Integrated Performance Evaluation Criteria for Network Traffic Control

Integrated Performance Evaluation Criteria for Network Traffic Control
复制标题

网络流量控制综合性能评价标准

DOI:
--
复制
发表时间:
2002
期刊:
IEICE Trans. on Communications, E85-B (11):
影响因子:
--
通讯作者:
Zhou W
Zhou W
中科院分区:
其他
文献类型:
--
作者:
Lin C;Jiang Y;Zhou W

文献摘要

相似文献

本文对网络流量控制模型进行了研究,提出了一套有效集成的性能评价准则,该准则综合了网络性能、用户和应用的服务质量要求以及系统公平性等多个目标。首先,为了评估网络流量控制方案的有效性,作者考虑了网络的两个主要指标:吞吐量和时延,并在原有网络能力的基础上进行了改进,将其扩展到多个竞争连接。其次,在相对区分服务方法中,提出了比例公平原则,并给出了相应的综合评价准则,该准则同时考虑了分组时延、丢失率等服务质量要求和系统公平性。比例公平原则将某些类别的性能度量与网络运营商选择的区分参数成比例地“隔开”。最后,定义了一套完整的网络流量控制性能评价标准,可用于实际的多目标性能评价。由于多目标性能研究的复杂性,到目前为止,还没有一个有效的综合评价网络流量控制策略的标准。本文做了有益的理论探索,其结果可应用于这类复杂系统的性能评价、选型和优化。
This paper studies the network traffic control model,and proposes a set of effectively integrated performance evaluation criteria,and the criteria integrates several objects,such as the network performance,user's and application's QoS requirement and system fairness. First,for evaluating the effectiveness of a network traffic control scheme, authors consider the two principal metrics of networking: throughput and delay,and improve on the original power of the network,and extend it to the multiple,competitive connections. Second,in the relative differentiated service approach, authors propose the proportional fairness principle,and the corresponding integrated evaluation criteria that consider QoS requirements,such as packet delay and loss rate,and system fairness simultaneously. The proportional fairness principle 'spaces' certain class performance metrics proportionally to the differentiation parameters that the network operator chooses. At last,a set of integrated performance evaluation criteria for network traffic control is defined; which could be used in practice to evaluate multiple-object performance. Because the complexity of the research on multiple-object performance,up to now,there is still no effective integrated evaluation criteria for network traffic control strategy. This paper made a useful theoretical pilot study,and the result can be applied to performance evaluation,selection and optimization of this kind of various complex systems.