Analysis And Resource Management Of Web Video Traffic

The fast-growing Internet has become an indispensable part of many people's lives.Filtering and identifying massive network video data and locating specific service modules are of great significance for mastering the traffic and user behavior characteristics of network and promoting the development of network video services.First of all,the thesis introduces the development and technical background of online video stream transmission on the HTTP protocol.The continuing content contain the following topics:Ⅰ.Summary of the operating process of video service and conclusions from video traffic feature analysis.Ⅱ.Features for machine learning algorithms to recognize the video traffic extracted after analyzing video traffic data.The first set of the features are a group of 6 numerical features constructed after statistical analysis of data.The second set of features are extracted from the text fields of the HTTP protocol of data.After feeding the features into traditional machine learning and deep learning algorithms that are effective in the realm of common classification and text classification,the results prove that our methods work well on the realistic dataset.Ⅲ.For the convenience of IP address data management,a system is designed for managing IP addresses of video delivery servers in a semi-automatic way,with less mastery barriers.The essential points of the technology proposal are covered in the thesis.