Illia Fedorovych
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Illia Fedorovych has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Andrii Voloshchuk
Ternopil Ivan Puluj National Technical University
1 shared publication - Halyna Osukhivska
Ternopil Ivan Puluj National Technical University
1 shared publication - Ivan Borodii
Ternopil Ivan Puluj National Technical University
1 shared publication - Maksym Drogobytskyi
Ternopil Ivan Puluj National Technical University
1 shared publication Г.м. Осухівська
1 shared publication
Latest Publications
- Open Access
- 30 June 2026
- Published by Yuriy Fedkovych Chernivtsi National University in Security of Infocommunication Systems and Internet of Things
- Vol. 4 (1) , 01002
Methods for Selecting Data Transmission Network Protocols in Energy Systems
This paper presents a comprehensive study aimed at enhancing data transmission stability and efficiency in modern energy systems, specifically Smart Grids. The research addresses the critical challenge of optimizing communication protocols within the context of evolving infrastructure and the increasing integration of Distributed Energy Resources. The primary focus of this work is the development of an intelligent communication control system featuring a decision-making mechanism that, based on the analysis of a set of input parameters including network latency, packet loss, and throughput, determines the feasibility of switching between data transmission protocols such as MQTT, CoAP, and HTTPS in real-time. The study employs an ensemble of machine learning classifiers to analyze real-time network telemetry to develop a mechanism for adaptive protocol switching. The proposed methodology integrates the analysis of network traffic dynamics with real-time performance metrics to facilitate dynamic protocol selection. An approach is proposed wherein an ensemble of machine learning models processes a vector of input metrics and classifies the current network state, initiating a protocol switch only when critical threshold values are reached. Experimental validation demonstrates that the adaptive protocol switching system provides improved network traffic scheduling efficiency compared to static implementations. The results indicate significant potential for reducing communication overhead while maintaining high standards of reliability and security in energy system operations. Specifically, the approach prevents data transmission over degraded channels by automatically switching to lightweight protocols like CoAP when connection parameters deteriorate, thereby ensuring compliance with timing requirements. The developed methodology establishes a foundation for the implementation of intelligent communication systems in Smart Grids, contributing to enhanced energy efficiency and grid stability. The findings hold practical value for energy utilities seeking to modernize their infrastructure and optimize data transmission within increasingly complex energy networks.
- 30 June 2026
- Published by Yuriy Fedkovych Chernivtsi National University in Security of Infocommunication Systems and Internet of Things
- Vol. 4 (1) , 01002
Open AccessMethods for Selecting Data Transmission Network Protocols in Energy Systems
This paper presents a comprehensive study aimed at enhancing data transmission stability and efficiency in modern energy systems, specifically Smart Grids. The research addresses the critical challenge of optimizing communication protocols within the context of evolving infrastructure and the increasing integration of Distributed Energy Resources. The primary focus of this work is the development of an intelligent communication control system featuring a decision-making mechanism that, based on the analysis of a set of input parameters including network latency, packet loss, and throughput, determines the feasibility of switching between data transmission protocols such as MQTT, CoAP, and HTTPS in real-time. The study employs an ensemble of machine learning classifiers to analyze real-time network telemetry to develop a mechanism for adaptive protocol switching. The proposed methodology integrates the analysis of network traffic dynamics with real-time performance metrics to facilitate dynamic protocol selection. An approach is proposed wherein an ensemble of machine learning models processes a vector of input metrics and classifies the current network state, initiating a protocol switch only when critical threshold values are reached. Experimental validation demonstrates that the adaptive protocol switching system provides improved network traffic scheduling efficiency compared to static implementations. The results indicate significant potential for reducing communication overhead while maintaining high standards of reliability and security in energy system operations. Specifically, the approach prevents data transmission over degraded channels by automatically switching to lightweight protocols like CoAP when connection parameters deteriorate, thereby ensuring compliance with timing requirements. The developed methodology establishes a foundation for the implementation of intelligent communication systems in Smart Grids, contributing to enhanced energy efficiency and grid stability. The findings hold practical value for energy utilities seeking to modernize their infrastructure and optimize data transmission within increasingly complex energy networks.
- 28 March 2026
- Published by Lutsk National Technical University in COMPUTER-INTEGRATED TECHNOLOGIES: EDUCATION, SCIENCE, PRODUCTION
- No. 62,p. 89-98
GPU-accelerated lemmatization with CUDA and RAPIDS accelerator for Spark.
Lemmatization is one of the few preprocessing stages in NLP pipelines that is still predominantly CPU-bound, even as tokenization, vectorization, and model inference increasingly rely on high-throughput GPU execution. Deterministic, dictionary-based morphological analyzers deliver stable and linguistically correct results, but their sequential nature, complex branch divergence, and interoperability overhead significantly limit processing throughput in large-scale distributed systems. In this work, we propose a GPU-native deterministic lemmatizer for Ukrainian, implemented as a custom CUDA kernel and integrated into Apache Spark via a RAPIDS-accelerated JNI UDF. A large morphological dictionary (VESUM/dict_uk) is compiled into a compact trie layout optimized for coalesced memory traversal and fully parallel token processing, preserving accuracy while eliminating CPU constraints and reducing end-to-end latency. We evaluate three Spark pipelines — Python with pymorphy3, Java with Morfologik, and our CUDA implementation — on corpora up to 175.6M tokens. The GPU approach achieves 35.77M tokens/s, outperforming the optimized Java baseline by 8.8× and the Python baseline by up to 120×. Our results demonstrate that classical morphological analysis — historically treated as inherently sequential — can be re-architected into an efficient, scalable, and bandwidth-bound GPU operation suitable for real-world big-data workflows.
- 28 March 2026
- Published by Lutsk National Technical University in COMPUTER-INTEGRATED TECHNOLOGIES: EDUCATION, SCIENCE, PRODUCTION
- No. 62,p. 89-98
GPU-accelerated lemmatization with CUDA and RAPIDS accelerator for Spark.
Lemmatization is one of the few preprocessing stages in NLP pipelines that is still predominantly CPU-bound, even as tokenization, vectorization, and model inference increasingly rely on high-throughput GPU execution. Deterministic, dictionary-based morphological analyzers deliver stable and linguistically correct results, but their sequential nature, complex branch divergence, and interoperability overhead significantly limit processing throughput in large-scale distributed systems. In this work, we propose a GPU-native deterministic lemmatizer for Ukrainian, implemented as a custom CUDA kernel and integrated into Apache Spark via a RAPIDS-accelerated JNI UDF. A large morphological dictionary (VESUM/dict_uk) is compiled into a compact trie layout optimized for coalesced memory traversal and fully parallel token processing, preserving accuracy while eliminating CPU constraints and reducing end-to-end latency. We evaluate three Spark pipelines — Python with pymorphy3, Java with Morfologik, and our CUDA implementation — on corpora up to 175.6M tokens. The GPU approach achieves 35.77M tokens/s, outperforming the optimized Java baseline by 8.8× and the Python baseline by up to 120×. Our results demonstrate that classical morphological analysis — historically treated as inherently sequential — can be re-architected into an efficient, scalable, and bandwidth-bound GPU operation suitable for real-world big-data workflows.