The Thesis is devoted to create a new connection admission control (CAC) method capable of providing a differentiated service strategies, which would be suitable for modern high-speed networks, serving thousands of new connections per second. The developed AggSessAC method is based on short-term delay (of µs-ms range) of the decision-making process, in order to collect multiple connection initialization packets, thereby obtaining a new connection request set for further evaluation. Such access control method works as opposed to the traditional network resource allocation approach. A new connection request aggregation allows to bring the virtualization of parallel processing into connection admission management. In the Thesis a broad contemporary Internet traffic analysis is given, which is based on the thesis author and other researchers collected measurement results. A comprehensive analysis of Internet data flows from different perspectives – packets, flow and protocol – has allowed justifying the topicality of research and served as the basis for the development of new approach for admission management. For the verification and validation of the developed AggSessAC admission management solution, OMNeT++ discrete event simulation tool was used. During the development of this thesis, OMNeT++ environment INET framework was supplemented with AggSessAC functional unit (agent) which is able to collect a new connection requests and to realize priority-based admission management. The developed access control agent was added to INET framework available router software bringing the AggSessAC functionality, thereby allowing the solution experimental evaluation in OMNeT++ simulation environment. The obtained results show the efficiency of the AggSessAC method in comparison with the classic manner of the request processing, demonstrating the benefits of inter-evaluative and selective demand handling, which improves the management policy implementation. The research also includes estimation of AggSessAC method parameters and gives recommendations for their selection. The amount of work is 162 pp., 78 figures, 28 tables and 116 sources of literature.