Reducing computational complexity of coded caching by partitioning users into groups: poster

In this paper, we show how to partition K users in a coded caching domain into L groups to reduce the complexity from O(2K) to [EQUATION]. We first show that the negative effects of the heterogeneity of cache sizes and partitioning on coded caching performance neutralize each other, which is the theoretical foundation of our partitioning schemes. We then prove the submodularity of the coded caching traffic volume function, based on which we develop a partition algorithm for minimizing the traffic volume. We show that the approximation ratio of proposed algorithm is better than the state-of-art result. Moreover, we develop a rounding algorithm to obtain the final partition based on Lovász extension, which resolves the issue of unbalanced partition. We implement our schemes in a testbed and validate our analysis and design with experiment results.