The inherent hysteresis nonlinearity of piezoactuated positioning stages (piezo-stages) is very difficult to deal with due to the amplitude-and frequency-dependent characteristics, which severely limits the tracking accuracy for high speed trajectories. In this work, we first develop a deep serial model to describe the dynamics of the piezo-stage by using historical voltage-displacement data over a period of time. It achieves relative prediction errors less than 0.16% on sinusoidal trajectories with frequencies greater than 72% of the resonance frequency of the piezo-stage through an elaborately designed network structure that includes direct connections between the input layer and the output layer. Then, we design an integral model predictive control (iMPC) and build a feedforward neural network (FNN) to learn its optimal solution offline. This forms the proposed FNN-iMPC and ensures the feasibility of evaluating the control law within the sampling time of 0.1ms. It achieves a maximum positioning error of 0.02µm for a ±32µm staircase reference signal and a maximum tracking error of 0.19µm for a sinusoidal reference signal with a range of 10µm and a frequency of 500Hz in real experiments.

True time delay circuits are generally realized using variable delay all-pass filters due to their flat delay-frequency response. However, they are power hungry and practically, variations in transistor characteristics and parasitic impedances result in delay variations with frequency. This paper discusses a first-order RC-low pass filter as a TTD circuit, which leverages the fact that the phase response starts changing linearly 1 decade before the pole location. The variable delay is accomplished by changing the circuit's effective impedance using the Miller effect obviating the need for varactors or variable resistors. A bandwidth extension of the delay cell is achieved using a delay correction circuit. A 4-bit TTD circuit realized by cascading 15 stages of variable delay LPFs has been designed in TSMC 65 nm CMOS technology and post-layout simulation results show that the TTD has a power consumption of 36 mW and a delay range of 400 ps with a delay variation of ±5 ps across a bandwidth of 1 GHz.

"Solar circuits" is a nomenclature for a photovoltaic panel, power manager, and microprocessor integrated on a System on a Chip that meet a predefined thermal design power envelope and volumetric design envelope for energy and space-constrained mobile devices such as cell phones, tablets, and laptops. The emerging market of Internet of Things, for over two decades, has prioritized interfaceless devices to either operate within extreme energy budgets or to rely on remote clients/terminals for offloading compute-intensive tasks. An emphasis on designing with energy generation and integration with interfaces in solar circuits expands the accessibility of IoT and mobile computers to developing markets as well as solutions requiring a "local first" design ethos. Solar circuit design will be contextualized within the precedent of Software Defined Hardware to specify an autonomous or batteryless capability requirement that utilizes energy harvesting or energy scavenging to meet "just enough" application needs. Furthermore, setting limits on system power consumption allows a solar circuit to serve as a proxy or "litmus test" for measuring (and subsequently demonstrating) the progress of Moore's Law within solar constraints. Performance of general purpose microprocessors that can be solar powered in mobile form factors is estimated to be consistently 25-30 years behind (in MIPS/W/mm^2) leading edge chips such as AMD Instinct and Apple A17. A number of heuristic methods are described in this paper to estimate the solar circuit requirements (in transistors, memory, and OPS) to run general purpose operating systems such as Microsoft Windows 95 and Linux 5.1 in modern, commercially viable process nodes, such as Global Foundries 22nm FD-SOI, TSMC's 22 and 12nm ULL, or Intel's 16nm FF - processes that all utilize low leakage.

With the proliferation of Zero-Trust Architecture (ZTA) [1]-[3], a paradigm evolving from traditional, perimetercentric approaches to network security, there is a growing need for vocabulary that clearly identifies the nuances of existing design elements being implemented in ZTA. This paper introduces Zero-Trust Microsegmentation (ZTM) as a formal model to describe the practice of designating a gateway security component as the Policy Enforcement Point (PEP) within a ZTA. While the underlying technique is not new [4]-[6], defining it more explicitly mitigates conceptual overlap and clarifies how ZTM aligns with zero-trust principles by ensuring that east-west traffic is consistently inspected, logged, and subject to unified policy enforcement. Through a focused discussion on defining this practice-shifting routing to a gateway firewall capable of advanced, context-aware inspection-this paper demonstrates how ZTM simplifies rule management and auditing for operational teams, while providing organizational stakeholders with clearer insights into internal security measures. By offering a concise, standardized label for a widespread yet seldom-named practice, ZTM serves to streamline architectural discussions, strengthen regulatory compliance efforts, and support future developments in network segmentation strategies.

In this work, we extend the standard quasi-Helmholtz filters to high-order discretizations in the context of the boundary element method. This generalization allows preconditioning integral equations such as the electric field integral equation (EFIE) in the h-refinement regime while preserving the accuracy gain of high-order discretizations. The theoretical framework will be corroborated by numerical results that validate the effectiveness of the proposed strategy when stabilizing the refinement-dependent spectral behavior of the high-order discretized EFIE.

The advent of large language models (LLMs) in the education sector has provided impetus to automate grading short answer questions. LLMs make evaluating short answers very efficient, thus addressing issues like staff shortage. However, in the task of Automated Short Answer Grading (ASAG), LLM responses are influenced by diverse perspectives in their training dataset, leading to inaccuracies in evaluating nuanced or partially correct answers. To address this challenge, we propose a novel framework, Grade Guard. • To enhance the task-based specialization of the LLMs, the temperature parameter has been fine-tuned using Root Mean Square Error (RMSE). • Unlike traditional approaches, LLMs in Grade Guard compute an Indecisiveness Score (IS) along with the grade to reflect uncertainty in predicted grades. • Introduced Confidence-Aware Loss (CAL) to generate an optimized Indecisiveness Score (IS). • To improve reliability, self-reflection based on the optimized IS has been introduced into the framework, enabling human re-evaluation to minimize incorrect grade assignments. Our experimentation shows that the best setting of Grade Guard outperforms traditional methods by 19.16% RMSE in Upstage Solar Pro, 23.64% RMSE in Upstage Solar Mini, 4.00% RMSE in Gemini 1.5 Flash, and 10.20% RMSE in GPT 4-o Mini. Future work includes improving interpretability by generating rationales for grades to enhance accuracy. Expanding benchmark datasets and annotating them with domain-specific nuances will enhance grading accuracy. Finally, analyzing feedback to enhance confidence in predicted grades, reduce biases, optimize grading criteria, and personalize learning while supporting multilingual grading systems will make the solution more accurate, adaptable, fair, and inclusive.

Explainable Artificial Intelligence (XAI) emerged to reveal the internal mechanism of machine learning models and how the features affect the prediction outcome. Collinearity is one of the big issues that XAI methods face when identifying the most informative features in the model. Current XAI approaches assume the features in the models are independent and calculate the effect of each feature toward model prediction independently from the rest of the features. However, such assumption is not realistic in real life applications. We propose an Additive Effects of Collinearity (AEC) as a novel XAI method that aim to considers the collinearity issue when it models the effect of each feature in the model on the outcome. AEC is based on the idea of dividing multivariate models into several univariate models in order to examine their impact on each other and consequently on the outcome. The proposed method is implemented using simulated and real data to validate its efficiency comparing with the a state of arts XAI method. The results indicate that AEC is more robust and stable against the impact of collinearity when it explains AI models compared with the state of arts XAI method.

The increasing complexity and diversity of natural language inputs have posed significant challenges to the contextual understanding capabilities of contemporary language models. Dynamic Pattern Alignment (DPA) emerges as a novel approach that dynamically aligns detected semantic patterns within input data with the model's internal representations, thereby enhancing semantic transfer and context-aware processing. This study presents a comprehensive examination of DPA's integration into an open-source language model, detailing the algorithmic framework and modifications to the model architecture. An automated testing environment was employed to evaluate performance gains and context consistency, utilizing diverse datasets and standardized preprocessing techniques. The experimental results demonstrate that DPA significantly improves semantic retention across contextually dynamic inputs, reduces computational load, and enhances response times. Comparative analyses with baseline models demonstrate DPA's superiority in semantic transfer and contextual relevance. These findings suggest that DPA offers a promising advancement in the development of context-aware language models, with potential applications extending beyond the current scope.

Current probes have long been used for Electromagnetic Compatibility (EMC) pre-compliance and diagnostic purposes. Dipole and transmission line-based models are often used to relate Common-Mode (CM) current levels to potential Radiated Emission (RE) levels. This relationship enables RE estimates to be performed without the use of accepted certification methodologies – reducing project costs and timelines. These models are based on a set of assumptions which cannot fully match real-world scenarios. This article expands the discussion by examining the relationship between the CM current and RE of a representative metallic structure using both measurements and a Finite Element Method (FEM) code. The relationship is shown to be frequency dependent, resulting in both over and under prediction of the RE levels. A practical recommendation is made on how to exploit the equivalence of CM current-based estimations of system RE.

Observations made using antenna arrays are often limited by the noise coming from the antenna system itself and from the environment. In demanding applications such as radioastronomy, the thermal noise emitted by the multilayered substrate below the antenna array may become significant and needs to be properly characterized. Both the intensity of the noise at each port and the correlation of the noise at pairs of ports need to be quantified. In this paper, we provide a spectral formulation to calculate the noise correlation matrix of antenna arrays. The method is based on the propagative and evanescent plane wave spectrum emitted by each pair of active antenna ports into the lossy layered medium, and can thus be implemented as a post-processing step combined with any full-wave numerical solver. This formulation allows us to account for layered media with different physical temperatures at each layer. Numerical examples are provided using Square Kilometer Array (SKA) stations lying on a finite ground plane, itself lying on a layered semi-infinite soil. The behaviours of noise power and noise correlation coefficients are studied for different positions of the antennas on the ground plane. This is done considering a multilayered soil with different moisture levels. A very good agreement with the commercial software FEKO is observed regarding the evaluation of the noise correlation coefficients. Moreover, the efficiency of the algorithm in terms of computational time is shown by comparison with FEKO. The direction-dependent system noise temperature picked up by a full SKA station in presence of the soil is also given.

The ever growing demand for services with high requirements is a significant force behind the evolution of cellular networks. End-to-End (E2E) analysis is a key methodology for the assesment of service performance, often paired with Machine Learning (ML) algorithms. However, ML comes with its own security challenges. This work will study attaks on the data, some of the major security challenges in ML systems. Namely, it will study poisoning and evasion in ML regressors, offering different perspectives on their impact in the outcome of the system within a mobile network and the effects on user perception. Moreover, new metrics for computing the statistical variation of model predictions are proposed to evaluate the attack effects. A realistic testbed is used to evaluate the proposed mechanisms.

We propose in this paper a novel hierarchical multi-grid orthogonal matching pursuit based angle estimation scheme. The proposed scheme accurately estimates the angles of departure (AoDs) and the angles of arrival (AoAs) of a point-to-point ultra massive multiple input multiple output (UM-MIMO) system, ensuring ultra-resolution estimation accuracy. Specifically, we define the on-grid error range in the conventional compressive sensing (CS) estimators and build high-resolution grids centered at the estimated discrete on-grid angles. The proposed scheme proceeds hierarchically through multiple levels to refine the angle estimations. Numerical simulations demonstrate the efficacy and superiority of the proposed scheme over recent state-of-the-art literature, where sub-millidegree angle estimation accuracy can be attained.

A document by JASJIT SINGH. Click on the document to view its contents.

This paper proposes a stacked ensemble framework for path loss prediction in complex urban wireless environments. The approach integrates five base learners-Linear Regression, Random Forest, Decision Tree, Gradient Boosting, and Support Vector Regression-combined using a neural network meta-learner to enhance predictive accuracy. Drive test data from two Nigerian cities were preprocessed and used for model training, with k-fold cross-validation and random search applied for hyperparameter tuning. Evaluation metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R²), confirm that the stacked model improves prediction accuracy by 9.5% over the best-performing base model. The model's generalization was further validated on an independent dataset. These findings demonstrate the advantage of ensemble learning in improving path loss prediction accuracy for 5G network planning.

Adversarial attacks on Natural Language Processing (NLP) models expose vulnerabilities by introducing subtle perturbations to input text, often leading to misclassification while maintaining human readability. Existing methods typically focus on word-level or local text segment alterations, overlooking the broader context, which results in detectable or semantically inconsistent perturbations. We propose a novel adversarial text attack scheme named Dynamic Contextual Perturbation (DCP). DCP dynamically generates context-aware perturbations across sentences, paragraphs, and documents, ensuring semantic fidelity and fluency. Leveraging the capabilities of pre-trained language models, DCP iteratively refines perturbations through an adversarial objective function that balances the dual objectives of inducing model misclassification and preserving the naturalness of the text. This comprehensive approach allows DCP to produce more sophisticated and effective adversarial examples that better mimic natural language patterns. Our experimental results, conducted on various NLP models and datasets, demonstrate the efficacy of DCP in challenging the robustness of state-of-theart NLP systems. By integrating dynamic contextual analysis, DCP significantly enhances the subtlety and impact of adversarial attacks. This study highlights the critical role of context in adversarial attacks and lays the groundwork for creating more robust NLP systems capable of withstanding sophisticated adversarial strategies.

In this paper we continue our previous investigation about the use of stress function in the flow of generalized Newtonian fluids through conduits of circular and noncircular (or/and multiply connected) cross sections where we visualize the stages of yield in the process of flow of viscoplastic fluids through tubes of elliptical, rectangular, triangular and annular cross sections. The purpose of this qualitative investigation is to provide an initial idea about the expected yield development in the process of flow of yield-stress fluids through tubes of some of the most common non-circular (and non-simply-connected) cross sectional geometries. [1]

Handwritten document recognition (HDR) is one of the most challenging tasks in the field of computer vision, due to the various writing styles and complex layouts inherent in handwritten texts. Traditionally, this problem has been approached as two separate tasks, handwritten text recognition and layout analysis, and struggled to integrate the two processes effectively. This paper introduces HAND (Hierarchical Attention Network for Multi-Scale Document), a novel end-to-end and segmentation-free architecture for simultaneous text recognition and layout analysis tasks. Our model's key components include an advanced convolutional encoder integrating Gated Depth-wise Separable and Octave Convolutions for robust feature extraction, a Multi-Scale Adaptive Processing (MSAP) framework that dynamically adjusts to document complexity and a hierarchical attention decoder with memoryaugmented and sparse attention mechanisms. These components enable our model to scale effectively from single-line to triple-column pages while maintaining computational efficiency. Additionally, HAND adopts curriculum learning across five complexity levels. To improve the recognition accuracy of complex ancient manuscripts, we fine-tune and integrate a Domain-Adaptive Pre-trained mT5 model for postprocessing refinement. Extensive evaluations on the READ 2016 dataset demonstrate the superior performance of HAND, achieving up to 59. 8% reduction in CER for line-level recognition and 31. 2% for page-level recognition compared to stateof-the-art methods. The model also maintains a compact size of 5.60M parameters while establishing new benchmarks in both text recognition and layout analysis. Source code and pre-trained models are available at https://github.com/MHHamdan/HAND.

This paper presents a method for locating a wireless signal source using signal strength measurements taken along the border of a 256x256-m2 area. The method leverages a deep Residual Neural Network (ResNet) to predict the location of the transmitter within the area of interest. This approach reduces data collection and computational overhead associated with traditional localization methods. The method is validated through simulated data as well as measurements at 2.7 GHz, with an average error of 7.23 m and a standard deviation of 3.32 m.