USC Mobility Observations and Surveys
These data were collected as part of the EE-579, EE-599 and EE-555
courses under the upervision of Prof. Ahmed Helmy by various groups of
graduate students at USC.
Database of mobility:
In this project the participating students are asked to provide their physical location on campus
daily. We used a database to facilitate the manipulation of data. The database dump can be downloaded
HERE.
Some more explanation:
For each day, the students input their mobility data in the following format:
(1)The location visited (select from a given set of locations).
(2)Time of arrival at the location.
(3)Reason to visit the location ("scheduled" for events that come from a pre-determined daily schedule, "planned" for events one plans for the day without specific timing, or "random" for events that one just decides on the spot).
Also they are asked to evaluate their social "closeness" (i.e. friendship) from a scale of 1 to 5 with other known users when they
created an account in the database, and the data is also available in the database.
All the participants are from two EE graduate classes.
The time span of the experiment was from March 23 to April 21, 2004.
Survey of Mobility and WWP
project:
In this experiment, we went out to campus and asked mobility-related questions to randomly sampled students.
In particular, we asked:
(1)Where were you before coming to the current location?
(2)Where are you going toafter the stay at the current location?
(3)How long did you (or will you) stay at these locations?
From the above simple survey, we could derive two "movement events" from each survey (i.e. the movement from
the last location to the current location, and from the current location to the future location). Working out the
statistics, we create a time-dependent transition probability table between different types of locations. There are
five types of locations: classrooms, libreries, cafeterias, other locations on campus, and off-campus. We also
obtained the different pause time distribution for different types of building.
Based on the above data, we build a trace generator which uses a partial map of USC campus, containing 3 classrooms,
2 libraries, and 2 cafe's. Nodes choose the next location based on the (time-variant) transition probability table, and draws a
pause time from the corresponding distribution when it reaches the destination. We have separate transition probability tables
for morning and afternoon.
The survey was conducted between March 22 to April 16, 2004. We obtained totally 268 surveys.
More information can be obtained at the WWP webpage, or in the
related papers below, or the documentation for the trace generator (my appology for the
messy write-up), or email one of the authors (Wei-jen Hsu). The trace generator can be download HERE.
Wei-jen Hsu, Kashyap Merchant, Haw-wei Shu, Chih-hsin Hsu, and Ahmed Helmy, "Weighted Waypoint Mobility Model and its Impact on Ad Hoc Networks - Mobicom 2004 Poster Abstract," Mobile Computing and Communication Review, Jan 2005
(paper (.pdf))
(a longer version report, with some detailed data (.pdf))
Wei-jen Hsu,Kashyap Merchant, Haw-wei Shu, Chih-hsin Hsu, and Almed Helmy, "Preference-based Mobility Modeling and the Case for Congestion Relief in WLANs using Ad hoc Networks," VTC, Sep. 2004, Los Angeles
(paper (.pdf))
Kashyap Merchant, Wei-jen Hsu, Haw-wei Shu, Chih-hsin Hsu, and Almed Helmy, "Weighted Waypoint Mobility Model and Its Impacts on Ad Hoc Networks," Mobicom poster, Sep. 2004, Pheladelphia
(Poster (.ppt))
In this experiments, we identified the important intersections from the USC campus map, and went to those intersections
to observe how people make direction changes when they reach the place. Based on extensive observation, we were able to
estimate the turning probability at these intersections, and built a Markov model using these locations as states. The
model captures how would a node moves around on campus, if they move indefinately.
The raw traces (in Excel worksheets) can be downloaded HERE. A trace generator based on the above data (incorporating real dimension of USC campus) can be download HERE. More information
about the projects can be found at the webpage for the observation project, or
in the related papers below:
Deepankar Bhattacharjee, Ashwin Rao, Chintan Shah, Manan Shah, Ahmed Helmy, "Trace-based Mobility Modeling for Campus-wide Mobile Ad-Hoc Networks", Sigcomm 2004 poster.
(Poster abstract(.pdf))
Created March 2006 by Weijen Hsu. Maintained by Ahmed Helmy and Weijen
Hsu.