基于统计网络演算的端到端时延上界研究

为分析融合网络端到端统计时延性能,根据最小加代数理论中的统计网络演算相关知识,提出了一种改进的统计服务曲线并应用GPS调度系统模型,然后利用网络演算理论求解串型网络节点统计时延上界。数值分析结果表明,改进的调度模型有效改善了自相关特性对时延上界所造成的影响,比未改进的模型相比统计时延上界更具有良好的紧致性,为在融合网络环境中对端到端节点延迟上界研究提供有效控制、调度和管理具有一定的参考价值。

Research on end-to-end delay performance based on statistical network

In order to analyze the statistical delay performance for end-to-end converged network,according to the know-ledge of statistical network calculus in minimum plus algebra theory,a scheduling system based on improved model of GPS is proposed,and the network calculus is used to solve the upper bound of statistical delay.Numerical results show that the improved scheduling model improves the impact from correlation properties on the delay.It has a certain reference value for end-to-end delay upper bound of the node to provide effective control,scheduling and management in the integration network environment for the next step.