Optimizing Caching Placement For D2D Assisted Wireless Caching Networks

The mobile multimedia data generated by smart phones and mobile devices is undergoing an unprecedented growth and expected to continue grow at a compound annual growth rate of more than 60%.Caching popular contents at various nodes in the network is a promising approach to alleviate the network bottleneck by offloading data traffic from the cellular network.Numerous recently works exploit the caching capacity in wireless networks and show the performance depends heavily on the caching placement.In this paper,we devise the optimal caching placement to maximize the offloading probability for a two-tier wireless caching system,where the helpers and a part of users have caching ability.The offloading comes from the local caching,D2 D sharing and the helper transmission.In particular,to maximize the offloading probability we reformulate the caching placement problem for users and helpers into a difference of convex(DC)problem which can be effectively solved by DC programming.Moreover,we analyze the two extreme cases where there is only help-tier caching network and only user-tier.Simulation results indicate users and helpers prefer to cache popular contents under low node density and prefer to cache different contents evenly under high node density.And then we investigate the effect of user mobility on the offloading probability for D2 D assisted wireless caching systems.Specifically,we model the mobility pattern of users as the contact time and connect frequency,and the offloading probability of both entire caching scheme and coded caching scheme are then derived.In order to maximize the offloading probability,the optimal caching placement problem for helpers and users is formulated and the corresponding solutions are proposed for both entire caching and coded caching schemes.We show the proposed scheme achieves both mobility gain and computation gain,in addition to the traditional caching gain,by increasing caching efficiency and distributing content across users caches.Finally,we find that the mobility of user not only provides mobility gain,but also improves computation gain.