Large-scale Characterization of Comprehensive Online Video Service in Mobile Network

Video service in mobile network generates massive Internet traffic and consumes a large share of bandwidtb. Nowadays, more and more video portals have upgraded tbeir service to a comprehensive one-stop solution, providing UGC, VoD and Live streaming at the same time. Understanding the characteristics of such service is crucial for tbe network traffic engineering and tbe sustainable development of service. In this paper we present the first characterization work in mobile network on a leading comprehensive online video service in China, Youku. Our study is based on large-scale datasets containing 16 billion network traffic traces and long-term crawling data (over 1 month). We examine tbe user activity in botb temporal and spacial dimensions and reveal tbe relatively fixed service usage. We demonstrate the strong correlations between users, which is reflected by tbe small-world phenomenon. We also investigate the static properties of Youku mobile videos and observe noticeably differences from otber UGC videos. Eventually, we study the distribution of view count and find out four growtb patterns for tbe new published videos. Our analysis result provides a insight into how Youku is utilized by mobile users and demonstrates promising practical applications for both network operators and service providers.