from the clock group. After Fault Detection and Exclusion (FDE), the particle swarm optimization (PSO) algorithm is introduced to calculate the optimal weights of the two clock groups using the ensembled time of the opposite group as references respectively. Based on the optimized weights, the final Inter-satellite Link Time (ISLT) is obtained. The simulation implements ISL measurement precision and clock performance parameters of 8 BDS-3 satellites. Simulation results show that when the stability of the ensembled time is significantly higher than that of the evaluated clock, the evaluation results are very close to the true values. In the case of one abnormal clock, the long-term stability of the ISLT obtained by the proposed scheme reaches $1.6\times10^{-14}$ at $\tau=1\;\text{day}$, which is obviously better than $2.3\times10^{-14}$ at $\tau=1\;\text{day}$ using default weights. After 30 days of autonomous operation, the detrended clock difference is 7.8 ns. The research results provide a reference scheme for the autonomous timekeeping of navigation constellations.

Wideband inverse synthetic aperture radar imaging of targets with rotating components suffers from severe defocusing due to high-order nonlinear phase modulation and range cell migration (RCM). To address this problem, a wideband subband fusion single-range Doppler interferometry (WSF-SRDI) framework is proposed. The wideband echo is first partitioned into multipe narrowband subbands via a polyphase filter bank. A theoretical constraint relating the subband number to RCM confinement is derived, ensuring that the RCM remains within a single range cell per subband. Within each subband, a SRDI path-integral model is then constructed to simultaneously achieve subband imaging and rotation parameter estimation, followed by a subband fusion process to enhance the overall image quality. Finally, utilizing the estimated parameters, the rotating components are reconstructed and suppressed through least squares estimation, yielding a well-focused image of the stationary main body. Simulation results and contrast analysis demonstrate that the WSF-SRDI maintains robust performance under low signal-to-noise ratio conditions. The proposed strategy provides an efficient solution for ISAR imaging of complex targets with multiple motion components.

Addressing the accuracy and stability issues of target tracking in radar network systems, an improved Fuzzy C-Means (FCM) clustering track fusion algorithm is proposed. The method maps local tracks into a fuzzy sample space and introduces sensor measurement quality factors into the objective function. By integrating Interacting Multiple Model (IMM) filtering, information entropy is employed to quantify the uncertainty and reliability of local data, enabling adaptive non-linear weight allocation. Unlike traditional FCM, this approach effectively identifies and suppresses low-quality measurement sources without requiring precise prior noise statistics. Simulation and experimental results demonstrate that the proposed algorithm significantly enhances fusion precision and global stability, particularly during target maneuvers, providing a robust solution for multi-radar cooperative surveillance.

This study characterises the identifiability limits of individual insulation admittances in an ungrounded network from phase-to-earth voltage data alone. The Millman nodal balance, recast as a homogeneous real-valued regression, is analysed through singular value decomposition (SVD). The Millman ratio is scale-invariant, so the full admittance vector cannot be uniquely recovered. The SVD exposes instead a near-null subspace whose dimension and stability reflect the insulation state. A two-stage gate on zero-sequence excitation and angular diversity selects which observation windows enter the regression, and five spectral indicators are extracted from the admitted set: effective rank, null-space dimension, conditioning, tail-gap separation, and informative-window share. On a single feeder whose phase-A insulation resistance is swept across several decades, the tail gap widens, the admitted-window fraction grows and the near-null subspace progressively captures the true admittance direction. The onset shifts earlier under healthy-pair symmetry and requires at least two non-triplen harmonics. Cable attenuation, voltage symmetrisation and source-side zero-sequence contamination each impose hard limits that longer records alone do not remove. The framework is diagnostic and refuses to claim identifiability from data that lack the requisite structure.

A compact, broadband, two-way Wilkinson power divider/combiner is presented, created in 0.18-𝜇m CMOS for use between 60 and 90 GHz. The suggested design uses stacked M5–M6 slow-wave trans- mission lines to shorten the physical quarter-wave length without need- ing extra matching capacitors, which allows for a condensed, meander- ing arrangement. Because of the strong broadside coupling between the stacked conductors, the suggested line has a phase constant about 43% greater than the standard structure at 75 GHz. This lets the necessary 90◦ electrical length be achieved with a considerably shorter physical line. Rather than using M1 as a direct shielding ground, a recurring M1 structure is put under the signal lines as an electromagnetic field-control layer, while the actual RF ground reference is made by the conductive ground under the substrate. Furthermore, an RC isolation network offers frequency-shaped odd-mode damping for wideband isolation. A back- to-back setup is used for precise on-wafer characterization. The mea- surements show an extra insertion loss of 0.5 dB at 75 GHz, a return loss exceeding 20 dB at all ports, and an isolation greater than 34 dB from 60 to 90 GHz within a small 0.086 mm2 core area. The suggested structure works well for small mmWave CMOS front-end uses.

The flexible direct current traction power supply system (DC-TPSS) is a nonlinear system involving kinematics and electrics. Automatic train supervision (ATS) systems are prone to positioning errors and integrate supervision and control systems (ISCS) tend to have missing electrical data of train, making it challenging to accurately obtain the system state in real time. To address these issues, this paper presents a train positioning method that combines the electrical with kinematic measurements for an accurate estimation using a Kalman filter. By solving the real-time power equations for each interval, this method generates pseudo-measured values for the train, addressing the lack of train information in ISCS. Subsequently, a measurement matrix and equations are dynamically generated from the real-time state of the network topology. The weighted least squares (WLS) method is used to filter out bad data for accurate estimates. a flexible DC-TPSS model is finally constructed to verify the accuracy of the algorithm, with the algorithm embedded in the ISCS for in-loop data validation. The results show that the proposed method achieves high-accuracy estimations of the train electrical quantities and position, with significantly reducing variance in the voltage estimation results.

Low Earth orbit (LEO) satellite-assisted mobile edge computing (MEC) is a promising solution for supporting computation-intensive and delay-sensitive services in satellite-terrestrial integrated networks (STINs). In this letter, we study a joint task offloading and computing resource allocation problem in LEO satellite MEC systems, aiming to minimize a weighted sum of system energy consumption and execution delay under resource constraints. The formulated problem jointly optimizes user-satellite association, task offloading ratio, and satellite CPU allocation, and is a mixed-integer nonconvex optimization. To efficiently solve it, we develop a low-complexity alternating optimization framework based on block coordinate descent, where each subproblem admits an efficient solution. Simulation results demonstrate that the proposed scheme consistently outperforms baseline schemes in terms of both energy efficiency and latency.

This manuscript presents a 12-bit 100-MS/s nonbinary successive approximation register (SAR)-assisted analog-to-digital converter (ADC) implemented in a 65-nm CMOS process. To achieve an optimal trade-off among conversion speed, power consumption, and chip area, this work proposes key design techniques, including a nonbinary capacitive digital-to-analog converter (CDAC) with multi-reference and Vcm-based switching, a coarse-fine comparator integrated with on-chip offset calibration, and optimized dynamic asynchronous loop (DAL) and dynamic SAR logic (DSL). Measurement results demonstrate that at 100 MS/s, the ADC achieves a peak effective number of bits (ENOB) of 10.11 bits, a peak signal-to-noise and distortion ratio (SNDR) of 62.64 dB, and a Nyquist-Walden figure of merit (FoM) of 21.87 fJ/conversion-step. The core area is merely 0.034 mm², with a total power consumption of 2.4 mW under a 1.0-V supply.

In vegetation-rich areas, tree-line flash faults pose challenges for protection systems due to their high-impedance characteristics, including low fault current amplitude and high harmonic content. To address detection difficulties, this study employs experiments and modeling to explore the relationship between arc characteristics and environmental factors. A controllable test platform was built to analyze the effects of branch moisture content, discharge gap, and tree species on arc electrical parameters. Based on experimental observations of a significant arc resistance transition near 100 Ω, a mechanism-guided, piecewise adaptive Mayr arc model is proposed, where the time constant varies with conductance. Simulation results show that the model accurately reproduces zero-crossing features, amplitude variations, and nonlinear resistance, with an overall error below 8%. This physically clear and environmentally adaptive model provides a foundation for developing high-sensitivity protection technologies for tree-line flash faults under diverse global conditions.

A two-stage learning framework based on a long-range structure-enhanced masked autoencoder is developed for few-shot radar specific emitter identification (SEI). To improve the modeling of long-range fingerprint structures and the discriminative stability of embeddings under limited labeled samples, a self-supervised masked reconstruction task is first constructed for one-dimensional axially integrated bispectrum (AIB) sequences to learn structural priors from unlabeled data. A ResLKA-TS encoder, consisting of a residual backbone, large-kernel attention, and soft-threshold shrinkage, is then employed to enhance the representation of distributed fingerprint structures. In the few-shot fine-tuning stage, center loss is further introduced to improve intra-class compactness and inter-class separability in the embedding space. Experiments on a measured dataset collected from eight ADALM-PLUTO devices of the same model show that, at an SNR of 20 dB, identification accuracies of 88.00\% and 91.87\% are achieved under 10-shot and 15-shot settings, respectively, exceeding those of asymmetric masked autoencoder (AMAE) by 5.12 and 4.69 percentage points. The method also maintains advantages under low-shot and different SNR conditions.

Vision-Language-Action (VLA) models condition robot actions on natural language, yet sensitivity to instruction wording has not been characterised. This letter evaluates OpenVLA-7B on three manipulation tasks, comparing action differences from synonymous rephrasing (e.g., ”put” vs. ”place” vs. ”set”) against differences from specificity variation (brief vs. step-by-step). Across 5 scenes per task with balanced comparisons (n = 50 pairs each), phrasing sensitivity is task-dependent: one task shows significantly larger phrasing than specificity differences (1.6x, p = 0.018), one shows no difference (p = 0.957), and one trends in the opposite direction (p = 0.092). In aggregate, phrasing and specificity produce comparable action differences (p = 0.395). Both exceed the stochastic noise floor by 2-4x. The results indicate that VLA instruction sensitivity is real but task-specific, and that deployment robustness cannot be assumed from single-task evaluation.

Manual grading of Ribbed Smoked Sheets (RSS) is labour-intensive and experience-dependent, which can lead to inconsistent grade assignment and reduced transparency in quality evaluation. This paper presents a low-cost machine-vision inspection system for full-spectrum RSS grading (RSS1–RSS5) that integrates standardised image acquisition and HSV-based colour representation with a lightweight convolutional neural network (CNN) classifier. Images were captured using a custom-built acquisition setup with fixed illumination and camera geometry to improve repeatability across samples. To enhance grading-relevant cue representation, RGB images were transformed into the Hue–Saturation–Value (HSV) colour space, separating chromatic information from brightness prior to classification. Using a balanced dataset of expert-labelled images, the HSV-based system achieved 91.87% classification accuracy, outperforming an RGB baseline (87.81%). Confusion-matrix analysis indicated strong performance across all five grades, with remaining errors concentrated between visually similar adjacent grades. The results demonstrate the feasibility of combining low-cost standardised illumination with HSV-based colour processing and lightweight learning for practical RSS inspection and grading.

Mental health disorders represent a growing global health burden, contributing significantly to reduced quality of life, decreased productivity, and escalating healthcare costs. Recent advancements in wearable electronic devices (WEDs), and sensor technologies have created new opportunities for continuous, unobtrusive monitoring of physiological and behavioral indicators associated with mental health conditions. Concurrently, machine learning (ML) techniques have emerged as powerful tools for transforming raw multimodal wearable data into meaningful clinical insights. This review presents a comprehensive analysis of the integration of wearable technologies and ML for mental health monitoring. It provides an overview of wearable sensing technologies, and bio signal analysis that are commonly used for mental health assessment and discusses their relevance to autonomic nervous system responses linked to psychological states. The review further analyzes traditional and neural network based learning approaches, highlighting their roles in feature extraction, classification, and prediction of mental health conditions such as stress, depression, emotion dysregulation, bipolar disorder, and schizophrenia. The review concludes by identifying key research gaps and outlining future directions toward scalable, clinically reliable, and ethically responsible ML-enabled wearable systems for mental healthcare.

Addressing the demand for clean heating transition in northern rural areas, this study tackles the challenges of coordinating source-load bilateral uncertainty from high-proportion electric heating and low-carbon operation. It constructs an optimal scheduling model for a rural IES incorporating P2G and CCS coupling. Environmental costs are quantified via a tiered carbon trading mechanism, while carbon resource recycling and methane production are achieved through CCS-P2G coupling. On this basis, a two-stage "day-ahead robust optimization - intra-day rolling adjustment" strategy is proposed. The day-ahead layer employs robust optimization to handle uncertainties in wind and PV power fluctuations, while the intra-day layer performs minute-level rolling corrections based on MPC. Finally, case simulations demonstrate that the proposed electric heating model incorporating tiered carbon trading and carbon capture effectively guides low-carbon operation, providing theoretical support and methodological reference for the planning and operation of rural clean energy systems.

To evaluate the impact of digital interventions used to support care transitions on health outcomes and hospital costs compared with usual clinical care for patients with chronic diseases, we conducted a systematic review and meta-analysis. A systematic search of Cochrane, ACM, Embase, MEDLINE, and PsycINFO was performed from database inception until October 6, 2023. Meta-analysis was conducted for hospitalization outcomes, mortality rates, and cost effectiveness, while quality of life and qualitative outcomes were summarized narratively. The review included 25 studies reporting outcomes for 29,724 patients, with 14 studies eligible for meta-analysis. The meta-analysis showed that digital interventions were associated with small but statistically significant reduction in hospitalization rates (SMD = −0.07, 95% CI −0.13 to −0.01, P = .02) and a statistically significant reduction in mortality (SMD = 0.70, 95% CI 0.58 to 0.85, P = .0003), favoring digital interventions. A positive trend was also observed for improvements in quality of life and reductions in care costs, although evidence was not conclusive. Digital supported transitional care interventions appear to improve clinical outcomes, reduce costs, and enhance quality of life. Future research should focus on digitalizing established care transition models and evaluating technology’s added value for effectiveness and user satisfaction.

In offshore wind farm systems, the submarine cable topology architecture consists primarily of two core components: collector submarine cables and transmission submarine cables, which undertake the functions of power collection and transmission, respectively. For the collector submarine cables, this study established a COMSOL electromagnetic-thermal coupling model to calculate the structural losses and ampacity characteristics of the cable layers under multi-circuit laying conditions. The analysis shows that losses are directly related to wind turbine current and losses in the direct-buriedsection. For the transmission submarine cables, electromagnetic-thermal-fluid multiphysics coupling models were constructed for both direct-buried and J-tube laying conditions. The study systematically investigated the influence mechanisms of ambient temperature, wind speed, and solar radiation intensity on ampacity, and established a multivariate linear regression prediction model for ampacity under environmental constraints in the J-tube section.

Electricity markets are experiencing increasing volatility due to the growing integration of renewable energy, which results in more complex planning and decision-making and highlights the importance of accurate forecasting. Medium-term load forecasting poses challenges, such as lead-time-dependent error propagation, evolving seasonal effects, and cross-market structural heterogeneity. This paper evaluates three deep learning architectures as base learners: Long Short-Term Memory, Gated Recurrent Unit, and Convolutional Neural Network integrated with Long Short-Term Memory. Given the lead-time-dependent performance variability of individual models, a lead-time-conditioned convex stacked ensemble framework is proposed. The ensemble learns horizon-specific blending weights from out-of-fold predictions, constrained through a Softmax transformation to ensure a convex combination of base forecasts. The framework is evaluated using rolling weekly forecasts on datasets from Spain, France, Greece, and Portugal. The results show that the proposed ensemble model consistently outperforms the standalone models. It reduces the average error considering all error metrics and countries by about 10%. Furthermore, it provides improved forecast stability across lead times, adapts to different market conditions, and offers reliable predictions for medium-term load forecasting.

This paper presents a fully-integrated frequency-modulated continuous-wave (FMCW) transceiver chip operating from 55 to 66 GHz, fabricated in a 65 nm silicon-on-insulator (SOI) CMOS process. The chip incorporates a wide-tuning-range voltage-controlled oscillator (VCO) with four digitally selectable states, a driver amplifier, a power amplifier, a low-noise amplifier, and passive mixers, providing a compact solution for high-resolution millimeter-wave radar and short-range communication systems. An innovative hybrid digital-analog VCO tuning architecture, combined with co-optimized transmit/receive front-end design, enables robust noise and linearity performance within a small core area. Measurement results demonstrate a transmitter saturated output power of 12 dBm and an output return loss better than 8 dB across the band. The receiver achieves 10 dB of gain, a 9.8 dB noise figure, and an input 1‑dB compression point of ‑18 dBm. The integrated VCO exhibits a phase noise better than –87.71 dBc/Hz at 1 MHz offset.