Research On Content Scheduling Technologies In Content Networks

The rapid development of information and communication technology brings an explosive growth to network traffic, as well as giving rise to changes in personal Internet consumption habits. Internet has changed from information communication to the largest information resource database in the world. Meanwhile, users demand for information is increasing rapidly, ranging from simple resource sharing to intelligent content access and distribution. However,due to the lack of content awareness and the insufficient management mechanism, the problems in the network management and content management are becoming more and more prominent, which leads to an unreasonable utilization in network resources and a lower efficiency in accessing to the content. As a consequence, it is pressing to find a more intelligent and efficient access to the content.Within this context, the concept of content networks which is composed of distributed caching system and content scheduling system is proposed. In content networks, contents are distributed to the edge of the network, easing the pressure of the backbone network traffic and accelerating the content access speed. However, nowadays the caching system and scheduling system in traditional content network could not meet users’ growing demand for network traffic. Therefore, it is important to find an appropriate scheduling method to reduce the network pressure for network operators and content providers as well as to guarantee the users experience in the content network.This paper focuses on the content scheduling problem in the content network, along with a summary and analysis of the current situation and development trend of the content network. We carried out the research on the optimization of the caching system based on game theory, the research on the content based on the social relations, and the research on content recommendation. It aims to achieve the optimization of network resource utilization and user experience.The roadmap of this paper is organized as follows:① This paper proposes a game model to optimize the efficiency of the users’ requests in logical domains and the content scheduling efficiency among logical domains, which will address the inefficiency both users’ requests and in content scheduling. Initially, based on the characteristics of distributed storage and scheduling in the content network, the paper analyzes the storage cost and transmission cost, as well as defining the constraints of network optimization. Secondly, the paper proposes a master-slave game model by targeting the issues of users’ difficulty in obtaining the optimal content in the logical domain. It analyzes the uniqueness of the optimal solution of the model is analyzed, and the optimal content acquisition speed of the user is obtained via the Nash equilibrium. The changing trend of network optimal state under different network parameters is analyzed through the simulation experiment. Lastly, according to the cooperative mechanism of the cache server as well as the link state, the paper offers a method to optimize the efficiency of the content scheduling. The user experience of the game equilibrium and the improvement of the efficiency of the cache are verified through the simulation experiment.② In order to solve the traffic impact caused by UGC(User Generated Content) in the network, the paper presents a new content scheduling algorithm based on social relations. Firstly, the user influence model is proposed from the perspective that most of the content in the network are merely related to a few active users, so the model could find the active users by analyzing users’ social relations. In addition, to predict the users’ demands based on social relations, a mathematical modeling of users’ content preferences is proposed to establish the correlation between the users and the content. Consequently, including the hot content scheduling policy and the user preference content policy is addressed. The predicted hot content and user requirements will be deployed to the cache server in the user area via the content scheduling. The experimental results show that through the prediction of the network hot content and user needs from the point of social relations, it can effectively improve the cache hit rate, thus improving the user experience, as well as reducing the load of the source server and backbone networks.③ The paper proposes a new method based on recommendation, which addresses the phenomenon of information overload caused by network information explosion. In order to quickly filter out the users’ needs among a large number of information, the users’ direct recommendation is calculated in accordance to the historical information of users’ request. Then, a global recommendation algorithm is applied based on the correlation of the contents of the nodes. An intersection and clustering algorithm is used to cluster the global recommendation. With this theory, a content scheduling policy is proposed. The recommended content is scheduled to the cache server to increase the cache-hit ratio by recommending and gathering users’ similar preferences in groups. The simulation results demonstrate that the recommendation and scheduling of the content can effectively reduce the server load and ensure the quality of user experience.④ By conducting the above research, the information collection module and content mechanism of the cache server are designed in detail, and the experimental verification system is implemented. The experiments demonstrate the correctness and feasibility of some key technology. Modular design has been integrated into the system so that each functional module can be extended in order to make the system more compatible and extensive.The research of this paper not only provides some new research ideas within content management and network management of content network, but also offers certain references for the further research on content scheduling in content network.