In this paper

the variation of the number of cell’s users within the cluster is studied based on Queuing theory and Markov transition model in C-RAN network.We proposes a prediction scheme to calculate the cell’s arriving rate and leaving rate by considering the history regular and the current trend.By solving the transient solution of the Markov transition model

the statistical average of the every cell’s users is achieved at different times.Considering the prediction error and the system overhead

the optimal period of the prediction is obtained.The simulation result shows that better prediction period can accurately reflect the variation of the users in the realistic environment.