A Change-Point Detection Scheme Based on Subspace Tracking for Mobile Access Traffic
Senri Hirabaru, Takahiro Matsuda, Yuki Hirota, Haruki Izumikawa, Hiroshi Hanano, Chihiro Ono, Tetsuya Takine · 2016
Because a huge number of base stations are deployed due to the recent development of mobile access services, network monitoring in mobile networks becomes more challenging. In this paper, we propose a novel change-point detection scheme to efficiently detect a time instant (change-point) when the spatial distribution of mobile terminals connecting to base stations is changed dramatically. In the proposed scheme, we regard the numbers of connections at multiple base stations as multi-dimensional data, and from the correlation matrix of the multi-dimensional data, the eigenvector associated with the largest eigenvalue is computed sequentially by means of subspace tracking. The eigenvector is transformed to a one-dimensional sequence and change-points are detected by an existing one-dimensional change-point detection scheme. We evaluate the performance of the proposed scheme with simulation experiments.