The elemental composition and structure of coatings made of Fe–Ni and Fe–Co alloys, formed by electron beam evaporation, were investigated. The azeotropic composition of the Fe–Co alloy was calculated. It was shown that the alloy containing 54 wt.% (∼55 at.%) Fe and 46 wt.% (∼45 at.%) Co evaporates congruently. It was found that at a deposition rate of coatings made of the above alloys on substrates made of polymer materials of 5 nm/s or more, the surface layer of these substrates is heated to a temperature above 100 °C. This leads to the destruction of the surface layer of the polymer with the release of volatile components and the formation of a relief coating surface. At coating deposition rates from 0.5 to 1.0 nm/s, the substrate temperature does not exceed 50 °C, while the surface roughness of the coating decreases and its reflectivity increases. It was shown that 150 nm thick coatings of Fe₅₆Ni₄₄ and Fe₅₅Co₄₅ alloys with azeotropic compositions obtained at a deposition rate of 1.0 nm/s have an X-ray amorphous (Fe–Ni system alloy) or predominantly amorphous (Fe–Co system alloy) structure. The elemental composition of the coatings was studied, and it was found that in the Fe–Ni alloy, the deviation towards a decrease in the iron content in the coating relative to the azeotropic one was less than 5 %, while in the Fe–Co alloy, the deviation towards an increase was 18 %.

An analysis of current standards for maximum permissible levels of exposure to radiofrequency electromagnetic fields from 4G/5G mobile communications equipment via the enclosure port of various technical devices was conducted. It was found that for many technical devices, these requirements are comparable to or more stringent than the requirements for the general public’s exposure to these fields. Emissions from 4G/5G base and subscriber equipment are potentially hazardous to the operation of a large number of technical devices, including those used in critical infrastructure facilities. Compliance with standard requirements for susceptibility to radiofrequency electromagnetic interference via the enclosure port for most technical devices does not ensure reliable protection for their operation in the complex electromagnetic environment created by 4G/5G radio network emissions, indicating the need to revise these requirements. Calculated data on the required spatial separation between 4G/5G equipment and technical devices of various groups to ensure their electromagnetic compatibility are provided.

This paper examines variants of a reflected signal angular coordinate meter using artificial neural networks for a frequency-agile radar based on an antenna array with subarrays, generating sum and difference channels, and an array of adapted receiving channels near the observation direction. A model of amplitude-phase- frequency non-identities of receiving channels during frequency agility and angular scanning is developed. Neural network architectures are proposed based on a multilayer perceptron for the case of creating sum-and-difference channels and a complex convolutional neural network and a multilayer perceptron for generating an array of adapted receiving channels. Modeling shows that the proposed architectures ensure the almost complete elimination of angular coordinate estimation errors associated with the presence of amplitude-phase-frequency non-identities of receiving channels, with an estimation accuracy close to the potentially achievable one.

With increasing integration density and decreasing feature sizes, there is a need to optimize photolithography processes. Off-axis illumination is effective in increasing the resolution and depth of focus of lenses in projection equipment operating in the UV range using mercury-vapor discharge lamps. Off-axis illumination reduces diffraction limitations and improves resolution within design limits of approximately 250–350 nm. This article presents the results of a comprehensive analysis of methods for developing and optimizing off-axis illumination systems in projection photolithography. A method for simulating off-axis ring illumination for systems lacking standard resolution enhancement systems is proposed and tested. The effectiveness and adequacy of this method are confirmed by experimental studies.

The performance of adaptive modulation and coding in second-generation digital broadcasting systems (DVB-T2) is investigated in terms of the reliability of the transmitted signal, the efficiency of channel bandwidth utilization, and the quality of the supported service. The performance of quadrature phase-shift keying (QPSK) modulation schemes (16QAM, 64QAM, 256QAM) using low-density parity-check (LDPC) codes with rates of 1/2–9/10 is assessed in the context of their application to provide high-quality digital terrestrial television broadcasting based on adaptive modulation and coding methods. The simulation results yielded bit error char acteristics for various modulation schemes using LDPC codes at different rates in channels with additive white Gaussian noise and multipath fading simultaneously. It was found that simpler modulation schemes, such as 1/2 rate QPSK, can provide a bit error rate of less than 10^–5 at a signal-to-noise ratio of 5 dB, while more complex ones, such as 9/10 rate 256QAM, require a signal-to-noise ratio of 25 dB to provide a bit error rate of less than 10^–6. An improvement in channel throughput was noted from 1.0 bps/Hz for QPSK modulation to 6.7 bps/Hz for 256QAM under high signal-to-noise ratio conditions.

The use of modern information security tools to protect information infrastructure objects interacting with other information systems does not guarantee cybersecurity in the event of complex and targeted cyber attacks. To address this issue, it is advisable to use unidirectional data transmission tools. This article pre sents the results of the substantiation and development of an improved unidirectional data transmission model, in which guaranteed unidirectional data transmission is ensured by hardware, and application-level operation is ensured by special software. To ensure fault tolerance, redundant proxy servers are used to organize the ope ration of special software. An algorithm for the improved model is presented, which provides for the use of a reverse data transmission channel to confirm successful transmission, and also implements a probabilistic mechanism, which reduces the risk of an attacker exploiting the reverse channel to transmit information using frequency or time modulation.

This article presents a software and hardware system developed for a comprehensive multimodal assessment of children’s motor abilities based on an analysis of the kinematic and dynamic characteristics of running locomotion. A detailed algorithm for the system’s operation is considered, describing the stages of its initialization and registration, as well as the processing of multimodal data. The kinematic and dynamic parameters of movement, as well as the generation and presentation of diagnostic results, are calculated. A structural diagram of the system based on a three-tier architecture is presented, describing its main functional blocks. The developed system implements an end-to-end diagnostic process for children, from recording biomechanical signals to generating an interpretable result – an individual motor profile and a final assessment of motor potential on a threepoint scale. The obtained results will improve the accuracy and reliability of the early identification of gifted children with motor development and the development of individual trajectories for their physical development.

Methods for the efficient hardware implementation of neural networks on FPGA-type programm able logic integrated circuits are investigated. A key aspect is the influence of the choice of activation functions on the characteristics of the developed device. An approach to the use of activation functions that allow for efficient hardware implementation is proposed. In particular, the use of LeakyReLU as a compromise between computational simplicity and classification accuracy is justified. To test the approach, a two-layer feedforward network architecture was developed, its hyperparameters were optimized, and hardware implementation was carried out on a PYNQ Z2 board. An analysis of the impact of the bit depth of the coefficients in fixed-point format on the re cognition accuracy of the MNIST database and on hardware costs is conducted. The optimal bit depth of the fractional part (9 bits) was experimentally determined, ensuring an accuracy of 95.27 % while economically using the resources of programmable logic integrated circuits. Additionally, using the Kullback – Leibler divergence, the distortion of the weight distribution during quantization was estimated, on the basis of which a regression model was constructed to predict the accuracy of a neural network with quantized coefficients.

The integration of the internet of medical things and artificial intelligence technologies is creating a new paradigm in diagnostic medicine. Taking into account growing demands for data security, processing speed, and regulatory restrictions, this article examines the architecture of a local (edge-based) internet of medical things system for analyzing multimodal medical images, such as magnetic resonance imaging and positron emission tomography scans of patients with Alzheimer’s disease. Unlike cloud-based solutions, the proposed system processes data directly at the medical facility, minimizing delays, reducing the risks associated with the transfer of confidential data, and providing complete control over the information. The system utilizes modern convolutional neural networks for automatic segmentation, classification, and multimodal analysis, demonstrating a 15–30 % increase in diagnostic accuracy for neurodegenerative diseases in research settings. The structure and details of the system are developed, and its key components are described. Neural network training and Alzheimer’s disease recognition using these networks are discussed. The advantages of the local approach and the prospects for implementing the internet of medical things system are highlighted. 

With the rapid growth of text data, there is a need for methods capable of effectively comparing text fragments by meaning, including cases of paraphrasing, synonymization, and sentence restructuring. One of the pressing challenges is comparing the results of semantic comparison methods based on various models with the human perception of semantic similarity. This article discusses an expert method for assessing the semantic similarity of text fragments based on the assessments of survey participants. The method consists of creating an interpretable semantic similarity scale derived from human perception of text content and used to analyze the consistency of various methods. To develop a “human” assessment, a survey of 138 participants was conducted. A comparative analysis revealed that various semantic similarity assessment methods demonstrate varying degrees of consistency with the human perception of text semantic similarity.