Abstract: The traditional green networks are usually achieved by minimizing the transmit power under some quality of service (QoS) constraints. In this paper, admission control is integrated with power allocation to pursue further performance improvement in green communication. This technique enables the network to serve as many users as possible when the network cannot guarantee all the users' QoS requirements. Moreover, it classifies the users and picks out a subset of admissible users whose QoS constraints can be easily satisfied with relatively low power consumption; thus the power efficiency is improved. After approximating the original joint power and admission control problem by a convex sparse optimization problem, an efficient distributed algorithm is developed for the approximated problem by fitting it into the alternating direction method of multipliers (ADMM) framework. Specifically, each iteration of ADMM can be computed in closed form, thus giving it very low complexity. The effectiveness of the proposed algorithm is validated by a series of numerical simulations.