TeamTrak: A Test Bed for Mobile Ad-Hoc Networks

1
TeamTrak A Test Bed forMobile Ad-Hoc Networks

• Hardware/software test bed to enable a variety of
  projects in wireless, mobile, social, and geo-
  computing.
• Hardware 32 tablet PCs plus with sensor helmet
  (GPS compass camera) and accelerometer on the
  foot.
• Software Collects sensor data, shares data with
  neighbors via multi-hop ad-hoc network over WiFi.
• TeamTrak allows us to explore concepts relevant
  to current and proposed mobile computing systems
• Cellular phones reporting sensor data.
• Mobile cartography data collection units.
• US Army Future Force Warrior.
• Our focus is on the algorithms, systems, and
  software, using simple commodity hardware.

2
USB Hub
Tablet PC
Garmin GPS-18
PNI V2Xe Compass
Watchport USB Camera
Pedometer (3-axis accel)
TeamTrak uses cheap commodity equipment and
software, so it is easy to swap in a higher
quality camera, newer PC, etc.
3
Research Challenges in TeamTrak

• Robust Navigation
• Problem GPS works fine on the open road, but is
  very inaccurate when obstructed by trees and
  buildings.
• Solution Share multiple sources of location data
  over the network to improve location quality
  e.g. peer GPS, pedometer, compass, fixed bases,
  (road signs?)
• Mining Mobile Social Networks
• Problem How do humans self-organize, share
  information? How do emergencies influence human
  behavior? What patterns can be inferred for an
  autonomic, dynamic, and reactive system?
• Solution Design efficient learning and
  predictive algorithms to discover community
  structures and anomalous. Integrate data
  collection, analysis and discovery into an
  action-oriented predictive framework.
• Managing Large-Scale Image Sets
• Problem It is very easy to acquire TB of image
  data, but it is much harder to store, manage, and
  explore it. Bottleneck is I/O bandwidth.
• Solution Employ massively parallel active
  storage clusters to archive, index, and search
  large datasets. Move small code to large data,
  instead of vice versa. Provide new languages for
  manipulation

4
People Involved in TeamTrak

• Prof. Douglas Thain
• Faculty in distributed systems and storage
  systems.
• Prof. Christian Poellabauer
• Faculty in mobile and real time systems.
• Prof. Nitesh Chawla
• Faculty in machine learning and data mining.
• Maj. Jeffrey Hemmes, USAF
• Ph.D student studying robust navigation.
• Rory Carmichael
• B.S. student working on testing and image
  acquisition.
• http//www.nd.edu/teamtrak