A framework has been developed that processes and analyses data from multiple mobile network operators, while respecting subscribers’ privacy. One application can be more accurate population density estimates, for better policies and planning.
Managers of buildings such as schools and offices often keep tabs on who is in the building and where, for safety and security. How can policy-makers know where whole populations are to manage services or resources? The traditional census is useful, but lacks reliable detail on how population densities vary over short periods – such as the working day – and is expensive to update.
Together with the Austrian Institute of Technology, researchers at the European Commission’s Joint Research Centre (JRC) are developing a framework that analyses data from different mobile network operators to learn about people’s behaviour. Because most people now carry a mobile device, records from these can reveal much about populations, such as where people are concentrated and how they move around.
The framework processes data from different network layers from multiple operators to recover this information, using methods that respect subscribers’ privacy. Data include anonymised call data records (CDRs), visitor location registers (logs of phones near a cell site), network topology maps and service coverage maps, creating a detailed geographic model of subscriber densities across whole cell sites.
As different operators provide their services and create and store data in different ways, the team applied EU interoperability guidelines to the framework to harmonise data processing and analysis. And because the framework is interoperable, any operator can add data to the framework without changing their configurations.
The model is based on a map-reference grid that can scale data up or down and is compatible with the Infrastructure for Spatial Information in the European Community (INSPIRE) Directive. Its layered structure makes it possible to compare data from different operators and combine it with traditional geographic information – such as land cover or official population data – to estimate changes in population densities.