Lavriv O. Models and algorithms of service quality assurance considering traffic features for multiservice network

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0412U004209

Applicant for

Specialization

  • 05.12.02 - Телекомунікаційні системи та мережі

21-09-2012

Specialized Academic Board

Д35.052.10

Essay

The thesis is devoted to modeling and research of multiservice networks and their components considering network traffic features to ensure service quality parameters. In this paper special attention is paid to the planning of network resources and methods of traffic prioritizing. The problem of service quality parameters assuring is solved in the design stage and in the stage of network management. The paper shows all existing models of multiservice telecommunication networks are based on hierarchical structure and should consider the main priority of service quality. The arrival traffic model and service, and service-oriented resource planning models are designed. Planning and management task are solved agreed by dynamic prioritizing of network traffic classes. Time parameters of service quality dependence on the resources provided by a particular traffic are studied. Option architectures Softswitch and IMS are proposed with upgrading some their control features for the implementation of models and algorithms proposed in the thesis for the multiservice network deployment. The conducted experiment has shown the packet jitter decreases approximately on 9-17% using designed models and algorithms. This result shows the service quality increases for real-time data. In the case of fixed quality parameters the network recourses capacity for its providing decreases on 5-10%. To confirm the adequacy of the developed simulation models it was analyzed aggregate multiservice network traffic of the large organization. The use of dynamic prioritizing of traffic classes for multiservice node resources management is shown to provide service for three groups of users.

Files

Similar theses