Modeling and optimization of bandwidth supply performance for cloud-assisted video systems under flash crowd

Abstract

The large-scale and sudden video content access such as flash crowds results in huge bandwidth demand, which severely influence user quality of experience and quality of service of video systems. In this paper, we firstly discuss the main reason of generation of flash crowds for video streaming services and analyze key factor for balance recovery between supply and demand of upload bandwidth. We construct two models: bandwidth supply capacity model of video systems and bandwidth demand model of users, which measures usage amount of bandwidth of the cloud. Based on the built models, we further employ a community-based cooperative caching strategy of video resources to promote supply capacity of upload bandwidth of video systems. Extensive tests show how the proposed cooperative caching strategy achieves much better performance results in comparison with original solution.