Diachenko V.O. Model and methods for improving distributed sensor networks based on modified Kohonen maps. – Qualification scientific work on the rights of the manuscript.
Dissertation for obtaining the Doctor of Philosophy degree by 123 Computer Engineering specialty – Kharkiv National University of Radio Electronics, Ministry of Education and Science of Ukraine, Kharkiv, 2025.
The dissertation is devoted to the development of a model and improvement of methods for the functioning of distributed sensor networks based on modified Kohonen maps. Within the work an approach was formed that combines the self-organization mechanisms of the Kohonen map with the energy parameters of the nodes, which made it possible to form clustering and routing structures that are aligned with the network topology and optimized for energy consumption. The proposed modifications cover both an improved learning mechanism for the Kohonen map and means of preliminary energy selection and routing list composition, thanks to which the model ensures rational load distribution, reduces the probability of energy depletion of individual nodes, and extends the effective operation of the sensor network.
The relevance of the chosen topic is determined by the important role of distributed sensor networks in the tasks of autonomous monitoring of various objects and processes in real time. One of the main prerequisites for the functioning of distributed sensor networks is the rational use of energy resources of network nodes that operate using autonomous power sources. Energy resource limitations have a significant impact not only on the stability of communication and network bandwidth, but also on the duration of the network functioning. In this regard the question arises of improving energy efficiency when developing routing algorithms and topology of distributed sensor networks. In traditional approaches to routing and clustering, the current energy status of nodes, their functional role in the network topology, and their position relative to the base station are not considered sufficiently. This, of course, leads to a decrease in network reliability. It should also be noted that with ever-increasing amounts of data, it is necessary to develop new or improve existing models and methods that will significantly reduce energy consumption in monitoring and control systems. This can be done using machine learning tools, in particular, self-organizing Kohonen maps, which are capable of adaptively reflecting the structure of input data and identifying hidden patterns. The approaches themselves, which combine Kohonen self-organizing maps with energy models of nodes, open new opportunities for creating more reliable and durable sensor networks. That’s why the research regarding effective methods of reducing the energy consumption of distributed sensor network nodes using adaptive self-organization mechanisms makes this work relevant.
The aim of the dissertation is to improve the efficiency of distributed sensor networks and reduce the energy consumption of their nodes by developing models and methods of intelligent data processing based on a modified Kohonen map in limited resource conditions.
The object of the study is data processing using machine learning tools in distributed sensor networks under conditions of limited energy resources.
The subject of the study is models and methods of data processing by nodes of distributed sensor networks, which are aimed at the accuracy of input information analysis, stability of functioning, and reduction of energy consumption.
.
