Communications engineering最新文献

Silicon-based technologies have been researched extensively over the past few decades, but one ongoing problem has been bringing these technologies into the sub-THz regime. For wireless communications, these bands exhibit potential for massive improvements to bandwidth, data rates, etc., but overcoming the limits of operating frequency to allow devices to enter the sub-THz regime has proven challenging when designing working systems. Despite this, researchers have been developing novel ways to construct transceiver systems above 200 GHz. In this article, we showcase interesting designs that push the envelope towards more powerful, faster, and more useful silicon-based transceiver systems. We discuss transmitter systems grouped by modulation schemes as well as incoherent and coherent receiver systems. This allows us to point out the specific difficulties seen throughout these works and describe the direction needed to improve these systems. Finally, we discuss the future direction and application of silicon-based wireless communication systems as they move towards sub-THz regions.

Superconducting motors offer high power density, compactness, and efficiency for hydrogen-powered cryo-electric aircraft, but AC operation in cryogenic temperatures produces thermal losses that must be estimated accurately and rapidly at the design stage to optimize efficiency, minimize cryogenic heat load, and maximize specific power density. Traditional modeling approaches fall short-Finite-element is too slow/costly for system-level models, analytical models and look-up tables lack accuracy/flexibility, and earlier intelligent models gave only cycle-averaged (static) losses. Here we demonstrate AI can rapidly and accurately predict dynamic AC losses for superconducting propulsion motors. Using a large dataset of motor configurations, our AI-driven approach both predicts cycle-averaged and time-dependent morphology of instantaneous AC loss waveforms across various operating conditions and generalizes to unseen designs. Integrated into system-level model-based design, these AI-surrogate models enable rapid model trials, compliance checks, and the discovery of integration issues within simulated environments before propulsion motor deployment. Our deep learning-based model achieves a prediction time of less than 9 ms with a 99.97% accuracy (R²), making it suitable for system-level modeling of electric powertrains in hydrogen-powered cryo-electric aircraft. Furthermore, we benchmarked 14 AI and 2 mathematical fitting techniques for estimating average AC losses, providing comparative performance analysis. The results highlight that AI-based surrogate models enable high-accuracy, low-latency loss predictions to achieve optimal performance in superconducting propulsion motors in aircraft powertrain design.

Mechanical stress during the cycling process notably impacts the performance of lithium-ion batteries (LIBs), making it crucial to accurately monitor stress generation and propagation during battery operation. Traditional electrochemical-mechanical models are limited to the particle and electrode scales, and their parameter identification relies solely on voltage. Here, a multi-scale electrochemical-mechanical-thermal modelling framework with non-destructive parameter identification capabilities is proposed. This numerical model couples electrochemical reactions with thermal effects and links particle-scale strain to electrode-scale displacement. Diffusion-induced stress (DIS) is selected as a key indicator, combined with voltage, to analyze the sensitivity of 23 parameters. A voltage-strain multi-objective parameter identification strategy based on the Pareto front is employed to determine the key parameters. The framework demonstrates high fidelity, with the mean absolute percentage error for voltage and strain predictions below 1% and 3.6%, respectively. This work enables high-fidelity simulation of multi-physics behavior, provides an effective method for calibrating key parameters, and holds potential for establishing a reliable digital twin of LIBs.

In the past decades, active exoskeletons have been dedicated to reducing human effort, in particular to assist workers in occupational environments. However, this approach does not promote the learning of more ergonomic postures by workers, which is critical for the long-term prevention of musculoskeletal disorders. Alternatively, we propose the use of exoskeletons as biofeedback systems, generating task-relevant perturbations guiding users towards ergonomic postures. To test this approach, participants performed reach-to-hold movements towards a redundant target, allowing multiple final postures. We then introduced vibrations with posture-dependent intensity, generating a sensorimotor disturbance that canceled out either above or below each participant's nominal preferred posture. Interestingly, participants adapted to minimize the vibrations, whether it increased or decreased the gravity efforts, and retained the novel posture when it induced lower effort. Finally, all participants significantly reduced effort post-exposure. This work demonstrates the feasibility of using exoskeletons as biofeedback systems to improve posture, paving the path for applications in musculoskeletal disorders prevention.

Single-molecule localization microscopy achieves nanometer-scale resolution but is compromised by sample drift during image acquisition. Here we present reinforced optical cage systems, a novel approach that eliminates drift at its mechanical source rather than correcting it through complex image post-processing or fiducial markers. Reinforced optical cage systems employ perforated optomechanical components interconnected by tungsten-steel rods in a design proven by mechanical stability simulations. Our bench-top microscope, built with reinforced optical cage systems, demonstrated exceptional three-dimensional stability, with mean cumulative lateral drift of approximately 5 nanometers over 2 h in widefield fluorescence microscopy and 11-16 nanometers over 15 min in single-molecule localization microscopy, free from measurable axial drift. This development allows super-resolution microscopy to reach its full resolution without the necessity of sample drift correction, offering a straightforward, cost-effective, low-maintenance, and readily accessible solution to high-performance super-resolution microscopy. By addressing the fundamental issue of mechanical instability, reinforced optical cage systems enable improved precision instrumentation for the broader scientific and engineering community.

Buckling is a common failure mode in lattice structures, limiting their use in some applications. The tendency of a strut to buckle is related to the local nodal connectivity. In this work, we introduce a pneumatic actuation strategy to actively tune the mechanical behavior of lattice structures by locally reconfiguring their effective nodal connectivity. By selectively inflating pneumatic actuators embedded in the lattice into spatial patterns with varying levels of connectivity, we demonstrate a method to modulate mechanical properties, including stiffness and buckling response. The most reinforced pattern can lead to 121.6% improvement in buckling strength relative to the regular lattice itself. Additionally, the post-buckling behavior of pneumatically controlled lattices can be programmably tuned by varying the input air pressure signals. The pneumatically controlled lattices reduced the peak acceleration by 50.9%, demonstrating enhanced impact mitigation capability. These results show that pneumatic actuation provides a versatile approach to enhancing structural performance under both static and dynamic loading. Since this strategy does not rely on multi-material interfaces or specific cell topologies, it can be broadly applied to optimize a wide range of lattice architectures.

Stress concentrations at geometric irregularities such as reentrant corners make it challenging to efficiently simulate localized plastic deformation in engineering materials. Fully nonlinear models capture these effects accurately but are computationally costly, whereas simplified elastic analyses neglect important nonlinearities. Here, we present NeuberNet, a Multi-Task Nonlinear Manifold Decoder that learns mappings between far-field displacement boundary conditions from low-fidelity elastic simulations and the corresponding high-resolution stress and strain fields derived from elastic-plastic axisymmetric solid mechanics, under assumptions of small-scale plasticity and bilinear isotropic hardening. NeuberNet serves as a data-driven implementation of the substructuring principle, designed to model complex geometries by activating plastic behavior only near stress raisers where nonlinearities arise. We provide guidelines for mesh resolution in low-fidelity simulations, demonstrate NeuberNet's ability to identify violations of the small-scale plasticity assumption, and assess its robustness to nonlinear hardening laws. We also show that NeuberNet generalizes to 3D problems with axisymmetric geometries and non-symmetric boundary conditions. Overall, NeuberNet provides a reliable and computationally efficient framework for small-scale plasticity analysis.

Extremely low frequency (ELF, 3-30 Hz) signals possess strong cross-medium communication capabilities, making them particularly well-suited for underground and underwater environments. However, traditional low-frequency (LF) transmission systems are large and inefficient, posing significant limitations in practical applications. In recent studies, mechanical antennas have been explored to generate LF signals, but current approaches rely on bulky equipment with limited range, making them unsuitable for personal use or integration into small unmanned devices. To address this challenge, this study introduces a flexible, magnet-based miniaturized LF mechanical antenna, fabricated using 3D printing. The antenna consists of a macro-fiber composite layer and a flexible permanent magnet film, and features an extremely compact volume (<6.8 cm³) and low weight (<50 g). It is also highly flexible, allowing for easy integration into diverse applications. Its transmitted signal can reach 60 m before the magnetic field strength attenuates to 1 pT. Mounted on an unmanned aerial vehicle (UAV), the antenna facilitates reliable communication between quadruped robots operating outside caves and aerial robots located deep within cave interiors, where high-frequency (HF) signals cannot penetrate. This study demonstrates robust LF cross-medium communication between UAV and ground robots in cave environments, paving the way for unmanned collaboration in scenarios inaccessible to HF wireless signals.

This work investigates the thermal stability of Cu-Cu bonding using a thin Ag passivation layer in applications targeting advanced packaging. Conventional Cu-Cu bonding often requires elevated temperatures (≥250 °C) that can exacerbate thermal stress and limit process flexibility, making multi-chip stacking more challenging. By introducing a 3 nm Ag passivation layer, we demonstrate reliable bonding at lower temperatures with improved durability against high-humidity and high-temperature environments, as confirmed by both Highly Accelerated Stress Tests (HAST) and burn-in measurements. In-situ transmission electron microscopy (TEM) and 4D-STEM strain mapping reveal that Ag diffusion along Cu grain boundaries not only retards abnormal grain growth but also reduces interfacial void formation at elevated temperatures. These enhancements collectively maintain a stable interface and superior mechanical strength relative to that for non-passivated Cu-Cu bonding. The results highlight the importance of metal passivation in enabling low-temperature Cu-Cu bonding technologies with robust thermal stability, providing the feasibility for next-generation advanced packaging platforms.

Dynamic weighing systems are crucial for bridge safety, yet current weighing systems are either too costly for widespread urban use or unable to track vehicles reliably in dense traffic. Here we present a video-strain inverse estimation system that combines computer vision with structural strain sensing to measure dynamic vehicle loads on bridges. The method employs a lightweight deep learning detector to recognize vehicle features, a multi-object tracking model to capture trajectories and lane positions, and a developed analytical algorithm to estimate vehicle weight from measured strains. We validated the system with field data from a heavily trafficked urban bridge. The system achieves vehicle recognition with near-baseline accuracy while using only 17% of the original model parameters and running 1.72 times faster, identifies lanes with complete accuracy with a missed detection rate of just 0.56%, and estimates total vehicle weights within 2% error. This low-cost and reliable approach advances intelligent bridge monitoring and supports digital twins of critical urban infrastructure.