Place Lab: Bootstrapping Whereware

1
Place LabBootstrapping Where-ware

• A. LaMarca, B. Schilit, A. Balachandran
• Intel Research Seattle
• V. Iverson
• Intel
• G. Borriello, E. Cadag, D. McDonald, J. Tabert,
  E. Lazowska
• University of Washington
• W. G. Griswold
• UC San Diego
• J. Hong
• UC Berkeley

2
Location, Location, Location

• Ubicomp requires context
• Location is important to context
• Many ways to track objects and people
• Infrared proximity
• Ultrasonic time-of-flight
• Radio signal strength
• Radio time-of-flight

3
Past Location Results
Motionstar .001 m
MSR RADAR 2 m
Active Bat .03 m
Cost per ft2
MIT Cricket .2 m
Olivetti Active badge 4 m
E911 300 m
GPS 3 m
Place Lab
Cost per User
4
Riding the WiFi Wave

• WiFi density increasing to point of overlap
• WiFi base stations broadcast unique IDs
• With a map of BSSIDs-to-locations any WiFi device
  can locate itself

5
Making Ubicomp Ubiquitous
Little investment in infrastructure
6
Research Challenges

• Build a global database of WiFi hotspots
• Move from location to place
• Associate web pages with places (aka geo-coding
  the web)
• Create trust model acceptable to user and AP
  owners

7
Challenge Bootstrapping a global RF Database

• Ekahau offline entry and offline refinement
• Single centralized authority
• Active Campus offline entry and online
  refinement
• Users can correct location for themselves
• Place Lab online entry and online refinement
• a distributed contributor model
• (ala IMDB, CDDB, Amazons ratings)

8
Research Opportunities

• Building viable reputation-based scheme
• Moving beyond war driving
• Using inference to refine beacon positions
• Data management challenges
• Central authority, P2P system or hybrid?
• Capacity issues on phones/PDAs
• Trading privacy for capacity via regional
  prefetching
• Cross-device P2P database sharing

9
Challenge Moving from location to place
Hey Joe! Youll never guess who I bumped into at
47.232434, -122.454563!

• Meaningful place names greatly improve services
• Multiple hierarchies
• Geographic, operational, organizational,
  legal/ownership, activity
• Could be user-contributed
• Opportunity for inferencing
• Purely location based
• Other information like content consumed at a
  location

10
Challenge Geo-Coding the Web

• Inferring locations relevant to existing content
• Content itself can be analyzed
• Can also observe where content is accessed from
• Keeping location from becoming another SPAM
  vehicle
• Develop location-aware web services
• Form fill-in, navigation, blogs, diaries,
  scrapbooks, ad-hoc communities, etc.
• Proactive services that learn from user

11
Challenge Creating Trust

• People dont trust location technologies

12
Protecting the Client

• What am I revealing when I use this system?
• Part HCI, part privacy problem
• Client-side positioning 75 of the battle
• Clients reveal only what they want to
• Most interesting services have a non-local
  component
• Multi-grain blurring
• Send location on case by case by service/host
• Utilize multiple grains country, state, city,
  zip, street, lat/long
• Grain negotiation between hosts and clients

13
What are we revealing about APs

• We want to answer Here is an AP, where am I?
• DB can also answer Where is a nearby AP?
• Naïve users dont configure their APs.
• Dont want to enable service theft and hacking
• Encrypt APs location
• Encrypt location data using BSSID as key
• Cant determine APs location unless you are
  actually there

14
Engage Academic Community
Research Faculty
Web programming courses
Students developers and users
15
Engaging Others UbiComp 2003 Demo

• Client-side conference companion application
• Hotspots images, commentary, web content
• Available to conference attendees
• Demonstration and an invitation

16
Summary
Place Lab is a grassroots effort to create a
global indoor/outdoor positioning system with a
low barrier to participation

• Opportunities for research
• Hotspot database development
• Geo-coding web content
• Privacy models and interfaces
• Location aware services

http//placelab.org
17
The end
18
Multi-grain Privacy Mechanism
When I use this what am I revealing about my
location?
PLACE BAR Zip Revealed ?

• Place Lab Google like toolbar for revealing
  location data zip, city, state, etc.
• A location privacy firewall on your notebook
• First time a particular location-aware page is
  referenced the user might choose to accept the
  requested level or provide a blurred location

19
Challenge Create Global RF Database

• Many previous example of distributed contributor
• IMDB
• Amazons product ratings
• CDDB
• People entered 1.8 million CDs by hand!