Making cities smart through anonymised mobile data

By , 4 April 2014 at 17:25
Making cities smart through anonymised mobile data
Future Trends

Making cities smart through anonymised mobile data

By , 4 April 2014 at 17:25

4 April 2014: Traditionally, urban analysis and the study environment by understanding the city of urban environments have used data obtained from surveys to characterise specific geographical areas or the behavior of groups of individuals. However, new data sources (including GPS, bluetooth, WiFi hotspots, geo-tagged resources, etc.) are becoming more relevant as traditional techniques face important limitations, mainly: (1) the complexity and cost of capturing survey data; (2) the lack of granularity of the data given that is typically of aggregated nature; (3) the data is static and represents a snapshot of the situation in a specific moment in time; and (4) the increasing unwillingness of individuals to provide (what they perceived to be) personal information.

Some of the applications of smart cities and social dynamics include traffic forecasting, modelling of the spread of biological viruses, urban and transportation design and location-based services.

One of the new data sources relevant for the study of urban environments are cell phone records, as they contain a wide range of human dynamics information (ranging from mobility, to social context and social networks) that can be used to characterise individuals or geographical areas.

In this research project we use the information obtained from call detailed records to characterise and model urban landscapes in order to provide complementary approaches to traditional urban analysis techniques. The areas on which we have focussed so far include: (1) the automatic identification of dense areas; (2) the automatic segmentation of the city according to its real use; and (3) the identification of routes and mobility patterns.

The identification of areas with high density of people and/or activity is of paramount importance for e.g. urban and transport planners or emergency relief and public health officials. Urban planners can use this information to improve the public transport system by identifying dense areas that are not well covered by the current infrastructure, and determine at which specific times the service is more needed. In addition, public health officials can use the information to identify the geographical areas in which epidemics can spread faster and thus prioritise preventive and relief plans accordingly.

The spatial layout of a city has an obvious influence on the movement patterns and social behaviors found therein. Most western cities have a mixture of residential, commercial, and recreational areas connected via narrow streets, one- way avenues and a multitude of public transportation options and topographic features. Each of these areas has its own patterns of behavior which to date have only been elucidated by means of surveys and questionnaires. We have developed clustering algorithms to automatically segment the city in areas with similar behavior from the data available in anonymised and aggregated cell-phone records. Given the inherently fuzzy nature of both human behavior and urban landscapes, we propose a method to obtain robust behavioral segmentation using fuzzy clustering techniques. Using this method, only sections of the city with a given minimum similarity in their behavior will be labeled. This technique could also be applied to data obtained from other ubiquitous data sources, like geo-localized tweets, Flickr or the logs of any service that includes geo-localisation.

Finally, we are working on variations of temporal association rules and Markov chains to characterize the movements in the city. This approach enables the identification of the main mobility routes and the characterisation of each geographical area according to the mobility of its individuals. This knowledge can be used in a variety of domains including urban planning (proposing new public transport routes based on real movements) and efficient car-pooling.

previous article

Solving information overload in a digital age

Solving information overload in a digital age
next article

Extending Amazon’s Dynamo key-value store architecture

Extending Amazon’s Dynamo key-value store architecture