Spatial traffic prediction for wireless cellular system based on base stations social network

Understanding the spatial traffic patterns of wireless cellular system could facilitate the performance analysis and system design. In this paper, a novel method based on base stations social network (BSSN) is proposed for the prediction of spatial traffic in wireless cellular system. Firstly a BSSN, which can describe the spatial traffic relationship between base stations (BSs), is established with the real spatial-temporal cellular system traffic data. Then, the very important base stations (VIBS) can be selected based on BSSN from complex networks perspective. Finally, with the acquired traffic data of these VIBS, all the other BSs traffic can be predicted with the help of Support Vector Regression (SVR). The analytical results show that our proposed prediction method can effective predict the characteristics of traffics of the entire wireless cellular system; and by applying this method, only the traffic of the 8% of the entire BSs are required for an effective prediction with a mean error ratio less than 20%.