Research On The Key Technologies Of Multi-outlet Campus Network

Reviewing the development of the Internet, educational research network is actually an innovative base, so colleges and universities have become the birthplace of these technologies. Campus network provides the necessary infrastructure and environment guarantee for the researchers who undertake scientific research and technological innovation work. With the rapid development of Internet technology and emergence of new network application, the development of the campus network is confronted with many problems such as network security, network congestion and the shortage of funds etc.Campus network is the essential infrastructure of information construction for colleges and universities, and it is responsible for providing information services of teaching, researching, official work and other to the majority of teachers and students in addition to developing and testing new network technologies for researches and other tasks. This dissertation starts from the current situation of the key technologies of multi-outlet campus network, focuses on the prediction of LAN traffic flow based on chaos theory and application of content delivery network and high-performance NAT technologies. Furthermore, it, taking the campus network in Northeast Normal University for example, summarizes some problems like security, congestion and some others, and then designs some relevant strategies and feasible technical solutions.The LAN traffic flow based on chaos theory is studied in the dissertation; the time series corresponding to the campus network traffic flow are applied to predict the system; phase space is reconstructed for time series, based on the Takens theory; embedding dimension and delay time are calculated via C-C algorithm. And the largest Lyapunov exponent of the time series is calculated by the small data method. The judge of chaotic characteristics of time series and predicted calculation are based on the largest Lyapunov exponent. The neighborhood weighted prediction method based on the largest Lyapunov exponent is proposed and it makes prediction more accurate. The average period is then calculated via the power of the frequency weighting and averaging method, and it helps make the small data method more perfect and calculation of the largest Lyapunov exponent more accurate. The simulation experiment, which makes a systematic analysis of the Lorenz system and the Rossler system, shows that the improved method is more accurate and faster, and can also predict more points. The improved method and the method before are respectively applied to the calculation of the Unicom outlet traffic flow of the campus network in Northeast Normal University. As the comparative analysis shows, the improved method is valid.The dissertation makes a in-depth study on the content delivery network technology. The content acceleration system is deployed in the campus network based on the technology; the performance analysis method is then proposed. The deployment of the content acceleration system reduces the repetitive and entertaining traffic flow on the campus network links, and improves the utilization of bandwidth. Through performance analysis, the application benefit of content acceleration system is raised, and it provides a basis for the rational allocation of funds on the construction of the campus network. Meanwhile, it saves a lot of cost. The benefit of content acceleration system varies as time going. The numbers of days are taken as a unit, and it is directly proportional to the user traffic flow. The average is 18.17% of the total traffic flow. During the running of content acceleration system, if users download the resources in content acceleration system, then they can achieve the theoretical top speed. The experience of downloading spotlight videos resources and downloading local resources can achieve the same speed. Download speed can be achieved single port 100 Mbps. Every person can be assigned to 16 Mbps in bedroom of six. User experience is significantly improved.The high-performance NAT technology is studied. The NAT server is deployed on campus network boundaries based on the iptables tool of Linux system. Linux system with NAT is powerful, but the application ability of the default system parameter is not fully developed. The NAT transformation capacity is increased several times by optimizing the parameters of Linux kernel. The ethernet interface bottleneck for forwarding data is settled down via using bonding technology. The NAT transformation of 1.7 million concurrent connections and 1.4Gbps data forwarding capability are taken into practice. The NAT server is deployed in the campus network boundary, which operate stably. Above all, it proves effective and practical.