Research Of Mobile Education Synthesis System With Dynamic Cache Based On Time Series Prediction

Need for mobile education become more and more urgent with the rapid development of mobile Internet, and mobile education platform has a lot of functions, complex business and management. So for the goal of mobile education service that provide comprehensive functions with high quality, this paper designs and implements a mobile education content browsing and management system. It is a large system with a variety of business, includes client and server clusters; and by using the time series prediction of reverse proxy for dynamic cache configuration, the system performs better.Considering the factor of business expansion, capacity expansion along with users number's increasing,dynamic adaption and high flexibility, this paper design the overall structure of the system including streaming media, real-time interactive live, payment system, content sharing, voice translation, mobile management and console forum platform, and show implementation details of main functions. In order to support a large number of concurrent users' request, the background server use cluster based on tomcat and the apply Dubbo distributed architecture decoupling optimization service, as a result that make the function of the whole system in form of fragmentation with high flexibility, expansion and scalability. Both client and background server apply business combination mode which is compatible and expandable,then create mass customization features and services. New cross-platform technology are adopted at client application, combined with the deployment of nationwide media server for streaming media, thus frequent streaming media request can achieve rapid response. Meanwhile the system use Nginx as a reverse proxy, which has the function of cluster load balance and cache configuration to optimize the system. Finally, by testing overall function of the system and performance, this paper achieve the desired results.In order to enhance the system performance, the system itself collects a large number of user behavior statistics in education industry and does data analysis, then according to different types of content resources get modeled time sequence. With the sequence data, this paper put forward the improved dynamic prediction algorithm based on the ideas of weighted moving average and exponential smoothing, and get more accurate prediction for the traffic time series. Subsequently, the algorithm is applied in the dynamic configuration of the reverse proxy and achieve the positive effect that the content of the cache can be chosen dynamically and the size of the cache can be adapted dynamically, which can eventually shorten the response time of the user request. Then by testing prediction algorithm in ways of prediction precision, shorten effect to response time and comparing with other prediction algorithm, it is concluded that the improved algorithm performs better.