In the context of the issue of increasing the cyber resilience of information and communication systems (ICS), the scientific task of predicting structural changes in cyberattack techniques in the space of their possible implementations is solved using the MITRE ATT&CK taxonomy as an example. The relevance of the above is due to the objective limitations of existing approaches to hunting for new ways of implementing cyberthreats (preventing cyberattacks before their implementation), in particular by means of machine learning. Thus, existing types of artificial neural networks used to predict cyberattacks (cyberattack techniques), such as recurrent, transformer, convolutional, graph, temporal graph and autoencoders, although they take into account various aspects of the data structure (feature space, temporal dependencies, global context, graph structure, latent representations, data distribution), approximating the function of interdependence between data, reveal only a statistical structure, which does not allow fully taking into account stable structural evolutionary patterns. In this regard, a model for predicting structural changes in cyberattack techniques in the space of possible implementations based on topological constraints on their evolution using temporal graph neural networks has been developed. The essence of the proposed model is to adjust the results of predicting a temporal graph network based on topological analysis of versions of cyberattack techniques by determining their structural compatibility through joint participation in connectivity components and cyclic structures that reflect stable topological characteristics of their evolution. The feasibility of this approach is due to the value of identifying potential vectors of cyberattack transformation in the space of their possible implementations, which makes it possible to increase the cyber resilience of the ICS to future classes of cyberattacks. An assessment of the effectiveness of the proposed model demonstrates a 15% increase in prediction accuracy and an 8% increase in F1-measure while maintaining a completeness level of 70%, indicating a significant reduction in the number of false positives.
This article is devoted to solving an urgent scientific and applied problem — increasing the efficiency of compression and ensuring the integrity of video information resources under conditions of limited channel bandwidth and the influence of interference. The introductory part substantiates the need for developing new approaches to image segmentation, as traditional methods (JPEG, H.264/AVC, HEVC) and modern formats (JPEG XL) have limited ability to adapt to complex structural features and high vulnerability to channel errors. It is determined that the loss of even a small part of the encoded stream in existing systems leads to disruptions in the decoding process of entire sections due to the lack of structural localization mechanisms. The theoretical section of the research focuses on the transition from the standard ARGB model to the XYB color space, which better matches the characteristics of human perception and avoids ignoring important color components (specifically red). An approach for independent processing of each layer of the color model is proposed, which provides a more efficient division into regions with homogeneous properties. The key contribution of the work is the development of a method for constructing dynamic apertures for asymmetric numeral systems (ANS). Instead of static partitioning, flexible segmentation is introduced, which takes into account local color and structural parameters of the image by dynamically changing the aperture "tube" parameters. The results section presents a comparative analysis of standard 8x8 partitioning, simple aperture partitioning, and the proposed adaptive method. Threshold values for aperture size (from 5 to 100 elements) were experimentally established, at which it is advisable to perform dynamic adaptation of the reference value. This allowed for the optimization of computational costs: encoding time was reduced from minutes to a few seconds, which is crucial for real-time systems. The obtained results confirm that the use of adaptive apertures together with strict value approximation allows for increasing the efficiency of compression using ANS by almost 5 times compared to base approaches, from 228,453 to 40,796 bytes, while maintaining the clarity of significant objects and high noise immunity.
The relevance of the study is determined by the recognised problem of heterogeneity and fragmentation of the cyber defence terminology in Ukraine. The lack of a unified, systematic terminology hinders effective interagency communication, the development of organizational and regulatory documentation, and operational coordination in countering cyber threats. The interdisciplinary nature of cyber defence, which encompasses national security, technology, law, and other fields, reinforces the need to unify the conceptual apparatus. Until now, this area has been studied selectively, without considering terminology as a coherent system. The purpose of this article is to highlight a methodological approach to creating a cyber defense terminology system. The methodological basis of the work is the theoretical foundations of terminology, in particular the distinction between the concepts of “terminology” (as a spontaneously formed set of terms) and “terminology system” (as a consciously ordered, coherent, and structured system of concepts based on a specific concept). An analysis of the current state of cyber defence in Ukraine shows that, in parallel with the development of the cyber defence system itself, cyber defence terminology is in the process of active formation and is undergoing stages of coordination and consolidation in scientific publications and legislative acts. A methodological approach is proposed, which consists in systematising and structuring the existing terminological fund to give it the characteristics of a coherent terminology system. Following this approach, key components have been identified that form the basis of the terminology system structure. The proposed model consists of six main subsystems (groups of terms): fundamental terms and concepts, cyber defence objects, cyber threats and vulnerabilities, the functioning of the cyber defence system and its components, cyber defence entities, as well as the legal and regulatory environment. It is noted that the cyber defence terminology system is synthetic in nature, integrating concepts in the fields of information technology, politics, economics, law, and defence. The proposed methodological approach has made it possible to create a comprehensive and structured terminology system that allows for the unambiguous classification of any term related to cyber defence. It is emphasised that this system is not static and will evolve along with changes in technologies and forms of cyber warfare. The practical value of the research results lies in the development of methodological recommendations for the further formation and improvement of terms based on the principles of consistency and unambiguous interpretation, which will contribute to ensuring effective communication and cooperation between different sectors and their operational coordination in the field of cyber defence.
The article presents a comprehensive study of current password policy issues in modern information and communication systems. It is substantiated that in the face of rapid computer technology development and the growing computing capabilities of cyber adversaries, traditional authentication methods require a significant overhaul. The authors identify key shortcomings of existing approaches, including the use of outdated hashing algorithms, the complexity of implementing multi-factor authentication (MFA) across all workstations, and the critical impact of the human factor (password reuse, storing credentials in plaintext). Particular attention is paid to the analysis of password entropy as the primary indicator of resistance to brute-force attacks. The paper provides a classification of entropy levels based on the sensitivity of the protected information: from 40–64 bits for public data to over 112–128 bits for critical infrastructure objects and restricted access information. The authors demonstrate that the use of modern graphics processing units (e.g., NVIDIA RTX 4090) allows attackers to crack weak passwords (based on MD5 or SHA-1) in mere minutes, making the transition to long and complex password combinations vital for security. It is proven that meeting the requirements of modern password policies is practically impossible for the average user without the use of specialized software. In this regard, the functional capabilities and security architecture of leading password managers – 1Password, Bitwarden, and LastPass – are analyzed in detail. Their encryption algorithms (AES-256, Argon2id, PBKDF2) and the "zero-knowledge" concept, which guarantees that only the master password holder can access the data, are thoroughly examined. The article proposes recommendations for selecting the optimal password length depending on the character set used to achieve target entropy indicators. The authors emphasize that the implementation of automated password management tools combined with multi-factor authentication is a fundamental condition for strengthening national security and increasing the cyber resilience of information and communication systems of state organizations and institutions.
Modern battlefields demand real-time intelligence and rapid decision-making in complex environments. This paper proposes an AI-assisted multi-sensor fusion architecture deployed on FPGA-based Unmanned Aerial Vehicles (UAVs) for enhanced situational awareness. The system integrates data from hetero- geneous sensors—thermal, acoustic, visual, and radar—through a lightweight fusion algorithm optimized for FPGA implementa- tion. The use of adaptive AI-driven fusion enables low-latency, power-efficient and reliable detection of enemy drones, impro- vised explosive devices (IEDs) and human activity. Experimental simulations demonstrate significant improvements in detection accuracy and response time compared to conventional centralized systems.
