Beshlei M. Synthesis and implementation of intent-based infocommunication networks for adaptive service provision

The thesis is devoted to solving the actual scientific problem of developing the methodology of analysis and synthesis of the complex heterogeneous infocommunication systems in order to develop a new intent-based software-defined network, which is constantly learning, adapting, automating and protecting against potential cyber-attacks based on changing user requirements regarding the quality of service and infrastructure deployment by using the new methods of resource allocation, traffic engineering, network analytics and machine learning algorithms. A conceptual model for the construction of a heterogeneous intent-based software-defined network has been proposed. This model, unlike the existing ones, provides an efficient allocation and redistribution of common resources, adapting to the changing requirements of business users regarding to the quality of service provision. A mathematical model for determining the subjective level of user satisfaction, depending on changes of the objective parameters of the quality of service of the network was developed. An energy-efficient flow routing model for intent-based networks has been developed. This model allowed to achieve a better balance of network channel resources load due to the rational choice of paths for heterogeneous traffic and to reduce to 3 times the average latency of real-time flows service. The model also allowed for a 53.56% reduction in network energy consumption under low load conditions. A new method for adaptive customer-oriented quality of service management for intent-based networks has been proposed. This method allowed the end customers of services to influence the functional configuration of the network, and with the use of machine learning techniques to respond to the degradation of the quality of service and prevent situations when the user is dissatisfied with the received quality of services for adaptive prediction of the network reconfiguration moment. The method for detecting anomalies in network traffic has been improved, which differs from the known ones by forming a set of informative features that characterize normal and abnormal behavior of the infocommunication system based on the Hurst parameter estimation with the possibility of self-learning. An intelligent system for network traffic monitoring and analysis was developed. This system made it possible to effectively manage traffic and service policies in real time, automatically create or optimize service offers, improve the quality of services and provide network protection based on the developed method of detecting anomalies. Methods of time-frequency resource allocation and load balancing in modern mobile networks for adaptive provision of IoT services have been developed. An adaptive intent-based method of resource allocation and formation of radio access level structure was developed. This method allowed a 25% more efficient use of available time-frequency resources and reduced the energy consumption of radio access level network by 8.7% to provide the ordered quality of service for users, as compared to traditional methods. The methodology of synthesis of intent-based networks for stage-by-stage transformation of modern infocommunication networks is proposed. Prototypes of intent-based networks based on the technology of software-defined networks and automation of the proposed management decisions are developed. The algorithm for selecting a wireless access network in a heterogeneous environment using Big Data has been developed. The developed algorithm takes into account and analyzes the assessments of the ordered quality of service perception and allows to improve the quality of service of high-priority services on demand. The prototypes of mobile and operator application for adaptive client-oriented service provision in a heterogeneous network are developed, which allows to obtain the ordered quality of service based on the feedback between the customer and the network operator.