請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/7729
標題: | 缺陷性霧運算網路之虛擬網路嵌入 Virtual Network Embedding in Faulty Fog Networks |
作者: | Kuo-Liang Chang Chien 張簡國良 |
指導教授: | 周俊廷(Chun-Ting Chou) |
關鍵字: | 霧運算,虛擬網路嵌入,負載平衡, fog computing,virtual network embedding,load balance, |
出版年 : | 2019 |
學位: | 碩士 |
摘要: | 霧運算將雲端運算與服務延伸至網路的邊緣,允許應用程式在較靠近使用者的地區運行,具有高度的地理分散性並支援使用者的移動性。作為霧運算可實踐的技術之一,網路功能虛擬化 (Network Function Virtualization, NFV) 允許應用程式被設計成虛擬網路,而異質性的虛擬網路則被視為是要求,被稱為虛擬網路要求 (Virtual Network Request, VNR) 並嵌入在設備提供者的實體網路中,這個將虛擬網路要求嵌入在實體網路的過程就被指為是虛擬網路嵌入 (Virtual Network Embedding, VNE) 問題。
在霧運算網路中的伺服器可能因為沉重的工作負載量和過熱問題而導致當機,而在其上的服務也隨之而中斷,從一個霧運算網路設備提供者的角度而言,由服務中斷而產生的懲罰會傷害其淨利。為了避免伺服器當機,設備提供者可以運用採取分散式佈建作法的虛擬網路嵌入演算法來保持負載平衡,然而分散式布建的方法會導致額外的能量消耗,而這同樣會減少設備提供者的淨利。在這篇論文裡,我們目標是在考慮懲罰及能量消耗的情況下最大化霧運算網路設備提供者的淨利,我們提出了搭配新穎點排名的貪婪演算法當作一個基準,而這個方法比現存虛擬網路嵌入的貪婪演算法在淨利上高了約9%,然後我們再運用meta-heuristic的方法發展成一個加強版的解法來進一步改善淨利,而加強版的解法比基準解法在淨利上高了約9%。 Fog computing extends Cloud computing and services to the edge of the network. It allows applications to run in close proximity of users, so it is highly geo-distributed and supports user mobility. As an enabler of fog computing, network function virtualization (NFV) allows applications to be designed as virtual networks. Heterogeneous virtual networks are considered as requests, so called virtual network request (VNRs) and embedded in a substrate network (SN) owned by an infrastructure provider (InP). The process of embedding VNRs in an SN is referred to as the virtual network embedding (VNE) problem. Servers in fog networks may crash due to heavy workload and overheating, and services on them are interrupted. From the view of a fog network InP, the penalty incurred from service interruption hurts its net profit. To avoid server crash, an InP can adopt VNE algorithms with distributed deployment approach to achieve load balance. However, a distributed deployment approach results in additional energy consumption and also reduces the InP’s net profit. In this thesis, we aim to maximize the net profit of a fog network InP with consideration of penalty and energy cost. We propose a greedy method with a novel node ranking as a baseline, and whose performance is about 9% in terms of profit higher than an existing VNE greedy method. Then we develop an enhanced solution with a meta-heuristic method to further improve the net profit of a fog network InP. The performance is about 9% in terms of profit higher than the baseline. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/7729 |
DOI: | 10.6342/NTU201904179 |
全文授權: | 同意授權(全球公開) |
電子全文公開日期: | 2029-10-02 |
顯示於系所單位: | 電信工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-108-1.pdf 此日期後於網路公開 2029-10-02 | 1.92 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。