A brief video summarizing the research from the NOMADS group at University of Florida


Part 1: User clustering (Wei-jen's MOBICOM 2007 extended abstract) and gender-based user analysis (Udayan's work)



Part 2: Profile-cast (Wei-jen's WCNC 2008 paper and ongoing work) and some future directions in the NOMADS group (See Sungwook's and Jeeyoung's page)

You can download the full-size video clips HERE (Warning: very large files, total of 200MB)



kml generator to display mobile network users on Google earth


Displaying mobile users on Google earth can help researchers find specific patterns of mobile users. Another advantage is visual validation of found pattern. We would like to assist other researchers in mobility as well as ourselves. Here, we show sample files for download. Please check our demo papers link down below(IEEE INFOCOM, IEEE SECON, ICNP) for further details.

Part 1: Sample kml files

kml is a special mark-up language to be used in Google earth. It is similar to xml and easy to read and write. For more details about kml: http://code.google.com/apis/kml/documentation/

1) nodes_black.kmz: kmz is a zipped file(simply renaming of file extension from zip to kmz). Users can either open this zipped file in the Google earth directly or unzip it and open the files. 120 kml files are included in this zipped file. Each kml file corresponds to one anonymous mobile user at USC campus during 1/24/2006 - 2/20/2006. This data is collected from [USC WLAN trace]. After running Google earth, open all the files by either dragging all the files to Google earth or opening all the files from the menu. By clicking the play button in the bar on the upper-right, you will be able to see black dots showing around. Each black dot indicates one user.

2) nodes_colored.kmz: 120 kml files are included in this zipped file. However, unlike the previous black dot representation, each node is colored according to their group. Grouping is preformed by the singular value decomposition technique depending on similarity of location visiting preference. [Wei-jen et al.] You can see how well this grouping techique works.

3) nodes_polygon.kmz: 5 kml files are included. When deploying thousands of kml files on Google earth, it stops for serveral minutes before displaying them on the screen due to too many files. To overcome this issue, we merged mobile users into their corresponding groups to put in one file for each group. Therefore, one kml file here represents behaviors of the same group of nodes.

Further, unlike 1), 2) in which one dot indicates one node, here, one polygon bar represents the same group of nodes appearing in the same location at the same time. This kind of representation can help users easily to observe the population change of group of nodes in various locations.

4) privacy issue: [Udayan et al] revealed that campus WLAN users are not safe from privacy issue in their paper. This is presented in our demo and also available from the video in part2. We are not releasing kml file for this due to potential privacy issue. Please contact Udayan for privacy research.

>> please contact Sungwook for questions regarding kml files and further details.


Part 2: Kml generator (parsing Wireless LAN trace)

Each WLAN trace has different format. Here, we assume the WLAN trace is filtered and refined to a format that has followings: identity(e.g. MAC address) of individual node; time stamp of association with access point(or location); duration of association with access point(or location).

GPS coordinates of each location is also required.

1) Using database

Given the above trace data in the database, the following sample C language code can be used:

[sample code download]

Note that this sample code is not a complete code that is compilable. It will work with database communication code. We intentionally removed most of the database code as different database may require different codes.

2) Using data files:

We do not provide a source code for using data files at this point as the most of code is the same as using database and some files are specific to our own purpose. If you want to use data files instead of databvase, necessary files include,

* list of mobile user's ID

* group or cluster ID list (if needed to display by groups or clusters)

* group color (e.g. blue for group 1)

* GPS coordinates for each building or location

* infomation file for each user (ID, timestamp, name of location such as access point)

We hope this information is useful in developing another tool for mobility analysis. For questions about codes, please contact Sungwook.

>> last updated, 9/25/2008




* Previous presentations

(Note: For slides presentation, some of the animations are not converted into PDF files perfectly. For best viewing please take the PPT files)