Seventh Joint Workshop on Coding and Communications: booklet of abstracts
Guiseppe Caire · Repository for Publications and Research Data (ETH Zurich) · 2014
Caching is a well-known general principle for which, if some information message is likely to be requested by a user (or network node) in the future, this can be pre-stored in the node itself or in its "vicinity" (in some topological sense) at a favorable time, such that when the request comes, it can be satisfied with low latency and/or without causing congestion in the network.Building on the fact that media content requests are highly predictable, caching has allowed the implementation of content distribution networks (CDNs), which are at the basis of vastly popular video streaming services over the Internet, such as Netflix, iTune and Amazon Instant Video.More recently (in the past 2-3 years), caching has attracted also a significant attention in information theory.In this talk, we shall review in a tutorial fashion some popular "basic" information theoretic models for caching networks and the corresponding known results on their fundamental limits.Interestingly, the achievability results are constructive and have an algebraic network coding flavor, and come at a fixed multiplicative penalty factor (independent of the size of the network and size of the message library) from information theoretic outer-bounds.