An Optimization Approach to Heterogeneous Coded Caching
Alexander Michael Daniel · TSpace (University of Toronto) · 2017
This thesis aims to develop an optimization framework for the development of coded caching schemes that can account for various heterogeneous aspects of practical communication systems. This framework is first used to develop an optimization-based perspective on the seminal work on the fundamental limits of caching by Maddah-Ali and Niesen, and is then extended to the case where files have non-uniform popularity and length, while users have non-uniform cache sizes. The resulting optimization problem scales exponentially in the system parameters, and so simplifications of the original problem that both perform well and scale as a polynomial function of the systems parameters are developed. By considering the aforementioned heterogeneities both individually and in conjunction with one another, insights into the influence of their interactions on optimal cache content are obtained.