Mathematical Analysis of caching policies and cooperation in YouTube-like services

Abstract

Currently, most video on-demand services offered over the Internet do not exploit the idle resources available from end-users. We discuss the benefits of user-assistance in video on-demand systems, where users are both clients and servers, helping with the task of video distribution. The mathematical machinery for the systematic analysis of video on-demand services is not mature yet. In this paper we develop a deterministic fluid model to determine the expected evolution of user-assisted on-demand video streaming services. We theoretically prove that cooperative systems always outperform non-cooperative solutions. A combinatorial optimization problem is proposed, where the goal is to distribute a set of video items into repositories trying to offer the minimum waiting times to end-users. This combinatorial problem is proved to be in the class of NP-Complete computational problem, and is heuristically solved with a GRASP methodology. Predictions inspired in YouTube scenarios suggest the introduction of cooperation is both robust and extremely attractive from an economical viewpoint as well.