Research On The QoS Model Of DiffServ Based On Fairness

Integrated services(IntServ),differentiated services(DiffServ) and Multiprotocol Label Switching(MPLS) are adopted to improve the QoS of IP networks. Differentiated Services is widely recognized since its scalability. Network congestion is an important issue of network QoS. With the development of multimedia application, Fairness among different flows emerges. Consequently, network congestion can not be solved only depending on TCP congestion control which is implemented in source node. Router should actively play an important role in congestion control, so congestion control based on the middle node gradually becomes the main direction in current research. For DiffServ, the fairness means that every micro-flow in the same aggregated flow can fairly share the resources. Queue management is a component in the congestion control mechanism. The distributed active queue management has a similar architecture to DiffServ and can assure fair bandwidth allocation among flows. Integrating distributed active queue management with DiffServ model to solve the fairness problem is proposed.Fairness problem in the DiffServ model is mainly discussed in this paper. Firstly the concept of network congestion and the reasons is introduced; then the congestion control mechanisms based on source node and on middle nodes are briefly described. Four active queue management algorithms are presented and the architecture of DiffServ and its characteristics are discussed in detail. Finally, a modified model merging SCALE-WFS algorithm with DiffServ model is proposed to achieve fairness in DiffServ model. Simulation result and analysis show that this model can improve the fairness in DiffServ networks.