当前位置: X-MOL 学术Opt. Switch. Netw. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Bounds for two static optimization problems on routing and spectrum allocation of anycasting
Optical Switching and Networking ( IF 1.9 ) Pub Date : 2018-10-24 , DOI: 10.1016/j.osn.2018.10.008
Yasutaka Miyagawa , Yosuke Watanabe , Maiko Shigeno , Kiyo Ishii , Atsuko Takefusa , Akiko Yoshise

Elastic optical networks with optical-orthogonal frequency division multiplexing have been addressed enthusiastically for communication networks in the last decade because they result in high bandwidth efficiency. Routing and spectrum allocation (RSA) problems need to be solved when we transmit demands in an elastic optical network. This research deals with static RSA models for anycast transmission, which is one-to-one-of-many transmission in inter-datacenter networks. Two static RSA optimization models are considered. One minimizes the maximum number of spectrum slots needed to allocate given demands. The other maximizes the traffic volume of demands served under a given spectrum slot number. For both models, lower and upper bounds are developed in order to obtain exact optimal solutions. One-side bounds of the problems are evaluated by relaxing spectrum continuity constraints. For the other side bounds, several greedy algorithms are investigated. We conducted computational experiments to confirm whether relaxation problems can give tight bounds and to determine greedy algorithmic behaviors by using each of route selection criterion and each of demand ordering policies. The results show that the solutions obtained by relaxing spectrum continuity constraints are almost optimal. They also indicate that exact optimal solutions are obtained efficiently by using these bounds.



中文翻译:

关于任播路由和频谱分配的两个静态优化问题的界限

在过去的十年中,具有光正交频分复用的弹性光网络已经引起了通信网络的热烈关注,因为它们带来了高带宽效率。在弹性光网络中传输需求时,需要解决路由和频谱分配(RSA)问题。这项研究涉及用于任意播传输的静态RSA模型,该模型在数据中心间网络中是一对多的传输。考虑了两个静态RSA优化模型。一个最小化分配给定需求所需的频谱时隙的最大数量。另一个最大化在给定频谱时隙号下服务的需求的业务量。对于这两个模型,都将开发上下限,以获取精确的最佳解。通过放宽频谱连续性约束来评估问题的单边范围。对于另一个边界,研究了几种贪婪算法。我们进行了计算实验,以确认松弛问题是否可以给出严格的界限,并通过使用每种路线选择标准和每种需求排序策略来确定贪婪的算法行为。结果表明,通过放松频谱连续性约束获得的解几乎是最优的。他们还表明,使用这些界限可以有效地获得精确的最优解。我们进行了计算实验,以确认松弛问题是否可以给出严格的界限,并通过使用每种路线选择标准和每种需求排序策略来确定贪婪的算法行为。结果表明,通过放松频谱连续性约束获得的解几乎是最优的。他们还表明,使用这些界限可以有效地获得精确的最优解。我们进行了计算实验,以确认松弛问题是否可以给出严格的界限,并通过使用每种路线选择标准和每种需求排序策略来确定贪婪的算法行为。结果表明,通过放松频谱连续性约束获得的解几乎是最优的。他们还表明,使用这些界限可以有效地获得精确的最优解。

更新日期:2018-10-24
down
wechat
bug