NFV 中的 vDPI 功能放置问题研究

网络功能虚拟化 (NFV) 是视频流应用的重要技术。在视频流使用场景下,已有的工作研究 NFV 网络中的虚拟深度包检测 (vDPI) 放置问题时,仅考虑减少 vDPI 放置的数量,但没有考虑放置 vDPI 功能带来的 NFV 网络性能稳定性问题。针对上述不足,本文在减小 vDPI 放置数量的基础上,考虑提高网络稳定性,提出了一个多目标线性整数规划模型,并设计了一种贪心近似放置算法。该算法在 NFV 中放置 vDPI 功能时,可以降低 vDPI 放置数量、减少流量经过的平均网络跳数、降低传输时延、保证网络性能的稳定性。本文采用 Lingo 求出数学模型的最优解,并用贪心近似放置算法进行实验,对比算法实验结果与模型所求最优解可知：所提算法正确性较高,有较好的时间复杂度,适用于不同规模的 NFV 网络。

A Research About Placement of vDPI Functions in NFV Infrastructures

Network Function Virtualization (NFV) is an important technology for video streaming applications. In the sense of using video stream, previous related work only considered placing the virtualized deep packet inspection function under the network function virtualization infrastructures to reduce the expenditure of vDPI. However, performance of the network (such as network hops) didn’t be taken into account. Focused on the issue that realizes a trade-off between these possibly conflicting goals (expenditure of vDPI and network hops), a Multi-objective Integer Linear Programming model and a greedy approximate optimal placement algorithm were devised to solve the problem. The devised algorithm can find suitable position to place vDPI in NFV and realize a trade-off between conflicting goals (expenditure of vDPI and network hops). Compared the results with mathematical model, the algorithm has high correctness and better time complexity and can be used in different size of NFV network.