The Research And Application Of Internet Congestion Control Based On Petri Net

When there are too many data packets transferring in the network, the performance of the network declines. This is called congestion. With the explosive growth of the users and multimedia application, the number of data flows in the network increases rapidly, which leads to severe congestion problems. The existing solutions cannot satisfy the demand of the current and future development of Internet. The congestion of network became one of the important problems that the Internet is faced with.According to the operation place, there are two kinds of congestion control strategies. One is the TCP congestion control based on the end system, the other is the IP congestion control strategy based on the middle nodes of network. This task analyses the current situation of the two strategies. They have two problems of unfairness and efficiency. And then three tasks are accomplished. The first one is in order to settle the unfairness of different RTT values in the competition of bandwidth in the TCP congestion control strategy, a Petri net model for the behaviors of the Slow-Start algorithm is built. By the analysis of the model, a modified Slow-Start algorithm is presented and implemented in the NS2 network simulator. The simulation results indicate that the modified Slow-Start algorithm can reduce the aggression of the TCP flow which has the smaller RTT in the competition of bandwidth and slow the increasing rate of the congestion window in the later phase when the network delay is large. The second one is analysis the current main several active queue management algorithms in IP congestion control strategies, and contrast the performance difference by NS2. The third one is a Petri net model for the behaviors of RED algorithm are built. By the analysis of the model, the shortcoming of the RED is found. Then a modified Petri net model of RED algorithm is built. After analysis the modified model, an modified RED algorithm is presented. It is implemented in the NS2. The simulation results indicate that the updated RED algorithm can reduce the packet loss ratio and the delay of End-to-End time. This means that the performance of the network is raised and the problem of efficiency is settled to a certain degree.