Policy robustness in queueing networks
Policy robustness in queueing networks
复制标题
排队网络中的策略稳健性
DOI：
10.1007/s11134-022-09776-5
复制
发表时间：
2022
期刊：
影响因子：
1.2
通讯作者：
Hasenbein, John J.
中科院分区：
文献类型：
--
作者：
Gurvich, Itai;Hasenbein, John J.
几十年来，鲁棒性一直是优化文献中的一个重要主题。然而，只是在过去的几年中，强大的技术已经进口和进一步发展的随机模型，特别是随机处理网络。这种延迟采用的原因可能是一般随机网络的“非鲁棒”分析已经提出了重大的数学挑战。因此，在前80年的大部分时间里，在文献中出现的大多数模型中，都没有明确考虑模型的鲁棒性。论文解决的鲁棒性在嵌入式网络集中在两个主要的鲁棒性角度：参数鲁棒性和数据鲁棒性。在前一组中，到达和服务过程的统计结构是已知的（例如，它是一个泊松过程），但到达和服务率（参数）只知道位于某个不确定集内;[4]是该领域的早期论文之一。最近，[5]研究了一个M/M/s排队，其中只有到达率分布的均值和支持度为必须做出人员配置决策的经理所知。相反，数据鲁棒性不强加参数统计结构，并认为实现的过程（例如，到达时间间隔的序列）属于适当定义的不确定性集，并使用鲁棒优化来获得性能指标的近似值和界限;参见[1，3]。沿着这些方面的进一步的最新发展出现在[2，10]。
Robustness has been, for decades now, a prominent topic in the optimization literature. Yet, it is only in the past few years that robust techniques have been imported and further developed for stochastic models, particularly stochastic processing networks. A reason for this late adoption might be that “non-robust” analysis of general stochastic networks already presents significant mathematical challenges. Hence, for much of the first 80 years of the study of queueing models robustness was not explicitly considered in most models that appeared in the literature. Papers that do address robustness in queueing networks focus on two major robustness angles: parameter robustness, and data robustness. In the former group, the statistical structure for the arrival and service processes is known (eg, it is a Poisson process) but the arrival and service rates (the parameters) are only known to lie within some uncertainty set;[4] is one of the earlier papers in this space. More recently,[5] studies an M/M/s queue where only the mean and support of the arrival-rate distribution is known to the manager who must make staffing decisions. Data robustness, in contrast, does not impose a parametric statistical structure and takes the view that the realized process (eg, the sequence of inter-arrival times) belongs to a suitably defined uncertainty set and uses robust optimization to obtain approximations and bounds for performance metrics; see [1, 3]. Further recent development along these lines appears in [2, 10].
为具有分布不确定性的大规模服务系统配备人员
DOI：
--
发表时间：
2017
期刊：
Queueing Syst. Theory Appl.
影响因子：
--
作者：
Ying Chen;J. Hasenbein
通讯作者：
J. Hasenbein
通过鲁棒优化对排队网络进行性能分析
DOI：
10.1287/opre.1100.0879
发表时间：
2010
期刊：
Oper. Res.
影响因子：
--
作者：
D. Bertsimas;D. Gamarnik;Alexander Anatoliy Rikun
通讯作者：
Alexander Anatoliy Rikun