This thesis aimed to create a new fuzzy-CAC design method, as well to evaluate the possible application of built fuzzy-CAC solution in the MPLS-TE (GMPLS) traffic transmission/management system, modifying the RSVP-TE protocol, thus providing a dynamic and selective LSP setup. Based on the defined fuzzy-CAC fuzzy inference system (FIS) model and the MPLS-TE network model, as well as several major international telecommunications service provider SLAs, by using the heuristic research method, FIS is found, which ensures the desired fuzzy-CAC performance. For FIS heuristic selection MPLS-TE model simulation tool is used. Within the framework of this thesis, MPLS-TE experimental network was made on the basis of a Cisco 2800 series routers, as well as the router interface and control modules, fuzzy-CAC control module for RSVP-TE protocol management, traffic generation and analysis modules. As the result, fuzzy-CAC effectiveness is proved, if compared to the classic threshold CAC algorithm that is currently being used in RSVP-TE for MPLS-TE network management. The doctoral thesis is based on the experimental results of fuzzy-CAC application to the RSVP-TE protocol, while managing traffic in MPLS-TE network. The amount of work is 174 pp., 90 figures, 20 tables, 2 appendices and 148 sources of literature.