In this paper, a new communication model is built named grouping D2D (GD2D). Different from the traditional D2D coordination, we proposed GD2D communication in licensed and unlicensed spectrum simultaneously. We formulate a resource allocation problem, which aims at maximizing the energy ef-ficiency (EE) of the system while guaranteeing the quality-of-service (Qos) of users. To efficiently solve this problem, the non-convex optimization problem is first transformed into a convex optimization prob-lem. By transforming the fractional-form problem into an equivalent subtractive-form problem, an iter-ative power allocation algorithm is proposed to max-imize the system EE. Moreover, the optimal closed-form power allocation expressions are derived by the Lagrangian approach. Simulation results show that our algorithm achieves higher EE performance than the traditional D2D communication scheme.