Anomaly Detection Framework for Tracing Problems in Radio Networks
This paper shows a novel concept of using diffusion maps for dimensionality reduction when tracing problems in 3G radio networks. The main goal of the paper is to identify abnormally behaving base station from a large set of data and find out reasons why the identified base stations behave differently. The paper describes an algorithm consisting of pre-processing, detection and analysis phases which were applied for RRC (Radio Resource Control) connection data gathered from the live radio networks. The results show that the proposed approach of using dimensionality reduction and anomaly detection techniques can be used to detect irregularly behaving base stations from a large set of data in a more self-organized manner.