Project Description

All around the World we experience the trends of the last decades on increased urbanization as more and more people shift their living to cities. However, many cities lack the resources to respond to the magnitude of the change in their urban areas, which forces people to compete for the use of land, roads, public transport, and other urban facilities.

As a result of the increasing number of people, cities face an increasing number of private vehicles and commuters which in turn cause various problems such as traffic congestion, parking difficulties, traffic accidents, loss of space for productive activities, public transport inadequacy and undesirable environmental impacts. In the past, public authorities followed approaches that nowadays are financially unsustainable, focused mainly on expanding the road network to alleviate the problem. However, many analysts argue that the solution for these problems is better addressed through intelligent planning and management of the existing urban and transportation systems.

Planning of the urban and transportation system traditionally relied on the knowledge of present and future problems that are associated to the urban growth such as how much travel will be generated, where these trips will take place, by which mode and on which routes. Creating such plans requires information regarding the movement of people and vehicles, knowledge of constituents of the urban system, and understanding the nature of activities at different places.

There are various traditional methods for gathering the raw data necessary for urban and transportation planning. Although these methods have the advantage of providing detailed information, their limited coverage and expensive costs of implementation often make them insufficient. More recently, the spread of massive sensoring, namely through the generalized use of cellphone, is producing massive amounts of data with spatiotemporal detail about our daily activities and traveling patterns, which could be important to the planning of urban and transportation systems given their pervasiveness, low cost, and real time nature.

In this thesis we explore the use of cellphone data for profiling the dynamics of urban activities and characterizing flows of people for planning of urban and transportation systems in cities. Three types of passive mobile positioning data were used: (1) Call Volume, which is the number of calls; (2) Erlang, which is used to measure the equivalent cellphone traffic per hour; and (3) Handover, which is the process of transferring an ongoing call from one base station to another without interruption of service. Our observations are based on hourly aggregated cellphone data obtained from a dataset from a telecom company in Lisbon, Portugal.

Though passive mobile positioning data is extracted without incurring additional costs and operational risks for the network, it has two main limitations. Firstly, location acquired by this method is at the granularity of a cell sector, which gives uncertainty on the exact location of the collected variables; secondly, it is only acquired when a phone is engaged in a call or short message service. However, we argue that the aggregate cellphone data used in this study remains useful for our analysis, which is at a scale where the lack of a detailed level of precision is not essential. For validation of our results, we collaborated with other data providers in Lisbon to gather different ground truth datasets that could improve our understanding of urban dynamics such as census data, taxi movement, bus movement, traffic count, points of interest, and presence of people.

We proposed new approaches to reflect the potential of passive mobile positioning data for urban and transportation planning. Our approach comprises three stages: (1) exploratory data analysis aimed to discover the kind of relationship that emerges between cellular networks data and urban characteristics, activities, and dynamics at a city-scale; (2) use of cellphone data to detect activities associated to the urban areas in what respects to two aspects of activities: spatial patterns of urban activities, and intensities of urban activities along the hours of a day; and (3) extraction of cellular network data for development of models that predict hourly traffic status.

Our results confirm that passive mobile positioning data, taking the advantage of its pervasiveness and availability with reasonably less cost, can provide ways to analyse the dynamics of urban activities at a larger scale. In addition, our approach complements traditional urban data collection methods, which are usually made available less frequently to urban and transportation planners, and is especially useful for developing countries where other approaches are too expensive.

Research Team

• Merkebe Demissie
• Gonçalo Correia (supervisor)

Financial Support

• FCT

Stage of Progress

• Finished in 2014