Poster: Cluster analysis based traffic prediction method from real traffic traces in LTE networks

In this paper, the relationship between traffic and radio resources is analyzed combined with the real LTE network data. We found that the correlation between traffic and resources in LTE networks is much more complex than that in traditional networks like 2G and 3G, which can be exploited to predict their trends. So we employ a prediction method based on multidimensional correlation and cluster analysis to predict traffic and resources. By analyzing a large number of historical data from current LTE systems, this method can not only improve the prediction accuracy but also greatly reduce the complexity of the algorithm.