Научный журнал «ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ И ТЕЛЕКОММУНИКАЦИИ»
Сообщение
2025, Vol. 13, Iss. 1 |
|
A. Paramonov, Ph. N. Hoang
|
|
Abstract: Purpose: The article addresses the task of dynamic subchannel selection in heterogeneous Internet of Things (IoT) networks, considering network parameter changes and the limited computational resources of devices. The subject of the study: Heterogeneous IoT networks that utilize various data transmission technologies. Methods used: The study employs a reinforcement learning method for dynamic subchannel selection based on the analysis of historical data and the current network state. A tug-of-war algorithm is also used for resource allocation among subchannels. Results: A method for dynamic subchannel selection has been developed, which allows for consideration of the probability of successful data transmission, subchannel usage frequency, and failure probability, thereby balancing transmission efficiency and computational costs. Theoretical /Practical relevance: The practical significance of the results lies in improving the performance and reliability of heterogeneous IoT networks under high load and with limited device resources.
Keywords: heterogeneous networks, communication channel, Internet of Things, tug-of-war algorithm, reinforcement learning.
DOI 10.31854/2307-1303-2025-13-1-1-13
EDN DYOMML
|
|
V. Tsap, G. Fokin
|
|
Abstract: The paper considers the applicability of machine learning models and methods in spectral probing to increase the speed of scanning and analyzing LTE signals. It describes the operating procedure of the software module for scanning LTE signals using the spectral probing algorithm in a wide frequency range. Methods. The research method is a full-scale experiment using software-defined radio boards. The result of scanning and analysis is the detection of signals from LTE base stations operating on transmission in a given area. The efficiency of detecting base stations is estimated by classifying spectrum range using machine learning methods. Practical relevance. The combination of a software module for panoramic scanning in a wide range and a software module for analyzing in the information frequency band allows to significantly reduce the detection time of LTE base stations in a given area.
Keywords: spectral probing, LTE standard, software-defined radio, machine learning.
DOI 10.31854/2307-1303-2025-13-1-14-22
EDN XYAPHF
|
|
A. Kalachikov, I. Popovich, V. Pushnitsa
|
|
Abstract: Statement of the problem. Software radio allows flexible implementation of algorithms for processing signals in radio communications. Reception of signals is possible only with synchronization in the time and frequency domains, taking into account the properties of the signals. The paper presents a prototype of a communication system with orthogonal frequency multiplexing, implemented on the Adalm Pluto platform using the libiio library. The aim of the study is to analyze and implement software algorithms for symbolic and frequency synchronization when receiving signals with orthogonal frequency multiplexing. For this purpose, a preamble based on the Zadoff – Chu sequence is used. The frequency shift was estimated using two methods: using a cyclic prefix of symbols and using the Zadoff – Chu preamble. Novelty. The developed algorithms are implemented as programs, without using specialized libraries of ready-made modules and tested on the Adalm Pluto hardware platform. The obtained results confirm the operability of the proposed solutions, which allows them to be used in software radio systems when implementing communication channels of various autonomous systems. The practical significance lies in the experimental confirmation of the functionality of the proposed solutions, which allows their use in software radio systems when implementing communication channels for various autonomous systems.
Keywords: software radio, time synchronization of signal reception, frequency synchronization of signal reception.
DOI 10.31854/2307-1303-2025-13-1-23-39
EDN TOKWFI
|
|
M. Vinitsky, E. Dustalev, D. Minin, V. Babich, V. Bobrovsky
|
|
Abstract: Problem statement. Machine learning methods and neural networks are a promising tool for forecasting and identifying objects in real time, which allows for the application of such technologies in ensuring road traffic safety. The aim of the work is to develop a solution capable of detecting and classifying objects using artificial intelligence methods, implementing the functions of an intelligent driver assistance system onboard a vehicle. Methods used: creating an intelligent assistance service based on convolutional neural networks. An element of novelty in the presented solution is the implementation of a decision support service for the driver based on a compact low-power computing platform. Result. The selected service of the intelligent decision support system for drivers is implemented on a compact low-power computing device with an accuracy of 87 % based on the mean average precision (mAP) at an average frame rate of 32 frames per second. Practical significance. The presented solution allows for the implementation of a system using artificial intelligence algorithms on a vehicle base due to low energy consumption and a neuroprocessor module capable of working with video streams in real-time.
Keywords: Advanced Driver’s Assistance System, convolutional neural networks, artificial intelligence, machine learning, intelligent transportation systems, object detection.
DOI 10.31854/2307-1303-2025-13-1-40-46
EDN ZZGRGL
|
|
D. Fazylov, E. Kravets
|
|
Abstract: Objective: IoT-based street lighting control is motivated by society's desire to build energetically efficient systems that underlie the concept of a smart city. The basic functions of automated lighting are: street and/or outdoor lighting control; individual and group dimming; adjusting light to specific weather conditions, the presence of pedestrians and traffic. Fulfilling the task requires access to each individual light source, which can be achieved with the help of IoT wireless technology. This work aims to provide calculations of the parameters of a system of remote street lighting control. Methods used: The Okamura – Hata radio wave propagation model is used to determine the number of base stations required for remote control of luminaires. This model allows determining the range of a base station in areas with typical urban development. Novelty: a comparison of different techniques applied in low-power large-coverage energy networks designed for lighting control. It has been shown that network capacity can increase using seven uplink channels and one downlink channel at a fixed frequency. The attained package delivery success rate is 99 %. The result presented is a RadioPlanner design of a LoRa-based initial approximation network.
Keywords: street lighting control, IoT wireless technology, LoRa, smart city.
DOI 10.31854/2307-1303-2025-13-1-47-58
EDN TQUHQG
|
© SPbSUT © Authors
Авторизация
