Journal of Problems in Computer Science and Information Technologies

NEW AUTONOMOUS SYSTEM FOR SPATIOTEMPORAL CLUSTERING AND VISUALIZATION OF DEVICE TRAJECTORIES IN FORENSIC INVESTIGATIONS

2025-08-22T11:50:24+00:00

This study presents «trajectory_analyzer», a Python-based system designed for the forensic analysis and visualization of geolocation data extracted from mobile devices. With the increasing volume of spatial-temporal data collected from sources such as GPS, Wi-Fi, and image metadata, forensic professionals face growing challenges in structuring and interpreting mobility patterns. Existing solutions often lack flexibility, require supervised models, or depend on proprietary infrastructure. Our approach applies an unsupervised DBSCAN-based trajectory clustering method, temporal ordering, and a real-time web map interface to reveal behavioral insights without the need for manual labeling or cloud services. Compared to prior research, the system improves spatial accuracy, source transparency, and visual clarity. Experimental results show that the proposed clustering method identifies movement clusters and transitions with high precision and responsiveness, while maintaining full offline operability. However, this improvement comes at the expense of more local storage because of embedded map tiles. Overall, this work provides a practical, understandable, and independent foundation for investigators dealing with unstructured multi-source geolocation data.

CVD PREDICTION FROM HRV DERIVED FROM WEARABLE PPG

2025-11-03T12:16:20+00:00

Cardiovascular disease is the leading global cause of death; ischemic heart disease (IHD) is its most common and lethal form, motivating scalable, non-invasive screening. We tested whether a single 60-minute photoplethysmography (PPG) recording from the Zhurek fingertip wearable can distinguish healthy autonomic control from IHD-related dysregulation. Agreement with a three-lead Holter reference was clinically acceptable (HR −0.601 bpm; SDNN +33.1 ms; RMSSD −4.8 ms). Forty hour-long sessions were analyzed (20 healthy, 18–22 years; 20 angiography-confirmed IHD) using eight HRV/demographic features. Mann–Whitney tests showed significant differences for SDNN, LF, HF, Max_HR, BMI, and age (p<0.05), and a two-component PCA (49.5% variance) separated cohorts without labels. SHAP for a CatBoost model highlighted LF and age as strongest positive contributors and HF as protective. Thus, one-hour PPG preserves diagnostically useful autonomic signatures, enabling ~24× shorter monitoring than Holter and supporting scalable ambulatory IHD risk stratification.

SMART BUILDING CLIMATE CONTROL: MACHINE LEARNING APPROACH FOR INDIVIDUAL THERMAL PREFERENCE PREDICTION

2025-12-15T10:28:38+00:00

Modern building management systems rely on uniform climate settings that fail to accommodate individual occupant preferences, resulting in energy waste and reduced comfort satisfaction. This study presents a data-driven approach for personalized thermal comfort prediction using machine learning algorithms integrated with multimodal sensor networks. We developed and evaluated three classification models (Random Forest, XGBoost, and Artificial Neural Network) using environmental parameters (air temperature, humidity, CO₂ concentration) and physiological measurements (heart rate variability, blood pressure, oxygen saturation) collected from controlled experiments with eight participants under various thermal conditions. The optimized Random Forest model achieved 95% accuracy in predicting seven-level thermal sensation votes using only ten key features identified through SHAP analysis. Indoor air temperature emerged as the dominant predictor, while physiological parameters provided complementary information for personalized comfort assessment. The proposed system demonstrates significant potential for integration into smart building automation, enabling dynamic climate control that adapts to individual preferences while optimizing energy consumption. Implementation of such personalized HVAC systems could reduce energy usage by up to 20% compared to conventional static temperature control, while simultaneously improving occupant satisfaction and productivity in commercial buildings.

IMPLEMENTATION OF A REPRODUCIBLE 5G STANDALONE TESTBED USING OPEN-SOURCE COMPONENTS

2025-11-20T05:42:14+00:00

Deploying a 5G Standalone (SA) network is often constrained by cost, complexity, and limited access to RF hardware. This paper describes a practical, software-defined 5G SA implementation assembled entirely from open-source components Open5GS for the 5G Core, srsRAN for gNB and UE, MongoDB for subscriber data, and ZeroMQ for RF emulation to enable end-to-end connectivity without radios. The blueprint details service initialization order, subscriber provisioning, and configuration alignment across PLMN, TAC, DNN, and key material, followed by validation of UE registration, PDU session establishment, and user-plane data transfer. On the reference setup, the system achieves low round-trip latency (≈1.34 ms), high throughput (≈847 Mbps TCP downlink and ≈823 Mbps uplink), and rapid PDU session setup (≈0.22 s), supporting repeatable functional and performance testing in a laboratory environment. The described approach lowers the barrier to 5G experimentation for teaching, prototyping, and research while providing a reproducible path from basic bring-up to performance evaluation. Limitations include the use of emulated RF and a single-cell scenario; nevertheless, the workflow can be extended to over-the-air SDR tests, mobility, and slicing when needed.

INTELLIGENT SYSTEM FOR AUTOMATIC DETECTION AND SCORING OF SHOOTING TARGETS BASED ON COMPUTER VISION AND MICROCONTROLLER TECHNOLOGIES

2025-11-19T11:25:27+00:00

This paper presents an intelligent system for the automatic detection and scoring of shooting targets based on the Raspberry Pi 3 microcontroller platform and computer vision technologies. The objective of the study is to develop an autonomous and highly accurate yet low-cost complex capable of recording and analyzing shooting results without human intervention.

The system integrates mechatronic and algorithmic components, including Nema 17 stepper motors, color sensors, a webcam, and a server-side image processing module, forming a unified cyber-physical architecture. The algorithmic core is based on geometric calibration using homography, adaptive illumination equalization via CLAHE, and a radial precision evaluation model. To detect bullet holes, a modified YOLOv8-Nano neural network architecture was employed, optimized for recognizing low-contrast circular targets.

Experimental results confirmed the high accuracy and robustness of the proposed approach: under stable lighting conditions, the system achieved a spatial recognition precision of ±2 mm with a response time below 0.2 seconds. The training and validation curves of the model demonstrate smooth convergence and stable generalization, confirming the correctness of the architectural modifications and the optimization of the loss function.

The scientific novelty of this work lies in the integration of a mechatronic framework and deep-learning algorithms into a unified real-time system that enables automatic target replacement, image processing, and result visualization through a web interface. The practical significance is in the potential application of the system in sports schools, mechatronics laboratories, training centers, and research test ranges requiring accurate and autonomous shooting evaluation.

Future work will focus on extending system capabilities through the integration of advanced neural network algorithms (YOLOv8, Detectron2), cloud-based technologies, and automatic camera stabilization, further improving accuracy and autonomy while maintaining low implementation cost.

INTEGRATED ENVIRONMENTAL AND PHYSIOLOGICAL MONITORING FOR CARDIOVASCULAR RISK DETECTION USING IOT AND MACHINE LEARNING

2025-08-08T13:19:46+00:00

This study investigates the impact of air pollution on heart rate variability (HRV), a key physiological marker reflecting the state of the autonomic nervous and cardiovascular systems. Despite growing interest, the complex relationship between environmental exposure and HRV, especially in the context of early cardiovascular disease (CVD) detection, remains insufficiently explored. An integrated real-time monitoring system was developed using Internet of Things (IoT) devices and machine learning (ML) methods to collect and analyze data from 10 healthy participants (aged 18–22) in three different environments: a controlled laboratory, an urban roadside (Al-Farabi Avenue), and a natural setting (botanical garden). Physiological signals (RMSSD, SDNN, LF, HF) were obtained using Polar H10 ECG sensors and Zhurek PPG devices, while environmental data (PM2.5, PM10, CO₂) were recorded via Tynys and Qingping sensors. Three supervised ML models—deep neural networks (DNN), random forest (RF), and XGBoost—were used to classify HRV levels based on environmental parameters. Among them, XGBoost achieved the best performance with 91.92% accuracy, 91.82% precision, and a 90.42% F1-score. The results revealed a consistent negative correlation between higher levels of PM2.5 and PM10 and reduced HRV metrics, particularly SDNN and RMSSD, indicating potential autonomic dysfunction and increased cardiovascular risk. Although CO₂ levels showed weaker associations, their influence was still noted. These findings emphasize the importance of considering environmental factors in health monitoring and demonstrate the potential of IoT and ML technologies in enabling early detection of cardiovascular stress and supporting personalized healthcare strategies.

SIMULATION MODELING OF DATA INTEGRITY VIOLATIONS IN INTELLIGENT SOCIAL SYSTEMS

2025-11-21T12:47:25+00:00

The article examines the issues of ensuring the integrity and confidentiality of electronic data through the use of post-quantum cryptographic algorithms Kyber and Dilithium. The rapid development of quantum computing threatens the security of traditional cryptographic systems (RSA, ECC), making the protection of digital infrastructures—such as government platforms, financial services, educational ecosystems, and intelligent social systems—an urgent priority.

The Kyber algorithm (for encryption and key exchange) and Dilithium algorithm (for digital signatures) are part of the post-quantum cryptography standards recommended by NIST. Their implementation significantly enhances resistance to quantum attacks, ensures the integrity of electronic transactions, and provides reliable authentication within digital ecosystems.

Particular attention is paid to the integration of post-quantum algorithms into trusted electronic systems, including e-voting, state registries, digital document management, and cloud storage infrastructures. Architectures based on the combination of Kyber and Dilithium guarantee not only protection from external interference but also verifiability, authenticity, and consistency of data throughout its entire life cycle.

Thus, the application of post-quantum cryptographic algorithms represents a crucial step in the development of secure digital ecosystems and the formation of a new paradigm of digital trust

UNDERSTANDING DIGITAL TRANSFORMATION AND AGENTIC SYSTEM THROUGH BIBLIOMETRICS: DEVELOPING AN AGENTIC MODEL OF HUMAN–AI INTERACTION

2025-12-19T11:26:11+00:00

This study conducts a comprehensive bibliometric analysis of research on digital transformation within the social sciences from 1997 to 2024 and integrated these findings to develop an Agentic Model of Human–AI Interaction. Drawing on 389 articles indexed in the Web of Science database, the analysis examines publication dynamics, influential authors and institutions, citation structures, and thematic research clusters using RStudio-based bibliometric tools. Results reveal that digital transformation has evolved into a central driver of societal, economic, and cultural change, with research output peaking in 2021–2022. Prominent themes include artificial intelligence, innovation processes, digital media technologies, and the societal implications of emerging technologies. Beyond mapping the knowledge landscape, the study proposes a conceptual agentic system model that explains how AI agents process, interpret, and operationalize human queries through structured stages of planning, retrieval, reasoning, and response generation. This integration of bibliometric insights with system conceptualization contributes to a deeper understanding of the evolving human–AI relationship and highlights key gaps and future research opportunities in the study of agentic systems within digital transformation.

COMPUTATIONAL SIMULATION OF INSTITUTIONAL-INVESTMENT DYNAMICS USING PANEL VECTOR AUTOREGRESSION

2025-12-26T13:39:53+00:00

This study develops a computational framework for simulating dynamic interactions between institutional quality indicators and foreign direct investment using a panel Vector Autoregression model applied to a multi-country dataset. The work emphasizes the algorithmic structure of the modeling pipeline, including preprocessing of heterogeneous panel time series, numerical stationarity diagnostics and cointegration testing. Impulse-response simulations are used to examine system behavior following institutional shocks, illustrating the dynamic propagation of disturbances in a high-dimensional environment. Although the empirical application concerns institutional governance, the contribution of this study lies primarily in its computational architecture, numerical considerations, and reproducible design of a data-driven simulation environment. The presented framework demonstrates how computational finance and applied computer science can integrate econometric modeling to analyze complex, interdependent systems.

MASS-CONSERVING PHYSICS-INFORMED AUGMENTATION AND FOURIER FEATURE NETWORKS FOR SMALL-DATA PREDICTION OF MOLYBDENITE (MO₂S) LEACHING KINETICS

2025-12-17T13:13:01+00:00

Molybdenum remains a strategic metal for advanced steels and catalysis, while environmental and energy pressures are accelerating interest in hydrometallurgical leaching routes for molybdenite (MoS₂). Predicting leaching kinetics is difficult because the process is highly nonlinear and strongly influenced by reagent chemistry and gas–liquid conditions, yet experimental datasets in metallurgical laboratories are often extremely small. This manuscript develops a hybrid, data-efficient machine-learning approach designed specifically for small-data settings. The method combines physics-informed data augmentation that enforces strict mass conservation with a Fourier Feature Network intended to reduce spectral bias and better capture sharp kinetic transitions. Using only six experimental measurements, the resulting model achieves high predictive accuracy on held-out data (R² = 0.9793, MAE = 1.61%) and maintains stable generalization without evidence of train–test divergence. The study concludes that physically admissible augmentation coupled with Fourier-enriched representations can produce reliable kinetic surrogates from minimal data, supporting in-silico screening and optimization of leaching conditions for process design and control.