Approach to problem of minimizing network power consumption based on robust optimization
Approach to problem of minimizing network power consumption based on robust optimization
复制标题
基于鲁棒优化的网络功耗最小化问题的解决方法
DOI：
10.1002/dac.3891
复制
发表时间：
2019
影响因子：
2.1
通讯作者：
Muramatsu Masakazu
中科院分区：
文献类型：
--
作者：
Das Bimal Chandra;Takahashi Satoshi;Oki Eiji;Muramatsu Masakazu
本文引入了一种绿色稳健的优化模型来最小化网络功耗，该模型允许网络中源-目的节点对之间的流量需求波动。我们的模型基于绿色软管模型，其中流量仅受每个节点的总传出和传入数量的限制。除了绿色软管模型外，我们还使用椭球不确定性集来允许交通需求的不同类型的波动。这里，交通需求中的平方误差的总量由一个常数限定，该常数控制网络上的总允许波动。应用二次锥对偶，我们将模型表示为混合整数二阶锥规划(MISOCP)问题。此外，我们建立了我们的模型与绿色软管模型之间的关系，该模型具有链接流量界限(HLT)模型，该模型是被称为HLT的软管模型的发展版本。数值结果表明，每一个MISOCP问题都可以用一个通用的MISOCP求解器在合理的时间内解到最优解，并且该模型与绿色软管和绿色HLT模型在性能上有不同的趋势。
In this paper, a green and robust optimization model is introduced to minimize network power consumption, which allows fluctuations of traffic demands between source‐destination pairs in the network. Our model is based on the green hose model, where the traffic is bounded by just total outgoing and incoming amount at each node. In addition to the green hose model, we use the ellipsoidal uncertainty set to allow a different type of fluctuations in traffic demands. Here, the total amount of squared errors in traffic demands is bounded by a constant which controls the total admissible fluctuations over the network. Applying the conic duality, we formulate our model in the form of mixed‐integer second order cone programming (MISOCP) problem. Furthermore, we establish a relationship between our model and the green hose model with bound of link traffic (HLT) model, an developed version of the hose model known as the HLT. Numerical results demonstrate that each of the MISOCP problems can be solved to its optimality in a reasonable time by a general MISOCP solver, and that the proposed model has different tendency in performance with the green hose and green HLT models.