User-level Caching In Device-to-Device Networks

The mobile data caused by content delivery has been explosively growing while users' tolerance to the data acquirement delay has been decreasing.To handle these requirements,the forthcoming 5G communications devote to establish a high-capacity low-delay network to satisfy the above two aspects of user requirements.An important method to meet the requirements of both capacity and delay is to cache contents into user devices in advance so as to enable users to acquire contents through the fast short-range device-to-device(D2D)commnications in peak hours.However,user devices have relatively limited storage spaces.Thus,the acquirable contents of users are also limited.Moreover,wireless channels and the reachable resources(e.g.,contents in other users' devices)differ from user to user.These features result in the non-linear cache-performance relationship and thus make it difficult to cache contents into devices with a high network performance.As a consequence,it is of great significance to come up with appropriate user-level cache placement schemes to match the wireless channels and reachable resources in D2 D networks.First,in order to enable users to acquire contents directly from local devices(i.e.,devices in the same cell),we discuss how to place contents into user devices with maximized cache hit ratio.The maximization is realized by the procedures as follows.First,we establish the sharing relationship through the direct D2 D communications and the indirect multicast communications among users.Then,we derive the propability that users can acquire contents according to the above relationship.Finally,we propose a heuristic algorithm to maximize this probability,which is exactly the cache hit ratio.Note that apart from our proposed scheme,other researchers also have proposed a coded caching framework to guarantee that users can acquire content indirectly from other user devices.However,this framework is established on the the low-rate multicast communication and thus cannot achieve a low delay.To further improve the delay performance under this coded caching framework,we first propose to decrease users' receive delay by replacing some multicast communications with D2 D communications and then derive the relationship between delay and subfile sizes under these communication modes.Finally,we propose a linear programming model to indirectly reduce D2 D transmit delay.According to the simulation results,the above two proposed schemes are both proved to outperform the other schemes in terms of the cache hit ratio and delay,respectively.