Research On IP QoS Issue Based On DiffServ-Aware MPLS Networks

With the blooming of Internet and the increasing of multi-media application and real-time application in networks, the requirement of network bandwidth, quality of service, scalability and adaptability to new services from the application layer became stricter. The IP QoS mechanism now using cannot meet the need, thus it is necessary to make further research on the mechanism of QoS realization. Based on DiffServ-aware MPLS networks, this thesis made a deep research on the related techniques which can solve quality of service problems.This thesis analyzed several major mechanisms which could realize QoS guarantee, including main principles and advantages of MPLS, characteristics and applications of IntServ model, DiffServ model and kinds of related techniques, then pointed out their advantages and disadvantages separately. Through research, we got to know that only a single kind of QoS mechanism cannot solve problems of quality of service. The combination of DiffServ and MPLS mechanism makes use of the explicit route ability of MPLS and the scalability of DiffServ, and it has preponderance in implement and scalability. It is the effective technique to solve QoS problems of backbone networks. The networks that use this mechanism can make optimal use of the resource according to user's requirement. Based on the principles of MPLS and DiffServ, a backbone networks' solution, which support DiffServ in MPLS networks, has been proposed in this thesis, and put out the mapping relation of DSCP and EXP field of MPLS label. The performance of the solution was tested by using network simulator and the result verified its feasibility and correctness.Besides this, a new LSP policy for resource preemption was proposed based on the research of DS-TE while the resources were limited. The policy's basic idea is sorting three criterions according to their importance and these three criterions are avoiding cascading effect, decreasing the numbers of affected traffic and avoiding waste of bandwidth. The policy can optimize preemption objectives in turn. Especially for avoiding waste of bandwidth, the policy used a kind of principle called hierarchical approach, which can enhance restriction degree for bandwidth preemption. Finally the experiment result shows that the new policy can implement LSP preemption according to network status and the requirement of applications,...