摘要
介绍了一种基于流量预测的上行带宽动态分配算法(P-DBA)。P-DBA算法根据短相关业务(SRD)和自相似、长相关业务(LRD)的流量特征建立了不同的线性预测模型,并在流量变化较快时放弃预测,以减小带宽浪费;同时,在光网络单元中提出一种配合预测机制的公平调度策略。仿真表明P-DBA算法比DBAM算法在端到端延时、丢包率等方面有明显改善。
P-DBA based on traffic prediction was presented, which established different linear prediction models for the Short-Range Dependence (SRD) and Song-Range Dependence (LRD) traffic and drop prediction to reduce the bandwidth waste when the variation of traffic was rapid, while Optical Network Unit (ONU) used a new local fair scheduling policy adapted to the prediction strategy. The simulation result shows that end-to-end delay, and packet loss rate has been improved obviously, compared with DBAM algorithm.
出处
《计算机应用》
CSCD
北大核心
2010年第1期124-126,共3页
journal of Computer Applications
关键词
以太无源光网络
动态带宽分配
流量预测
服务质量
公平性
Ethernet Passive Optical Network (EPON)
Dynamic Bandwidth Allocation (DBA)
traffic prediction
Quality of Service (QoS)
fairness