Communications engineering最新文献

Event-based imaging is at the forefront of high-speed sensing applications due to the low latency of asynchronous detection and low data volume. Conventionally, events used in this form of sensing correspond to changes in intensity on the microsecond scale, this inherently makes the approach incompatible with scenarios where single photon detection is required such as single photon Light Detection and Ranging (LiDAR), and low-light-level imaging. Here we propose a new imaging modality which is driven instead by events generated from the detection of individual photons. We use a form of single-pixel imaging in which information from a 3-Dimensional scene is encoded entirely in the time-of-arrival of the photons such that each detected photon can be used to update an estimation of all transverse positions simultaneously. The image reconstruction is performed by a Spiking Convolutional Neural Network (SCNN) which has a natural complementarity with single photon detection that allows the scheme to run fully asynchronously and be driven by the detection of each individual photon. Our approach for processing LiDAR information asynchronously, driven by each detected photon, has the potential for minimal latency 3D imaging and sensing even in weak light conditions with applications in high speed target detection and robotics.

Transcranial magnetic stimulation combined with electroencephalography enables direct assessment of cortical excitability and connectivity via stimulation-evoked responses. While passive electrodes remain the gold standard, they require extensive preparation and are sensitive to contact quality, whereas active electrodes are easier to use but limited by increased height and larger decay artifacts. Here we introduce an ultra-thin active electrode system combined with hardware- and software-based artifact suppression. We collected electroencephalographic data from 10 healthy adults in Austria in 2024, recording brain responses with both active and passive electrodes during stimulation of the left primary motor cortex. We analyzed early and late responses for similarity and amplitude variability and found high consistency between electrode types, with stable waveforms after 20-30 trials. Response amplitudes did not differ significantly between electrode types. These findings demonstrate that active electrode systems can provide reliable and efficient recordings, supporting their broader use in stimulation-electroencephalography research.

Automatic object detection is increasingly used in the medical field to enhance clinical workflows before, during, and after diagnosis of various conditions. One example is prostate detection and prostate volume estimation, which can aid in triaging patients for prostate cancer through risk-stratification using prostate-specific antigen density. In this paper, a baseline prostate detection framework is presented, highlighting that current state-of-the-art object detection models can detect the prostate in difficult to interpret surface-based ultrasound images. A 5-fold cross-validation study returned intersection-over-union, precision, recall, F1, and average-precision values above $\mathrm{0.7}$ with real-time capabilities possible. Additionally, a simple size calculation based on the detection results showed high correlation with ground truth measurements, with Pearson Correlation Coefficients ranging from $\mathrm{0.55}$ to $\mathrm{0.84}$ for prostate volume estimates. These findings will contribute to the development of a real-time prostate detection and size estimation platform for prostate cancer risk-stratification to reduce unnecessary biopsy rates in healthcare systems.

Low-grade thermal gradients in marine environments represent an underexploited energy source for autonomous sensing and monitoring. Converting such small temperature differences into usable electrical power remains a key challenge for ocean-deployed systems. We present a deployable thermomagnetic generator thoroughly characterized for marine-relevant energy harvesting. The device powers an internet-connected sensor and harvests ultra-low temperature differences akin to those at the ocean surface. It draws heat from water and rejects it to ambient air, operating optimally at a temperature difference (ΔT) of ~7.5 °C. Laboratory prototypes generated up to 6.7 mW at ΔT ~ 10 °C with gentle airflow (~1 m s^-1). A separate controlled wave-tank demonstration validated stable operation and sensor powering under marine-like boundary conditions. Given its voltage and power margins, the generator could sustain multiple sensor nodes. Scalability and material assessments identify modular deployment and non-rare-earth alternatives as pathways toward practical marine energy harvesting and low-grade waste-heat recovery.

The primary challenge in active object tracking (AOT) lies in maintaining robust and accurate tracking performance in the complex physical scenarios. Existing end-to-end frameworks based on deep learning and reinforcement learning often struggle with high computational costs, data dependency, and limited generalization, hindering their performance in practical applications. Although embodied intelligence (EI) is promising to enable agents to learn from physical interactions, it cannot tackle severe anomalies happened in the complex scenarios. In order to address this issue, here we propose a novel embodied learning method, called the Cognitive Embodied Learning (CEL), which is inspired by the dual decision-making system of the human brain. The CEL can dynamically switch between normal tracking and anomaly handling modes, supported by specialized modules including the anomaly cognition module, the rule reasoning module, and the anomaly elimination module. Moreover, we further introduce the categorical objective function to address function non-measurability and data confusion caused by severe anomalies. Extensive unmanned aerial vehicle anomaly active target tracking experiments in both simulated and real-world scenarios demonstrate the superior performance of our method. Compared to the state-of-the-art methods, the CEL achieves a 361.4% increase in the success rate and a 54.4% improvement of the task completion efficiency, which highlights the potential of CEL to advance the field of AOT and open new avenues for more robust and intelligent tracking systems in the challenging environments.

Swarms of drones offer increased sensing aperture. When these swarms mimic natural behaviors, sampling is enhanced by adapting the aperture to local conditions. We demonstrate that this enables detection and tracking of heavily occluded targets. Object classification in conventional aerial images generalizes poorly due to occlusion randomness and is inefficient even under minimal occlusion. In contrast, anomaly detection applied to synthetic-aperture integral images remains robust in dense vegetation and independent of pre-trained classes. Our autonomous, centralized swarm searches for unknown or unexpected occurrences, tracking them while continuously adapting its sampling pattern to optimize local viewing conditions. We achieved average positional accuracies of 0.39 m with average precisions of 93.2% and average recalls of 95.9%. Here, adapted particle swarm optimization considers detection confidences and predicted target appearance. We present a new confidence metric that identifies the most abnormal targets and show that sensor noise can be effectively included in the synthetic aperture process, removing the need for costly optimization of high-dimensional parameter spaces. Finally, we provide a hardware-software framework enabling low-latency transmission and fast processing of video and telemetry data. Although our field experiments involved six drones, ongoing technological advances will soon enable larger, faster swarms for military and civil applications.

Free-space optical communication with high transmission bandwidth and small antenna size has been progressively deployed for ground-air-space communications in recent years. However, current mitigation methods, including adaptive optics and transmitter-side beam shaping, are often limited by design complexity and insufficient bandwidth. Here, we propose a receiver-side wavefront correction scheme based on optical pin beam-a ring-shaped, self-healing beam structure-formed via a static phase mask in front of the coupling lens. Unlike traditional optical pin beam methods requiring transmitter-side modulation, our design proactively reshapes the aberrated receiving beam into a stable optical pin beam with extended Rayleigh length, thereby improving mode matching and enhancing coupling resilience under turbulence. In a kilometer-scale outdoor experiment, we demonstrate a 100 Gbps free-space laser link, with coupled power stability increased by 26% and bit error rate decreased by up to 2 orders of magnitude compared with a Gaussian receiver. This receiver-side-only solution simplifies system architecture, ensures high-power compatibility, and offers a low-cost and scalable pathway for future optical ground stations, paving the way for ultra-long-distance, high-speed, and compact Free-space optical communication systems.

Efficient production of monoclonal antibodies (mAb) using Chinese Hamster Ovary (CHO) cells is central to pharmaceutical biomanufacturing. The clone selection process traditionally requires lengthy 7-to-14-day assessments to evaluate performance, which extends development timelines. Here we introduce a hybrid Luedeking-Piret Regression model that integrates mechanistic insights with machine learning to more accurately predict mAb yields in fed-batch CHO cultures. Using experimental data from the early growth stages (up to day 9) of seven (n=7) distinct CHO cultures, the model performed multi-step-ahead forecasting to predict final production. The model predicted monoclonal antibody titers on day 16 with a mean percentage error of 5.85%, correctly selected higher-performing clones in 76.2% of trials from leave-two-out cross-validation and accurately forecasted daily production trajectories from day 10 to day 16. The model's multi-step-ahead forecasting capabilities have the potential to accelerate clone selection, providing the biomanufacturing community with a computationally straightforward algorithm for predicting production yields.

With the continuous integration of advanced information and communication technologies into smart grids, the distribution network is undergoing a digital transformation, making the power distribution system increasingly complex. Edge computing shifts computation from the central control station to the distribution substations, thus enabling true distributed autonomy in power system operations. Taking an edge-computing-based digital substation as an example, this paper proposes a deep neural networks-based voltage regulation strategy for PV-rich distribution networks. However, executing tasks on resource-constrained edge devices faces several challenges, including data flow congestion, the inapplicability of conventional modelling and algorithm, and low computational efficiency. Therefore, we employ a unified weights neural network for Volt-Var control to achieve compression of the network parameters while still achieving differentiated action output. Furthermore, a carefully designed pipeline parallel computing structure is employed to simultaneously perform computations at different levels, further improving computational efficiency. The tested results show that, compared with existing methods, the proposed approach effectively mitigates voltage violations, improves storage efficiency and computational speed, and maintains robust performance under communication failures with partial observation, highlighting its resilience and potential for edge deployment.

Drone flight safety and operational efficiency are challenged in the landing phase, especially in windy conditions. While flight control should be a key focus, flight trajectory also plays a critical role in landing. Inspired by the multiple objectives, wind sensory capability, and skill learning of avian species, this study proposes a reinforcement learning-based trajectory planner for a drone to perform trajectory planning in the wind, aiming to balance multiple landing-related objectives based on onboard wind sensory capability and address the safety-operational efficiency dilemma of a landing site. Through four key experiments, this study demonstrates successful training, balanced landing performance, and strong generalization capability of the trajectory planner. The experiments highlight the importance of velocity sensory capability while indicating that wind sensory capability is less critical to the trajectory planner. The proposed framework with multiple objectives, wind sensory capability, and skill learning can benefit applications such as improving drone performance.