Clustering in data mining involves grouping similar objects into categories based on their characteristics. As the volume of data continues to grow and advancements in highperformance computing evolve, a critical need has emerged for algorithms that can efficiently process these computations and exploit the various levels of parallelism offered by modern supercomputing systems. Exploiting Single Instruction Multiple Data (SIMD) instructions enhances parallelism at the instruction level and minimizes data movement within the memory hierarchy. To fully harness a processor's SIMD capabilities and achieve optimal performance, adapting algorithms for better compatibility with vector operations is necessary. In this paper, we introduce a vectorized implementation of the Density-based Clustering for Applications with Noise (DBSCAN) algorithm suitable for the execution on both shared and distributed memory systems. By leveraging SIMD, we enhance the performance of distance computations. Our proposed Vectorized HPDBSCAN (VHPDBSCAN) demonstrates a performance improvement of up to two times over the state-of-the-art parallel version, Highly Parallel DBSCAN (HPDBSCAN), on the ARM-based A64FX processor on two different datasets with varying dimensions. Additionally, we evaluate VHPDBSCAN's energy consumption on the A64FX and Intel Xeon processors. The results show that the proposed implementation reduces energy consumption by a factor of two on the A64FX Central Processing Unit (CPU) and by approximately 19.5% on the Intel Xeon 8368 CPU compared to previous methods.

Ultra-Wideband (UWB) technology re-emerges as a groundbreaking ranging technology with its precise microlocation capabilities and robustness. However, the security aspects of UWB technology demand thorough scrutiny due to its widespread use in both consumer and industrial sectors. This white paper highlights the security dimensions of UWB technology, focusing in particular on the intricacies of device fingerprinting for authentication, examined through the lens of state-of-the-art machine learning techniques. Furthermore, we explore various potential enhancements to the UWB standard that could realize a sovereign UWB data network. We argue that UWB data communication holds significant potential in healthcare and ultra-secure environments, where the use of the common unlicensed 2.4 GHz band-centric wireless technology is limited or prohibited. A sovereign UWB network could serve as an alternative, providing secure localization and short-range data communication in such environments.

This study introduces a sophisticated supervised machine learning method for electric theft detection utilizing a customized Histogram Gradient Boosting (HGB) algorithm. Comprehensive preprocessing, including imputation, normalization, outlier management, and resampling, ensures the timeseries data is accurately prepared for analysis. The SMOTE-ENN algorithm corrects class imbalances, preparing the data for the feature optimization stage where crucial features are selected and extracted. The HGB algorithm, enhanced through Bayesian optimization, is central to the training process, resulting in a model that precisely classifies electricity consumption patterns as genuine or fraudulent. The robustness of the model is assessed against other recognized boosting methods, such as Adaptive Boosting (ADB), Gradient Boosting Decision Tree (GBDT), and LightGBM, alongside various ensemble and traditional machine learning models. Utilizing key performance metrics like accuracy, F1 score, and AUC for validation, the proposed model yields very promising results, with a 93% accuracy, 95% F1 score, and 98% AUC, outperforming the comparison group under similar dataset and hyperparameter conditions. This underscores the model's potential as a highly accurate tool for combating electricity theft within an advanced metering infrastructure (AMI).

Modern software systems intensively use the Agile software development processes for their development and maintenance. The Agile development methodology encourages customer satisfaction, early incremental delivery, and overall development simplicity. Agile development methods accept changes in requirements and technology and use a more adaptive or iterative approach to planning. With the adaptation of Agile process models in the development of modern software systems, there is a need for continuous improvement in the Agile processes. Agile development processes are modified and upgraded by utilizing software metrics. There are several proposed software metrics for measuring performance and quality in Agile software development. These include customer satisfaction, story point estimation, velocity, test coverage, defects in production, and other metrics. Each software metric has its own merits and demerits. This research aims to provide a comprehensive review of published research work on software metrics. Specifically, it summarizes the software metrics used for measuring the performance of Agile process models. This research paper will help understand the usefulness of various software metrics used in Agile development, and it will serve as a foundation for future research in software metrics for Agile software development.

Engineering education is a complex that involves multiple stakeholders, including students, educators, administrators, and industry partners. It is continuously evolving to meet the demands of modern industry and society. The traditional teaching and learning methodologies are being replaced by a more integrated skillset that focuses on developing students' cognitive, social, and emotional skills. The shift towards this integrated approach is gaining momentum, and it is important to have a framework that has been proven to solve complex systems. The usage of systems engineering tools to model engineering education systems is not often seen. In this paper, a novel application of Model-Based Systems Engineering using Systems Modeling Language (SysML) to develop an Engineering Learning Analytic System (ELAS) framework that consists of multi-dimensional elements related to educational systems. The core of this study involves a rigorous Requirements Verification and Validation (V&V) process to ensure stakeholder needs which systematically were map with system capabilities. ELAS model simulations provided predictive insights into soft skill development, enabling decision-making via targeted interventions that could significantly enhance students' skill sets. ELAS highlights that a data-driven approach, enabled by SysML, significantly enhances the ability to enact timely by relevant interventions at various levels of the educational management process. The proposed ELAS model offers a strategic blueprint for continuous improvement within educational institutions, demonstrating a pathway toward a responsive and self-improving educational system. The refining of the ELAS model, for broadening simulation scopes, and further integrating predictive analytics into administrative decision-making processes is an ongoing endeavor.

Passive channel sounding in live cellular network, where downlink signals of in-service base stations (BSs) are acquired to extract multipath propagation parameters, has attracted great interest from academia and industry in recent years. In this work, downlink signals of live 5G new radio (NR) multi-cell BSs were measured with a recent home-developed multi-antenna passive channel sounder. It was found out that the performance of multipath parameter extraction algorithm is severely affected by the existence of many interfering neighbor cells (which share the same frequency band as the serving cell) in the multi-cell measurements. For this reason, a novel algorithm based on the interference cancellation principle is proposed, which can effectively mitigate the impact of multicell interference and thus improve the parameter extraction performance. The effectiveness and robustness of the proposed algorithm is finally demonstrated by the passive channel sounding campaigns in live 5G NR multi-cell cellular network.

This study presents a generalized model predictive torque control (GMPTC) scheme that applies to all types of synchronous machines (SMs), which are used for vehicle traction. The GMPTC problem is formulated to simultaneously minimize the torque error and the performance index (PI) while satisfying the voltage and current constraints. The PI can be defined by any function to be minimized to enhance the SM drive's performance, such as copper loss and inverter loss. When the torque command is not achievable due to either of the constraints, the problem is defined differently to maximize the torque magnitude under the two constraints. The GMPTC problem, a nonlinear optimization problem, is solved based on either a continuous control set (CCS) or finite control set (FCS) using the augmented Lagrangian method. The GMPTC scheme guarantees optimal operation under all operating regions, including the maximum torque per ampere, flux weakening, maximum current, and maximum torque per voltage, even without additional controllers. The GMPTC method is practical and easy to implement because it involves one optimization problem with a small number (four to five) of tuning parameters. This aspect differs from existing model predictive control schemes for SMs that involve an additional optimization, usually solved offline, to obtain optimal references for the stator currents or stator flux linkages. The effectiveness of the proposed GMPTC method is demonstrated numerically and experimentally by controlling interior permanent magnet synchronous machines based on both the CCS and FCS.

In this work, we address the key industry challenge of accurately capturing the propagation environment in live 5G commercial networks, which is essential for predicting radio device performance in realistic propagation conditions during the design phase. An advanced passive channel sounding system is designed to decode 5G downlink signals, such as the channel state information-reference signal (CSI-RS), and to resolve multipath parameters including delays, arrival angles, polarization, and complex amplitudes. To ensure measurement accuracy, the designed passive channel sounding system is rigorously calibrated and validated within a compact antenna test range (CATR) prior to field deployment. Furthermore, we propose a digital twin (DT) framework to create precise digital representations of electrical signals received on antennas of investigation through highprecision simulations, ensuring alignment with actual conditions among various locations and orientations for the antennas. The methodology begins with the precise digitization of 5G fading channel environments via the passive channel sounding process. Utilizing the acquired channel parameters and the complex radiation patterns of the targeted antennas, the DTs of the received electrical signals for the antennas can be computed. Field measurements, conducted using a standard dipole for electrical field scanning, validate the consistency between the DT predictions and measured results, thereby affirming the accuracy of both the passive channel sounding results and the proposed DT framework.

In the dynamic world of financial markets, investors and traders alike seek tools and methodologies that can provide insights into potential price movements and help them make informed decisions. Technical analysis stands at the forefront of such methodologies, offering a systematic approach to interpreting historical price data and identifying trends and patterns that may guide future market behavior.

Recently myokinetic interfaces have been proposed to exploit magnet tracking for controlling bionic prostheses. This interface derives information about muscle contractions from permanent magnets implanted into the amputee's forearm muscles. Machine learning models have been mapped on Field Programmable Gate Arrays (FPGAs) to track a single magnet, achieving good precision and computational efficiency, but consuming a large area and hardware resources. To track several magnets, here we propose a novel solution based on dynamic partial reconfiguration, switching three prediction models: a linear regressor, a radial basis function neural network, and a multi-layer perceptron neural network. A system with five magnets and 128 magnetic sensor inputs was used and experimental data were collected to train a system with five hardware predictors. To reduce the complexity of the models, we applied principal component analysis, ranking by correlation the number of inputs of each model. This run-time reconfigurable solution allows the circuits to be reconfigured in order to select the most reliable predictor model for each magnet while the rest of the circuit continues to operate extracting the most significant information from the captured signals. Thus, the proposed solution remarkably reduces the hardware occupation and improves the computational efficiency compared to previous solutions.

This study investigates AC electrical aging induced by partial discharge (PD) in fluoropolymer cable insulation materials, namely fluorinated ethylene propylene (FEP) and perfluoroalkoxy (PFA), with the cross-linked polyethylene (XLPE) serving as the reference material. Utilizing a specialized electrical aging setup and comprehensive surface characterization techniques, we compared the aging behavior of FEP and PFA to that of XLPE under identical electrical discharge conditions. Over a three-week aging process, the stability of phase-resolved partial discharge (PRPD) patterns for FEP and PFA suggested minimal changes in surface chemistry and roughness, unlike XLPE. FTIR and Raman spectroscopy results confirmed minimal changes in the surface chemistry of FEP and PFA, while microscopic and profilometric analyses demonstrated notably fewer PD-induced alterations on their surfaces. These findings suggest a gradual aging effect on the electrical insulating properties of fluoropolymers, likely due to their resilient carbon-fluorine bonds, contrasting with the immediate and significant erosion and by-product formation observed in XLPE upon PD exposure.

Vision language models (VLMs) are transforming how we perceive and interact with visual data by bridging the gap between natural language understanding and visual perception. This paper provides a comprehensive overview of VLMs and their applications in text-based video retrieval and manipulation. It examines how these models leverage transformer architectures and self-attention mechanisms to learn joint representations of text and visual inputs. The paper traces the evolution of VLM pretraining techniques like ViLBERT, LXMERT, VisualBERT, and approaches for key tasks such as cross-modal retrieval, text-driven video manipulation, and zeroshot learning to generalize to unseen domains. It explores how VLMs can be integrated with other technologies like GANs and reinforcement learning to enhance capabilities. The survey also covers emerging areas like multimodal architectures that combine multiple modalities, video-language pretraining on video and text data, and generative VLMs for applications like text-to-image synthesis. Additionally, it discusses challenges and future directions related to multimodal fusion, interpretability, reasoning, and the ethical implications of developing and deploying VLMs. Overall, this paper provides a timely and comprehensive perspective on the rapidly evolving field of VLMs and their role in enabling multimodal AI systems.

Numerical analysis of high-temperature superconductors (HTS) commonly addresses AC losses in the superconducting REBCO layer. However, alternating magnetic fields also induce eddy current losses in the metallic non-superconducting (non-SC) substrate, silver, and copper layers of the HTS, which cannot be overlooked for certain operating temperatures and frequencies. Previously, non-SC losses have been estimated numerically for simple geometries at 77 K. Using a multilayer H-A-formulated model, this paper estimates the eddy current losses for a full-scale 2.5 MW superconducting aviation motor for frequencies between 50 and 1000 Hz and temperatures between 25 and 60 K. Furthermore, a sensitivity study is presented to show how AC losses are impacted by specific design parameters such as the copper purity, angle of the external field, tapes per turn, tape height, and copper layer thickness. The results show that eddy current losses are negligible for high HTS temperatures and low frequencies. However, at low temperatures and high frequencies, these losses represent a significant share of the total, meaning that while they can be neglected for low-speed machinery, they must be considered for power-dense machinery, particularly if the planned cooling method involves liquid hydrogen (LH2) temperatures.

Over the past years, Fall Detection System (FDS) has undergone extensive research to improve living risk, especially for the elderly who are vulnerable to these fall events. Devices employing sensors are crucial components of FDS in achieving high accuracy and sensitivity. This article overviews different sensor modalities, such as ambient-based and vision-based systems, as well as commonly used wearable devices for fall detection, along with the associated data processing algorithms. The critical elements of fall detection, such as architectures and algorithms for processing sensor data, machine learning and deep learning methodologies, and validation of FDS performance, are considered. The article also delves into safety aspects and presents technical challenges and concerns in FDS for researchers in the field to identify areas requiring further improvement. Finally, future research opportunities to improve fall detection for widespread use are outlined.

The rapidly evolving landscape of Generative AI (GenAI) tools necessitate continuous vigilance and adaptation by educators. This dynamism requires stakeholders to stay up to date with developments to address emerging issues effectively, creating complexity in managing the responsible and ethical use of GenAI. This paper presents a pilot study involving the use of student learning agreements for governing GenAI use in a first-year engineering degree course. The learning agreement contains ethical and social considerations students agree to make if they decide to use GenAI in the course. As part of the pilot study, pre-post surveys and student artefacts – a group assignment adapted to accommodate GenAI use – were analysed. Results show that the vast majority of students were in favour of the learning agreement approach both at the start and upon completion of the course. However, 7 of 17 groups did not use GenAI in their assignment. Of the 10 groups that did, only 1 acknowledged GenAI limitations in adherence with the learning agreement. A thematic analysis of student suggestions for improving the learning agreement approach include suggestions for making the agreement easier to understand and adhere to (e.g., providing specific examples, reengaging with the agreement during the course). Overall, findings suggest that learning agreements have the potential to offer an interface through which student decision-making can be supported and interactions among students, educators, researchers, and policy makers related to the ethical and societal challenges of GenAI can take place.

This article presents a comprehensive investigation into the adaptability of state-of-the-art language models (LMs) to diverse domains through transfer learning techniques, evaluated using the General Language Understanding Evaluation (GLUE) benchmark. Our study systematically examines the effectiveness of various transfer learning strategies, including fine-tuning and data augmentation, in enhancing the performance of selected LMs across the spectrum of GLUE tasks. Findings reveal significant improvements in domain adaptability, though the degree of effectiveness varies across models, highlighting the influence of model architecture and pre-training depth. The analysis provides insights into the complexities of transfer learning, suggesting a nuanced understanding of its application for optimal model performance. The study contributes to the discourse on the potential and limitations of current LMs in generalizing learned knowledge to new domains, underscoring the need for more sophisticated transfer learning frameworks, diverse and comprehensive evaluation benchmarks, and future research directions aimed at improving model adaptability and inclusivity in natural language processing.

Recent advances in large language models (LLMs) have demonstrated their remarkable abilities in complex language tasks. However, achieving the depth and flexibility of human reasoning remains a challenge. Inspired by the power of step-by-step reasoning highlighted in "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models," this paper introduces the "Pyramid of Thought" framework. This framework leverages principles of Euclidean geometry, the Fibonacci sequence, and psychological engagement to create structured, adaptable AI responses while promoting transparent and intuitive communication.The "Pyramid of Thought" uses a dynamic layer mechanism, adjustable via  linear, logarithmic, or Fibonacci-based scaling.  This aligns the AI's thought process with natural mathematical patterns. Mirroring psychological principles, this approach aims to promote user engagement.  Each layer builds on the previous one, starting with foundational facts ("Five Ws") and progressing through reasoning steps ("How," "Then"), culminating in an "Apex" of analysis or insight. An inner thought journal maintains contextual richness.The flexible and iterative nature of the "Pyramid of Thought" reflects the emphasis on refinement and evolution found in conceptual modeling. Experiments will evaluate its effectiveness against traditional AI response methods. Metrics will include user engagement, response coherence, satisfaction, and the model's ability to communicate its reasoning process transparently. We anticipate that the "Pyramid of Thought" will produce more intuitive, meaningful, and thought-provoking interactions, aligning with the goals of conceptual modeling in simulation. This work contributes to the development of AI systems that better emulate human reasoning patterns and foster clearer communication.

Reliable analysis of the electromagnetic situation in a particular area is critical for wireless technology applications. With the increased number of devices and exploration of the new frequency bands, sensing helps, inter alia, in interference and radio resource management. Additionally, it facilitates the classification of signals to detect jammers and distinguish between legal and anomalous transmitters. This paper discusses the cooperative sensor network as a technology candidate for 6G. Specifically, we describe a possible architectural extension of the future cellular networks to obtain and utilize spectrum awareness. Several applications are systematically reviewed in this context, including dynamic spectrum sharing for coexistence in unlicensed bands, spectrum monitoring for signal detection and classification, and emitter localization. Qualitative and quantitative examples of possible performance improvements in cellular networks with the proposed approach are presented. Finally, we highlight the open research topics and challenges to solve for such a cooperative sensor network.