Analysis of large call data records with big data

Abstract

Mobile communication flows describe the on-going traffic on the network and are therefore a good indication of what is happening. Analysing these flows can improve the overall quality offered to the users and it can enable operators to detect abnormal patterns and react. This paper will focus on the analysis of cellular communications records. By using the collected call and message exchanges we present a method based on the PageRank algorithm that detects abnormal communications events. Taking the number of calls and the total call duration as parameters we use a weighted version of the PageRank algorithm to further investigate the influence of these parameters on the connected network graph. We proceed by correlating the results obtained with events happening in the respective region and at that time.