Research On Traffic Control And Dynamic Routing Algorithm Of Multi-service IP Network

In recent years, with the development of Internet, Internet has become multi-service bearer network from a single data transmission network. Those data transmission services which are mainly E-mail and files transmission can not meet the needs of users, and real-time applications which include VOD, VOIP, and distance learning are emerging quickly. The most important features of real-time applications are sensitive to delay and jitter. However, traditional IP network adopts best-effort packet forwarding manner, which affects the network performance indicators such as delay, jitter and packet loss rate lack. Therefore, how to provide users with differentiated end-to-end QoS guarantees becomes a hot topic in the industry.This paper first introduces those existing QoS models in IP network, analyses essential principles and main features, we can find that each model can not provide good QoS guarantees alone. In order to find a solution, we give an improved hybrid end-to-end QoS model and implementation method through combination of IntServ, DiffServ, and MPLS in practice, and complete the service mapping. This model not only reserves advantages of each model, but also provides reliable end-to-end QoS guarantees to users and has good scalability and robustness.Then we discuss traffic control model of edge network and analyze the service curve of real-time traffic based on network calculus, and give the maximum end-to-end delay expression. Through using improved service curve, we can reserve fair network resource for more users. By comparing E-LSP and L-LSP, we use extended E-LSP to solve the problem of expansibility and localization by modifying the definition of label switching. Finally, a new QoS dynamic routing algorithm named CPRA is proposed, we analyze several dynamic routing algorithms through MATLAB, and verify the efficiency of CPRA in load balancing.