This brief presents a novel monolithic ultra-low leakage current sensor with large dynamic range. The sensor features an integrated digitizing circuit that generates digital output corresponding to changing leakage current levels, eliminating the need for an explicit ADC. Fabricated in a 180nm BCD process, the sensor occupies just 0.0234 mm 2 of area, consumes 3.1 mW of power and operates at 3.125MHz. It achieves a dynamic range of 25 pA to 5.5 µA, outperforming previous publications by several orders of magnitude. Measurement results demonstrate the sensor's accuracy, with INL and DNL remaining below 0.4 LSB across a temperature range of −40 • C to 125 • C. The sensor's ultra-low sensing capability and wide dynamic range make it suitable for time-dependent dielectric breakdown (TDDB) characterization and in-field reliability monitoring of circuits, enhancing reliability assessments in safety-critical systems.

In this paper we present an approach to estimate the epicenter of an earthquake and calculate the Peak Ground Motion (PGA) at the epicenter. This approach is a first step in designing a low-cost Earthquake Early Warning (EEW) system using fixed-location smartphone sensors. In the proposed approach, multiple potential earthquake epicenters are estimated based on the earthquake signals recorded by the fixed-location smart phones. Then, the actual epicenter is estimated and the PGA is calculated. These results can be used to determine possible disaster areas for early warning initiation. We tested the proposed approach on actual earthquake data to simulate the proposed approach. We achieved an Area Under the Curve (AUC) of 88% in estimating PGA. Simulations demonstrate the feasibility of using smartphone sensors for realtime earthquake detection using the proposed approach, offering an accessible alternative for regions with limited resources, albeit with slightly reduced precision compared to the existing systems using high precision accelerometers.

This paper explores the integration of uncertainty frameworks such as Fuzzy, Neutrosophic, and Plithogenic sets into Large Language Models (LLMs) and Natural Language Processing (NLP). We propose novel models, including Large Uncertain Language Models and Natural Uncertain Language Processing, to enhance linguistic representations and processing capabilities. Furthermore, we extend the theoretical foundation of LLMs and NLP by incorporating Hyperstructures and Superhyperstructures, enabling higher-order generalizations and hierarchical modeling. These advancements provide new perspectives for addressing uncertainty and complexity in language understanding and processing. While the paper focuses on theoretical generalizations, practical validation through computational experiments remains an important direction for future work.

Epilepsy is a common neurological disorder, and electroencephalogram (EEG) is widely used for its detection and analysis. EEG-based functional brain networks play a key role in revealing epileptic seizure patterns. Multilayer networks, which can capture multiple types of interactions, have shown great potential in this domain. However, previous analysis frameworks often simplified multilayer networks into single-layer subnetworks, ignoring the multilayer structure, or adopted a multiplex analysis framework, ignoring the importance of inter-layer connections. In this study, we proposed a multi-level analysis framework for frequency-based multilayer networks, ranging from supra-adjacency matrix, subnetwork, node, layer, and global levels. We applied this framework to the TUH EEG Seizure Corpus (TUSZ) to investigate the phase synchronization changes between the interictal (between seizures) and ictal (during seizures) phases of epileptic seizures. Compared to the interictal phase, the ictal phase showed increased local connectivity, decreased overall network connectivity, and a shift from random to regular network organization. Meanwhile, cross-frequency coupling (CFC) analysis suggested that δ-γ (frontal and parietal-occipital regions) and θ-α (frontal and right frontal-temporal regions) bands are potentially associated with seizure propagation or termination. These findings provide insights for future analysis of CFC in seizures and intervention strategies by interpreting seizure propagation networks. In conclusion, our framework enables efficient multilayer network analysis in epilepsy, facilitating the identification of key frequency bands and brain regions in CFC analysis.  

This article presents a comprehensive exploration of the architecture and various approaches in the domain of cloud computing and software-defined networks. The salient points addressed in this article encompass: Foundational Concepts: An overview of the foundational concepts and technologies of cloud computing, including software-defined cloud computing. Algorithm Evaluation: An introduction and evaluation of various algorithms aimed at enhancing network performance. These algorithms include Intelligent Rule-Based Metaheuristic Task Scheduling (IRMTS), reinforcement learning algorithms, task scheduling algorithms, and Priority-aware Semi-Greedy (PSG). Each of these algorithms contributes uniquely to optimizing Quality of Service (QoS) and data center efficiency. Resource Optimization: An introduction and examination of cloud network resource optimization based on presented results and practical experiments, including a comparison of the performance of different algorithms and approaches. Future Challenges: An investigation and presentation of challenges and future scenarios in the realm of cloud computing and software-defined networks. In conclusion, by introducing and analyzing simulators like Mininet and CloudSim, the article guides the reader in choosing the most suitable simulation tool for their project. Through its comprehensive analysis of the architecture, methodologies, and prevalent algorithms in cloud computing and software-defined networking, this article aids the reader in achieving a deeper understanding of the domain. Additionally, by presenting the findings and results of conducted research, it facilitates the discovery of the most effective and practical solutions for optimizing cloud network resources.

Contemporary control methods for interior permanent magnet synchronous motors frequently depend on mathematical equations or finite element analysis to model motor behavior. However, these approaches are challenging to accurately capture real-world nonlinearities and manufacturing variations, such as manufacturing tolerances and material uncertainties. Therefore, this study proposes a data-driven deep learning approach using long short-term memory networks, trained directly from experimental measurements.The proposed model maintains high accuracy even under challenging driving conditions and successfully captures complex harmonic characteristics. This approach signifies a substantial advancement by training directly from actual motor behavior as opposed to relying on idealized analysis models.The results demonstrate a promising pathway toward accurate and efficient motor control systems that better align with real-world conditions, providing a foundation for improving the design and control of electric drives across various applications.

Large-scale virtual reality (VR) enables novel art forms in large physical environments. A key technical solution in such cases is to split the rendering process between a client and a streamer. In this article, we focused on the peculiarities of VR split rendering and particularly on the performance of Wi-Fi. First, we calculated the network delay budget and then conducted a set of experiments to evaluate the single-user and multi-user performance of VR split rendering systems over Wi-Fi in terms of latency. Our findings were summarized as recommendations on: i) the frame rate, ii) the encoding bitrate, iii) the signal strength, iv) the channel bandwidth, and v) the multiple access techniques. In particular, our multi-user tests highlighted cases where using the highest settings or latest features of Wi-Fi can compromise performance. Finally, we applied our recommendations to a large-scale multi-user case study that took place physically in the multi-purpose hall of our university campus and virtually in the Turret Singers environment of SteamVR Home. This article serves as a reference to the critical aspects of VR split rendering systems over Wi-Fi.

Microfabrication processes and I-V characteristics are reported for a set of free-floating cubic microparticles with electrodes biased by one or two onboard photovoltaic cells. The anodes of such microparticles can serve as a site for the growth of metal reducing bacteria, and the ability of the microparticles to flow freely in aqueous solutions means that diffusion limitations on the nutrient supply for bacterial growth are partially overcome. One or two pn junctions are formed using electrically isolated mesas on the particles with a junction depth of 1 µm and a doping concentration of higher than 1020 cm-3. An AlSi (1%) metallization is used to avoid spiking during ohmic contact formation within the shallow n-type junctions. The photoresponse of the etch-released microparticles demonstrates an open circuit voltage of 0.4 V , appropriate for assisting electron transfer across an electrode-bacteria interface. The short circuit current density reaches 5-50 A/m2 under low light conditions (about 9.0 nW), which is greater than typical maximum microbial biofilm current densities.

Self-supervised monocular depth estimation (MDE) relies solely on sequential image frames for supervision. However, its performance in dynamic scenes is suboptimal, as the model's depth estimation of dynamic objects deteriorates over the course of training. This paper proposes an innovative Depth Self-Inhibition (DSI) training framework to suppress this process, allowing the MDE model to maintain strong performance in dynamic scenes. This method consists of two stages: the raw stage and the self-inhibition stage. In the raw stage, we train the MDE model without considering dynamic objects, aiming to identify unreliable depth regions where the depth values change significantly during training. In the self-inhibition stage, we retrain the MDE model and use the novel Disparity Difference Mask (DD-mask) method to exclude these unreliable regions from the loss. Additionally, the Ground-contact-prior Disparity Smoothness Loss (GDS-Loss) is employed to supervise depth learning in these areas. In this stage, we use a novel Normal Distribution-based Loss (ND-Loss), which outperforms L1 loss in dynamic scenes. During the evaluation, we introduce the Ratio Median Scaling (RM-scaling) method to address scale ambiguity in the estimated depth, providing a more reliable performance evaluation than the existing median scaling method. Experimental results show that the DSI training framework can be conveniently applied into existing MDE models, significantly improving their performance on the Cityscapes and KITTI datasets, particularly in moving object regions. The code will be released.

Encryption is a cornerstone of security for Softwareas-a-Service (SaaS) platforms, especially in the healthcare industry[1] where sensitive patient data is at stake. Protecting data in transit (as it travels between client devices, services, or cloud components) is crucial to prevent eavesdropping and man-in-themiddle attacks. Statistics show that organizations employing strong encryption experience significantly fewer data breaches – companies using robust encryption reported 70% fewer breaches than those that did not. This underscores that encryption is not just a compliance checkbox but a vital practice for reducing security incidents.

Artificial Intelligence (AI) and Cloud Computing are transforming healthcare by enabling scalable, real-time, and data-driven decision-making. While AI offers powerful analytics, it needs the cloud for scalability and accessibility. Similarly, cloud platforms store vast amounts of healthcare data but require AI to extract insights from them. Together, AI and Cloud enhance disease prediction, automate clinical workflows, and improve medical imaging analysis. This paper explores how Google Cloud AI tools-including Vertex AI, BigQuery ML, Document AI, and Medical Imaging Suite-enable faster diagnoses, predictive healthcare, and AI-assisted clinical documentation. It discusses why AI alone is insufficient and how Cloud-based AI provides the necessary infrastructure for scalable, costeffective, and secure AI-driven healthcare solutions. Using real-world applications such as AI-powered stroke prediction, telemedicine automation, and disease surveillance, this paper demonstrates how Google Cloud's AI ecosystem is revolutionizing modern healthcare. It also addresses data privacy, explainability, and ethical concerns, paving the way for future advancements in AI-driven medicine.

5G and beyond provides connectivity for a variety of heterogeneous, often mission critical services, placing stringent performance requirements on these systems. Providing satisfactory Quality-of-Experience (QoE) for diverse, coexisting applications prompts the network operators to enforce application-aware, efficient resource allocation schemes that can increase user-satisfaction, system utilization, and decrease the operational costs. For these purposes, QoS Flows and network slicing have been identified as key enablers. Those concepts move away from economy of scale, towards a fine-grained slice and flow handling with customized resource control for each application, application type, or slice. Implications of this change on the scaling of network resources and how to optimally allocate the available resources have been largely ignored so far. Furthermore, while capacity has been recognized as a key resource, selecting the appropriate queue size, granularity of the resource allocation scheme, and their relations with the number of clients are often neglected in the process of resource dimensioning. To address these shortcomings, we perform an in-depth evaluation of the effects that impact factors have on the overall QoE and system utilization using the OMNeT++ simulator. We show the optimization potential for QoE and resource utilization, and further formulate guidelines for efficient and QoE-aware resource allocation.  

Chronic kidney disease (CKD) affects millions of people worldwide, making early diagnosis critical for effective management and healthcare cost reduction. Monitoring blood urea levels is essential for this purpose. In this paper, we present a microwave sensor capable of detecting urea concentrations within the clinical range required for CKD diagnosis, a sensitivity not achieved by previous microwave (MW) based approaches. The proposed device operates without reagents or enzymes, requires no specialized personnel and offers an extended lifespan, ease of use, and sensitivity comparable to conventional optical sensors. Our work consists of two main steps. First, we conducted a dielectric characterization of artificial blood plasma (ABPS) and urea-enriched ABPS (ABPSU, 320 mg/dL) over a frequency range of 500 MHz to 10 GHz. The results revealed that the most significant differentiation between samples occurs at low frequencies, particularly in the loss tangent rather than in the real part of the dielectric constant, thereby identifying the optimal frequency range for sensor design. In the second step, we designed a double concentric circular split-ring resonator (DCCSRR) on a 49 mm × 45 mm Rogers 5880™ substrate. A dedicated optimization protocol enhanced both the resonance peak amplitude and the quality factor of the sensor’s fundamental mode. After fabrication, the DCCSRR was tested for urea concentrations ranging from 10 to 320 mg/dL, exhibiting measurable variations in S21 (dB) and frequency shifts around 575 MHz. This study establishes the first microwave sensor operating within the urea concentration range necessary for CKD diagnosis, paving the way for practical microwave-based urea monitoring.

Today, in the accelerating pace of cyber threats, new approaches and adaptive protection are required, specifically on risk analysis and fraud. Therefore, Adversarial Machine Learning with Artificial Intelligence provides a complete framework for fast-in-time cybersecurity threats assessment and advanced fraud detection capability. AML techniques would be critical for representing potential online attacks, ensuring AI-enabled defense systems are stronger and more resilient, and ultimately mitigating an adversarial attack. The fact that AML has both defensive and attacking uses shows that it's the requirement for developing models resistant to adversaries that is being made with a proper defense against emerging cyber threats. It introduces an adaptive risk identification framework, using predictive modeling, machine learning algorithms, and real-time data analysis for detecting, ranking, and mitigating risks dynamically. This is the method that keeps security protocols one step ahead of possible threats-they make this happen through a continuous evolution in their capacities to keep up with the speed of the digital world. Examples of using AI as one of the impressive tools in analyzing and owning huge datasets to detect anomalies and discover patterns of fraudulent behavior are, among others, the use of AI to do online fraud detection. It shows practical use of AI for fraud detection, from secure online authentication to preventing financial fraud. Evidence from real life helps demonstrate how AI-driven solutions can improve current risk assessment methodologies and fraud detection systems in response to increasingly complex cyber threats. The results provide a holistic view of how AML and AI techniques strengthen cyber defenses while contributing to effective cybersecurity practices. The demand for AI to take up mobility in AML has, however, begun to be very pertinent regarding the safety of digital assets, integrity of online systems, and the walls of defense erected against increasingly sophisticated cyber threats in a highly connected digital terrain. This study reflects the critical adoption of intelligent and adaptive solutions in meeting the demands posed by modern cybersecurity.

Network virtualization offers enormous flexibility to construct network services tailored to the needs of network tenants and users. Network services can be implemented through composition of virtual network functions, which can be deployed in multiple sites, forming multi-site network services. However, network virtualization further induces the increasing complexity of the network, which, in turn, increases the difficulty of error avoidance in management operations. Thus, network management must be automated to tackle such complexity and eliminate errors. In this paper, we propose the intelligent network service automation system (INSA), and we demonstrate its high efficiency and accuracy in automated network service construction and management. INSA is designed to transparently orchestrate all elements of the management system using a control plane based on publish/subscribe messaging, known as a management service bus. This enables INSA to make highly accurate decisions using vast telemetry information without overloading the control plane. Additionally, INSA integrates key enablers such as efficient and scalable control plane networking, telemetry compression, distributed processing, hidden fault detection, and autonomic resource control. We evaluated INSA both theoretically and through a prototype implementation. The former proves our claims regarding the efficiency improvement provided by compression and distribution. The latter showcases INSA management accuracy and performance, and it also demonstrates how INSA manages multi-site network services. The evaluation results show that, in comparison with the state-of-the-art alternatives, INSA uses 15-36% less CPU time, possesses 80% higher scalability, and improves accuracy by 33%.

After decades of evolution, the financial system has increasingly deviated from an idealized framework based on precise theorems. It necessitates accurate projections of complex market dynamics and human behavioral patterns. With the development of data science and machine intelligence, researchers are trying to digitalize and automate market prediction. However, existing methodologies struggle to represent the diversity of individuals and are regardless of the domino effects of interactions on market dynamics, leading to the poor performance facing abnormal market conditions where non-quantitative information dominates the market. To alleviate these disadvantages requires the introduction of knowledge about how non-quantitative information, like news and policy, affects market dynamics. This study investigates overcoming these challenges through rehearsing potential market trends based on the financial large language model agents whose behaviors are aligned with their cognition and analyses in markets. We propose a hierarchical knowledge architecture for financial large language model agents, integrating fine-tuned language models and specialized generators optimized for trading scenarios. For financial market, we develop an advanced interactive behavioral simulation system that enables users to configure agents and automate market simulations. Our real-world case simulation succeeds in rehearsing abnormal market dynamics under geopolitical events and reaches an average accuracy of 3.4% across various points in time after the event on predicting futures price. In this work, we take commodity futures as an example to research the effectiveness of our methodologies. Under normal market conditions, with corresponding news, our simulator also exhibits lower mean square error than series deep learning models and large language models in predicting three-day futures price of specific commodities. All experimental results demonstrates our method effectively leverages diverse markets information to simulate behaviors and their impact on market dynamics through systematic interaction.

A fundamental objective of the forthcoming sixthgeneration wireless networks is to concurrently serve a vast array of devices many of which, such as Internet-of-Things (IoT) sensors, are projected to have low power requirements or even operate in a battery-free manner. To achieve this goal, nonorthogonal multiple access (NOMA) and ambient backscatter communications (AmBC) are regarded as two pivotal and promising technologies. In this work, we present a novel analytical framework for studying the reliability and security of uplink NOMA-based AmBC systems. Specifically, closed-form analytical expressions for both NOMA-users' and IoT backscatter device's (BD's) outage probabilities (OPs) are derived for both cases of perfect and imperfect successive interference cancellation (SIC). In addition, assuming that one NOMA-user transmits an artificial noise in order to enhance system's security, the physical layer security (PLS) of the system is investigated by extracting analytical expressions for NOMA-users' and BD's intercept probabilities (IPs). To gain insightful understandings, an asymptotic analysis is carried out by focusing on the high signal-to-noise (SNR) regime, which reveals that NOMA-users and BDs face outage floors in the high SNR regime as well as that IPs reach constant values at high SNR. Additionally, practical insights regarding how different system parameters affect these OP floors and IP constant values are extracted. Numerical results verify the accuracy of othe developed theoretical framework, offer performance comparisons between the presented NOMAbased AmBC system and a conventional orthogonal multiple access-based AmBC system, and reveal the impact of different system parameters on the reliability and security of NOMA-based AmBC networks.

Current trajectory datasets of traffic participants often lack detailed environmental information, which is crucial for developing effective data-driven methods for future mobility solutions. To address this gap, we introduce the comprehensive DLR Urban Traffic dataset version 1.2.0. The dataset includes 32,296 trajectories of traffic participants, along with traffic light data, local weather data, air quality data, and road condition data collected at the Application Platform for Intelligent Mobility Research Intersection during a single day of recording. A comparison with other publicly available datasets reveals that our dataset offers more comprehensive information about the traffic environment than existing alternatives. An analysis of our dataset shows that trajectories are available for all possible 16 routes at the intersection, with the number of trajectories per route varying significantly, ranging from 11 to 3,344. Most interactions between motorized road users occur during unprotected left turns with oncoming traffic. However, there are also interactions between motorized and vulnerable road users, particularly during right turns. All in all, the dataset provides researchers with the resources needed to improve urban mobility solutions.