Cache Storage Optimization for Locality-Aware Peer-to-Peer Multimedia Distribution

In this paper we present an optimization algorithm to minimize caching storage occupation in a peer-to-peer locality-aware multimedia distribution service. Our proposed solution uses observed historical data on video requests to predict future popularity of the elements in the catalog; the expected request rates computed in this way are then used to guide the caching decisions of the optimization algorithm. Simulation results show that our solution is able to achieve local hit rates that are within 0.6% of those of a traditional Least Frequently Used caching policy, but with a fraction of the storage occupation.