Providing Location Abstract Information in a Ubiquitous Computing Environment

1
Providing LocationAbstract Information in
a Ubiquitous Computing Environment

• Mike Spreitzer and Marvin TheimerXerox Palo Alto

2
Problem

• How do we keep applications useful and
  efficient while protecting users privacy needs?
• Solution
• Develop an architecture that allows users to
  control how much to reveal/hide about their
  identity and location depending on the situation

3
Location based applications

• Visitor Guidance
• Migrating windows
• Note distribution
• Ubiquitous message distribution (UMD)
• Media Call
• Scoreboard
• Responsive environment
• FindNearest
• Locations

4
Assumptions/ key issues

• The system will span more than one administrative
  domain
• Cannot trust all components equally
• Cannot keep all users location info in one
  central database
• Possibility of traffic analysis attempts
• Users priorities change based on current context
  
• Privacy compromises
• Better privacy gt poorer functionality

5
Architecture

• Components
• User Agents
• Location Query Service (LQS)
• LocationBroker
• Terminal agent
• Decentralized increased overhead

6
Architecture
7
Design Principles

• Design should scale well
• Users should have the ability choice to be
  invisible to sensors
• Personal info should not be required to be stored
  on servers
• Information should be encrypted and allow for
  anonymous handles

8
Tracking Systems

• Focused on UMD and Locations program
• Problems user not always near display or out of
  range
• Recipient context important
• Message delivery alerts intrusive
• Active badge system
• Tracks location/movement of user
• Computer Input system
• Results of recent activity forwarded to user
  agents
• Variations due to different work patterns
• Finding The two systems rarely overlap, seem to
  complement each other instead

9
Issues

• Need a way to negotiate when both querier and
  responder refuse to reveal info
• Knowing which sensor systems are at a given
  location
• How do we prevent privacy from being compromised?
• How accurate does location information need to
  be?
• How many interruptions are people willing to deal
  with?

10
Conclusion

• Not all location-based sensors can be used if
  users can hide
• Uncertainty is a given
• More research needs to be done to reveal how
  users actually use privacy controls
• Future considerations adding multicast to the
  system

11
Discussion Questions

• How slow is this querying process?
• Why bother having applications such as
  FindNearest if people can hide from it?
• Who decides the visibility of shared objects?
• Is the technology smart enough to learn user
  preferences and apply them automatically over
  time?