The influence of technological parameters of deposition of dielectric passivating films by the ICP CVD method (gas ratio, inductively coupled plasma power, substrate temperature, operating pressure) on the structural and electrophysical properties of AlGaN surface layers was studied. It is shown that preliminary plasma-chemical treatment of AlGaN in an oxygen environment leads to the formation of a controlled oxide interface, which reduces surface states and suppresses drain-source leakage currents in the closed state by two to three orders of magnitude. A correlation was established between silicon nitride (SiN) deposition conditions and the degree of plasma-induced damage to AlGaN/GaN heterostructures. To minimize degradation, a composite dielectric based on oxygen-containing silicon nitride (SiON) is proposed. The key result is the development of a combined approach: preliminary modification of the AlGaN surface with oxygen plasma followed by SiON deposition in a low-power ICP CVD regime (power ≤300 W, pressure ≥14 Pa). The obtained results demonstrate the promise of the combined approach (surface pre-modification + controlled passivation) for minimizing the degradation of the electrophysical properties of HEMT structures.

Modern machine and deep learning methods effectively handle direct sensor data processing tasks and identify complex dependencies, ensuring robustness to noise and data variations. The most pressing problem is the synthesis of the results of processing multidimensional data using hybrid learning methods to solve inverse problems, such as determining the components of dynamically changing complex heterogeneous systems without violating their properties in multisensory measurement platforms. This paper presents a hybrid deep learning method integrated into an optical sensor platform that combines convolutional neural networks, bidirectional long short-term memory networks, and transformer encoders to enable dynamic monitoring of heterogeneous system components. The hybrid method, when applied to a biosensor platform, demonstrates classification accuracy of over 98 % by solution type and recovery of protein component concentration values with a median accuracy of up to 2 ng/ml.

Mathematical models for the analysis of probabilistic characteristics of the spectral efficiency of downlink radio channels in macro-sites of mobile (cellular) radio communications have been developed, and estimates of the probability distributions and average values of the potential and real spectral efficiency of downlink radio channels have been made for various variants of the typical IMT-2020 scenarios “Rural-eMBB”, “Urban Macro–mMTC” and “Urban Macro–URLLC”. Obtained models and estimates take into account the randomness of the user’s equipment distribution on the macro-site territory and the possibility of user’s equipment location in areas of both free and interference (multi-ray) propagation of radio waves between the base station and user’s equipment. These results allow us to conclude that the declarations of new generations of mobile communications in terms of increasing the spectral efficiency of radio channels at the 4G→5G→6G evolution are feasible.

The article presents the results of antenna directivity characteristics measurement using a near-field scanner. A comparison is made between the radiation patterns of the manufactured antenna array obtained using the scanner and the classical method using a turntable in an anechoic chamber. A brief description of the scanner developed by the authors, the algorithms and methods used in the process of converting the measured near field of the antenna into a directional pattern are provided. The scanner’s advantages include the ability to assess defective array elements; significantly reduce the size requirements for the measurement room; the ability to obtain a three-dimensional radiation pattern in the forward hemisphere; and the ability to perform primary calibration of the antenna system.

A hardware and software model for constructing data transmission signal parameters in a noisy environment has been developed. Approaches to increasing the throughput of information channels, applied in specialized scientific and practical research, are analyzed. A scheme for identifying quasi-harmonic and rectangular signals for modification into a common group signal is presented. Experimental confirmation of the stability of the computational process when modeling digital streams using overlapping quasi-harmonic signals is obtained. The fundamental principles of calculating the basic components of a channel pulse enable the design of devices with high data throughput.

A method for measuring the complex dielectric constant of materials using a resonant method based on a biconic resonator is presented. An improved design of the measuring chamber is proposed, which ensures high quality and stability of resonant frequencies. An algorithm for processing the results has been developed, including identification of resonant modes, calculation of the parameters ε′ and ε′′, as well as correction of errors related to sample geometry and field inhomogeneity. Numerical simulation was performed in CST Microwave Studio. The data obtained confirm the accuracy and reproducibility of the proposed resonant method in the microwave frequency range.

An approach to analyzing classification features for radar recognition of small unmanned aerial vehic­ les based on spectral radar portraits generated from real S-band pulsed radar recordings is developed and stu­ died. The problem of extracting and assessing the information content of signal features available for practical implementation in detection and tracking radars is considered. Experimental data were processed, and a set of physically interpretable features is generated, reflecting the properties of the airframe component and secondary micro-Doppler modulation of the reflected signal. To quantitatively evaluate the separating ability of features, the Fisher criterion, the amount of information and normalized interclass distances were used. It is established that micro-Doppler features possess the greatest information content, ensuring stable separation of multicopter and hybrid types of unmanned aerial vehicles.

This article analyzes information security events in a virtual infrastructure-based electronic digital signature system in accordance with the requirements of the regulations on the procedure for information exchange between elements of the national cybersecurity system. Key system modules are considered: the authentication module, the electronic digital signature generation module, and the verification module, including their submodules for data protection, signature generation, and verification. Examples of events for logging are provided, such as user authentication, electronic document generation, hash value transfer, and indicators of cyber incidents, including abnormal user actions. Information security events logged in virtual infrastructure-based electronic digital signature systems are presented. An assessment is made of the compliance of the logging results for these events with the requirements set out in Order No 130 of the Operational and Analytical Center under the President of the Republic of Belarus dated July 25, 2023.

The article discusses the problem of organizing effective component interaction in client web applications based on component architectures. It is shown that as the scale of interface systems increases, traditio­ nal approaches to data exchange between components lead to redundant calculations, more complicated support, and lower performance. Using the example of frameworks, the existing mechanisms for synchronizing the state and updating the user interface are analyzed, their limitations related to global change tracking and the lack of builtin support for two-way data transfer between components are identified. A Bidirectional Component Linking Model (BCLM) is proposed, based on the direct exchange of relevant state changes between logically connected components. A synchronization mechanism has been formalized through the display of publication and application of data, providing adaptive interface updates without the involvement of centralized repositories. It is shown that the proposed approach makes it possible to reduce the number of re-renderings and simplify the application architecture while maintaining reactivity.

Visual speech recognition is the task of classifying spoken words or letters from a video stream capturing lip movements. This paper presents the synthesis and study of a neural network architecture for visual speech recognition based on a combination of convolutional and recurrent neural networks with an attention mechanism. The model was trained and evaluated on the AVLetters2 dataset in the most challenging speakerindependent mode. The model architecture includes an encoder based on convolutional layers for extracting spatial features, recurrent layers based on GRU units for modeling temporal dependencies, and an attention mechanism for highlighting informative fragments of the speech sequence. To assess the accuracy of the model, five-fold cross-validation was performed. Model hyperparameters were selected using Bayesian optimization, which allowed us to determine the optimal configuration of the model parameters and the training process. As a result of the experiments, an average recognition accuracy of 14.3 % was achieved. Analysis of the results revealed signi­ ficant variability in recognition quality depending on the characteristics of the speakers (accuracy ranged from 3.9 to 31.9 %), which indicates the need to further improve the invariance of the model to inter-speaker differences.

This article examines methods for improving the energy efficiency of IoT edge devices, including hardware solutions, wireless powering via Wi-Fi, communication protocols, energy management algorithms, and the use of renewable energy sources. A hybrid power system architecture is proposed, and an adaptive control sequence algorithm is developed as a diagram. Together, these factors ensure efficient power supply to IoT system sensors. The main challenge was identifying the high nonlinearity of the rectifier system impedance, which depends on the input power, frequency, and DC-DC converter parameters, making traditional approaches to mat­ ching circuits ineffective over the wide Wi-Fi frequency band.

The application of plasma technologies in engineering, and particularly in healthcare, requires assessing the impact and interaction of plasma with material surfaces. This article presents developed methods for reliably monitoring the plasma state and its interaction with the surface of objects, ensuring the reproducibility of technological processes for surface treatment of materials and therapeutic procedures, as well as expanded opportunities for increasing the efficiency of plasma technologies in healthcare. An assessment of the impact of plasma on the surface of various organic and inorganic materials, biological tissues, and an assessment of the interaction with them was carried out. For a qualitative and quantitative assessment, it is proposed to use the volt-ampere characteristics and the power released inthe “plasma-object” load. It has been established that the power released at the load during interaction with cold plasma depends on the material being exposed to the plasma. This parameter must be monitored when processing materials and biological tissues. It is shown that there are differences in the values of active power released at the electrodes of the discharge system for biological tissue samples (subcutaneous fat and muscle tissue). Moreover, this difference increases with increasing output voltage.