Ip Network Congestion Control And Aqm Is

During the recent few years, there has been widely spread research on networkcongestion control. TCP protocol adopts windows based control strategy whichplays a critical role for the robustness of Internet. However, as the networkcontinues to expand in size, diversity, and application domain, it becomes morecrucial to understand how these fundmental resources are controlled. Trational TCPwindow control can not provide us the required quality of service. FIFO queue hasinherent limitation, which will lead to low QoS, such as high round trip delay andjitter. Given the scale and complexity of network, and the heuristic intricate natureof network protocols, only till recently can we apply analytical modeling andfeedback control theory to this area. Key to these advances has been the explicitlymodeling the congestion measure that communicates back to data sources theinformation on congestion in network. Two types of studies are of fundamental interest. On the one hand, it isimportant to characterize the equilibrium condition under certain congestion controlprotocols from the point of view of fairness, throughput, and robustness, etc. Asecond line of inquiry concerns that dynamics of congestion control protocolsdirectly in the domain of control theory. In particular, we are interested in thestability of the postulated equilibiria, especially in the presence of feedback delay.Recently analysis has shown that predominate TCP implantation are prone toinstability when combined with network delay. The two most popular strategies arenew transport control protocols and active queue management. In this thesis, we first describe congestion control protocols and relatednetwork performance from the perspective of network architecture. Throughmathematics model, we analyze interoperation of network protocols and the impactof procols' parameter on network performance, such as throughput, delay, fairnessand stability etc. We concentrate our research on AQM and TCP protocols fromcontrol theory and simulate network behavior. Both theory analysis and simulationdemonstrate improvement of network performance and better QoS under support ofintermedian nodes.