ABSTRACT. Telecommunications, healthcare, and global supply chains operate under constant pressure to process vast amounts of data while maintaining reliability, security, and regulatory compliance. When scalability is pursued without parallel ethical safeguards, the risk of privacy breaches, biased outcomes, and opaque automation increases. To address this, we propose a six- pillar framework that places human oversight, explainability, privacy-by-design, modularity, interoperability, and accessibility at the core of data system design. The framework draws on more than 17 years of practical experience across multinational pro- grams and a self-funded research initiative in digital health. Its applicability is demonstrated through three operational contexts: the shift to paperless workflows and automated Key Performance Indicator (KPI) pipelines in telecom networks, the harmonization of performance models in supply chains, and a confidential decision-support assistant for Type 1 Diabetes (T1D). A conceptual simulation of the latter indicated a 15–20% reduction in insulin dosing errors when human oversight was preserved, with response times under two seconds. The findings show that embedding ethical design principles into system architecture strengthens transparency, trust, and adoption, and provides a transferable blueprint for responsible Artificial Intelligence (AI) deployment across regulated, data-intensive domains.

ABSTRACT. Continuous Glucose Monitoring (CGM) provides detailed, minute-level insights into metabolic processes and holds significant promise for supporting overall well-being beyond traditional diabetes management. Despite its potential, CGM remains underused in detecting subtle physiological and lifestyle-related changes, such as variations in diet, physical activity, or hormonal cycles. To address this gap, an unsupervised anomaly detection framework is introduced to identify atypical daily glucose patterns without the need for labeled data. The approach encodes each day’s CGM profile as a high-dimensional vector, applies UMAP for dimensionality reduction, and uses DBSCAN to detect outliers. Applied to a two-month dataset from a single individual, the method revealed anomalies aligned with unlogged lifestyle changes. One event was independently confirmed through clinical records, underscoring the potential of CGM for passive, personalized monitoring and broader health tracking.

ABSTRACT. Future Railway Mobile Communication System (FRMCS) is a successor of the legacy railway communication system. Based on fifth generation mobile communication system, it enables automated and more efficient railway operation and enhances passenger services. The FRMCS critical and performance applications could be isolated from its underlying operating system and hardware using the technology of Network Function Virtualization (NFV). The paper presents an approach to the centralized management of virtualized FRMCS applications which simplifies deployment, improve security, and flexibility to access the FRMCS applications from different devices and locations. Basic management functionality related to FRMCS application lifecycle is identified and designed as a RESTful service.

ABSTRACT. Multicast broadcast solution can improve communication efficiency and provide the required reliability, coverage, latency, mobility, and scalability in the railway transport. The paper presents a method for the design of information recording and broadcasting as services for railway applications. The method feasibility is proved by modeling the state of a multicast broadcast session.

ABSTRACT. Quantum communication brings novel ideas and protocols to the field of infocommunications and networks, which imposes new challenges for various networking tasks such routing and link management. In our work we extend a previous model on the behavior of quantum networks and show that this modification reduces the necessary resources for managing a network of this kind.

ABSTRACT. In secure communications we are interested in protecting the content of the message. On the other hand, in the low probability of intercept (LPI) communications we are rather interested in the privacy of the communication process. The most restrictive version of LPI communications, in which the adversary shuts down the communication, once he detects the presence of the communication, is known as the covert communication. To hide the transmitted signal in solar radiation, we propose using the incoherent Gaussian source modulated with the phase-shift keying (PSK). To detect the transmitted signal, we propose transmitting portion of unmodulated carrier. To establish the link, we use the spatial light modulator (SLM)-based beam steering. To study the proposed concepts, we have developed a 1.5 km free-space optical link at the University of Arizona campus. We demonstrate the record number of covert bits transmitted in the saturation turbulence regime.

ABSTRACT. The article analyzes an integrated approach to calculating the threat to the information security of an organization. Based on the analysis, a mathematical model for assessing the risk of cyber threats to the information system has been built. The construction of this model allows for the calculation the security assessment of the information system. An application has been developed to calculate the level of threat to an organization using data from registered incidents.

ABSTRACT. This paper presents enhancements to the precision of the DOEC projection and improved control of transformation errors, enabling accurate spatial transformations on graphics processing units using single-precision arithmetic. The proposed contributions include: a systematic method for accuracy verification, a refined formulation of the transformation, and an approximation scheme for regions prone to high numerical error. The approach reduces mean transformation error by more than a factor of three and maximum error by two orders of magnitude. The results demonstrate that DOEC transformations can be reliably executed on GPUs with a mean error below 1m and a worst-case error of approximately 50m in polar regions, making this approach suitable for real-time satellite image reprojection and other geospatial applications.

ABSTRACT. The significant evolution of technologies over the last decades, especially those used in computer network design, has impacted the transition from a traditional network paradigm to a more modern and effective cloud-based network that overcomes the challenges and problems inherent in traditional network structures. This research provides a comprehensive systematic review of the evolution of cloud networking based on technological changes underlying these networks. A review of the developmental stages, from virtualization, SDN, NFV, and CNF, leading to cloud-based networks, is conducted. The analysis demonstrates the new concepts' impact on overcoming challenges within rigid traditional structures, including scalability, hardware costs, operational efficiency, and the adoption of innovation and new technologies. Despite the benefits of cloud networking, several challenges remain, such as latency, operability, cost-benefit analysis, and security concerns. Enterprises need to develop a precise plan for implementing the cloud concept, considering the extent of their need for cloud services, and must stay updated on emerging solutions and procedures for cloud deployment.

ABSTRACT. Document covers a short overview of activities needed to adopt MS Teams Direct Routing solution as a calling service for PSTN telcos, e.g. Telekom Serbia. Direct Routing scenario for PSTN connectivity through Teams Phone provides the opportunity for client companies to connect existing SIP trunking services and legacy voice platforms to Microsoft Teams Phone via session border controllers (SBCs). This document systematizes requirements, tests, and validation procedures for the Direct Routing solution in Telekom Serbia (MTS) in the PoC phase. Concept developed for MTS is planned for a multitenant environment, where it can support many customers of Office 365 and traditional services that are unified: calling plans, voice services, SIP trunks, Internet, MS Teams etc.

ABSTRACT. This paper describes top down design of a two-bit encrypted register that exhibits high immunity to Side Channel Attack (SCA). The schematic level and post-layout simulation results show the validity of cryptographic shift register design by observing a Normalized Standard Deviation (NSD) of the consumed energy. This encrypted block is designed for CMOS TSMC 035 um technology.

ABSTRACT. This paper presents a comparison between the efficiency of non-learning and machine learning approaches on edge devices, using a typical NFC system centred around commercial ESP32 microcontroller as an example. NFC tag data is read by the NFC reader and forwarded to two ESP32 modules, one using rule-based non-learning approach, and the other one using pre-trained model for text correction. Read success rate as well as internal and external heating are analyzed on the basis of 100 data processing attempts. Machine learning method shows 4 % better read success rate, but also suffers from more self-heating.

ABSTRACT. Electrooculography (EOG) is a non-invasive technique used to measure eye movements by detecting corneo-retinal potential differences. In addition to its wide application in medicine, EOG is increasingly used for practical applications such as assistive device control (artificial arms, wheelchairs, virtual keyboards, etc), driver monitoring (fatigue and sleepiness detection, pilot training, etc), and gaze-based human–computer interaction (HCI) systems. A cost-effective acquisition front-end incorporating lightweight hardware has been developed. The proposed system is suitable for use in student laboratory exercises for educational and training purposes, as well as for further research and development. A model of such an acquisition front-end that fulfills the requirements is proposed in this paper. The results obtained so far are presented, and directions for further research are proposed.

ABSTRACT. This paper deals with subjective assessment of visual quality of synthetic aperture radar (SAR) images despeckled using two conventional filters. Depending on image properties, different filters or filters with different parameters occur to be the best. We show that the results of subjective assessment are in good agreement with visual quality metrics that can be employed in objective characterization whilst analysis of traditional metrics might lead to partly erroneous conclusions. The tests are carried out for three images of different complexity, whereas 15 observers participated in quality assessment.

ABSTRACT. Rate-distortion curves are a fundamental tool in image compression for evaluating the trade-off between the compressed file size and the resulting image quality. However, these curves can exhibit unexpected irregularities, such as non-monotonic behavior or “wiggles”, which challenge the conventional rate-distortion theory. In JPEG compression, these irregularities are primarily a consequence of the quantization process and are particularly pronounced in images containing large uniform regions. So, this paper investigates these irregularities firstly for uniform colorless blocks by analyzing the JPEG quantization error, and then for grayscale JPEG compressed images employing a range of objective image quality assessment measures. Our experimental results demonstrate that for such irregular images, selecting a lower quality factor that controls JPEG compression can paradoxically yield better visual quality than a higher quality factor, as the latter can introduce more noticeable blocking artifacts. This finding highlights the critical importance of quality factor selection in JPEG compression for images with uniform areas.

ABSTRACT. This paper presents a data-driven approach for predicting congestion at electric vehicle charging stations using XGBoost. By combining real-world usage data with synthetically generated sessions and modeling personalized, time-based usage patterns, our method improves accuracy and interpretability. Cross-validation and residual analysis confirm strong performance comparing to similar models. This approach led to useful predictions that are practical and user-centered, and extends the possibilities for forecasting within emerging forms of mobility and energy systems.

ABSTRACT. Kernel ridge regression is applied to predict normalized magnetic field values around a ring-shaped permanent magnet with a trapezoidal cross-section. Results are compared with a semi-numerical method based on fictitious magnetic charges. The close agreement between methods confirms the accuracy and potential of machine learning for magnetic field prediction.

ABSTRACT. We discuss how ChatGPT can be utilized to iteratively generate computational electrostatics code and visualizations. Specifically, we consider axisymmetric conductors and dielectrics and calculate and visualize charge densities, potential and electrostatic fields.

ABSTRACT. The aim of this study is to determine the impact of titanium mesh used in cranial reconstruction, on the magnetic field distribution from a mobile phone within the biological tissues and organs of the user's head. This assessment requires creating realistic three-dimensional models of the user’s head, titanium mesh implant and a cellphone followed by the numerical solution of the electromagnetic propagation equation. Numerical calculations were carried out for the mobile network frequency of 1800 MHz. Finally, a comparative evaluation was performed for both models, with and without a titanium mesh

ABSTRACT. This study investigates the extent to which the size of the dataset used for training may affect neural network model accuracy. An experiment was performed on models predicting solar plant production during clear summer days based on numerical weather data, specifically the estimated amount of global horizontal irradiance and surrounding air temperature. Results show that model accuracy does not change significantly with decreasing training set size, which suggests that strategic data selection may be more critical than dataset size for optimal model performance.

ABSTRACT. The study reviews data fusion methods in multi-agent systems (MAS), categorised into low-, mid, and high-level fusion. Using the PRISMA methodology, we analyse recent literature, highlighting trends such as probabilistic filters, deep learning, and federated decision-making. Challenges remain in scalability, interoperability, and privacy, warranting future research into adaptive, secure fusion models.

ABSTRACT. This paper reviews theoretical cybersecurity approaches in smart cities, categorizing classical, system-oriented, and innovative models. It highlights current gaps, such as limited Al integration and regulatory fragmentation, and proposes a novel framework combining Game Theory, Complex Systems Theory, and Blockchain to enhance resilience and adaptive security in smart city infrastructures.

ABSTRACT. This paper analyzes the outage probability (Pout) as a key reliability metric for wireless systems under Hoyt fading and co-channel interference (CCI). A receiver with Selection Combining (SC) selects the best of L independent branches based on signal quality. The Hoyt fading model, as a generalization of Rayleigh fading, enables accurate modeling of non-line-of-sight (NLoS) environments. Closed-form expressions for Pout are derived to evaluate the effects of channel parameters, number of branches, and interference. In the second part, we propose an agentic AI-based workflow for network management, leveraging the previously derived outage probability expression and associated channel parameters for the presented case study.

ABSTRACT. This paper presents a methodology for topic -specific analyzing of discussions on the decentralized, federated and open-sourced social network. The methodology integrates NLP, sentiment analysis, rule-based NER, semantic enrichment via DBpedia, RDF modeling, and SPARQL querying to enable structured, machine-interpretable analysis. Results of experiments focused on AI topic show that semantic enrichment improves understanding compared to basic methods while supporting knowledge extraction in intelligent systems.

ABSTRACT. This study examines global ethical guidelines for AI and cybersecurity, highlighting key principles like fairness, transparency, privacy, and accountability. It identifies inconsistencies across sectors and emphasizes the need for harmonized standards. The study stresses balancing innovation with public trust, focusing on responsible practices in AI decision-making and cybersecurity protocols.

ABSTRACT. In this paper, the design and analysis of a harmonic tag with an RF energy harvesting circuit is performed. In the first research stage, a low power passive harmonic tag has been designed at 2.45/4.9 GHz whose main function is to receive the fundamental signal and transmit harmonics. In the second stage, an RF energy harvesting circuit is designed at 2.45 GHz by using voltage doubler RF rectifier. Additionally, a diplexer was inserted between the harmonic tag and the RF energy harvesting circuit to separate the fundamental signal from the second harmonic. Impedance matching is performed between the antennas and the harmonic tag, as well as between the harmonic tag and the RF rectifier, with aim to obtain maximum power transfer. Results accomplished by ADS simulations for different input power levels provide a performance evaluation of the harmonic tag and the RF energy harvesting circuit in terms of RF-DC conversion efficiency, DC power, and second harmonic power. For the observed input power levels (-20 dBm to 10 dBm), the gained RF-DC conversion efficiency varies between 7 % and 17 %, the output DC voltage varies between 0.1 V and 3.6 V, while the second harmonic power goes from -61 dBm to -3 dBm.

ABSTRACT. The rise of quantum computers threatens classical cryptosystems through Shor’s and Grover’s algorithms. Post-quantum cryptography (PQC) addresses this challenge, with lattice-based schemes emerging as leading candidates. This paper presents an overview in this emerging field with main focus on Kyber method for key exchange and Dilithium method for digital signatures. the paper show why these method are good candidates for future implementation in real-world protocols. However, the paper also presents concerns regarding different types of applications of such a methodology, especially because of the attacks in IoT systems.