Geovisual analytics for self-organizing network data

Cellular radio networks are continually growing in both node count and complexity. It therefore becomes more difficult to manage the networks and necessary to use time and cost effective automatic algorithms to organize the networks neighbor cell relations. There have been a number of attempts to develop such automatic algorithms. Network operators, however, may not trust them because they need to have an understanding of their behavior and of their reliability and performance, which is not easily perceived. This paper presents a novel Web-enabled geovisual analytics approach to exploration and understanding of self-organizing network data related to cells and neighbor cell relations. A demonstrator and case study are presented in this paper, developed in close collaboration with the Swedish telecom company Ericsson and based on large multivariate, time-varying and geospatial data provided by the company. It allows the operators to follow, interact with and analyze the evolution of a self-organizing network and enhance their understanding of how an automatic algorithm configures locally-unique physical cell identities and organizes neighbor cell relations of the network. The geovisual analytics tool is tested with a self-organizing network that is operated by the automatic neighbor relations (ANR) algorithm. The demonstrator has been tested with positive results by a group of domain experts from Ericsson and will be tested in production.