Improving Quality of Service in Software Defined Network via Prediction Methodology

碩士 === 國立交通大學 === 資訊管理研究所 === 107 === With the huge demands for communication devices, network traffic will increase dramatically year by year. At the same time, the probability of network congestion has increased significantly. In order to ensure that users can be guaranteed to reach a certain qual...

Full description

Bibliographic Details
Main Authors: Hsieh, Yun-Shyuan, 謝昀璇
Other Authors: Ku, Cheng-Yuan
Format: Others
Language:en_US
Published: 2019
Online Access:http://ndltd.ncl.edu.tw/handle/6ycc2q
Description
Summary:碩士 === 國立交通大學 === 資訊管理研究所 === 107 === With the huge demands for communication devices, network traffic will increase dramatically year by year. At the same time, the probability of network congestion has increased significantly. In order to ensure that users can be guaranteed to reach a certain quality of service (Quality of Service, QoS) under limited resources, there are several criteria of QoS to be accomplished. Software Defined Network (SDN) is an architecture that separates the control logic from the data plane, to manage the entire network flexibly. The Genetic Algorithm (GA), its spirit is inspired by nature's evolution laws, is a widely used meta-heuristic algorithm. It can be used to solve routing problems with multiple restrictions. In this paper, the software-defined network architecture is used to build the experimental environment, and machine learning is used to predict the network traffic, and then the genetic algorithm is used to solve the routing problem. Through the above mechanism, which can reduce the probability of congestion and improve the performance of QoS , such as reducing the packet loss rate.