The dissertation solves the scientific and practical problem of increasing the efficiency of cloud systems when they are integrated into modern information and communication service-oriented networks by developing a method for clustering server nodes, improving models for assessing the efficiency of cloud systems in terms of servicing group request flows and developing systems for intelligent detection of network attacks. The first chapter, “Approaches and features of integration of cloud architectures into information and communication networks,” considers the basic principles of construction, approaches and features of integration of cloud systems and information and communication networks of telecommunications providers. The main requirements for the quality of service provision in service-oriented telecommunication networks are analyzed. It is established that despite all the advantages that enterprises and providers receive when integrating cloud systems, their efficiency is deteriorating due to the dynamic nature of the cloud environment and the variety of user requests.In the second section “Models and methods for improving the efficiency of cloud systems as components of service-oriented networks”, a mathematical model for assessing the efficiency of cloud systems in the conditions of group requests from users is improved. The model is based on a two-stage approximation technique and provides a complete probabilistic distribution of request waiting time, response time for request fulfillment, and the number of requests in the system. The improved model takes into account the possibility of servicing requests in groups and distributed request service time, which will allow not to disrupt the performance of a network with a high degree of virtualization at peak times by reducing the time requests are waiting in the queue. The model allows you to find a balance between the number of virtual machines that service group request flows and the number of requests already being served in the system. The method of clustering cloud system nodes in terms of servicing group request flows has been improved, which will increase the efficiency of cloud systems in which a large number of virtual machines simultaneously serve group request flows. This approach is based on grouping into a cluster a pool of virtual machines that will be used to service a clearly defined group flow, taking into account the geographical proximity of nodes to each other and monitoring the available software, hardware, and telecommunications resources, which will increase the life cycle of physical machines by increasing the level of residual energy. An algorithm for clustering and selecting the cluster head is developed, the feature of which is the determination of the cluster head taking into account the centrality of Voronoi diagrams, fuzzy logic rules, and monitoring of node parameters. The decision on the choice of the main node is made using fuzzy logic methods using the Mamdani rule and the center of gravity, as well as taking into account sufficient software and hardware resources for the further functioning of the node as the cluster centroid. In the third chapter, “Modeling and studying the efficiency of cloud systems in information and communication service-oriented networks,” we model and study the efficiency of the proposed clustering algorithm based on a comprehensive efficiency criterion: route search duration, network energy consumption, and network life cycle. Thanks to the implementation of the developed method of clustering cloud system nodes, it was possible to reduce the duration of route search between an arbitrary pair of virtualized nodes within one cluster by 1.3 times. Due to the determination of the main node of the cluster, taking into account the centrality of Voronoi diagrams, fuzzy logic rules and monitoring of node parameters, it was possible to reduce the duration of the route search between any pair of virtualized nodes serving the same group stream by 2.5 times. It is proved that the developed algorithm reduces energy consumption by 45% and extends the network life cycle by 8% compared to the known algorithms. This method allows to maintain the quality of service to users, especially in the conditions of group requests. In the fourth chapter, “Implementation of network protection of cloud web services in software-configurable service-oriented networks,” an integrated architecture of a system for intelligent detection of DDoS attacks in information and communication networks is developed. The main feature of the architecture is the implementation of a log file analysis subsystem, which is an element of an integrated management system that analyzes potential problems and triggers preventive mechanisms, or can notify system administrators of network security breaches.
