University of Science and Technology Houari Boumediene
Measurement studies of YouTube show that there are some distinctive properties regarding to video sharing service, e.g., small clip size, suggestion links between related clips. This observation motivates them to explore the possibility to use peer-assisted approach to offload content providers. Specifically, the authors want to leverage the suggestion links between video clips to construct cooperative caching overlay. In this paper, taking advantage of this correlation between video clips, they design a distributed algorithm to adapt the overlay towards a loosely connected Small World Network (SWN), in which similar-interest peers are clustered together by short links to maximize streaming cooperation, and long links are introduced to reduce switching delay when one peer switches to an unrelated clip.