Papers

Our work introduces the DermETAS-SNA LLM Assistant that integrates Dermatology-focused Evolutionary Transformer Architecture Search with StackNet Augmented LLM. The assistant dynamically learns skin-disease classifiers and provides medically informed descriptions to facilitate clinician-patient interpretation. Contributions include: (1) Developed an ETAS framework on the SKINCON dataset to optimize a Vision Transformer (ViT) tailored for dermatological feature representation and then fine-tuned binary classifiers for each of the 23 skin disease categories in the DermNet dataset to enhance classification performance; (2) Designed a StackNet architecture that integrates multiple fine-tuned binary ViT classifiers to enhance predictive robustness and mitigate class imbalance issues; (3) Implemented a RAG pipeline, termed Diagnostic Explanation and Retrieval Model for Dermatology, which harnesses the capabilities of the Google Gemini 2.5 Pro LLM architecture to generate personalized, contextually informed diagnostic descriptions and explanations for patients, leveraging a repository of verified dermatological materials; (4) Performed extensive experimental evaluations on 23 skin disease categories to demonstrate performance increase, achieving an overall F1-score of 56.30% that surpasses SkinGPT-4 (48.51%) by a considerable margin, representing a performance increase of 16.06%; (5) Conducted a domain-expert evaluation, with eight licensed medical doctors, of the clinical responses generated by our AI assistant for seven dermatological conditions. Our results show a 92% agreement rate with the assessments provided by our AI assistant (6) Created a proof-of-concept prototype that fully integrates our DermETAS-SNA LLM into our AI assistant to demonstrate its practical feasibility for real-world clinical and educational applications.

Current-steering digital-to-analog converter (DAC) is a prominent architecture that is commonly used in high-speed applications such as optical communications. One of the shortcomings of this architecture is the output glitches that are input dependent and degrade the dynamic performance of the DAC. We investigate DAC glitches that arise from asymmetry in the fall/rise response of DAC switches. We formulate a glitch metric that defines the overall DAC performance, which is then used to find a novel DAC weighting scheme. Numerical simulations show that the proposed architecture can potentially provide a significant performance advantage compared to the segmented structure.

We propose a novel digital-to-analog converter (DAC) weighting architecture that statistically minimizes the distortion caused by random timing mismatches among current sources. To decode the DAC input codewords into corresponding DAC switches, we present three algorithms with varying computational complexities. We perform high-level Matlab simulations to illustrate the dynamic performance improvement over the segmented structure.

We present a new computationally efficient method for multi-beamforming in the broadband setting. Our "fast beamspace transformation" forms $B$ beams from $M$ sensor outputs using a number of operations per sample that scales linearly (to within logarithmic factors) with $M$ when $B\sim M$. While the narrowband version of this transformation can be performed efficiently with a spatial fast Fourier transform, the broadband setting requires coherent processing of multiple array snapshots simultaneously. Our algorithm works by taking $N$ samples off of each of $M$ sensors and encoding the sensor outputs into a set of coefficients using a special non-uniform spaced Fourier transform. From these coefficients, each beam is formed by solving a small system of equations that has Toeplitz structure. The total runtime complexity is $\mathcal{O}(M\log N+B\log N)$ operations per sample, exhibiting essentially the same scaling as in the narrowband case and vastly outperforming broadband beamformers based on delay and sum whose computations scale as $\mathcal{O}(MB)$. Alongside a careful mathematical formulation and analysis of our fast broadband beamspace transform, we provide a host of numerical experiments demonstrating the algorithm's favorable computational scaling and high accuracy. Finally, we demonstrate how tasks such as interpolating to “off-grid" angles and nulling an interferer are more computationally efficient when performed directly in beamspace.

Minimizing transmission delay in wireless multi-hop networks is a fundamental yet challenging task due to the complex coupling among interference, queue dynamics, and distributed control. Traditional scheduling algorithms, such as max-weight or queue-length-based policies, primarily aim to optimize throughput but often suffer from high latency, especially in heterogeneous or dynamically changing topologies. Recent learning-based approaches, particularly those employing Graph Neural Networks (GNNs), have shown promise in capturing spatial interference structures. However, conventional Graph Convolutional Networks (GCNs) remain limited by their local aggregation mechanism and their inability to model long-range dependencies within the conflict graph. To address these challenges, this paper proposes a delay-oriented distributed scheduling framework based on Transformer GNN. The proposed model employs an attention-based graph encoder to generate adaptive per-link utility scores that reflect both queue backlog and interference intensity. A Local Greedy Solver (LGS) then utilizes these utilities to construct a feasible independent set of links for transmission, ensuring distributed and conflict-free scheduling.

Interferometric Synthetic Aperture Radar (InSAR) is a powerful tool for monitoring surface deformation with high precision. However, low Signal-to-Noise Ratio (SNR) conditions, common in regions with low backscatter, can degrade phase coherence and compromise displacement accuracy. In this study, we quantify the impact of low-SNR conditions on InSAR-derived displacement using L-band UAVSAR data collected over the San Andreas Fault and Greenland ice sheet. We simulate low-SNR conditions by degrading the Noise-Equivalent Sigma Zero (NESZ) to $-15~\mathrm{dB}$ and assess the resulting effects on interferometric coherence, phase unwrapping, and time series inversion. The displacement accuracy of 4mm in single interferogram can be achieved by taking looks for the signal decorrelation of 0.6 and SNR between -9dB to -10dB. Our findings indicate that even under low-SNR conditions, a velocity precision of $0.5~\mathrm{cm/yr}$ can be achieved in comparison to high-SNR conditions. By applying multilooking with an 8x8 window, we significantly improve coherence and eliminate this bias, demonstrating that low-SNR systems can achieve comparable precision to high-SNR systems at the expense of spatial resolution. These results have important implications for the design of future cost-effective SAR missions, such as Surface Deformation and Change (SDC), and the optimization of InSAR processing techniques in challenging environments.

This paper studies robust optimal operation control problems for microgrids with a high share of renewable energy sources. The main goal is to ensure an optimal operation in the presence of a wide range of scenarios of uncertain infeed of renewable sources and uncertain load demand. We formally state a minimum-regret robust model predictive control (MPC) problem and address it by making effective use of a hierarchical microgrid control structure. In detail, we consider an enhanced primary control layer composed of droop control and an autonomous limitation of power and energy. We prove that this enables us to use constant power setpoints to achieve an optimal operation under certain conditions. To obtain a tractable controller, we then combine the abovementioned constant saturation-aware setpoints with an energy management system, which solves a robust unit commitment problem within a model predictive control framework. In a case study, we finally demonstrate the viability of the control design.

Post-harvest losses due to subjective quality assessment cause significant damage to the economy and food safety, especially in countries like Bangladesh. To mitigate such damages, objective decision-making backed by scientific methods is necessary. An IoT-based, cost-effective quality monitoring system can provide a solution by going beyond subjective quality monitoring and decision-making practices. Here, we propose a low-power, cost-effective fruit quality monitoring system with an array of MQ gas sensors, which can be used as an electronic nose. We track the volatile gas emissions, specifically ethanol, methane, and ammonia, encompassing both ripening and decomposition for a set of bananas. Based on the gas concentration thresholds, we develop a mathematical model to accurately assess fruit quality. We also integrate this information into a dashboard for prompt decision-making and monitoring to make it useful to the farmers. This approach has the potential to reduce economic losses, enhance food safety, and provide scalable solutions for the supply chain.

Radio Frequency (RF) sensing is attracting interest in research, standardization, and industry, especially for its potential in Internet of Things (IoT) applications. By leveraging the properties of the ElectroMagnetic (EM) waves used in wireless networks, RF sensing captures environmental information such as the presence and movement of people and objects, enabling passive localization and vision applications. This paper investigates the theoretical bounds on accuracy and resolution for RF sensing systems within dense networks. It employs an EM model to predict the effects of body blockage in various scenarios. To detect human movements, the paper proposes a deep graph neural network, trained on Received Signal Strength (RSS) samples generated from the EM model. These samples are structured as dense graphs, with nodes representing antennas and edges as radio links. Focusing on the problem of identifying the number of human subjects co-present in a monitored area over time, the paper analyzes the theoretical limits on the number of distinguishable subjects, exploring how these limits depend on factors such as the number of radio links, the size of the monitored area and the subjects physical dimensions. These bounds enable the prediction of the system performance during network pre-deployment stages. The paper also presents the results of an indoor case study, which demonstrate the effectiveness of the approach and confirm the model's predictive potential in the network design stages.

The success of large language models LLMs amplifies the need for highthroughput energyefficient inference at scale. 3DDRAMbased accelerators provide high memory bandwidth and therefore an opportunity to accelerate the bandwidthbound decode phase. However, how to adequately balance compute density for prefill with bandwidthcapacity for decode remains open. Moreover, most prior designs do not target endtoend serving, leaving the codesign of dataflow, parallel mapping, and scheduling underexplored. To bridge the gap, we present LaMoSys3.5D, to our knowledge the first scalable 3.5DIC architecture for LLM serving. LaMoSys3.5D composes heterogeneous 3DDRAM chiplets on a 2.5D interposer: computerich chiplets for prefill and bandwidthcapacityrich chiplets for decode. To realize efficient serving, we adopt a hardwaresoftware codesign spanning dataflow, parallel mapping, and introduce a thermalaware modeling and hierarchical designspace exploration framework. Across diverse LLMs and workloads, LaMoSys3.5D improves throughputperwatt over DGXA100 systems by 62 and achieves a 4.87 better endtoend latency geomean versus prior 3D designs. We further distill intriguing design guidelines for 3.5DIC architectures and endtoend inference serving.

This paper explores signal and image analysis by using the Singular Value Decomposition (SVD) and its extension, the Generalized Singular Value Decomposition (GSVD). A key strength of SVD lies in its ability to separate information into orthogonal subspaces. While SVD is a well-established tool in ECG analysis, particularly for source separation, this work proposes a refined method for selecting a threshold to distinguish between maternal and fetal components more effectively. In the first part of the paper, the focus is onmedical signal analysis,where the concepts of Energy Gap Variation (EGV) and Singular Energy are introduced to isolate fetal and maternal ECG signals, improving the known ones. Furthermore, the approach is significantly enhanced by the application of GSVD, which provides additional discriminative power for more accurate signal separation. The second part introduces a novel technique called Singular Smoothness, developed for image analysis. This method incorporates Singular Entropy and the Frobenius normto evaluate information density, and is applied to the detection of natural anomalies such asmountain fractures and burned forest regions. Numerical experiments are presented to demonstrate the effectiveness of the proposed approaches.

Preliminary mission design of low-thrust spacecraft trajectories in the Circular Restricted Three-Body Problem is a global search characterized by a complex objective landscape and numerous local minima. Formulating the problem as sampling from an unnormalized distribution supported on neighborhoods of locally optimal solutions, provides the opportunity to deploy Markov chain Monte Carlo methods and generative machine learning. In this work, we extend our previous self-supervised diffusion model fine-tuning framework to employ gradient-informed Markov chain Monte Carlo. We compare two algorithms - the Metropolis-Adjusted Langevin Algorithm and Hamiltonian Monte Carlo - both initialized from a distribution learned by a diffusion model. Derivatives of an objective function that balances fuel consumption, time of flight and constraint violations are computed analytically using state transition matrices. We show that incorporating the gradient drift term accelerates mixing and improves convergence of the Markov chain for a multi-revolution transfer in the Saturn-Titan system. Among the evaluated methods, MALA provides the best trade-off between performance and computational cost. Starting from samples generated by a baseline diffusion model trained on a related transfer, MALA explicitly targets Pareto-optimal solutions. Compared to a random walk Metropolis algorithm, it increases the feasibility rate from 17.34% to 63.01% and produces a denser, more diverse coverage of the Pareto front. By fine-tuning a diffusion model on the generated samples and associated reward values with reward-weighted likelihood maximization, we learn the global solution structure of the problem and eliminate the need for a tedious separate data generation phase.

Scaled model experiments are commonly used in various engineering fields to reduce experimentation costs and overcome constraints associated with full-scale systems. The relevance of such experiments relies on dimensional analysis and the principle of dynamic similarity. However, transferring controllers to full-scale systems often requires additional tuning. In this paper, we propose a method to enable a direct controller transfer using dimensionless model predictive control, tuned automatically for closed-loop performance. With this reformulation, the closed-loop behavior of an optimized controller transfers directly to a new, dynamically similar system. Additionally, the dimensionless formulation allows for the use of data from systems of different scales during parameter optimization. We demonstrate the method on a cartpole swing-up and a car racing problem, applying either reinforcement learning or Bayesian optimization for tuning the controller parameters. Software used to obtain the results in this paper is publicly available at https://github.com/josipkh/dimensionless-mpcrl.

Accurate outdoor localization in Non-Line-of-Sight (NLoS) environments remains a critical challenge for wireless communication and sensing systems. Existing methods, including positioning based on the Global Navigation Satellite System (GNSS) and triple Base Stations (BSs) techniques, cannot provide reliable performance under NLoS conditions, particularly in dense urban areas with strong multipath effects. To address this limitation, we propose a single BS localization framework that integrates sequential signal measurements with prior radio information embedded in the Radio Map (RM). Using temporal measurement features and matching them with radio maps, the proposed method effectively mitigates the adverse impact of multipath propagation and reduces the dependence on LoS paths. Simulation experiments further evaluate the impact of different radio map construction strategies and the varying lengths of the measurement sequence on localization accuracy. Results demonstrate that the proposed scheme achieves sub-meter positioning accuracy in typical NLoS environments, highlighting its potential as a practical and robust solution for single-base-station deployment.

This article presents a systematic method for designing time-varying Control Barrier Functions (CBF) composed of a time-invariant component and multiple time-dependent components, leveraging structural properties of the system dynamics. The method involves the construction of a specific class of time-invariant CBFs that encode the system's dynamic capabilities with respect to a given constraint, and augments them subsequently with appropriately designed time-dependent transformations. While transformations uniformly varying the time-invariant CBF can be applied to arbitrary systems, transformations exploiting structural properties in the dynamics - equivariances in particular - enable the handling of a broader and more expressive class of time-varying constraints. The article shows how to leverage such properties in the design of time-varying CBFs. The proposed method decouples the design of time variations from the computationally expensive construction of the underlying CBFs, thereby providing a computationally attractive method to the design of time-varying CBFs. The method accounts for input constraints and under-actuation, and requires only qualitative knowledge on the time-variation of the constraints making it suitable to the application in uncertain environments.

Artificial Intelligence (AI) systems are now an integral part of multiple industries. In clinical research, AI supports automated adverse event detection in clinical trials, patient eligibility screening for protocol enrollment, and data quality validation. Beyond healthcare, AI is transforming finance through real-time fraud detection, automated loan risk assessment, and algorithmic decision-making. Similarly, in manufacturing, AI enables predictive maintenance to reduce equipment downtime, enhances quality control through computer-vision inspection, and optimizes production workflows using real-time operational data. While these technologies enhance operational efficiency, they introduce new challenges regarding safety, accountability, and regulatory compliance. To address these concerns, we introduce the SMART+ Framework - a structured model built on the pillars of Safety, Monitoring, Accountability, Reliability, and Transparency, and further enhanced with Privacy & Security, Data Governance, Fairness & Bias, and Guardrails. SMART+ offers a practical, comprehensive approach to evaluating and governing AI systems across industries. This framework aligns with evolving mechanisms and regulatory guidance to integrate operational safeguards, oversight procedures, and strengthened privacy and governance controls. SMART+ demonstrates risk mitigation, trust-building, and compliance readiness. By enabling responsible AI adoption and ensuring auditability, SMART+ provides a robust foundation for effective AI governance in clinical research.

This article addresses the challenge of optimizing handover (HO) in next-generation wireless networks by integrating Reconfigurable Intelligent Surfaces (RIS), predicting received signal power, and utilizing learning-based decision-making. A conventional reactive HO mechanism, such as lower-layer triggered mobility (LTM), is enhanced through linear prediction to anticipate link degradation. Additionally, the use of RIS helps to mitigate signal blockage and extend coverage. An online trained non-linear Contextual Multi-Armed Bandit (CMAB) agent selects target gNBs based on context features, which reduces unnecessary HO and signaling overhead. Extensive simulations evaluate eight combinations of these techniques under realistic mobility and channel conditions. Results show that CMAB and RSRP prediction consistently reduce the number of HO, ping-pong rate and cell preparations, while RIS improves link reliability.

We consider a behavioral-feedback SIR epidemic model, in which the infection rate depends in feedback on the fractions of susceptible and infected agents. Contrary to the classic SIR dynamics, the considered model allows one to account for endogenous adaptation mechanisms of the agents in response to the epidemics, such as voluntary social distancing, or the adoption of face masks. For this model, we formulate an optimal control problem for a social planner that has the ability to reduce the infection rate to keep the infection curve below a certain threshold within an infinite time horizon, while minimizing the intervention cost. Based on the dynamic properties of the model, we prove that, under quite general conditions on the infection rate, the filling-the-box strategy is the optimal control. This strategy consists in letting the epidemics spread without intervention until the threshold is reached, then applying the minimum control that leaves the fraction of infected individuals constantly at the threshold until the reproduction number becomes less than one and the infection naturally fades out. Our contribution enhances the understanding of epidemic management while accounting for adaptive human behavior, offering novel insights for robust containment strategies.

Reconfigurable Intelligent Surfaces (RIS) have emerged as a transformative technology in wireless communications, offering unprecedented control over signal propagation. This study focuses on passive beyond diagonal reconfigurable intelligent surface (BD-RIS), which has been proposed to generalize conventional diagonal RIS, in Multiple-Input Multiple-Output (MIMO) downlink (DL) communication systems. We compare the performance of transmit beamforming (TxBF) and MIMO capacity transmission with waterfilling power allocation in the millimeter wave (mmWave) band, where propagation primarily occurs under line-of-sight (LOS) conditions. In the lack of closed-form expressions for the optimal RIS elements in either case, our approach adopts a gradient-based optimization approach requiring lower complexity than the solution in arXiv:2406.02170. Numerical results reveal that BD-RIS significantly outperforms traditional diagonal RIS in terms of spectral efficiency and coverage

Holographic multiple-input multiple-output (MIMO) enables electrically large continuous apertures, overcoming the physical scaling limits of conventional MIMO architectures with half-wavelength spacing. Their near-field operating regime requires channel models that jointly capture line-of-sight (LoS) and non-line-of-sight (NLoS) components in a physically consistent manner. Existing studies typically treat these components separately or rely on environment-specific multipath models. In this work, we present a comprehensive analysis of the wavenumber-division multiplexing (WDM) framework, originally proposed for purely LoS channels, applied to a unified electromagnetic (EM)-based LoS+NLoS channel model. We first formulate the EM-based channel model for holographic lines and then apply WDM to this model, enabling a comparison between the corresponding spatial-sampling-based and expansion-based formulations. Applying WDM to the NLoS component yields its angular-domain representation, enabling direct characterization through the power spectral factor and power spectral density. We further derive closed-form characterizations for isotropic and non-isotropic scattering, with the former recovering Jakes' isotropic model. Lastly, we evaluate the resulting degrees of freedom and ergodic capacity, showing that incorporating the NLoS component substantially improves the performance relative to the purely LoS case.

Next-generation communication and localization systems increasingly rely on extremely large-scale arrays (XL-arrays), which promise unprecedented spatial resolution and new functionalities. These gains arise from their inherent operation in the near field (NF) regime, where the spherical nature of the wavefront can no longer be ignored; consequently, characterizing the ambiguity function – which amounts to the matched beam pattern – is considerably more challenging. Implementing very wide apertures with half-wavelength element spacing is costly and complex. This motivates thinning the array (removing elements), which introduces intricate aliasing structures, i.e., grating lobes. Whereas prior work has addressed this challenge using approximations tailored to specific array geometries, this paper develops a general framework that reveals the fundamental origins and geometric behavior of grating lobes in near-field ambiguity functions. Using a local spatial-frequency analysis of steering signals, we derive a systematic methodology to model NF grating lobes as aliasing artifacts, quantifying their structure on the AF, and providing design guidelines for XL-arrays that operate within aliasing-safe regions. We further connect our framework to established far-field principles. Finally, we demonstrate the practical value of the approach by deriving closed-form expressions for aliasing-free regions in canonical uniform linear arrays and uniform circular arrays.

Modern power networks face increasing vulnerability to cascading failures due to high complexity and the growing penetration of intermittent resources, necessitating rigorous security assessment beyond the conventional $N-1$ criterion. Current approaches often struggle to achieve the computational tractability required for exhaustive $N-2$ contingency analysis integrated with complex stability evaluations like small-signal stability. Addressing this computational bottleneck and the limitations of deterministic screening, this paper presents a scalable methodology for the vulnerability assessment of modern power networks, integrating $N-2$ contingency analysis with small-signal stability evaluation. To prioritize critical components, we propose a probabilistic \textbf{Risk Index ($R_i$)} that weights the deterministic severity of a contingency (including optimal power flow divergence, islanding, and oscillatory instability) by the failure frequency of the involved elements based on reliability data. The proposed framework is implemented using High-Performance Computing (HPC) techniques through the PyCOMPSs parallel programming library, orchestrating optimal power flow simulations (VeraGrid) and small-signal analysis (STAMP) to enable the exhaustive exploration of massive contingency sets. The methodology is validated on the IEEE 118-bus test system, processing more than \num{57000} scenarios to identify components prone to triggering cascading failures. Results demonstrate that the risk-based approach effectively isolates critical assets that deterministic $N-1$ criteria often overlook. This work establishes a replicable and efficient workflow for probabilistic security assessment, suitable for large-scale networks and capable of supporting operator decision-making in near real-time environments.

This study proposes a systematic design procedure for determining the quantization gain and the security parameter in the Confidential Fictitious Reference Iterative Tuning (CFRIT), enabling overflow-free and accuracy-guaranteed encrypted controller tuning. Within an encrypted data-driven gain tuning, the range of quantization errors induced during the encoding (encryption) process can be estimated from operational data. Based on this insight, explicit analytical conditions on the quantization gain and the security parameter are derived to prevent overflow in computing over encrypted data. Furthermore, the analysis reveals a quantitative relationship between quantization-induced errors and the deviation between the gains obtained by CFRIT and non-confidential Fictitious Reference Iterative Tuning (FRIT), clarifying how parameter choice affects tuning accuracy. A numerical example verifies the proposed procedure by demonstrating that the designed parameters achieve accurate encrypted tuning within a prescribed tolerance while preventing overflow. In addition, the admissible region of parameter combinations is visualized to examine the characteristics of feasible and infeasible regions, providing practical insights into parameter design for encrypted data-driven control.

We study how privileged information about a physical system affects the discovery of high-performing policies when training a reinforcement learning agent directly in the physical world. We let the agent control a cylinder in a tabletop water channel to maximize or minimize drag. The flow is chaotic and difficult to model or simulate accurately and good strategies are not obvious beforehand. Decades-old experimental studies provide recipes for simple, high-performance, periodic open-loop policies that increase drag by 26.6% +/- 0.7% and reduce it by 29.7% +/- 1.3%. With dense observations of the cylinder wake, the agent learns within tens of minutes policies that increase drag by 25.5% +/- 0.9% and reduce it by 32.4% +/- 1.6%. We record action trajectories during online policy execution and replay them open loop as fixed sequences; these replays increase drag by 23.2% +/- 2.2% and reduce it by 32.1% +/- 3.0%. However, when we withhold flow observations during training, the agent still learns to decrease drag by 31.4% +/- 2.1%, but no run learns to increase it (1.8% +/- 4.3% drag increase). Our physical experiments demonstrate an extreme case: privileged observations can be decisive for policy discovery even when the resulting actions can be replayed in open loop.

In this paper, an MPC for tracking formulation is proposed for the control of anesthesia dynamics. It seamlessly enables the optimization of the steady-states pair that is not unique due to the MISO nature of the model. Anesthesia dynamics is a multi-time scale system with two types of states characterized, respectively, by fast and slow dynamics. In anesthesia control, the output equation depends only on the fast dynamics. Therefore, the slow states can be treated as disturbances, and compensation terms can be introduced. Subsequently, the system can be reformulated as a nominal one allowing the design of an MPC for tracking strategy. The presented framework ensures recursive feasibility and asymptotic stability, through the design of appropriate terminal ingredients in the MPC for tracking framework. The controller performance is then assessed on a patient in a simulation environment.

Learned image reconstruction has become a pillar in computational imaging and inverse problems. Among the most successful approaches are learned iterative networks, which are formulated by unrolling classical iterative optimisation algorithms for solving variational problems. While the underlying algorithm is usually formulated in the functional analytic setting, learned approaches are often viewed as purely discrete. In this survey we present a unified operator view for learned iterative networks. Specifically, we formulate a learned reconstruction operator, defining how to compute, and separately the learning problem, which defines what to compute. In this setting we present common approaches and show that many approaches are closely related in their core. We review linear as well as non-linear inverse problems in this framework and present a short numerical study to conclude.

Time delays in communication channels present significant challenges for bilateral teleoperation systems, affecting both transparency and stability. Although traditional wave variable-based methods for a four-channel architecture ensure stability via passivity, they remain vulnerable to wave reflections and disturbances like variable delays and environmental noise. This article presents a data-driven hybrid framework that replaces the conventional wave-variable transform with an ensemble of three advanced sequence models, each optimized separately via the state-of-the-art Optuna optimizer, and combined through a stacking meta-learner. The base predictors include an LSTM augmented with Prophet for trend correction, an LSTM-based feature extractor paired with clustering and a random forest for improved regression, and a CNN-LSTM model for localized and long-term dynamics. Experimental validation was performed in Python using data generated from the baseline system implemented in MATLAB/Simulink. The results show that our optimized ensemble achieves a transparency comparable to the baseline wave-variable system under varying delays and noise, while ensuring stability through passivity constraints.

Shading faults remain one of the most critical challenges affecting photovoltaic (PV) system efficiency, as they not only reduce power generation but also disturb maximum power point tracking (MPPT). To address this issue, this study introduces a hybrid optimization framework that combines Fuzzy Logic Control (FLC) with a Shading-Aware Particle Swarm Optimization (SA-PSO) method. The proposed scheme is designed to adapt dynamically to both partial shading (20%-80%) and complete shading events, ensuring reliable global maximum power point (GMPP) detection. In this approach, the fuzzy controller provides rapid decision support based on shading patterns, while SA-PSO accelerates the search process and prevents the system from becoming trapped in local minima. A comparative performance assessment with the conventional Perturb and Observe (P&O) algorithm highlights the advantages of the hybrid model, showing up to an 11.8% improvement in power output and a 62% reduction in tracking time. These results indicate that integrating intelligent control with shading-aware optimization can significantly enhance the resilience and energy yield of PV systems operating under complex real-world conditions.

This article investigates on the improvement and stabilization of alternating current (AC) and direct current (DC) output voltages in a Permanent Magnet Synchronous Generator (PMSG) driven by a vertical-axis tidal turbine using advanced control strategies. The research integrates artificial intelligence (AI)-based techniques to enhance voltage stability and efficiency. Initially, the Maximum Power Point Tracking (MPPT) approach based on Tip Speed Ratio (TSR) and Artificial Neural Network (ANN) Fuzzy logic controllers is explored. To further optimize the performance, Particle Swarm Optimization (PSO) and a hybrid ANN-PSO methodology are implemented. These strategies aim to refine the reference rotational speed of the turbine while minimizing deviations from optimal power extraction conditions. The simulation results of a tidal turbine operating at a water flow velocity of 1.5 m/s demonstrate that the PSO-based control approach significantly enhances the voltage stability compared to conventional MPPT-TSR and ANN-Fuzzy controllers. The hybrid ANN-PSO technique improves the voltage regulation by dynamically adapting to system variations and providing real-time reference speed adjustments. This research highlights the AI-based hybrid optimization benefit to stabilize the output voltage of tidal energy systems, thereby increasing reliability and efficiency in renewable energy applications.

The integration of AI into CAD systems transforms how engineers plan and develop infrastructure projects involving water and power transportation across industrial and remote landscapes. This paper discusses how AI-driven CAD systems improve the efficient, effective, and sustainable design of infrastructure by embedding automation, predictive modeling, and real-time data analytics. This study examines how AI-supported toolsets can enhance design workflows, minimize human error, and optimize resource allocation for projects in underdeveloped environments. It also addresses technical and organizational challenges to AI adoption, including data silos, interoperability issues, and workforce adaptation. The findings demonstrate that AI-powered CAD enables faster project delivery, enhanced design precision, and increased resilience to environmental and logistical constraints. AI helps connect CAD, GIS, and IoT technologies to develop self-learning, adaptive design systems that are needed to meet the increasing global demand for sustainable infrastructure.