Радиотехника и электроника

Abstract—There was significant recent progress in the field of deep learning, which has led to compelling advances in most tasks of semantic computer vision (e.g., classification, detection, and segmentation). Point cloud registration is a problem in which two or more different point clouds are aligned by estimation of the relative geometric transformation between them. This well-known problem plays an important role in many applications such as SLAM, 3D reconstruction, mapping, positioning, and localization. The complexity of the point cloud registration increases due to the difficulty of feature extraction related to a large difference in the appearances of a single object obtained by a laser scanner from different points of view. Millions of points created every second require high-efficiency algorithms and powerful computing devices. The well-known ICP algorithm for point cloud registration and its variants have relatively high computational efficiency, but are known to be immune to local minima and, therefore, rely on the quality of the initial rough alignment. Algorithm operation with the interference caused by noisy points on dynamic objects is usually critical for obtaining a satisfactory estimate, especially when using real LiDAR data. In this study, we propose a neural network algorithm to solve the problem of point cloud registration by estimating the soft alignment of the points of the source and target point clouds. The proposed algorithm efficiently works with incongruent noisy point clouds generated by LiDAR. Results of computer simulation are presented to illustrate the efficiency of the proposed algorithm.

Abstract—

The problem of depth estimation of a scene from a given light field can be reduced to the problem of classical stereo matching with the statement that matching pixels in stereo images have the same brightness values. However, this assumption is generally incorrect, given the presence of noise in the images and the different illumination of the left and right images in the stereo pair, as well as the existence of occlusions. In this regard, the luminous flux, representing 80 images pairwise epipolar to the central one, offers a number of advantages, especially in terms of occlusion handling. In this paper, we propose the principle of viewing symmetry: if a pixel of the central image lies in the occlusion zone relative to one of the peripheral images of the light field, then this pixel does not belong to the occlusion zone for an axisymmetric image of the peripheral field. Thus, it is possible to form a robust volume of discrepancy weights relative to the occlusion. As a result, the algorithm proposed in the article significantly improves the result of the scene depth reconstruction. The effectiveness of our approach is demonstrated using the main test database of the light field and comparing it with the best reconstruction algorithms in the efficiency of border recognition and in the speed of calculation.

Abstract—The energy minimization or maximum a posteriori probability (MAP) method is the basis for solving many computer vision problems, including the segmentation problem. However, it is assumed that the number of regions during segmentation is quite small. At the same time, in the problem of superpixel segmentation or otherwise excessive segmentation, the number of such areas exceeds 1000, which makes the computational optimization problem by the MAP method practically impossible. In this paper, we propose a solution that reduces segmentation with any number of areas to the problem of marking only nine labels. In addition, convolution with the geodesic distance kernel is used to enhance the robustness of optimization. This makes it possible to obtain single-linked superpixels at the output of the algorithm, unlike many other methods that require additional adjustments. The effectiveness of the proposed method is compared and measured by the precision-recall criteria, as well as by visual illustration.

Abstract

Mathematical and simulation models of network nodes and the simplest topologies of modern data transmission networks are developed. The functional model of a modern network node is described, and its mathematical model is developed. A simulation model of a topology consisting of network nodes is developed to evaluate probabilistic and temporal indicators of the service quality of a communication network. Based on the simulation model, the probability of packet loss is plotted versus the packet arrival rate for the topology under study.

Abstract—Construction of a combined model of time series (for two models of the same type that exhibit additivity, for example, ARIMA) or a combined forecast of models (in the absence of additivity, for example, for ETS models) providing minimization of the estimated forecast variance is considered. As distinct from alternative models of time series in which the forecast variance is estimated using the Student test, the ARIMA and ETS models allow construction of a function that is related to the parameters of model. Thus, it is possible to estimate the value of the confidence interval for the forecast and construct combinations of models with a minimum estimate of the width of the interval depending on the parameters of the combination. The theoretical part of the work studies linear combinations of forecasts of two models, in which the estimate of forecast variance is minimized (regardless of the type of model). The Hessian of the function for estimating the forecast variance is obtained for construction of a linear combination of forecasts. It is analyzed under the conditions for extremum (zero first derivatives of the function for estimating the variance of the forecast for the combined models). Then, the Hessian is estimated for several groups of ETS models, and the conditions for the presence of a minimum of the estimated forecast variance at a stationary point are considered versus parameters of models.

Abstract—Multi-link operation (MLO) is one of the main innovations of the upcoming IEEE 802.11be standard. MLO uses a common sliding window for all links to control the delivery of packets. Due to the finiteness of the sliding window, transmission of packets on one link may result in packet shortages on other links, reducing throughput. To avoid it, an aggregation algorithm is required that selects the number and packets to be transmitted. In this paper, the known algorithms are extended by sending duplicates. Simulations have confirmed that duplicates can increase the Multi-Link Device (MLD) throughput by accelerating the sliding window progression.

Abstract

An important step in solving the problem of classification and segmentation of 2D images is the extraction of local geometric features. Convolutional neural networks were widely used in recent years to solve problems in this field. Typically, the neighborhood of each pixel in an image is used to collect local geometric information. A convolutional neural network is used to extract the underlying geometric features of the neighborhood. In this work, we propose a neural network based on descriptor concatenation for two well-known neural networks to solve the problem of extracting local geometric features of mammographic images. To improve the accuracy of mammogram classification, feature filtering is used based on the calculation of joint information. Results of computer simulation are presented to illustrate the performance of the proposed method.

Abstract—The main task of the study is to improve the performance of existing computer diagnostic systems using new methods for classification of benign and malignant tumors using digital mammograms. Methods and algorithms for systems of computer diagnostics are being actively developed using deep neural networks. To achieve better results on the selected data set, we transform the data using autoencoders to obtain features with low intraclass and high interclass variance. The entire working cycle of the system consists of the following stages: extraction of features using a segmented part of the pathology, division of the data into two clusters, and feature transformations using linear discriminant analysis for minimization of intraclass variance and classification of pathologies. The results of this study show that the classification of pathologies using deep learning methods makes it possible to achieve high results.

Abstract—The paper addresses capacitated clustering problems: (a) basic capacitated clustering problem, (b) capacitated centered clustering problem, (c) multicapacity clustering problem, and (d) related problems. The paper is based on the combinatorial clustering viewpoint. A survey on the problems, solving approaches, and some applications are presented. The optimization models of the basic capacitated clustering problem and multicriteria capacitated clustering problem are considered. An application of capacitated clustering as the handover minimization problem in mobile wireless networks is briefly described. Numerical examples illustrate problems and applications.

Abstract—The rapid growth in the number of smart devices capable of exchanging data within a single network leads to the emergence of mechanisms that allow adapting data transmission technologies to the Internet of Things networks. One of them is the mechanism of the periodic restricted access window (PRAW) presented in the 802.11ah standard. A competent choice of parameters of the PRAW mechanism allows a large number of sensors to transmit data quickly and energy-efficiently, but the 802.11ah standard itself does not give recommendations on their choice. This article solves the following optimization problems: minimizing (a) the average delay, (b) the average energy consumption per transmitted packet when the average delay limit is met, and (c) the share of channel time consumed by the PRAW mechanism when the restrictions on both metrics are met. Based on the results of solving these problems, we give recommendations on the choice of PRAW parameters for different network loads determined by the intensity of packet generation and the number of stations.

Abstract—

Multi-User Multiple Input Multiple Output (MU-MIMO) technology allows you to increase the channel throughput. However, its efficiency is reduced by overhead induced by frequent channel sounding and transmission of channel feedback frames. This article examines the problems of channel state information (CSI) compression in Wi-Fi networks using MU-MIMO with channel aging. The research aims to experimentally test the effectiveness of MU-MIMO technology in real use cases, considering the channel sounding procedures and CSI feedback transmission. Using an experimental setup, the channel has been recorded to analyze its behavior and the evolution of the signal received power under different conditions. In addition, the limits of the applicability of the TGax MU-MIMO channel model in the WLAN Toolbox have been investigated by comparing it with experimental results. The findings of this study are particularly useful in optimizing MU-MIMO performance under channel time evolution and CSI compression.

Abstract—Reconfigurable intelligent surfaces (RISs) are a promising technology for increasing the information capacity and coverage of future wireless networks. Various available types of these devices consist of different elements that can be used for signal absorption, increasing the information capacity, and phase shift keying. This causes a lack of a single concept and misunderstanding of functionality of specific RISs. The aim of this study is to eliminate this gap by describing one of the most widely used RIS structures and its operating principles, which make it possible to formulate the main functionalities of the RIS.

Abstract—Hyperspectral analysis is a powerful tool in the precision agriculture arsenal that becomes increasingly accessible. The number of hyperspectral images obtained near the Earth surface is constantly growing. It is important to consistently use this data along with conventional data of multispectral monitoring. In this work, problem of harmonization of hyperspectral survey data obtained at the surface of the Earth and satellite multispectral monitoring data is investigated. The problem of spectral harmonization, which is insoluble in general case, is further complicated in this case by the heterogeneity of the available data. In this regard, a simplified formulation of the harmonization problem is considered, aimed at calculation of vegetation indices. A novel method has been developed that does not require pixelwise matching or calibration panels. The experimental part of the work shows that the proposed method allows significant compensation for shifts of the NDVI and WBI, observed in the absence of harmonization.

Abstract—Reconfigurable intelligent surfaces (RISs) are promising devices capable of increasing the information capacity and coverage of new and available wireless networks. By now, most of these devices have been presented in the form of prototypes that have no environmental protection and are not adapted for use in real communication systems. Meanwhile, a protective coating can significantly affect the characteristics of a RIS and reduce its efficiency. This study considers the effect of thickness, permittivity, and dissipation factor of the most common protective coating materials on the frequency and phase responses of a RIS unit cell (UC). Recommendations on choosing a material and its thickness and on correcting UC sizes at constant parameters of a protective coating are provided.

Abstract—Reconfigurable intelligent surface (RIS) is a promising technology that can increase the capacity and coverage of wireless networks. The effectiveness of a RIS is determined by its configuration, which can be made based on the information about location of the transceiver devices. In practice, there are two main types of RIS configurations: focusing the signal reflected from the RIS at the receiver location and redirecting the signal towards the receiver. Both types of RIS configuration become outdated in time due to changes in environmental parameters caused by the movement of transceiver devices and other objects in the surrounding environment. This paper examines outdating of RIS configurations made by focusing and redirection procedures in a system with spatially static transceiver devices. The paper shows that the difference in signal-to-noise ratio for the two types of configurations can reach up to 8 dB and has nonmonotonic features that can be explained by considering the near-field region of a RIS.

Abstract

To increase throughput of Wi-Fi networks, the IEEE 802.11be standard has introduced a multi-link operation feature that enables devices to transmit and receive data in multiple channels. The article studies an adaptive waiting control algorithm for channel access by multi-link devices (MLDs) incapable of simultaneous transmission and reception (NSTR) at different channels. The algorithm features accounting for the channel capacities, the number of single-link devices in the network, and the traffic intensity in the channels. The simulation shows high performance of the proposed algorithm.