With the improvement of wireless networks and the capacity of terminal devices, the relationship between people and networks has become closer. Device-to-Device (D2D), one of 5G's key technologies, has the potential to make full use of context-processing capabilities of network and user behaviors. Since handheld devices are closely related to the social behaviors, there have been many methods utilizing the data of users' daily life to optimize the performance of the D2D communication. Among these factors, the movement of users which may lead to the instability of the D2D communication is often ignored. We focus on the instability of the D2D communication caused by the movement of users and propose a model of mode switching of D2D pairs. In addition, We propose a mobility-aware joint resource allocation and power allocation algorithm (MARP) to optimize the channel resources allocation and power allocation and improve the performance of the system. The simulation results verify the effectiveness of the proposed algorithm,