On-demand network bandwidth reservation combining machine learning and linear programming
On-demand network bandwidth reservation combining machine learning and linear programming
复制标题
DOI：
10.23919/cnsm52442.2021.9615572
复制
发表时间：
2021-10
期刊：
影响因子：
--
通讯作者：
K. Genda
中科院分区：
文献类型：
--
作者：
K. Genda
网络带宽预留是一种代表性的服务，其中用户直接按需预留网络资源。它利用了软件定义网络的优势，如灵活性。为了广泛地提供带宽预留服务，对用户请求的即时响应（例如，小于1秒）和高用户请求接受率（例如，90%以上）。在这项研究中，我们提出了一种带宽预留方法，以满足这两个要求，结合机器学习（ML）和线性规划（LP），特别是不可预测的带宽需求，其中使用时间是严格指示。在所提出的方法中，用户请求通过ML即时判断，网络资源分配，包括流量路由，通过LP最优确定。我们证明，该方法提供了一个次优的接受率与最佳的解决方案和小于0.1毫秒的瞬时响应相比，在一般的计算环境中的差异小于1%。
Network bandwidth reservation is a representative service, where users directly reserve network resources on an on-demand basis. It utilizes the advantages of software defined networks, such as flexibility. To provide bandwidth reservation services extensively, an instantaneous response to user requests (e.g., less than 1 s) and a high user request acceptance ratio (e.g., over 90%) are required. In this study, we propose a bandwidth reservation method to meet these two requirements by combining machine learning (ML) and linear programming (LP), particularly for unpredictable bandwidth demands in which the usage time is strictly indicated. In the proposed method, a user request is instantaneously judged through ML, and network resource allocation, including traffic routing, is optimally determined through LP. We demonstrate that the proposed method provides a suboptimal acceptance ratio with a difference of less than 1% compared with the optimal solution and an instantaneous response of less than 0.1 ms under a general computation environment.