摘要
在研究现有网络数据传递拥塞控制问题的基础上,提出了一种基于路由器缓冲区队列长度的网络拥塞预测机制,并以该预测机制为基础提出了一种计费策略(PPBC,Per-Packet Billing Control)来控制网络拥塞。在拥塞控制模式中,将路由器缓冲区队列长度作为拥塞指标,通过马尔可夫链建模预测网络拥塞度。在服务器终端和数据发送方之间建立反馈机制,利用网络拥塞度,通过数据包计费方式控制网络拥塞。最后采用NS2仿真软件对PPBC性能进行仿真测评,仿真结果表明:PPBC不仅能够有效缓解网络拥塞,还能够使网络保持较低的丢包率。
Based on the study of data congestion control, a network congestion prediction mechanism based on router buff queue length is proposed, and furthermore, a pricing strategy （PPBC, Per-Packet Billing Control） for control of network congestion is suggested. This mechanism takes the router buffer queue length as the congestion indicator in congestion control mode. Moreover, server predicts the network congestion by Markov modeling in touters and thus acquires the data service cost. Thereupon, a feedback mechanism is constructed between server and data sender, and network congestion level is used to control congestion through pricing strategy. Finally the simulation on this mechanism with NS2 software indicates that the proposed mechanism could effectively relax the congestion and maintain fairly low network packet-loss rate.
出处
《通信技术》
2011年第12期69-71,共3页
Communications Technology
