Coded Caching under Heterogeneous User Preferences: An Effective Throughput Perspective

Proactive caching is a promising means to handle increasing wireless traffic. However, how heterogeneous user preferences impact the caching gain is still an open problem. In this paper, a two-phase cache-aided multicasting network is investigated, in which users with heterogeneous preferences are served by a base station through a shared link. It is shown that the achievable domain of effective throughput of the users is a convex set and can be characterized by its boundary in the positive orthant. A special type of caching schemes, named uncoded placement absolutely fair (UPAF) caching, is studied. For the two user case, the achievable domain of UPAF policies has a piecewise linear boundary. For the multiuser case, a feasible UPAF policy is proposed to obtain caching and multicasting gains. Simulation results demonstrate that users with more concentrated preferences can attain higher effective throughput.