An Implementation of Mobile Sensing for Large-Scale Urban Monitoring

The extent of human flow in the urban space has reached unprecedented levels, and is therefore the focus of the present study. Unlike common practice on urban monitoring which utilizes cameras or sensors, this research aims to introduce a new platform of urban analysis and monitoring by using mobile sensing to recognize urban behaviour. Population density and its moving pattern can thus be visualized from the usage of the mobile phone. The mobile sensing system is implemented on a web-based interface in order to maximize compatibility and interoperability. The aggregate mobile usages and antenna-mast positions are interpolated into griddensity surfaces. We then analyze urban patterns at a point of time to illustrate how people experience their city. Furthermore, the analytical results help to detect and to explain geographical ‘hot spots’ and clusterings of the unique characteristics of each urban space. Finally, we present visualizations of the results such as pseudo color and contour maps in order to demonstrate the urban dynamics. I.INTRODUCTION With the ubiquity and ever-increasing capabilities of mobile devices, cell phones and their locations could potentially become a powerful source to describe the pattern of the urban space. Traditionally, urban monitoring and analysis rely on a fixed location and a considerable amount of statistical data. This procedure does not permit the identification of multi-temporal events in wide areas. Another key problem is how to acquire and update the statistical data of the moving objects, for instance, humans and their activities in the whole city space. In this research, the usage of mobile devices will be treated as a medium for data collection. Erlang data that represent a distribution of call duration in the Global System for Mobile Communication (GSM) network could be performed as aggregate-data sources to estimate the population density of a city. For the large scale monitoring, clusters of Erlang data from mobile base stations are excellent at providing indirect interpretations of spatial patterns of urban life and its temporal dynamic. This aspect is very useful in the view of public monitoring. It could potentially become a new way to extract or identify invisible problem spots from the complex urbanized areas. Furthermore, mobile sensing is potentially applicable for public-marketing analysis. The distribution of a population at different points of time in each city space could be an ideal source to help people decide a place for urban advertising or opening a shop. In addition, an exploration of mobile sensing data would give the urban planner a better understanding on flowing patterns of people at specific times of the day. If we look at broader contexts, then population transfers from city to city during special events or public holidays could also be determined. Study area The research takes place at Bangkok Metropolitan, Thailand. Bangkok city has a well developed mobile network and has a high degree of mobile usage. We received the support of the Erlang data and base station locations from Advanced Info Service PLC (AIS), a leading mobile operator in Thailand. UrbanSense08 Nov. 4, 2008, Raleigh, NC, USA 51 II.SYSTEM ARCHITECTURE In the early research, the system was implemented as a fundamental tool for spatial exploration and visualization which permitted the data to be obtained, integrated and displayed quickly, easily and flexibly. Since the internet has now made distance virtually disappear, we questioned the traditional method of urban monitoring by enhancing the way how to instantaneously integrate and obtain large amounts of data via the network. Back end application servers To maximize compatibility and interoperability, open standards such as XML and web services are utilized for data exchange and sharing. Data analysis is assisted by the open source statistical package R (http://www.rproject.org), integrated into the PostgreSQL database via the PL/R procedural language. A browser front end The web browser we use as a universal front end, an Ajax mashup, is a hybrid web application which presents a rich UI to update and integrate contents asynchronously from multiple sources. This makes combining data easier, not only spatial data from the host server but also third-party sources from the services available on the internet. The calls can also be made directly to the third-party sources from the browser or back to the originating server, which acts as a proxy for the third-party contents. Fig. 1. System architecture of mobile sensing system Data manipulation We first simulate a connection with a mobile operator in our local environment instead of communicating directly to the mobile-operator system. The mobile-log data during the period from February to April 2008 that cover a part of the central Bangkok area are transformed and inputted into a database. The data established in the database mainly include cell-id, the base station geographic position, update time and Erlang data. The Erlang data