In this article a novel distributed cell-breathing algorithm based on a clustered architecture is proposed for energy-efficient cellular networks. Firstly, a brief state of the art of current cell-breathing techniques is provided. Afterwards a description of the system model is given followed by the algorithm description. The algorithm is executed in each BS of each cluster following a sequence, which is synchronized into the set of clusters. When a BS attempts to go to sleep mode, it tries to redistribute the traffic to its neighbors if it is not more loaded than any one of them. When the BS redistributes this traffic, it will prefer to send the traffic to the highest loaded neighbors first. The presented algorithm avoids a centralized coordination and initial setup load thresholds. Therefore single point failure problems are not a risk and a robust self-organized approach is provided. We compare the results of state-of-the-art algorithms with the ones of our algorithm and point out the drawbacks of the other algorithms.