摘要:
A system and methods for the optimal distribution of streaming data among cloud-based providers, Content Delivery Networks and P2P Networks is proposed to aid the provision of entertainment services over the Internet. The Forecasting Middleware Scheduler (FMS) collects monitoring data about the usage of the multimedia content and uses it as input to a Resource Prediction Engine that predicts the future usage demand. For that purpose the RPE utilizes: (1) a number of standard time series models in order to fit the data to the most suitable model and predict future values of the demand and (2) an innovative epidemic spread model and an innovative method to predict a possible upcoming epidemic spread of content. An algorithm is used within the FMS for the combination of the prediction methods in order to suggest the optimal delivery methods to be used.