Cooperative local caching and file sharing under heterogeneous file preferences

Local caching with device-to-device (D2D) communications has been recently introduced as an effective scheme for reducing the average download time of the mobile terminals (MTs). The MTs first cache the files in their local memories and then exchange the files with each other within the vicinity via D2D communications. Prior works have largely overlooked MTs' heterogeneity in file preferences and assume unselfish caching behaviors of the MTs. In this work, we practically divide the MTs into different groups according to their individual preferences over the files and propose optimal file caching strategies for self-interested MTs to reduce the average file download time. Assuming the knowledge of the social file preference for an intelligent group, we develop the optimal caching strategy for this group by formulating and solving a convex optimization problem. Closed-form solution for the problem is obtained, which is shown to follow a water-filling structure over the files. Finally, numerical examples are presented to show that the selfish caching of a group can be detrimental to both itself and the other intelligent groups.