Exploiting Sociological Orbits in Mobile Users

1
Exploiting Sociological Orbits in Mobile Users
Mobility Pattern

• Chunming Qiao
• http//www.cse.buffalo.edu/qiao
• LANDER
• University at Buffalo (SUNY)

2
The World of Mobility

• Deterministic (controllable or not)
• e.g., planets, satellites, robots
• research communications (MANET and DTN), data
  collection, autonomous and cooperative protocols
• Random (hard to predict)
• e.g., air/water pollutant (and micro-scale,
  fine-granular mobility)
• research sensing/detection, tracking
• Semi-deterministic (somewhat predictable)
• Wireless/mobile users, or user carried/operated
  devices (including cars)
• research localizations/directions,
  communications, social applications

3

• Mobile Users
• influenced by social routines
• visit a few hubs / places (outdoor/indoor)
  regularly
• orbit around (fine to coarse grained) hubs at
  several levels

4
Mobility Profiles

• A profile is a probabilistic list of hubs a
  user visits during a given time window (e.g., a
  day)
• E.g. P1A0.7, B0.5 and P2B0.9, C0.6
• A user may have a few profiles, each associated
  with a weight (probability)
• Different from AP-centric views (e.g., which AP
  gets visited most often and with what
  distribution)
• Can/should take adv of profiles to provide
  better, localization, routing and more
  personalized services
• Different from simple statistics (e.g., based on
  histograms), e.g. visited hub A 4 times, hub B 8
  times and hub C 3 times no correlation here!
• Profiling results in better prediction accuracy
  (gt20)
• Different from continuous tracking
• Profiles require infrequent updates/changes

5
Mobility Traces Analyzed

• Dartmouth traces of WLAN users
• Duration of 21 months from 4/1/2001 till
  12/31/2002
• 18,319 wireless users, 603 APs, 179 buildings
• Grouped users into 9 groups based on degree of
  activity
• Selected top two active users from each group
• ETH Zurich traces of WLAN users
• Duration of 1 year from 4/1/04 till 3/31/05
• 13,620 wireless users, 391 APs, 43 buildings
• Grouped users into 6 groups based on degree of
  activity
• Selected one sample (most active) user from each
  group

6
Profiling illustration
Translate to binary hub visitation vectors
Apply clustering algorithm to find mixture of
profiles
Obtain daily hub stay durations
7
Profile parameters for all sample users
8
Applications of Orbital Mobility Profiles

• Location Predictions and Routing within MANET and
  ICMAN/DTN ? route to destination hubs with
  storage devices or other users (SOLAR)
• Impact on social science ? study where people go
  and who do they meet, and acquaintance based soft
  location management (ABSoLoM)
• Customizable traffic/event alerts ? alert only
  the individuals who might be affected by a
  specific traffic/event condition in certain
  locations (MoPADS)
• Environmental/health monitoring/targeted
  inspection ? identify travelers who can relay
  data from remote locations with no APs, or who
  poses threat
• Anomaly based intrusion detection ? unexpected
  movement (in time or space) sets off an alarm
• Generate mobility traces based on orbital profiles

9
Summary

• Need a user-centric view (in addition or in
  place of to system or data centric views) and
  consider social-relevance in wireless/mobile
  networks
• Mobility profiles are useful in designing
  scalable networks with or without infrastructures
  (including MANET, DTN, VANET) as well as in
  integrated networks (e.g. iCAR)
• Research issues include user mobility trace
  collection, mobility profiling, profile-aware
  trace generation, profile access/sharing, and
  profile aided localization, routing and other
  applications/services

10
Select Publications on Mobility Related Papers

• A. Khan, C. Qiao, P. Sharma and S. Tripathi, An
  Energy-Efficient Mobile Triangulation-based
  Coverage Scheme" in ICC'07
• S. Yoon, C. Qiao, A New Search Algorithm using
  Autonomous and Cooperative Multiple Sensor
  Nodes, Infocom07
• J. Ghosh, H. Q. Ngo, S. Yoon, C. Qiao, "On a
  Routing Problem within Probabilistic Graphs,
  Infocom '07
• J. Ghosh, S J. Philip, C. Qiao, "SOLAR in MANET,
  Ad Hoc Networks, Mar. 2007 (previously a ACM
  MobiHoc05 poster)
• J. Ghosh, C. Westphal, H. Q. Ngo, C. Qiao,
  "Bridging Intermittently Connected Mobile Ad hoc
  Networks (ICMAN) with Sociological Orbits,
  Infocom '06 poster
• S.J. Philip, J. Ghosh, C. Qiao, "Performance
  Evaluation of Multilevel Hierarchical Location
  Management for Ad Hoc Networks, Computer
  Communications, June 2005
• J. Ghosh, S. J. Philip, C. Qiao, "Acquaintance
  Based Soft Location Management in MANET, IEEE
  WCNC 2004 (March)
• J. Ghosh, S. Yoon, H. Q. Ngo, C. Qiao, "On
  Profiling Mobility and Predicting Locations of
  Wireless Users, IEEE JSAC (under review)
  previously in ACM MobiHoc '06 REALMAN workshop
• S. Yoon, D.T. Ha, H.Q. Ngo and C. Qiao, MOPADS
  A Mobility Profile Aided File Downloading Service
  in Vehicular Networks, Infocom08 (under review)
  previously in Infocom07 MOVE workshop

Submitted