indoor outdoor
TRANSCRIPT
Indoor-Outdoor
Positioning and Lifelog
Experiment with
Mobile Phones
Reference
Hiroshi Mizuno, Ken Sasaki, Hiroshi Hosaka “Indoor-
Outdoor Positioning and Lifelog Experiment with
Mobile Phones” Proceedings of the 2007 workshop on
Multimodal interfaces in semantic interaction
Outline
Introduction
System Architecture
Lifelog Experiment
Conclusion
Introduction
With the rapid development of computers and networks,
there are many projects on lifelog
StartleCam: A Cybernetic Wearable Camera
Time-machine Computing: a Time-centric Approach for the
Information Environment
MyLifeBits: Fulfilling the Memex Vision
Analysis on lifelog data and some useful parameters for
predicting user’s next behavior are presented
System Architecture
Outdoor Positioning with GPS
Bluetooth Indoor Positioning System
Behavior Recording Software
Outdoor Positioning with GPS
Mobile phones’ GPS are used to track users’ positions in
outdoors
A user starts to find his/her current location, the phone
connects to a “recording server”, via the Internet
The recording server works as a web server ,it returns
the time at which the phone should send the next
positioning data
Outdoor Positioning with GPS
Bluetooth Indoor Positioning
System
Personal computers (PCs) in rooms and offices will
serve as base stations
When user’s Bluetooth signal from the mobile phone is
detected, the base stations make connections with the
user’s phone and measure the signal strength
Accuracy of positioning depends on how cluttered the
environment is
In ordinary office buildings, this system can identify
the room in which the user is present
Bluetooth Indoor Positioning
System
Behavior Recording Software
Behavior Recording Software is a Java program that
runs on mobile phones
Allows the user to record activities or input
supplementary data
When a user pushes a button, a menu of common daily
activities appears on the screen to prompt the user
History of the past inputs is shown at the bottom of the
screen for check and correction
Behavior Recording Software
Lifelog Experiment (1/5)
76 consecutive days
A college student
User’s activities were classified into five categories:
Sleeping
Working on a PC
Reading
Taking a bath
Went out
To analyze user’s lifestyle and to find parameters for predicting user’s next activity
Lifelog Experiment (2/5)
Lifelog Experiment (3/5)
Lifelog Experiment (4/5)
“Sleeping” and “Went out” are repetitive at period of 24 hours
“Working on PC”, “Reading”, and “Taking a bath” has no regular time interval
“Working on PC” is random and the time spent on this activity is relatively long
“Reading” is not a daily activity of this person
Interval of “Taking a bath” is longer than 12 hours
Finding: the activities are related with the following three factors:
Current time
Time from wake up
current activity
Lifelog Experiment (5/5)
Using a feature distance defined by equation (1) and
applying k-nearest neighbor method, we were able to
predict the user’s next activity with probability of 66%
Conclusion
Indoor-outdoor positioning system incorporating GPS
and Bluetooth has been developed for lifelog system
using mobile phones
Lifelog experiment of 76 days has proved the usability
of the system