The battery capacity of a terminal device is limited and the data transmission energy consumption is too large between devices in the Device to Device (D2D) cache communication, which lead to the decline of the file unloading rate. In order to solve the problem, a Caching communication content Sharing Algorithm for minimizing inner-Cluster node energy consumption (CCSA)was proposed. Firstly, the user nodes in the network were modeled as Poisson cluster process in view of the random distribution characteristics of user terminals. The unloading model was established based on the energy and communication distance of the node devices, and an adaptive cluster head selection weighting formula was designed. Secondly, the energy and distance weighted sum of nodes were traversed, and the local optimal principle of greedy algorithm was used to select cluster head node. Thus, the user node communication distance was optimized to ensure that users' energy consumption was the lowest to prolong their survival cycles, and the unloading rate of the system was improved. The experimental results show that, compared with the clustered Random cluster head (Random) and the non-clustered Energy Cost optimal (EC) energy consumption optimization algorithms, when the network energy consumption is optimal, the proposed algorithm prolongs the system survival cycle by about 60 percentage points and 72 percentage points. The proposed CCSA can improve the unloading rate and reduce the unloading energy consumption of the system.