Network Congestion Control Technology

Network congestion has become an important issue with the rapid development of the Internet. Further, AQM(Active Queue Management) is a hotspot in the current。 TCP/IP protocol can't ensure network transmit QOS (Quality of Service), but provide "better effort" service. Network Congestion is exist consequentially because the quantity of data been required to transmit is beyond the bottleneck bandwidth at present, so it is very important that congestion control mechanisms are used. This paper introduces the study of Active queue management (AQM) and the important parameter RTT(Round Trip time) in network congestion control.Firstly, this paper summarize the study status in quo and common algorithm of network congestion control in detail, analyzes its study trends in detail. Analyze many result may have been brought if unfit queue management algorithm used in management of queue in middle node, such as global synchronization, keep queue full long time and lock-out. This paper implemented an Active queue management- P~2I control algorithm, Simulation results are given to prove that the new AQM algorithm can control the queue effectively.Secondly, the paper introduces Socket program technique and some important functions; we realize the test of RTT by Socket program, and present its test results. RTT is the important parameter of network congestion control, it can reflect the network congestion earlier compare to the drop probability. Forecast it can knowcongestion earlier.Lastly, Make use of the neural network method to establish intelligent forecast model to estimate RTT. The experiment shows that make use of neural network to forecast RTT in the short time can gain relatively ideal result.