22nd International Conference on Advanced Information Networking and Applications - Workshops (aina workshops 2008)
Outlier Detection in Cellular Network Data Exploration
March 25-March 28
ISBN: 978-0-7695-3096-3
A cellular network like a GSM network is built up using a number of base stations to cover a large geographical area. The area covered by one base station can be seen as a cell in the network. Regardless of the location of a user the network should be able to provide the services. Therefore each cell should have enough resources to succeed in meeting the demand. Arising from this the analysis of the network can to a certain extent be divided into the analysis of separate cells. In this paper, methods for finding outlying base stations are examined by applying them on the analysis of real data from GSM network. As little as possible prior knowledge of the network is used in the analysis. This way the methods stay more portable into use with data from other networks. The results of the methods are analyzed by comparing them to each other.
Index Terms:
outlier detection, cellular network, nearest neighbors, Parzen windows, Gaussian mixture model, neural gas
Citation:
Mikko Multanen, Kimmo Raivio, Pasi Lehtim?ki, "Outlier Detection in Cellular Network Data Exploration," ainaw, pp.1323-1328, 22nd International Conference on Advanced Information Networking and Applications - Workshops (aina workshops 2008), 2008