Mining and modeling large scale cell phone data: invited talk

• Authors: Jean Bolot
• Venue: FOMC '11 Proceedings of the 7th ACM ACM SIGACT/SIGMOBILE International Workshop on Foundations of Mobile Computing
• Year: 2011

Cell phones are ubiquitous in modern life and the call records collected by network operators are a powerful tool to study the behavior of cell phone users, and how those users use network resources, at previously impossible-to-achieve scales. In this paper we report on results from the analysis of billions of call records at a large cellular operator and we describe how mining that data leads to new and extremely exciting research problems in the areas of social network analysis, privacy, and economics. We consider three kinds of data, namely social network data (who calls whom, how often, etc), location and mobility data (who is where) and spectrum data (who uses how much spectrum in which cell). We describe practical examples of insights derived from mining that data as well as interesting research questions such as i) why is the structure of the cell phone social network different from the power law found in many other networks, ii) how predictable and unique are mobility patterns and what does this mean for location privacy, or iii) how can we model and quantify the economic value of private user data such as location data.