Papers

Reliable anomaly detection in photovoltaic (PV) modules is critical for maintaining solar energy efficiency. However, developing robust computer vision models for PV inspection is constrained by the scarcity of large-scale, diverse, and balanced datasets. This study introduces PV-DDPM, a spatially conditioned denoising diffusion probabilistic model that generates anomalous electroluminescence (EL) images across four PV cell types: multi-crystalline silicon (multi-c-Si), mono-crystalline silicon (mono-c-Si), half-cut multi-c-Si, and interdigitated back contact (IBC) with dogbone interconnect. PV-DDPM enables controlled synthesis of single-defect and multi-defect scenarios by conditioning on binary masks representing structural features and defect positions. To the best of our knowledge, this is the first framework that jointly models multiple PV cell types while supporting simultaneous generation of diverse anomaly types. We also introduce E-SCDD, an enhanced version of the SCDD dataset, comprising 1,000 pixel-wise annotated EL images spanning 30 semantic classes, and 1,768 unlabeled synthetic samples. Quantitative evaluation shows our generated images achieve a Fréchet Inception Distance (FID) of 4.10 and Kernel Inception Distance (KID) of 0.0023 $\pm$ 0.0007 across all categories. Training the vision–language anomaly detection model AA-CLIP on E-SCDD, compared to the SCDD dataset, improves pixel-level AUC and average precision by 1.70 and 8.34 points, respectively.

Satellite missions provide valuable optical data for monitoring rivers at diverse spatial and temporal scales. However, accessibility remains a challenge: high-resolution imagery is ideal for fine-grained monitoring but is typically scarce and expensive compared to low-resolution imagery. To address this gap, we introduce SuperRivolution, a framework that improves river segmentation resolution by leveraging information from time series of low-resolution satellite images. We contribute a new benchmark dataset of 9,810 low-resolution temporal images paired with high-resolution labels from an existing river monitoring dataset. Using this benchmark, we investigate multiple strategies for river segmentation, including ensembling single-image models, applying image super-resolution, and developing end-to-end models trained on temporal sequences. SuperRivolution significantly outperforms single-image methods and baseline temporal approaches, narrowing the gap with supervised high-resolution models. For example, the F1 score for river segmentation improves from 60.9% to 80.5%, while the state-of-the-art model operating on high-resolution images achieves 94.1%. Similar improvements are also observed in river width estimation tasks. Our results highlight the potential of publicly available low-resolution satellite archives for fine-scale river monitoring.

Medical image segmentation using deep learning (DL) has enabled the development of automated analysis pipelines for large-scale population studies. However, state-of-the-art DL methods are prone to hallucinations, which can result in anatomically implausible segmentations. With manual correction impractical at scale, automated quality control (QC) techniques have to address the challenge. While promising, existing QC methods are organ-specific, limiting their generalizability and usability beyond their original intended task. To overcome this limitation, we propose no-new Quality Control (nnQC), a robust QC framework based on a diffusion-generative paradigm that self-adapts to any input organ dataset. Central to nnQC is a novel Team of Experts (ToE) architecture, where two specialized experts independently encode 3D spatial awareness, represented by the relative spatial position of an axial slice, and anatomical information derived from visual features from the original image. A weighted conditional module dynamically combines the pair of independent embeddings, or opinions to condition the sampling mechanism within a diffusion process, enabling the generation of a spatially aware pseudo-ground truth for predicting QC scores. Within its framework, nnQC integrates fingerprint adaptation to ensure adaptability across organs, datasets, and imaging modalities. We evaluated nnQC on seven organs using twelve publicly available datasets. Our results demonstrate that nnQC consistently outperforms state-of-the-art methods across all experiments, including cases where segmentation masks are highly degraded or completely missing, confirming its versatility and effectiveness across different organs.

Objective: Time-domain diffuse optical imaging (DOI) requires accurate forward models for photon propagation in scattering media. However, existing simulators lack comprehensive experimental validation, especially for non-contact configurations with oblique illumination. This study rigorously evaluates three widely used open-source simulators, including MMC, NIRFASTer, and Toast++, using time-resolved experimental data. Approach: All simulations employed a unified mesh and point-source illumination. Virtual source correction was applied to FEM solvers for oblique incidence. A time-resolved DOI system with a 32 $\times$ 32 single-photon avalanche diode (SPAD) array acquired transmission-mode data from 16 standardized phantoms with varying absorption coefficient $\mu_a$ and reduced scattering coefficient $\mu_s'$. The simulation results were quantified across five metrics: spatial-domain (SD) precision, time-domain (TD) precision, oblique beam accuracy, computational speed, and mesh-density independence. Results: Among three simulators, MMC achieves superior accuracy in SD and TD metrics, and shows robustness across all optical properties. NIRFASTer and Toast++ demonstrate comparable overall performance. In general, MMC is optimal for accuracy-critical TD-DOI applications, while NIRFASTer and Toast++ suit scenarios prioritizing speed with sufficiently large $\mu_s'$. Besides, virtual source correction is essential for non-contact FEM modeling, which reduced average errors by > 34% in large-angle scenarios. Significance: This work provides benchmarked guidelines for simulator selection during the development phase of next-generation TD-DOI systems. Our work represents the first study to systematically validate TD simulators against SPAD array-based data under clinically relevant non-contact conditions, bridging a critical gap in biomedical optical simulation standards.

Radiology is essential to modern healthcare, yet rising demand and staffing shortages continue to pose major challenges. Recent advances in artificial intelligence have the potential to support radiologists and help address these challenges. Given its widespread use and clinical importance, chest X-ray classification is well suited to augment radiologists' workflows. However, most existing approaches rely solely on single-view, image-level inputs, ignoring the structured clinical information and multi-image studies available at the time of reporting. In this work, we introduce CaMCheX, a multimodal transformer-based framework that aligns multi-view chest X-ray studies with structured clinical data to better reflect how clinicians make diagnostic decisions. Our architecture employs view-specific ConvNeXt encoders for frontal and lateral chest radiographs, whose features are fused with clinical indications, history, and vital signs using a transformer fusion module. This design enables the model to generate context-aware representations that mirror reasoning in clinical practice. Our results exceed the state of the art for both the original MIMIC-CXR dataset and the more recent CXR-LT benchmarks, highlighting the value of clinically grounded multimodal alignment for advancing chest X-ray classification.

We introduce HAMscope, a compact, snapshot hyperspectral autofluorescence miniscope that enables real-time, label-free molecular imaging in a wide range of biological systems. By integrating a thin polymer diffuser into a widefield miniscope, HAMscope spectrally encodes each frame and employs a probabilistic deep learning framework to reconstruct 30-channel hyperspectral stacks (452 to 703 nm) or directly infer molecular composition maps from single images. A scalable multi-pass U-Net architecture with transformer-based attention and per-pixel uncertainty estimation enables high spatio-spectral fidelity (mean absolute error ~ 0.0048) at video rates. While initially demonstrated in plant systems, including lignin, chlorophyll, and suberin imaging in intact poplar and cork tissues, the platform is readily adaptable to other applications such as neural activity mapping, metabolic profiling, and histopathology. We show that the system generalizes to out-of-distribution tissue types and supports direct molecular mapping without the need for spectral unmixing. HAMscope establishes a general framework for compact, uncertainty-aware spectral imaging that combines minimal optics with advanced deep learning, offering broad utility for real-time biochemical imaging across neuroscience, environmental monitoring, and biomedicine.

The concept of a security index quantifies the minimum number of components that must be compromised to carry out a stealth attack. This metric enables system operators to assess the security risk of each component and implement countermeasures accordingly. In this paper, we introduce a data-driven security index that can be computed solely from input/output data when the system model is unknown. We show a sufficient condition under which the data-driven security index coincides with the model-based security index, which implies that the exact risk level of each component can be identified solely from data. We also provide an algorithm for computing the data-driven security index.

In this paper, we propose a second-order deterministic actor-critic framework in reinforcement learning that extends the classical deterministic policy gradient method to exploit curvature information of the performance function. Building on the concept of compatible function approximation for the critic, we introduce a quadratic critic that simultaneously preserves the true policy gradient and an approximation of the performance Hessian. A least-squares temporal difference learning scheme is then developed to estimate the quadratic critic parameters efficiently. This construction enables a quasi-Newton actor update using information learned by the critic, yielding faster convergence compared to first-order methods. The proposed approach is general and applicable to any differentiable policy class. Numerical examples demonstrate that the method achieves improved convergence and performance over standard deterministic actor-critic baselines.

This paper investigates a fluid antenna system (FAS) where a single-antenna transmitter communicates with a receiver equipped with a fluid antenna (FA) over Rician fading channels. Two channel models are considered to incorporate the correlation among the ports into the fading gains, namely: i) the widely used physical reference port model, implemented by considering the first physical port as the reference port; and ii) the virtual reference port model, considering a common port correlation coefficient and implemented by assuming the presence of a virtual reference port. Assuming that multiple ports among the M available FA ports can be activated, the receiver selects the best K with the highest instantaneous signal-to-noise ratio (SNR) and combines the received signals at the selected ports using maximum ratio combining. For both port models, the statistics of the post-combining SNR are derived using a characteristic function approach, which allows to analyze the outage probabilities (OPs) of the FAS. Additional closed-form lower bounds on the OPs and the asymptotic OPs at high SNR are derived, revealing the diversity order of the FAS to be M. Numerical results validate the analytical framework and demonstrate the interplay of key system parameters on the performance of the considered MRC-based FAS. Specifically, the performance difference between the two reference port models becomes increasingly evident as M, K, and the average SNR of the FAS increase.

Fully digital massive multiple-input multiple-output (MIMO) systems with large numbers (1000+) of antennas offer capacity gains from spatial multiplexing and beamforming, but receivers that scale to these array dimensions face challenges in both channel estimation overhead and digital computation. Long-term beamforming addresses both by projecting the data onto a low-dimensional subspace that can be tracked at a slow time scale from the long-term channel parameters. In this setting, we show how to compute, in closed form, the projection matrix that maximizes a capacity upper bound, using a matrix inverse square root; the same projection is shown to maximize the mean post-projection signal-to-interference-plus-noise ratio (SINR) exactly. Computationally efficient methods are then presented for the matrix computation, realizable with matrix-matrix multiplies and hence amenable to systolic array implementations in hardware. Bounds on the SINR degradation are derived, and ray tracing simulations in a realistic rural uplink setting show a small loss relative to instantaneous minimum mean-square error (MMSE) beamforming when the covariance is accurately estimated. The efficient Gram-domain form of the instantaneous MMSE receiver applies the maximum-ratio combining reduction before an inverse whose dimension is the total number of streams. Against this baseline, the method estimates and refreshes beamforming coefficients three orders of magnitude less often and decouples the real-time path across users. With the conjugate-gradient solve and a rank-one projection, its total arithmetic cost is 2% higher at the ten-user operating point and lower above a stream-dimension crossover that we characterize.

To enable intelligent beam training, a large language model (LLM)-enabled beam training framework is proposed for the pinching antenna system (PASS) in downlink multi-user multiple-input multiple-output (MIMO) communications. A novel LLM-based beam training supervised learning mechanism is developed, allowing context-aware and environment-adaptive probing for PASS to reduce overheads. Both single-user and multi-user cases are considered. 1) For single-user case, the LLM-based pinching beamforming codebook generation problem is formulated to maximize the beamforming gain. Then, the optimal transmit beamforming is obtained by maximum ratio transmission (MRT). 2) For multi-user case, a joint codebook generation and beam selection problem is formulated based on the system sum rate under the minimum mean square error (MMSE) transmit beamforming. The training labels for pinching beamforming are constructed by selecting the beam combination that maximizes system performance from each user's Top-S candidate beams. Based on pretrained Generative Pre-trained Transformers (GPTs), the LLM is trained in an end-to-end fashion to minimize the cross-entropy loss. Simulation results demonstrate that: i) For single-user case, the proposed LLM-enabled PASS attains over 95% Top-1 accuracy in beam selection and achieves 51.92% improvements in beamforming gains compared to conventional method. ii) For multi-user case, the proposed LLM-enabled PASS framework significantly outperforms both the LLM-based massive MIMO and conventional PASS beam training, achieving up to 57.14% and 33.33% improvements in sum rate, respectively.

We study virtual energy storage services based on the aggregation of EV batteries in parking lots under time-varying, uncertain EV departures and state-of-charge limits. We propose a convex data-driven scheduling framework in which a parking lot manager provides storage services to a prosumer community while interacting with a retailer. The framework yields finite-sample, distribution-free guarantees on constraint violations and allows the parking lot manager to explicitly tune the trade-off between economic performance and operational safety. To enhance reliability under imperfect data, we extend the formulation to adversarial perturbations of the training samples and Wasserstein distributional shifts, obtaining robustness certificates against both corrupted data and out-of-distribution uncertainty. Numerical studies confirm the predicted profit-risk trade-off and show consistency between the theoretical certificates and the observed violation levels.

There is significant recent interest in designing new modulation schemes for doubly-selective channels with large delay and Doppler spreads, where legacy modulation schemes based on time-frequency signal representations underperform. Multiple modulation schemes, e.g., in the delay-Doppler, chirp, time-sequency, and other domains, have been proposed in the literature for this purpose, with varying implementation details. In this letter, we establish that all previously proposed modulation schemes for doubly-selective signaling are instances of a single family of complex Hadamard-modulated pulse trains. When the delay and Doppler spread of the doubly-selective channel is limited to a certain support, all modulation schemes in this waveform family offer equivalent, full diversity achieving performance with no symbol fading and low channel estimation overhead. The existence of this waveform family also enables flexible, multi-waveform co-existence – allowing a common transceiver architecture to generate multiple waveforms in the family, that may each be flexibly allocated to different users and services.

Smart radio environments (SREs) enhance wireless communications by allowing control over the channel. They have been enabled through surfaces with reconfigurable electromagnetic (EM) properties, known as reconfigurable intelligent surfaces (RISs), and through flexible antennas, which can be viewed as realizations of SREs in the EM domain and space domain, respectively. However, these technologies rely on electronically reconfigurable or movable components, introducing implementation challenges that could hinder commercialization. To overcome these challenges, we propose a new domain to enable SREs, the frequency domain, through the concept of movable signals, where the signal spectrum can be dynamically moved along the frequency axis. We first analyze movable signals in multiple-input single-output (MISO) systems under line-of-sight (LoS) conditions, showing that they can achieve higher average received power than quantized equal gain transmission (EGT). We then study movable signals under non-line-of-sight (NLoS) conditions, showing that they remain effective by leveraging reflections from surfaces made of uniformly spaced elements with fixed EM properties, denoted as fixed intelligent surfaces (FISs). Analytical results reveal that a FIS-aided system using movable signals can achieve up to four times the received power of a RIS-aided system using fixed-frequency signals.

The massive scale of Internet of Things (IoT) connectivity expected in 6G networks raises unprecedented challenges in energy use, battery waste, and lifecycle sustainability. Current cellular IoT solutions remain bound to the lifetime of underlying network generations and rely on billions of disposable batteries, creating unsustainable economic and environmental costs. This article proposes generation-agnostic zero-energy devices (XG-ZEDs), a new class of backscatter based IoT devices that are battery-less, spectrum-agnostic, and future-proof across successive network generations. XG-ZEDs exploit existing ambient wireless signals for communication, sensing, and localization, transforming infrastructure and user devices into universal enablers of ultra-low-power connectivity. We review architectural classifications, communication protocols, network integration, and representative applications such as sensing, localization, and radio-SLAM, while outlining the challenges ahead.

This paper adapts a Multiple-Model Coding (MMC) approach for sampled electrical signal waveforms to satisfy reconstructed signal quality constraints. The baseline MMC approach consists of two stages processing vectors of Voltage and Current Signal (VCS) of constant size and producing bitstreams of constant rate but varying quality. In the proposed approach, the parametric model and the rate allocated to the first stage, as well as the residual compression method of the second stage and its associated rate, are jointly optimized to achieve a target distortion of the reconstructed signal. Three approaches are proposed. An exhaustive search serves as a baseline for comparison. Then, an approach involving a Golden Section search is exploited to determine the rate of the first stage with reduced complexity. Finally, rate-distortion models of the compression efficiency for each model in the first stage are employed to obtain a subset of promising models in the first stage and reduced-size search intervals for the rate selection in both stages. Simulation results demonstrate that the proposed reduced-complexity MMC approach reduces the rate for a given distortion constraint compared to state-of-the-art solutions for VCS with equivalent complexity.

Ultra-low-field (ULF) MRI promises broader accessibility but suffers from low signal-to-noise ratio (SNR), reduced spatial resolution, and contrasts that deviate from high-field standards. Image-to-image translation can map ULF images to a high-field appearance, yet efficacy is limited by scarce paired training data. Working within the ULF-EnC challenge constraints (50 paired 3D volumes; no external data), we study how task-adapted data augmentations impact a standard deep model for ULF image enhancement. We show that strong, diverse augmentations, including auxiliary tasks on high-field data, substantially improve fidelity. Our submission ranked third by brain-masked SSIM on the public validation leaderboard and fourth by the official score on the final test leaderboard. Code is available at https://github.com/fzimmermann89/low-field-enhancement.

Safety verification of dynamical systems via barrier certificates is essential for ensuring correctness in autonomous applications. Synthesizing these certificates involves discovering mathematical functions with current methods suffering from poor scalability, dependence on carefully designed templates, and exhaustive or incremental function-space searches. They also demand substantial manual expertise–selecting templates, solvers, and hyperparameters, and designing sampling strategies–requiring both theoretical and practical knowledge traditionally shared through linguistic reasoning rather than formalized methods. This motivates a key question: can such expert reasoning be captured and operationalized by language models? We address this by introducing an LLM-based agentic framework for barrier certificate synthesis. The framework uses natural language reasoning to propose, refine, and validate candidate certificates, integrating LLM-driven template discovery with SMT-based verification, and supporting barrier-controller co-synthesis to ensure consistency between safety certificates and controllers. To evaluate this capability, we introduce BarrierBench, a benchmark of 100 dynamical systems spanning linear, nonlinear, discrete-time, and continuous-time settings. Our experiments assess not only the effectiveness of LLM-guided barrier synthesis but also the utility of retrieval-augmented generation and agentic coordination strategies in improving its reliability and performance. Across these tasks, the framework achieves more than 90% success in generating valid certificates. By releasing BarrierBench and the accompanying toolchain, we aim to establish a community testbed for advancing the integration of language-based reasoning with formal verification in dynamical systems. The benchmark is publicly available at https://hycodev.com/dataset/barrierbench

Fiber-optic distributed acoustic sensing (DAS) has emerged as a critical Internet-of-Things (IoT) sensing technology with broad industrial applications. However, the two-dimensional spatial-temporal morphology of DAS signals presents analytical challenges where conventional methods prove suboptimal, while being well-suited for deep learning approaches. Although our previous work, DAS Masked Autoencoder (DAS-MAE), established state-of-the-art performance and generalization without labels, it is not satisfactory in frequency analysis in temporal-dominated DAS data. Moreover, the limitation of effective training data fails to address the substantial data requirements inherent to Transformer architectures in DAS-MAE. To overcome these limitations, we present an enhanced framework incorporating short-time Fourier transform (STFT) for explicit temporal-frequency feature extraction and pioneering video-to-DAS cross-modal pre-training to mitigate data constraints. This approach learns high-level representations (e.g., event classification) through label-free reconstruction tasks. Experimental results demonstrate transformative improvements: 0.1% error rate in few-shot classification (90.9% relative improvement over DAS-MAE) and 4.7% recognition error in external damage prevention applications (75.4% improvement over from-scratch training). As the first work to pioneer video-to-DAS cross-modal pre-training, available training resources are expanded by bridging computer vision and distributed sensing areas. The enhanced performance and generalization facilitate DAS deployment across diverse industrial scenarios while advancing cross-modal representation learning for industrial IoT sensing.

The sensitivity of the acoustic detection subsystem in photoacoustic imaging (PAI) critically affects image quality. However, previous studies often focused only on front-end acoustic components or back-end electronic components, overlooking end-to-end coupling among the transducer, cable, and receiver. This work develops a complete analytical model for system-level sensitivity optimization based on the Krimholtz, Leedom, and Matthaei (KLM) model. The KLM model is rederived from first principles of linear piezoelectric constitutive equations, 1D wave equations and transmission line theory to clarify its physical basis and applicable conditions. By encapsulating the acoustic components into a controlled voltage source and extending the model to include lumped-parameter representations of cable and receiver, an end-to-end equivalent circuit is established. Analytical expressions for the system transfer functions are derived, revealing the coupling effects among key parameters such as transducer element area (EA), cable length (CL), and receiver impedance (RI). Experimental results validate the model with an average error below 5%. Additionally, a low-frequency tailing phenomenon arising from exceeding the 1D vibration assumption is identified and analyzed, illustrating the importance of understanding the model's applicable conditions and providing a potential pathway for artifact suppression. This work offers a comprehensive framework for optimizing detection sensitivity and improving image fidelity in PAI systems.

We propose decentralized k-hop Prescribed Performance State and Input Observers for heterogeneous multi-agent systems subject to bounded external disturbances. In the proposed input/state observer, each agent estimates the state and input of agents located two or more hops away using only local information exchanged with 1-hop neighbors, while guaranteeing that transient estimation errors satisfy predefined performance bounds. Conditions are established under which the input observer can be omitted, allowing the state observer convergence to be independent of the input estimates. Theoretical analysis demonstrates that if a closed-loop controller with full state knowledge achieves the control objective and the estimation-based closed-loop system is set-Input to State Stable (set-ISS) with respect to the goal set, then the estimated states can be used to achieve the system objective with an arbitrarily small worst-case error governed by the accuracy of the states estimates. Simulation results are provided to validate the proposed approach.

This paper presents a physics-consistent framework for bistatic sensing incorporating a 2-Dimensional (2D) waveguide-fed metasurface antenna array capable of realizing eXtremely-Large Multiple-Input Multiple-Output (XL MIMO) apertures. A coupled-dipole model is presented that captures the array's mutual coupling due to both waveguide and free-space interactions, and a novel passivity constraint on the corresponding magnetic polarizabilities is proposed. Focusing on a bistatic sensing setup, we leverage a Neumann-series approximation of the array response model and derive the Cramer-Rao bound for multi-target parameter estimation, which is then incorporated into a sensing optimization formulation with respect to the metasurface's per-element resonance strength configuration. Simulation results on the position error bound in the radiative near field with the proposed design quantify the critical role of metamaterial placement in strongly coupled metasurface-based XL MIMO bistatic sensing systems.

A novel electromagnetic (EM) structure termed flexible continuous aperture array (FCAPA) is proposed, which incorporates inherent surface flexibility into typical continuous aperture array (CAPA) systems, thereby enhancing the degrees-of-freedom (DoF) of multiple-input multiple-output (MIMO) systems equipped with this technology. By formulating and solving a downlink multi-user beamforming optimization problem to maximize the weighted sum rate (WSR) of the multiple users with FCAPA, it is shown that the proposed structure outperforms typical CAPA systems by a wide margin, with performance increasing with increasing morphability.

HVDC networks offer several advantages over traditional HVAC systems, particularly for long-distance power transmission and integration of renewable energy sources, such as reduced losses and enhanced stability and control, but also increase the risk of oscillations. This study investigates electrical resonant phenomena associated with HVDC stations through numerical EMT simulations. The findings indicate that electrical resonance is primarily pronounced in weak networks with long cables, as confirmed by the Nyquist criterion applied to frequency responses. Two real cases were successfully simulated in the time domain by introducing network changes, such as temporary faults and alterations in network's power strength, to activate the identified resonances. Notably, in a strong network with short cables, electrical resonance occurred alongside interactions between the network and the converter's protection system. The analysis of voltage waveforms revealed that the amplitude of the induced resonant harmonic dissipates quickly, indicating sufficient damping in the network configuration. Furthermore, the study confirmed the network's sensitivity to changes in converter parameters modeled using available MMC model.

This paper presents a maximum-likelihood detection framework that jointly mitigates hardware (HW) impairments in both amplitude and phase. By modeling transceiver distortions as residual amplitude and phase noise, we introduce the approximate phase-and-amplitude distortion detector (PAD-D), which operates in the polar domain and effectively mitigates both distortion components through distortion-aware weighting. The proposed detector performs reliable detection under generalized HW impairment conditions, achieving substantial performance gains over the conventional Euclidean detector (EUC-D) and the Gaussian-assumption phase noise detector (GAP-D), which is primarily designed to address phase distortions. In addition, we derive a closed-form high-SNR symbol error probability (SEP) approximation, which offers a generic analytical expression applicable to arbitrary constellations. Simulation results demonstrate that the PAD-D achieves up to an order-of-magnitude reduction in the error floor relative to EUC-D and GAP-D for both high-order quadrature amplitude modulation (QAM) and super amplitude phase-shift keying (SAPSK) constellations, establishing a unified and practical framework for detection under realistic transceiver impairments. Building on this framework, we further develop optimized constellations tailored to PAD-D, where the symbol positions are optimized in the complex plane to minimize SEP. The optimality of these constellations is confirmed through extensive simulations, which also verify the accuracy of the proposed analytical SEP approximation, even for the optimized designs.

Channel charting (CC) is a self-supervised positioning technique whose main limitation is that the estimated positions lie in an arbitrary coordinate system that is not aligned with true spatial coordinates. In this work, we propose a novel method to produce CC locations in true spatial coordinates with the aid of a digital twin (DT). Our main contribution is a new framework that (i) extracts large-scale channel-state information (CSI) features from estimated CSI and the DT and (ii) matches these features with a cosine-similarity loss function. The DT-aided loss function is then combined with a conventional CC loss to learn a positioning function that provides true spatial coordinates without relying on labeled data. Our results for a simulated indoor scenario demonstrate that the proposed framework reduces the relative mean distance error by 29% compared to the state of the art. We also show that the proposed approach is robust to DT modeling mismatches and a distribution shift in the testing data.

Recently, the pinching-antenna system (PASS) has emerged as a promising architecture owing to its ability to reconfigure large-scale path loss and signal phase by activating radiation points along a dielectric waveguide. However, existing studies mainly focus on line-shaped PASS architectures, whose limited spatial flexibility constrains their applicability in multiuser and indoor scenarios. In this paper, we propose a novel two-dimensional (2D) pinching-antenna system (2D-PASS) that extends the conventional line-shaped structure into a continuous dielectric waveguide plane, thereby forming a reconfigurable radiating plane capable of dynamic beam adaptation across a 2D spatial domain. An optimization framework is developed to maximize the minimum received signal-to-noise ratio (SNR) among user equipments (UEs) by adaptively adjusting the spatial configuration of pinching antennas (PAs), serving as an analog beamforming mechanism for dynamic spatial control. For the continuous-position scenario, a particle swarm optimization (PSO)-based algorithm is proposed to efficiently explore the nonconvex search space, while a discrete variant is introduced to accommodate practical hardware constraints with limited PA placement resolution. Simulation results demonstrate that the proposed 2D-PASS substantially improves the minimum SNR compared with conventional line-shaped PASS and fixed-position antenna (FPA) benchmarks, while maintaining robustness under varying user distributions and distances.

Stochastic dynamical systems have emerged as fundamental models across numerous application domains, providing powerful mathematical representations for capturing uncertain system behavior. In this paper, we address the problem of runtime safety and reach-avoid probability prediction for discrete-time stochastic systems with online observations, i.e., estimating the probability that the system satisfies a given safety or reach-avoid specification. Unlike traditional approaches that rely solely on offline models, we propose a framework that incorporates real-time observations to dynamically refine probability estimates for safety and reach-avoid events. By introducing observation-aware barrier functions, our method adaptively updates probability bounds as new observations are collected, combining efficient offline computation with online backward iteration. This approach enables rigorous and responsive prediction of safety and reach-avoid probabilities under uncertainty. In addition to the theoretical guarantees, experimental results on benchmark systems demonstrate the practical effectiveness of the proposed method.

The implementation of agentic AI systems has the potential of providing more helpful AI systems in a variety of applications. These systems work autonomously towards a defined goal with reduced external control. Despite their potential, one of their flaws is the insufficient reliability which makes them especially unsuitable for high-risk domains such as healthcare or process industry. Unreliable systems pose a risk in terms of unexpected behavior during operation and mitigation techniques are needed. In this work, we derive the main reliability challenges of agentic AI systems during operation based on their characteristics. We draw the connection to traditional AI systems and formulate a fundamental reliability challenge during operation which is inherent to traditional and agentic AI systems. As our main contribution, we propose a two-layered reliability monitoring framework for agentic AI systems which consists of a out-of-distribution detection layer for novel inputs and AI transparency layer to reveal internal operations. This two-layered monitoring approach gives a human operator the decision support which is needed to decide whether an output is potential unreliable or not and intervene. This framework provides a foundation for developing mitigation techniques to reduce risk stemming from uncertain reliability during operation.

In-air acoustic imaging systems demand beamforming techniques that offer a high dynamic range and spatial resolution while also remaining robust. Conventional Delay-and-Sum (DAS) beamforming fails to meet these quality demands due to high sidelobes, a wide main lobe and the resulting low contrast, whereas advanced adaptive methods are typically precluded by the computational cost and the single-snapshot constraint of real-time field operation. To overcome this trade-off, we propose and detail the implementation of higher-order non-linear beamforming methods using the Delay-Multiply-and-Sum technique, coupled with Coherence Factor weighting, specifically adapted for ultrasonic in-air microphone arrays. Our efficient implementation allows for enabling GPU-accelerated, real-time performance on embedded computing platforms. Through validation against the DAS baseline using simulated and real-world acoustic data, we demonstrate that the proposed method provides significant improvements in image contrast, establishing higher-order non-linear beamforming as a practical, high-performance solution for in-air acoustic imaging.