Cloud-Edge Computing Continuum (CEC) system, where edge and cloud nodes are seamlessly connected, is dedicated to handle substantial computational loads offloaded by end-users. These tasks can suffer from delays or be dropped entirely when deadlines are missed, particularly under fluctuating network conditions and resource limitations. The CEC is coupled with the need for hybrid task offloading, where the task placement decisions concern whether the tasks are processed locally, offloaded vertically to the cloud, or horizontally to interconnected edge servers. In this paper, we present a distributed hybrid task offloading scheme (HOODIE) designed to jointly optimize the tasks latency and drop rate, under dynamic CEC traffic. HOODIE employs a model-free deep reinforcement learning (DRL) framework, where distributed DRL agents at each edge server autonomously determine offloading decisions without global task distribution awareness. To further enhance the system pro-activity and learning stability, we incorporate techniques such as Long Short-term Memory (LSTM), Dueling deep Q-networks (DQN), and double-DQN. Extensive simulation results demonstrate that HOODIE effectively reduces task drop rates and average task processing delays, outperforming several baseline methods under changing CEC settings and dynamic conditions.

The trend towards sensor miniaturization has heightened interest in optimizing sensor array configurations across various scientific and industrial applications that imply multichannel measurements. In this article, we present an extended version of our previous approach, adapting it to the realworld conditions in magnetoencephalography (MEG). At each step the method selects the sensor location that maximizes the ROI-related signal-to-noise ratio (SNR), followed by the projection recursively applied to the gain matrix to orthogonalize the subsequent stereotypic steps with respect to the source subspace served by the already selected sensors. This algorithm offers high computational efficiency and flexibility with respect to physical constraints. We dub our approach as Optimal Layout Design via Recursively Applied Leadfield Elimination (OLD RALFE). Our analysis of the optimized MEG sensor layouts obtained with OLD RALFE shows that the proposed algorithm performs similarly to several other significantly more computationally demanding sensor layout optimization approaches in terms of both SNR and the Cramér-Rao lower bound (CRLB) on the spatial resolution of the inverse problem solution. OLD RALFE demonstrated in application to MEG, is broadly applicable to many other multichannel measurement challenges.

Reconfigurable Intelligent Surfaces (RIS) represent transformative technologies for next-generation wireless communications, offering advanced control over electromagnetic wave propagation. While RIS have been extensively studied, Stacked Intelligent Metasurfaces (SIM), which extend the RIS concept to multilayered systems, present significant modeling and optimization challenges. This work addresses these challenges by introducing an optimization framework for SIM that, unlike previous approaches, is based on a comprehensive model without relying on specific assumptions, allowing for broader applicability of the results. We first present a model based on multi-port network theory for characterizing a general electromagnetic collaborative object (ECO) and derive a general framework for ECO optimization. We then introduce the SIM as an ECO with a specific architecture, providing insights into SIM optimization for various architectures and discussing the complexity in each case. Finally, we analyze the impact of commonly used assumptions, demonstrating that employing a complete model-one that does not rely on the unilateral hypothesis and accounts for mutual coupling among SIM elements-can yield better performance, using the realization of a 2D DFT as a case study.

This paper presents channel measurements and modeling at 28 and 160 GHz in an industrial-style open office environment with angular information at the receiver. Channel impulse response snapshots are captured at 145 measurement positions for both frequencies and are evaluated in terms of power metrics, delay, and angular characteristics. Frequencydependent models for path loss, delay, and angular spread of arrival and K-factor are estimated, and large-scale parameter cross-correlations are evaluated. The results show that delay and angle spread are not frequency dependent in this environment although twice as many multi-path components were identified at 28 GHz compared to 160 GHz.

The IPv6 Routing Protocol for LLNs (RPL) is fundamental to the Internet of Things (IoT), providing essential routing capabilities for devices operating in constrained environments. At the heart of RPL's efficiency is the Trickle Algorithm (TA), designed to optimize the dissemination of control messages across networks, balancing timely updates with the imperative to conserve bandwidth and energy. This paper focuses on validating the implementation of the TA within the Latest version of Contiki-OS (Contiki-NG), a leading experimental platform for IoT networks. By comparing the results of theoretical models with practical simulations of Contiki-NG, we aim to determine the reliability of the Contiki-NG representation of the TA behaviour. Our analysis reveals a high degree of agreement between theoretical expectations and simulated results, underscoring Contiki-NG's robustness as a simulation tool for IoT research. The findings affirm the precision of Contiki-NG implementation and highlight the platform's value in facilitating the development and testing of IoT protocols. By ensuring the reliability of these simulations, our work strengthens the confidence of the IoT research community in using Contiki-NG as a foundation for exploring innovative solutions to the unique challenges of IoT networks.

Escalating threats from increasingly sophisticated ransomware attacks pose significant risks to data integrity, financial stability, and operational continuity for organizations across sectors. Addressing these threats requires a novel detection framework that overcomes the limitations of traditional signature-based and machine learning-based approaches, particularly their vulnerability to adaptive ransomware tactics and resource constraints. The proposed Dynamic Threat Imprint Analysis (DTIA) method leverages real-time behavioral analysis combined with adaptive learning mechanisms, enabling it to detect both known and previously unseen ransomware variants with high accuracy. Experimental evaluation demonstrates that DTIA achieves a 98.7% detection rate for known ransomware while effectively adapting to new variants, maintaining a 95% detection rate for previously unseen threats. Additionally, DTIA operates with minimal impact on system performance, showcasing scalable and efficient detection capabilities suited for deployment in environments with diverse computational resources. These findings demonstrate the potential of DTIA as a sophisticated and adaptable solution for the evolving challenge of ransomware detection, enhancing cybersecurity resilience through advanced behavioral analysis and dynamic learning capabilities.

This paper introduces Convergent Hierarchical Reasoning (CHR), a unified artificial intelligence (AI) framework that integrates four advanced reasoning paradigms-Chain-of-Thought (CoT), Forest-of-Thought (FoT), Trilevel Reasoning Graph (TRG), and Logarithmic Memory Networks (LMNs)-into a single, plug-and-play module. CHR offers end-to-end interpretability, multi-tree exploration, multi-tiered decision-making, and efficient long-sequence handling, making it especially suitable for high-stakes government applications. Agencies often struggle with strict compliance requirements, data sensitivity, and resource constraints when implementing AI for large-scale training, simulation, and decision support. By merging CoT, FoT, TRG, and LMNs into a seamless system, CHR alleviates the complexity and security risks associated with manually integrating multiple modules. Its architecture facilitates one-stop deployment within FedRAMP-approved cloud environments, incorporating real-time monitoring and automated checks aligned with NIST and FIPS standards. CHR A Preprint Early evaluations indicate significant improvements in both interpretability and computational efficiency, vital for applications ranging from military readiness to cybersecurity threat analysis. With parallel reasoning trees (FoT) and stepwise clarity (CoT) operating under a tiered decision-making structure (TRG), CHR ensures that high-level goals inform operational tactics. Meanwhile, LMNs efficiently manage extended historical data, minimizing computational overhead for long-sequence tasks. Although CHR is primarily designed for government and defense settings, its modular and transparent nature makes it equally valuable in industry and academia. By unifying advanced reasoning with compliance-focused MLOps, CHR demonstrates a scalable path for adopting next-generation AI, offering robust oversight and clear accountability in mission-critical scenarios.

Deep auto-encoders (AEs) are widely employed deep learning methods in the field of anomaly detection, especially detection of security malicious activities. Cybersecurity analysts face challenges in managing a large volume of alerts, often constrained by limited processing resources. In response, various strategies, including false positive reduction, human-in-the-loop and alert prioritization, are employed. This paper explores the integration of uncertainty quantification (UQ) methods into alert prioritization for anomaly detection using an ensemble of AEs. UQ models recognize doubtful classification decisions, aiding analysts in treating the most certain alerts first (as a more certain alert is more likely to be accurate). Our study reveals a nuanced issue where applying UQ to an ensemble of AEs can lead to skewed distributions of large reconstruction errors, falsely indicating large standard deviation uncertainties on certain classification decisions. Contrary to intuition, AEs suggest that large reconstruction errors could indicate small uncertainties. To address this, we propose an extension for obtaining a calibrated standard deviation uncertainty distribution, mitigating erroneous alert prioritization. Evaluation on 6 benchmark intrusion detection datasets demonstrates that our proposed calibration approach enhances UQ methods' ability to prioritize alerts with a favorable trade-off in other invaluable performance metrics.

The escalating sophistication of cyber threats necessitates innovative detection mechanisms to safeguard digital infrastructures. Dynamic Enclave Partitioning (DEP) emerges as a novel approach, offering real-time isolation and monitoring of suspicious code activities within virtualized environments. DEP's adaptive segmentation enables continuous adjustment of enclave boundaries based on observed behavioral patterns, effectively distinguishing between benign and malicious processes. This methodology enhances detection accuracy by minimizing false positives and provides a robust defense against evolving ransomware tactics. The modular design of DEP facilitates seamless integration with existing security frameworks, extending its applicability beyond ransomware to a broader spectrum of malware threats. Comprehensive evaluation demonstrates DEP's efficacy in identifying and isolating novel ransomware behaviors, showing its potential as a significant component in contemporary cybersecurity strategies.

Network Function Virtualization (NFV) enables the implementation of Service Function Chains (SFCs), where a set of ordered Virtual Network Functions (VNFs) support diverse network services over resources. However, deploying SFCs across multiple domains faces challenges due to resource limitations and the confidentiality of domain-specific data. Achieving lowlatency SFC placement, followed by the deployed network service performance guarantee, is under research. Maintaining low endto-end (E2E) latency of network service delivery, while preserving the privacy of configuration details among Network Function Providers (NFPs) is also challenging the cross-domain SFC deployment. This paper focuses on reducing the E2E latency of cross-domain SFCs deployment while considering VNF scaling with data privacy. First, a Federated Learning with Graph Neural Network (FLGNN) approach is introduced among different NFPs to optimise VNF placement and SFC deployment while preserving data privacy. Then, a Viterbi-based SFC partitioning (VSFCP) method is proposed, which segments the SFC into sub chains based on FLGNN predictions, optimising E2E latency. After deploying the cross-domain SFC, a VNF auto-scaling method with transfer learning (TLVS) is presented to handle dynamic service demands. Results show that cross-domain SFC deployment latency decreases by up to 30.8%. VNF resource prediction error reduces by 39.3% to avoid service latency.

At any given time, a battery unit can only be in either of two states: It is either charging or discharging. For a microgrid with a rolling time horizon partitioned into a fixed number of specific time intervals, an optimal battery power scheduling is therefore inevitably discrete in nature, and one may have to resort to very complex and time-consuming mixed integer programming techniques. In this paper, we show that this can in many cases be avoided by introducing small perturbing cost elements in an alternative convex formulation where the solution can be forced to take unique battery charge/discharge states (or configurations) without the need to employ integer variables. A formulation is then obtained which can be solved by using very efficient disciplined convex programming techniques. The new approach is demonstrated by formulating an explicit theory comprising of eight propositions with proofs which are associated with a small generic case consisting of one battery unit, two renewable energy sources, a general-purpose power grid, and one consumer. The general procedure can be straightforwardly extended to higher dimensions. Numerical examples are included to illustrate the theory based on a 24-hour spot price scenario. In particular, it is also demonstrated how a suitably chosen convex battery cost can provide an efficient threshold to avoid unnecessary battery use, and thereby promote a longer lifetime.

This research is in the domain of Natural Language Processing and delves into statistical methods that are intrinsic to the input text, that is not needing any external corpus or training. The methods are used to calculate similarity of sentences in Hindi. The study is performed on sentence pairs in English and Hindi and the differences are analysed with the structural and grammar attributes of the languages. The methods chosen mathematically operate at different levels from character to phrase. The purpose is to identify methods at which a language performs differently than the other, and how different they are with the actual truth. The conclusions can be elemental when considering building larger applications that may need to use one of these methods or hybrid of these methods. An analysis on various methods applied on a sentence pair dataset with both the languages Hindi and English is conducted using visualisation of the produced output of these methods against their actual scores.

A pure weight-based test case generator for context-free grammars is introduced, where weights are set for each rule and used as unique driver to identify which one to apply during the derivation process; we demonstrate why static flat weights generally cannot work whereas, introducing a simple penalty mechanism, a dynamic weight-based approach is always able to produce random but finite sentences; then, looking at the source grammar from an original perspective, we proceed towards the definition of constraints through which, while still guaranteeing the convergence of the derivation, static but balanced weights can be identified: balanced weights own the interesting capability to drive the generation of an inexhaustible number of extremely random, structurally complex but finite sentences by an automatic test case generator having a dramatically simple derivation logic. Thereafter, we illustrate how to design a weight-based system which, by exploiting the characteristics of balanced weights, can be easily configured to produce sentences that address any specific coverage metric, including those not having a definite generative strategy yet. A complete example of the application of the balanced concept and the weight-based system is then included and experimental results are eventually shown.

Rainfall forecasting has recently attained the highest level of academic relevance, due to its intricacies and enduring uses, including flood forecasting and pollutant concentration level monitoring. It is difficult to predict rainfall due to its complicated nature, which is impacted by a number of meteorological, oceanic, and geographical elements. In order to get over these limitations, methods that combine time-series data with machine learning techniques are being investigated. The goal of this project is to apply machine learning algorithms to reliably forecast rainfall. Current models rely on intricate statistical models that are frequently excessively expensive, both financially and computationally, or aren't used in downstream applications. Using simple rainfall estimating models based on traditional machine learning methods and Deep Learning architectures that are effective for various downstream applications, this work offers a comparative examination. In order to capture both short-term and long-term rainfall trends, we test a variety of machine learning models, such as Random Forest, Support Vector Machines, and Long Short-Term Memory (LSTM) networks. Additionally, the study investigates how data preprocessing methods and hyperparameter adjustment can improve the model's performance. Real-time meteorological data from several geographical locations was used to validate the model, and the findings showed that it could make predictions more accurately than conventional statistical techniques.

Silicon photonics is the most viable and cost-effective platform for next generation optical interconnects. One of the key optical devices in these interconnects is an optical modulator which, along with photodetectors, effectively defines performance of the whole optical link. However, due to its centrosymmetric structure, silicon does not offer linear electro-optic effects, leaving plasma dispersion effect as the only feasible modulation mechanism for high speed applications. The performance of plasma dispersion-based modulators is affected by several design variables, affecting various modulator figures of merit (FOMs) where performance trade-off exists. This necessitates design optimization, requiring reasonably accurate mathematical predictive models for all the FOMs. This paper presents a rigorous full 3D model for the junction capacitance of silicon-on-insulator interleaved junction optical phase modulators with submicron dimensions. Junction capacitance is instrumental in calculating power consumption per bit, modulation speed and efficiency of optical modulators. Solution of Drift-Diffusion and Poisson's equations were carried out on 3D finite-element-mesh and Maxwell's equations were solved using Finite-Difference-Frequency-Domain (FDFD) method on non-uniform rectangular grid to calculate confinement factor. Whole of the modeling process has been detailed and all the coefficients required in the model are presented. Model validation suggests < 10 % RMS error.

In this paper, we present "Comprehensify," a Chrome extension designed to enhance learning by providing personalized content summarization using large language models (LLMs). By analyzing user knowledge levels across different topics and reading web page content, Comprehensify generates summaries tailored to the user's comprehension level. We integrate various models, including GPT-4o, GPT-4o-mini, LLaMA 3, and Claude 3, and implement a client-side cache as well as a server-side cache to optimize performance. Additionally, user authentication is employed to ensure personalized experiences. Our evaluation shows promising results in providing effective learning aids, with potential future enhancements to further improve the system's functionality.

This paper develops a power-based microscopic energy consumption model for battery electric vehicles (BEVs) entitled Virginia Tech Comprehensive Power-based Energy Consumption Model 2.0 (VT-CPEM 2.0). This model avoids bang-bang control, is calibrated using US Environmental Protection Agency (EPA) city and highway energy consumption ratings; along with publicly available vehicle and roadway pavement parameters; and captures regenerative braking. The model validation considers three BEVs and three drive cycles. The test results show that the estimated energy consumption rates by the proposed model are consistent with in-laboratory data collected by the Argonne National Laboratory producing an average error of 3.9% and a maximum error of less than 7.5% for three vehicles across three drive cycles. Additionally, the model demonstrates that the optimal speed of BEVs increases with an increase in the auxiliary power.

Robotic devices offer assistance to individuals with upper-limb impairments. Among different control strategies, assistance should specifically account for the weight of both the device and the wearer's arm without introducing any distortion to the kinematic behaviour. This study investigates how varying weight support levels in the Harmony exoskeleton impact hand spatio-temporal features and inter-joint coordination during a pick-and-place task. Eight healthy subjects performed the selected functional movement across five distinct levels of arm support. Results indicated that movement smoothness and planning in the Cartesian space were not influenced by the Harmony exoskeleton. A decreased accuracy was noted only at high levels of weight support for movements against gravity, while movements propelled by gravity suffered an overall drop. Consistent inter-joint coordination was observed across different support levels, except for shoulder intra-extra rotation during movements against gravity. This study shed light on the effect of gravity compensation provided by the Harmony exoskeleton on kinematic performance, providing valuable insights for optimizing assistive devices for individuals with upper-limb impairments.