Radio Electronics, Computer Science, Control

A DESIGN PATTERN FOR ENABLING FUNCTIONAL STABILITY IN SOFTWARE SYSTEMS

O. S. Bychkov — 2026-03-27

Context. Modern software systems operate in dynamic and harsh environments where internal and external failures, unexpected disturbances, direct attacks, and resource constraints challenge the consistent provision of core functionalities. In these contexts, ensuring functional stability – where the quality of each system function remains within a predetermined stable range despite failures or environmental anomalies – is critical, especially for safety-critical and high-availability applications.
Objective. The primary objective of this work is to develop and justify an enabling design pattern that provides the architectural backbone for achieving functional stability in software systems. The main focus is to provide a flexible solution that facilitates dynamic adaptation while maintaining robust system behavior.
Method. We propose a novel pattern that combines the dynamic strategy selection capabilities with the loose coupling between components afforded by an event-driven approach. This enabling pattern decouples system components by enforcing communication solely through standardized event types and allows each module to select an appropriate adaptation strategy based on its current context. The described pattern was used to build a design of a real-life example that aims to implement stable object tracking functionality for autonomous quad-platforms. The proposed design was evaluated using design-level metrics alongside qualitative comparisons with existing adaptive approaches.
Results. Our analysis shows that the enabling pattern achieves significant modularity and adaptability. Key object-oriented metrics indicate minimal interdependencies among modules and a clear separation of concerns. The design proposal demonstrates that the pattern supports dynamic behavior adjustment through flexible strategy selection and serves as an enabler for functional stability by providing a robust architectural backbone for software systems.
Conclusions. The scientific novelty of this work is twofold: firstly, the novel pattern is obtained in our study, providing dynamic adaptation through context-aware strategy selection; secondly, functional stability received further development in the area of
software architecture. The proposed pattern offers a robust, scalable, and maintainable architectural solution, with significant practical implications for the design of adaptive, resilient software systems

COMPARISON OF SOFTWARE ARCHITECTURE EVALUATION METHODS APPLICABILITY IN THE CONTEXT OF CQRS WITH EVENT SOURCING ARCHITECTURAL VARIATIONS

D. L. Hruzin — 2026-03-27

Context. This study is conducted in the context of developing and justifying a methodology for software architecture (SA) evaluation in relation to the Command Query Responsibility Segregation (CQRS) with Event Sourcing (ES) architectural variations.
Objective. This work aims to evaluate and compare the applicability of SA evaluation methods to support the selection of an optimal CQRS with ES architectural variation for real-world projects.
Method. Various SA evaluation methods are applied to enhance objectivity in architectural decisions. However, these methods
are not universal; they vary in depth, focus, and required effort. The task considered in this work is the selection among CQRS with ES architectural variations, often structurally similar and thus difficult to distinguish using general-purpose evaluation methods. Comparing architectural variations requires in-depth analysis; however, for most methods, practical implementation is limited by time and resource constraints. The proposed approach identifies the most appropriate SA evaluation method for selecting between CQRS with ES architectural variations. It is based on a validated framework for classifying and comparing SA evaluation methods. In addition to qualitative analysis, the approach introduces a quantitative assessment of applicability to a specific case, allowing for supporting more informed decision-making.
Results. The approach was applied to compare several SA evaluation methods, including Information Technology for Decisionmaking Support regarding CQRS with ES Architectural Variations (DSAV-CQRSES), a method specifically designed for evaluating variations of the CQRS with ES architecture.
Conclusions. The existing framework of comparing Software Architectures cannot be directly applied to architectural variations
(the deviations of the architecture significant for customer). The proposed modifications of the framework are primarily focused on CQRS with ES variations assessment

ABOUT RATIONAL METHODS FOR FINDING OPTIMAL ROUTES IN FUZZY TRAVELING SALESMAN PROBLEMS

E. V. Ivohin — 2026-03-27

Context. This paper presents the results of a study on the use of triangular fuzzy numbers for determining time-optimal routes in the traveling salesman problem under fuzzy representations of travel duration in a transportation network. To formalize the uncertainty and imprecision of input data – associated with the subjectivity in estimating the time intervals required to travel between individual cities-triangular fuzzy numbers are employed. Various approaches to solving fuzzy traveling salesman problems are examined.
Objective. The goal of the work is to develop algorithms for solving the fuzzy traveling salesman problem based on the implementation of the Bellman-Zadeh parametric optimization methods, the use of a two-criteria approach with a given weight function and the refinement of the scheme for calculating the center of gravity of the membership function graph for a given curve density.
Method. The article considers methods for solving the fuzzy traveling salesman problem, which is formulated as the problem of finding a route to visit a given number of cities without repetitions with a minimum travel time. The parameters of the problem for formalizing the uncertainty and inaccuracy of input data associated with the influence of subjectivity in assessing the duration of time intervals required to travel between individual cities are presented as fuzzy triangular numbers. Different approaches to solving fuzzy traveling salesman problems are considered. The application of the Bellman-Zadeh method, methods taking into account refinements of defuzzified data, and methods based on a multicriteria approach is formalized. Computational experiments are carried out.
Results. Rational algorithms for solving the fuzzy traveling salesman problem based on the Bellman-Zadeh parametric optimization model, multicriteria approach and methods for refining the results of defuzzification of fuzzy data have been developed. In the conducted numerical experiments on solving the traveling salesman problem, fuzzy input data are based on the method for calculating the center of gravity (CoG), the center of gravity of homogeneous and non-homogeneous curves, which are determined by the membership function and the specified reliability values of subjective data. A comparison of the results obtained based on solving the crisp traveling salesman problem and the results based on defuzzified duration values for the fuzzy traveling salesman problem is carried out, according to the results of which the dependence of the solution on the defuzzification method is confirmed.
Conclusions. The article considers the method of formalizing the algorithm for solving the fuzzy traveling salesman problem with the minimum duration of movement along the route based on the Belman-Zadeh method, methods taking into account the refinements of defuzzified data and methods based on the multicriteria approach. Fuzzy triangular numbers are used to formalize the uncertainty of the input data when assessing the duration of movement between individual towns of the transport network. It was made a conclusion about the feasibility of using fuzzy numbers when solving fuzzy traveling salesman problems in real conditions of logistics transportation

A STUDY OF THE PERFORMANCE OF ANY-ANGLE THETA* ALGORITHMS ON WEIGHTED GRID MAPS FOR ROUTE PLANNING

Y. Kis — 2026-03-27

Context. The article addresses the study of free-direction pathfinding algorithms, in particular the Theta* algorithm, and evaluates their performance on weighted grid maps in order to determine optimal routes for delivering goods to a firearms store. This research is carried out in the broader context of developing an information system for tracking and managing arms sales and logistics under complex conditions. One of the main motivations is that any-angle methods can produce more realistic and natural-looking paths compared to the classical A* algorithm.
Objective. The purpose of the study is to analyze the performance of three Theta*-based algorithms – Basic Theta*, Lazy Theta*, and Strict Theta*– on both uniform and weighted square grids, with special emphasis on execution time and path cost metrics. The work aims to generalize the applicability of these algorithms to weighted environments and to propose improvements suitable for real-world route planning scenarios.
Method. The principles of A*, the three Theta* variants, and path post-processing smoothing techniques are presented. The research describes the transition from unweighted uniform square grids to weighted grids and highlights the complexity of calculating accurate path costs when applying any-angle approaches. Visual demonstrations of algorithmic behavior were implemented using the Unity game engine. Performance metrics were measured separately for uniform and weighted grids to ensure comparative analysis.
Results. The results include comparative evaluations of Basic Theta*, Lazy Theta*, Strict Theta*, and classical A* algorithms. The analysis identifies conditions under which each algorithm performs effectively, as well as factors that limit their applicability in weighted environments. It is shown that path length and path cost may differ substantially in weighted grids, leading to new considerations for cost-based optimization. Based on the experiments, a generalization of the Basic Theta* algorithm is proposed to enhance its suitability for weighted square grids, and a potential extension of the Strict Theta* algorithm to this context is outlined.
Conclusions. The findings demonstrate that while any-angle algorithms provide smoother and more realistic routes, their effectiveness in weighted environments depends on careful adaptation of cost functions. The research highlights their value not only for simulating complex virtual environments and agent behaviors in games and robotics but also for practical applications in logistics, particularly in the development of an information system for tracking and managing firearms sales. The proposed algorithmic adaptations may contribute to improving delivery planning and supply chain efficiency, including the modeling of weapons delivery routes under wartime conditions.

DEVELOPMENT OF A CLASS STORAGE REPOSITORY FOR OBJECT-ORIENTED SOFTWARE DEVELOPMENT TECHNOLOGIES

O. B. Kungurtsev — 2026-03-27

Context. Using previously developed code, particularly software classes and class groups related through inheritance, aggregation and composition in object-oriented technologies, significantly reduces software design time.
Objective. Problems arise when it is necessary to store classes for different purposes. In this case, the class name cannot serve as a characteristic for searching. Creating a special structure linking the class name with its purpose will significantly complicate the class repository design. Besides, if relations connect some classes with other classes, there is a need to store their relations along with the class itself, which can significantly complicate both the placement of the class in the repository and its further search. This paper aims to create a special repository of software classes, in which a class is represented by a model that defines its purpose, possible relations with other classes, and role in these relations.
Method. A mathematical model of a software class has been developed, which allows for determining the class designation and possible inter-class relationships. A method of automated placement and search of individual classes and class groups in the class repository is proposed. A storage model is proposed for placing both individual classes and groups of classes connected by inheritance, aggregation and composition relationships. A mechanism has been developed for the automated addition of a separate class or group of classes to the repository, as well as for searching and deleting a separate class or group of classes.
Result. The ClassCall programme has been developed to test the proposed solutions. Experiments were conducted to determine the time and quality of placement operations and search of classes and class groups in the repository. The results showed a significant reduction in time for class search compared to known libraries.
Conclusions. The proposed method of automated placement and search of classes based on the class model allows for maintaining class versions, significantly reducing the time to search for the required class and related classes. The method can be used for classes in various object-oriented languages.

ESTIMATION OF EFFORT OF MIGRATION AMONG DOMAIN-DRIVEN DESIGN ARCHITECTURAL VARIATIONS

O. A. Lytvynov — 2026-03-27

Context. The article addresses the issue of effort estimation of migration among variations of DDD architecture using a method based on specifications of requirements to increase the predictability of the software migration process.
Objective. The goal of the work is to propose an effective method of effort estimation based on Use Case analysis.
Method. First, a set of rules for rigorous Use Case description adapted for software effort estimation needs are provided. Second, the modified Use Case metamodel and the method of use cases classification based on frame-based knowledge representation model are also suggested. The rigorous description allows to make the estimation of the use cases more precise, using FUSP method, and to build individual predictors for each class of use cases. Thirdly, the method uses historical data taken from previous iterations of the same project, and is based on three trends (optimistic, pessimistic and mean based trend).
Results. The result is the collection of functions used to predict the effort required for the next iteration (measured in personhours) for each class of use cases.
Conclusions. FUSP method was adapted for task of gaining greater prediction accuracy of effort estimation for migration among variations of DDD architecture using a methodology based on specifications of requirements. The set of conditions to form the Use Case description rules adapted for software migration effort estimation needs is developed. The modified Use Case metamodel and the method of use cases classification based on frame-based knowledge representation model are suggested. It was proposed algorithm for building the individual predictors of each class and for corresponding effort estimation. The coefficient of FUSP personhours transformation is based on three trends achieved and updated considering the results from previous iterations: the most pessimistic prediction is based on the upper bound, the lower bound predictor plays the role of the optimistic predictor, and the main trend is the meaning among these. The coefficients are used to predict the effort in person-hours required for the next iteration for each class of use cases. The results of experiment, conducted using the test RTP project for this class of software, showed that MMRE for the proposal method is 0.0343, and for the standard method – 0.1094. The obtained results evidence that the classification of use cases along with their rigorous description according to provided rules, and modification of the method by separating prediction logic in accordance with the use case classes makes the prediction more accurate and can be effectively used for effort estimation for DDD architectural variations migration.

A FRAMEWORK FOR THE REMOTE MONITORING OF PATIENTS IN THE HEALTHCARE SYSTEM

H. I. Mafraq — 2026-03-27

Context. Remote patient monitoring (RPM) technology plays a vital role in developing healthcare services. The medical team can continuously monitor a patient’s health status, even outside of hospitals. It is considered one of the most important digital health services, as it facilitates patient care and reduces the spread of disease.
Objective. This paper aims to review current remote patient monitoring (RPM) systems for various diseases. Then proposes a new platform architecture to increase the effectiveness and quality of remote patient care.
Method. The paper analyzes systems for remote monitoring, focusing on the most common systems of several diseases such as diabetes, epilepsy, headache, cardiovascular and heart failure diseases, COVID-19, chronic kidney failure, fainting and unconsciousness, and cancer. Additionally, it provides an overview of the systems with contact and contact-less features, addressing them according to the system type, architectures, technology used, and services they provide.
Results. After analyzing remote patient monitoring (RPM) applications for a variety of diseases, the results highlighted the strengths and weaknesses of existing systems. We then demonstrated how the proposed architecture addresses these shortcomings and develops a scalable and effective solution.
Conclusions. This paper validates the effectiveness of RPM for healthcare development, offering an innovative ontology-based platform that improves service delivery and patient outcomes. This work offers valuable insights for healthcare providers, developers, and policymakers who are advancing remote care solutions.

MODIFIED BIOMETRIC TEMPLATE PROTECTION METHOD WITH NONLINEAR TRANSFORMATIONS

M. V. Onai — 2026-03-27

Context. Biometric data is a common option for authentication or identification. However, it is vulnerable and not replaceable in case of stealing. Several methods for constructing protected biometric templates are proposed in literature, one of them is biohashing. However, linearity of biohashing may be a vulnerability. MLP-hash is similar, but adds nonlinearity. It is modified in this work.
Objective. The goal of this work is to develop a modification of MLP-hash which is faster and allows better separation of users by their templates.
Method. This work focuses on modifying MLP-hash, a biohashing variation with nonlinear transformations. One of the proposed changes is the usage of normalization before applying nonlinear transformation in each layer of MLP-hash. Different normalization methods are investigated and compared. The other proposed change is the simplification of the pseudorandom matrices used in each layer of MLP-hash. Each such matrix is replaced by a block matrix in which blocks that are laying on the diagonal are orthonormal matrices and all other blocks are filled with zeros. Each nonzero block is generated from the user’s secret token. In order to make the effect of each nonzero block less localized, a pseudorandom permutation is added before each matrix multiplication and also after all layers. Pseudorandom permutations are also generated with the user’s secret token as seed. The proposed method can be used in a similar way to how original MLP-hash and biohashing methods are used: it takes the user’s secret token and biometric vector of fixed length and outputs a binary vector of fixed length with the same or smaller dimensionality. MLP-hash with block matrices is compared to the original while applying different normalization techniques and different nonlinear transformations.
Results. The proposed modifications, original MLP-hash and biohashing have been implemented in code. Speed and accuracy of user separation with the usage of these methods have been compared on feature vectors extracted from fingerprints with the usage of Gabor filters.
Conclusions. The conducted experiments have shown an increase of speed and ability to separate user templates from the substitution of proposed block matrices and an increase of ability to separate user templates from the usage of normalization.
Comparison of different normalizations and nonlinear transformation has also been conducted. The practical usefulness of the developed method is that it is faster and can be used in applications when users expect no delays while still being difficult to invert. The prospects for further research include testing this method with other biometric modalities, other nonlinear transformations and normalization techniques and an analysis of inversion difficulty of the developed method in comparison to MLP-hash and biohashing.

METHOD FOR CORRECTION OF MULTISUBJECTIVE MULTIFACTORIAL ENVIRONMENTS OF SOFTWARE COMPLEXES’ SUPPORT

A. I. Pukach — 2026-03-27

Context. The problem of correction of multisubjective multifactorial environments of software complexes’ support is considered in this research, necessary to provide the possibility(-ies) of adjusting the perception’s subjectivization of the support object (the supported software, as well as related processes of its complex support), caused by the influence of relevant impact factors. The object of research is a process of correction of multisubjective multifactorial environments of software complexes’ support. The subject of research are methods and means of correction of a multisubjective multifactorial environments of software complexes’ support, as well as methods of an artificial neural networks (in particular: a multilayer perceptron).
Objective – is the development of method for correction of multisubjective multifactorial environments of software complexes’
support.
Method. The development of a method for correction of multisubjective multifactorial environments of software’ support is proposed. which provides possibilities for the necessary adjustments of the perception subjectivization of the researched support objects (which could be either the supported software itself, as well as the related processes for its comprehensive support) relevant (directly or directly) interacting subjects, who provide and implement this comprehensive support of the researched supported software product, in order to provide the possibility(-ies) of further automation and intellectualization of its comprehensive support.
Results. The results of functioning of the developed method – are relevant models of adjusted multisubjective multifactorial environments of software complexes’ support, obtained in result of solving a relevant scientific and applied problem of adjusting such class of environments. The developed method provides the opportunity(-ies) for studying the processes of collective perception’s subjectivization (caused by the influence of existing impact factors) of the objects of comprehensive support by the appropriate related subjects, which directly provide and implement this support, and also facilitates and ensures for further automation and intellectualization of such complex support of various software products and complexes in this separate and exact functional and procedural segment. As a practical approbation of the developed method, – the results of solved applied practical task of determination and further correction the impact factors of maximum imbalance of the researched multisubjective multifactorial environment (representing the technician team of the supported software product) are given.
Conclusions. The developed method solves the declared problem of correction of multisubjective multifactorial environments of
software complexes’ support. At the same time, the obtained results of performed practical approbation of the developed method confirm its functionality in solving a range of scientific and applied tasks based on the processes of collective perception’s subjectivization of support objects (the supported software complexes, as well as the processes of their comprehensive support), which (those tasks), in turn, are included into the cluster of a more valuable scientific and applied problem of software products’ comprehensive support automation and intellectualization.

TUNABLE SQUASHING ACTIVATION FUNCTION FOR DEEP NEURAL NETWORKS

A. Yu. Shafronenko — 2026-03-27

Context. At present, artificial neural networks have become widely used to solve many problems of information processing of the most diverse nature and, above all, data mining, due to their universal approximating capabilities and the ability to learn their parameters – synaptic weights. The process of training a multilayer network consists of adjusting the synaptic weights of each neuron using the error backpropagation procedure, which is based on the chain rule of differentiation for complex functions and gradientbased optimization. Deep neural networks are based on multilayer perceptrons, which have proven their effectiveness in solving many very complex problems related to the processing and synthesis of images of various natures, natural language texts, multidimensional stochastic and chaotic sequences, including audio and video signals. Unlike classical three-layer perceptrons, DNNs contain dozens and hundreds of layers, and the number of their synaptic weights is commensurate with or even exceeds the number of synapses in the biological brain. It is clear that for these neural networks, the effect of a vanishing gradient is extremely undesirable; therefore, instead of traditional compression functions, piecewise-linear constructions are usually used here, the most popular of which is the so-called ReLU.
Objective. The purpose of the work is to introduce an adaptive activation function for deep neural networks based on the most common piecewise linear function, ReLU.
Method. A new tunable activation function for deep neural networks is proposed based on the most common piecewise linear function, ReLU, which, however, does not satisfy the conditions of G. Cybenko’s approximation theorem, but provides protection for the learning process against the undesirable effect of vanishing gradients.
Results. A new adaptive piecewise linear function based on ReLU is introduced, which is both compressive and protected against vanishing gradients. In this case, during the training process, not only are the synaptic weights adjusted in the network, but also the parameters of the activation function itself. Using the proposed function allows you to reduce the number of neurons and hidden layers in the neural network, the number of required training samples, and the time required to set up the network.
Conclusions. An adaptive squashing activation function based on the widely used ReLU for deep neural networks is introduced, providing both universal approximating properties and preventing vanishing gradients. A training procedure using this function is proposed, offering high performance and a simple numerical implementation. An additional circuit for tuning the parameters of the activation functions can be quite simply introduced into existing deep neural networks that use piecewise linear activation functions.

HYBRID SATELIT IMAGE RECOGNITION SYSTEM COMBINING NEURAL NETWORK FEATURE EXTRACTION AND AN INFORMATION-EXTREMAL CLASSIFIER

A. S. Dovbysh — 2026-03-27

Context. The study solves the relevant task of developing an interpretable and adaptive recognition system for semantic segmentation of satellite imagery by integrating neural network feature extractors with an information-extreme classifier.
Objective. To improve the accuracy of satellite land cover classification by developing a hybrid machine learning model that combines a deep convolutional neural network for extracting informative features with an information-extreme classifier, enabling the construction of highly reliable decision rules even in the presence of overlapping recognition classes in the feature space.
Method. A hybrid model is proposed that combines efficient spatial feature extraction using a convolutional neural network (CNN) with an information-extreme intelligent (IEI) technology for data analysis, based on maximizing the information capacity of the recognition system during machine learning. For feature aggregation, GlobalAveragePooling is applied instead of the classical Flatten operation. Additionally, regularization techniques such as weight decay and cyclical learning rate scheduling are implemented. The optimization of IEI model parameters is carried out using a modified Kullback information criterion, interpreted as a measure of recognition class diversity.
Results. The developed model achieves high classification accuracy (95%) on the test set and demonstrates stable performance, with improved efficiency of the neural feature extractor due to a reduced number of training epochs enabled by regularization techniques. As a result of the information-extreme machine learning process, the optimal geometric parameters of the hyperspherical recognition class containers were determined, allowing the construction of highly reliable decision rules even under conditions of recognition class overlap in the feature space.
Conclusions. The proposed hybrid model enables the construction of highly reliable decision rules through information-extreme machine learning, even in cases of a priori fuzzy partitioning of recognition classes in the feature space, based on the input training matrix formed during feature extraction.

WELER: A COMPLEX METRIC FOR TEXT QUALITY ASSESSMENT

A. R. Dumyn — 2026-03-27

Context. Assessing text quality is essential for reliable AI that processes language. In ASR, it reflects how faithfully speech becomes text; in OCR, how accurately images yield text; and in NLP, how correct and coherent outputs are.
Objective. The goal of the work is the creation of a complex metric for text quality assessment.
Method. Classic metrics WER and CER are narrow: they capture only lexical edits, weigh all changes equally, ignore context and semantics, and often skip punctuation and case, masking readability issues and error types. We propose WELER, a hybrid metric that blends weighted WER and CER with a semantic component based on contextual embeddings to measure meaning preservation. Weights can be set manually or learned (e.g., via PCA), adapting the metric to ASR, OCR, or NLP tasks. Key challenges include computational cost, choosing optimal weights through correlation with human judgments, and the need for high-quality reference data. Proposed WELER metric integrates accurate word and character level error counting, using Levenshtein distance as a basis, with advanced semantic similarity methods based on contextual embeddings. This allows WELER to take into account not only what was incorrectly recognized, but also how much this error affects the meaning and understanding of the text. The inclusion of selfadjusting weights depending on the text category is a key feature of WELER, which allows adapting the metric to the specific requirements of different applications and domains, prioritizing those aspects of quality that are most critical for a particular task.
Results. Proposed WELER metric is an alternative solution in this direction. It integrates accurate word and character level error counting, using Levenshtein distance as a basis, with advanced semantic similarity methods based on contextual embeddings.
Conclusions. WELER, like all metrics based on reference data, relies on accurate and consistent human-verified transcriptions. Errors in the reference data can affect the accuracy of the assessment. Therefore, for complex metrics, the quality and representativeness of these data are especially important, since semantic and weighted errors are much more sensitive to the quality of the annotation than simple word counts.

EVALUATION AND QUALITY ASSURANCE OF MIGRATED ABAP CODE USING AN INTEGRAL METRIC AND GENERATIVE ARTIFICIAL INTELLIGENCE MODELS

O. A. Pozdnyakov — 2026-03-27

Context. Migration automation of legacy custom code when transitioning to the new version of the SAP S/4HANA system using large language models (LLMs) is a promising option. However, the generated code quality assessment remains an unresolved issue, since existing approaches utilize fragmented metrics which do not allow for a comprehensive software code quality assessment and assurance for further use without additional revision.
Objective. The objective of this work is to improve the efficiency of the process of intelligent reengineering of a computer system based on the method of comprehensive assessment and quality assurance of migrated ABAP custom code.
Method. The developed method is based on two key components. The Integral ABAP Quality Score (IAQS) comprehensively takes into account the syntactic, functional, and semantic characteristics of the code and is based on the provisions of the international software quality standards ISO/IEC 25010, ISO/IEC 25040, as well as the theory of composite indicators. The threestage approach to LLM fine-tuning (Qwen 2.5 Coder 14B) includes continuous pre-training (CPT), parameter-efficient fine-tuning (PEFT), and alignment based on preferences using the ORPO algorithm. At the same time, the use of the developed IAQS metric to form a set of preference data at the alignment stage creates a mechanism for controlled improvement, namely, it determines the direction of LLM adaptation.
Results. The results of experimental studies demonstrate that the implementation of the developed method allows improving both individual indicators of software code quality and the integral metric of IAQS quality assessment as a whole. The final model, trained on the basis of the proposed three-stage approach, achieved a high IAQS value (0.756), which demonstrates a significant improvement compared to the baseline model (0.117).
Conclusions. The study presents a new problem-oriented approach to automated migration of ABAP code during intelligent reengineering of computer systems. The proposed IAQS integral metric is the basis for creating a formalized and objective system for evaluating the quality of software generated by LLM in the context of legacy custom code migrating. It has been demonstrated that consistent fine-tuning of LLM based on a three-stage approach using IAQS provides a significant improvement in the generated software code integral quality indicator.

DEEP LEARNING MODELS FOR PREDICTING HUMAN MOVEMENT IN VIDEO STREAMS

N. V. Bilous — 2026-03-27

Context. The problem of accurately predicting human movement in an environment is critical for applications in monitoring, search, and navigation systems. Existing approaches often struggle to integrate spatial and temporal dynamics of trajectories while processing real-time video streams.
Objective. The goal of this work is to develop a deep learning-based framework capable of predicting human motion by combining object-level features and spatio-temporal trajectory information extracted from video streams.
Method. The proposed method integrates YOLO11 for object detection, which extracts coordinates, velocity, movement direction, and position relative to the environment. A graph neural network models local and global relationships between environment nodes, aggregating features while considering terrain structure and obstacles. Spatio-temporal attention highlights the most relevant moments in the trajectory, enhancing prediction accuracy. The model processes sequences of frames from video streams to predict subsequent positions of each tracked object in real time.
Results. Experiments on video sequences with varying motion scenarios, trajectory lengths, and speed variations demonstrated high prediction accuracy. The proposed method effectively integrates spatial and temporal features, outperforming baseline models in tracking and motion prediction tasks.
Conclusions. The results confirm that the proposed deep learning framework is suitable for real-time human motion prediction in complex environments. Future research may focus on extending the approach to multi-agent scenarios, optimizing computational performance, and testing on larger and more diverse datasets

LONG-DISTANCE CABBAGE DAMAGE AND PEST DETECTION METHOD USING YOLO11

K. S. Khabarlak — 2026-03-27

Context. To ensure sustainable yield, plant health must be constantly monitored with timely measures applied to prevent disease spread. Traditional approaches rely on manual inspection of plants, while neural networks require large amounts of annotated data to train. Both manual inspection and data annotation require expert knowledge and are time-consuming. Close-up photos of leaves are often used for training as they are easier to collect from the Internet. However, this complicates disease spread estimation at a scale. Cabbage is one of the plants widely grown in Ukraine, but existing research focusing on cabbage health monitoring is limited.
Objective. The goal of this work is to build a neural-network-based cabbage disease and pest detection system, which can be trained in on a small number of training images. At inference the system should detect pests on plant images at a distance of a whole plant.
Method. Given that existing plant disease datasets, such as IP102 and PlantDoc mostly contain close-up images of diseased plants, the networks trained on such datasets suffer from lack of generalization to images at a distance. To select the best object detection model, state-of-the-art object detection architectures, namely YOLO 8, 9, 10, 11, and RT-DETR have been analyzed in the work. To increase detection distance multi-image loss is proposed to improve hyperparameter search using Tree-Structured Parzen Estimators. Also, to improve detection quality, a novel cabbage disease dataset has been collected in Dnipro region, Ukraine. The new classes include crucifier flea beetle (widespread pest in Dnipro region) and damaged leaf. When the pest is not visible, but leaf damage is taken, determining specific pest might not be possible. Therefore, we introduce additional damaged leaf class, that captures generic plant damage. This also enables tracking of plant healing rate, when measures to stop pest spread have been taken. We combine collected images with the larger IP102 dataset to increase the number of pests covered to form new Cabbage+IP102 dataset.
Results. 1) Tree-Structured Parzen Estimators search on the multi-image loss has improved the YOLO 11 M performance from 0.3642 to 0.3892 mAP50-95 on images taken at a distance. 2) Collected dataset has enabled detection of cabbage plant health problems at a distance, including cases when the pest is currently not visible, but the damage is present.
Conclusions. In this work, the cabbage pest and damaged leaf YOLO 11 M detection system has been presented. The detector
architecture has been selected as the best-found during analysis on 2 datasets. The developed system requires only 7 annotated cabbage images to be trained and to perform pest and damaged leaf detection on high resolution images (2016x2016) of whole cabbage plants. The final model can be used to monitor cabbage health problems, damage, and rate of healing using images taken at a distance.

OPTIMIZATION OF FUEL CONSUMPTION IN THE PROBLEM OF STABILIZING THE ANGULAR POSITION OF AN AXISYMMETRIC SPACECRAFT

А. A. Stenin — 2026-03-27

Context. The problem of maintaining the angular orientation of a spacecraft is critical, especially when subjected to impulsive external disturbances that cause sharp deviations in angular velocities. The relevance of solving this problem is determined by the limited fuel supply on board, particularly for the class of spacecraft designed to provide artificial gravity, where precise and efficient control is paramount.
Objective. The main objective of this work is to minimize the consumption of energy resources (fuel) for the stabilization of the angular position of a specific class of spacecraft. This goal is achieved through the sequential execution of two interrelated tasks: 1) damping sharp deviations in the spacecraft’s angular velocities; 2) stabilizing the final angular position.
Method. A two-stage approach is proposed. To solve the first task (damping), optimal control is synthesized using a combination
of Pontryagin’s maximum principle and the phase plane method. This allows for the creation of optimal switching curves that divide the phase plane into regions with corresponding optimal control values. To solve the second task (stabilization), a modal approach based on a proposed method of indeterminate coefficients is used, which ensures the specified dynamic characteristics of the transient stabilization processes.
Results. Modeling of the dynamics of the spacecraft’s angular motion was carried out. The simulation results confirm the high effectiveness of using the proposed combined approach for solving the problem of stabilizing the angular position of the spacecraft after significant external disturbances.
Conclusion. The joint application of Pontryagin’s maximum principle with the phase plane method for fuel-efficient damping of angular velocities, followed by the implementation of an optimal stabilization law based on the proposed method of indeterminate coefficients, represents an effective procedure for controlling the orientation and stabilization of a spacecraft with minimal fuel consumption

FUZZY-LOGIC ALGORITHM FOR RISK ASSESSMENT IN WI-FI NETWORKS

I. Antipov — 2026-03-27

Context. With the increasing use of Wi-Fi wireless networks, the risk of attacks specific to them is also rising. Traditional protection methods, which usually rely on precise thresholds, do not reflect the actual uncertainty of the conditions in which wireless networks operate. Due to the openness of the radio channel, its instability, dispersion, and the presence of noise, a promising direction is the use of fuzzy logic algorithms, which allow for taking into account the incompleteness and ambiguity of data when assessing the risks of Wi-Fi wireless networks.
Objective. Develop a fuzzy logic algorithm for assessing the state of Wi-Fi networks, which allows adaptively determining the level of risk by analyzing wireless network parameters and making decisions regarding security system actions.
Method. A fuzzy-logic-based algorithm for analyzing the operational state of a wireless Wi-Fi network is proposed. The algorithm is based on the integrated analysis of six network parameters using elements of fuzzy logic. It includes the construction of membership functions for the input variables, the formation of a fuzzy IF–THEN rule base, and a defuzzification mechanism that provides a continuous numerical assessment of the network risk level. To evaluate the effectiveness of the proposed approach, a comparative simulation study was conducted against the classical threshold-based decision-making method. The study was carried out in the MathCAD and MATLAB environments to enable cross-validation of the algorithm’s functionality. Three network operation scenarios were considered, with 100 network states simulated for each scenario.
Results. The simulation results obtained in the MathCAD and MATLAB environments coincide up to the third decimal place, confirming the correctness of the software implementation of the algorithm. Comparative analysis showed that the threshold-based method produces binary decisions and is highly sensitive to random fluctuations in network parameters, which leads to an increased number of false alarms. The proposed fuzzy-logic-based algorithm provides a continuous risk assessment, demonstrates lower result variance, and exhibits a stable response to changes in network conditions. Under unstable network operating conditions, the algorithm enables discrimination between noise and interference effects and the initial phases of attacks, while also ensuring a gradual increase in the risk level without abrupt transitions between linguistic levels. The obtained results confirm a reduction in Type I errors and an improvement in decision-making informativeness.
Conclusions. The fuzzy logic-based Wi-Fi network state analysis algorithm proposed in this work enables more adequate decision-making regarding the network’s condition. The use of fuzzy logic allows adjusting decisions depending on changes in network operating conditions in real time and can be integrated into intrusion detection systems or advanced wireless network cybersecurity tools.

METHOD FOR MINIMIZING MESSAGE DELIVERY TIME IN METEOR-BURST COMMUNICATION CHANNELS

O. V. Holovan — 2026-03-27

Context. In special conditions, particularly during emergencies, when satellite and terrestrial communication channels become vulnerable or completely unavailable, communication via meteor burst channels can effectively serve as a backup or even a primary path for information transmission. The operational range of such a radio channel can reach up to 2000 km, and the absence of “dead zones” ensures broad territorial coverage that is comparable to other types of long-range radio communication. Objective Improvement of the method for one-way message transmission via meteor burst communication channels and its implementation algorithm, enabling minimization of message delivery time at a given reliability level.
Method. Further development was achieved for the method based on minimizing the message structure through merging the address field with the synchronization flag. Additionally, a hybrid synchronization algorithm combining threshold and non -threshold signal processing is applied for the first time. To enhance reliability, the majority algorithm is utilized instead of classical ARQ methods through repeated message transmission.
Results. An improved method for alert transmission over meteor-burst channels has been proposed, ensuring minimized delivery time and high reception reliability. Based on this method, a transmission protocol was developed, the message delivery time was evaluated, and synchronization techniques were identified, confirming the method’s effectiveness. The practical value lies in the development of an implementation algorithm suitable for deployment on DSP and FPGA platforms in alert systems without relying on satellite communication channels.
Conclusions. The proposed method and implementation algorithm enable the minimization of short message delivery time at a specified reliability level and improve communication reliability under challenging conditions.