- Note: If you use any of these tools, code or data, please make sure to cite our correspnding paper(s)! Thanks!
- Note2: Please abide by all the licenses under which this data and code are released.
News
- (Dec 2018): Collectively, over 40TB of data have been processed and used in our research
(among the largest mobility/mobile-networking datasets at a university-based research group).
We are in the process
of sanitizing and preparing chunks of this data to share with the community and collaborators, along with the tools
and algorithms to utilize them. You will find initial releases of data and code in the links below. Feel free to contact
us at any of our emails if you are interested in other datasets or have feedback!
- (April 16, 2018): A new website with mobile data for the
Flutes vs. Cellos paper (IEEE INFOCOM 2018) is
coming on line. Please stay tuned for updates!
- A link to a closely related NSF-funded project, MobiBench,
is available here.
- (Oct 31, 2018): A new website for the 'En-Route'
paper (IEEE INFOCOM wkshp 2017) on scalable vehicular simulations with scenarios, code and
sample data is now available at this link.
- (Oct 31, 2018): The tools for reproducing the results in our two ACM MSWiM 2018 papers are now available
(also as 'docker' images) at: (published version of the papers available through the ACM digital library)
For the 'Encounter-Traffic' (tech report) paper the data/tools are on Github at:
This webpage aims to establish a community-wide library of mobile wireless
networks traces and measurements.
The goal is to have the traces, simulation code,
test-suites (e.g., benchmarks, test scenarios) and models (of mobility, traffic and user behavior)
established by the experts in the field, widely available
for everyone to use and compare against.
History: Initial idea for MobiLib started in 2003 by A. Helmy and his
group at USC. Actual effort for initial data collection and library setup started in 2004 by
A. Helmy and Fan Bai. First launch of the MobiLib website was in 2005 by A. Helmy,
F. Bai and Wei-Jen Hsu. Many Prof.s and researchers around the world encouraged this
work and provided links to datasets (acknowledged at the bottom of this page).
Around the same time frame Prof. D. Kotz (Dartmouth) was starting his
effort for CRAWDAD (the goal for which was to 'house' the data in addition to
collection from Dartmouth, unlike MobiLib which was to provide pointers to
existing datasets). Most wireless datasets contributed by various Prof.s, including our
group, are housed (wholly or partially) in CRAWDAD.
MobiLib moved to the University of Florida in '06/'07, where A. Helmy
and his group are conducting further research in the area of mobility and behavioral modeling for wireless network users.
Thanks to many major universities who agreed to provide traces
(or pointers to traces) including USC, UFL, MIT, UCSD, Dartmouth, UCSB, UIUC, GA Tech,
Purdue, UCLA, Rice, Boston U, Columbia, U Washington, UNC. Links are continuously updated.
If you are interested and can contribute, please
contact: helmy@ufl.edu.
UCSD: PDA trace of University of
California, San Diego
(contains access points seen by PDAs taken at a 20 seconds interval, for 3 months) [check CRAWDAD for availability].
- Here are selected publications from the NOMADS group at UF (and previously at USC):
[Note: For more recent publications check A. Helmy's home page and related publications links.]
- Further related work from the NOMADS group: (most papers available through this publications website)
- U. Kumar, N. Yadav, and A. Helmy, "Analyzing Gender-gaps in
Mobile Student Societies," CRAWDAD Workshop poster (colocated
with MOBICOM 2007)
- J. Kim, Y. Du, M. Chen, and A. Helmy,
"Comparing Mobility and Predictability of VoIP and WLAN Traces,"
CRAWDAD Workshop poster (colocated with MOBICOM 2007)
- U. Kumar, N. Yadav, and A. Helmy, "Gender-based feature
analysis in Campus-wide WLANS," MOBICOM 2007 poster and SRC. [A webpage for the results is available HERE]
- W. Hsu, D. Dutta, and A. Helmy, "Profile-cast: Behavior-Aware Mobile Networking," MOBICOM 2007 poster and SRC.
- W. Hsu, D. Dutta, and A. Helmy, "Extended abstract: Mining behavioral groups in large wireless LANs," to appear in Proceedings of MOBICOM 2007. [Longer version of technical report available HERE]
Highlights:
In this work we leverage unsupervised learning technique (i.e., clustering) to identify groups of users with distinct behavioral patterns in the WLAN traces. We develop the TRACE framework and use summarized mobility pattern as an example to show the applicability of the framework. We find that university campus is a diverse setting in which hundreds of groups with distinct behavioral modes exist, and the group sizes follow a power-law distribution.
The technique we propose in this paper could be used for better mobility models, behavioral norm establishment and abnormality detection, profile-based services such as advertisement and group-cast, to name a few.
- W. Hsu, T. Spyropoulos, K. Psounis, and A. Helmy, "Modeling Time-variant User Mobility in Wireless Mobile Networks," in Proceedings of IEEE INFOCOM, May 2007. [Webpage for the time-variant community model HERE, mobility trace generator and manual available]
Highlights:
The Time-Variant Community Mobiliy Model is a model we create to capture two important mobility characteristics we observed earlier from WLAN traces. These two
mobility characteristics are skewed location visiting preferences and periodical re-appearance.
While improving the realism of the mobility model, we also keep mathematical tractability as a requirement for the mobility model. We use random-direction mobility
model as the basic building block, modify it to incorporate fore-mentioned mobility chracteristics. We are able to derive two quantities of interest related
to mobility-assisted routing, the hitting time and the meeting time. We intereted in deriving other quantities in the future.
We make the code for our time-variant community model available here. The code has many parameters and provides full flexibility to match with various mobility
scenarios (for full details, refer to the manual). It simulates the hitting time, the meeting time, and prints the movement traces in two option formats:
(1) NS-2 compatible format, or (2) time, location (in x,y coordinates) format.
- W. Hsu, D. Dutta, and A. Helmy, "Mobicom Poster Abstract: On the Structure of User Association Patterns in Wireless LANs," to appear in Mobile Computing and Communication Review. Earlier version of poster abstract accepted by MOBICOM 2006.
Highlights:
This paper provides the most comprehensive study of WLAN traces to date.
Traces collected from four major universities (~12,000 users) are analyzed
using metrics for individual user and group behaviors. Similarities and
differences across campuses are studied. Conclusions provide great insight
into realistic behavior of wireless users. Most users are 'on' for a small
fraction of the time, number of access points visited (per user) is quite
low, and on-line user mobility is quite low. On average, a user encounters
only 2%-6% of the user population. Encounter-graphs and small worlds are
introduced to model encounter patterns between users. We find that number
of encounters follows a biPareto distribution and the frienship indexes
follow exponential distributions. A paradigm for 'encounter-based
information diffusion' is introduced for efficient data dissemination in
mobile networks.
- W. Hsu, A. Helmy, "Principal Component Analysis of User Association
Patterns in Wireless LAN Traces", IEEE INFOCOM poster, April
2006.
- W. Hsu, A. Helmy, "Capturing User Friendship in WLAN Traces", IEEE INFOCOM poster, April 2006.
- F. Bai, A. Helmy,
"Impact of Mobility on Mobility-Assisted
Information Diffusion (MAID) Protocols", IEEE SECON 2007. Highlights: This study analyzes a class of protocols, MAID, that utilize mobility for information diffusion. MAID uses encounter
information to create age gradients towards the target, and can be used
for discovering resources, routing or
locating nodes efficiently in future mobile networks.
Analytical models are developed to evaluate MAID's performance during its
various (transient and steady-state) phases of operation. Extensive
simulations are used to validate these models and to study the
sensitivity of MAID to a rich set of mobility models.
We find that although MAID is sensitive to the mobility pattern, its
steady state performance is, surprisingly, insensitive to velocity.
We identify the properties of the 'age gradient tree' as the key factor to
explain this interplay between mobility and the MAID protocols.
- F. Chinchilla, M. Lindsey, M. Papadopouli,
"Analysis of wireless information locality and association patterns in a
campus", IEEE Infocom '04.
- A. Balachandran, G. Voelker, P. Bahl, P. Rangan,
Characterizing
User Behavior and Network Performance in a Public Wireless
LAN", ACM Sigmetrics 02
- D. Tang, M. Baker, "Analysis of a Local-Area Wireless Network",
ACM Mobicom 2000.
Thanks to the people who have contributed traces and/or encouraged this
work:
Mostafa Ammar, Richard Fujimoto (Georgia Tech),
Kevin Almeroth, Elizabeth Royer (UCSB),
David Kotz, Andrew Campbell (Dartmouth),
Nitin Vaidya, Jennifer Hou (UIUC),
Ness Schroff, Sonia Fahmy (Purdue), Mario Gerla, Medy
Sanadidi (UCLA),
Tracy Camp (Colorado School of
Mines), David Wetherall (U. Washington), Victor Bahl (Microsoft
Research), Ed Knightly, David Johnson (Rice),
Rene Cruz (UCSD), Maria Papadopouli, Kevin Jeffay (U North Carolina),
Henning Schulzrinne
(Columbia), Azer Bestavros, Ibrahim Matta (Boston U), Dina Katabi (MIT),
Stefano Basagni (Northeastern U.), Michele Zorzi (U. Padova/UCSD), Eylem
Ekici (Ohio State U),
Jim Kurose, Brian Levine (U. Mass - Amherst), Srikanth Krishnamurthy,
Michalis Faloutsos (UC Riverside)
This material is based upon work supported in part by the National Science
Foundation under Grant No. 0134650.
Any opinions, findings and conclusions or recomendations expressed in this
material are those of the author(s) and do not necessarily reflect the
views of the National Science Foundation (NSF).
Privacy: Anonymization techniques have been used to remove information that may help identify MAC addresses
of devices. No private information is contained within these traces. The trace collection process was conducted
in adherance with the code of the corresponding universities, and proper permissions were granted as/when needed.
Number of visitors since Jul. 18 2005:
This page was created May '05 at USC. Moved to UF 2007. Updated intermittently, with the latest update 2011...
New, on-going updates, Dec. 2017... stay tuned!
[Apr 2013] This page has been translated by students/volunteers in the Czech Republic to make it available for
researchers and scientists there.
- Note: If you use any of these tools, code or data, please make sure to cite our correspnding paper(s)! Thanks!