Location Without GPS

1
Location Without GPS

• John KrummMicrosoft ResearchRedmond,
  Washington, USA

2
Location
3
Importance of Location

• Find your way
• Find nearby things
• Invoke location-based services
• Electronic graffiti, e.g. There is a better
  Mexican restaurant 0.2 km north of here.
• List of nearby events
• Part of context
• In lecture hall ? cell phone off
• At home ? use home network

4
IWMMS Location
Study of Structuring and Recalling Life Log
Experience Using Location Information, Y.
Aihara, R. Ueoka, K. Hirota and M. Hirose --
Already using location for activity inference
Active Wearable Vision Sensor Recognition of
Human Activities, K. Sumi, M. Toda, S. Tsukizawa
and T. Matsuyama Cooperative Dialogue Planning
with User and Situation Models via Example-based
Training, I. R. Lane, S. Ueno and T.
Kawahara -- Inferring context of user location
is part of context
A Hybrid Dynamical System for Event
Segmentation, Learning, and Recognition, H.
Kawashima, K. Tsutsumi and T. Matsuyama Time-Ser
ies Human-Motion Analysis with Kernels derived
from Learned Switching Linear Dynamics, T. Mori,
M. Shimosaka, T. Harada and T. Sato -- Apply
HDS/SLDS to infer location mode of
transportation destination?
5
Why Not Use GPS?

• Does not work indoors
• Needs view of satellites

6
Location Sensing
Hazas, Scott, Krumm, Location-Aware Computing,
IEEE Computer Magazine, February 2004.
7
Outline

• Introduction
• LOCADIO Wi-Fi triangulation
• NearMe Wi-Fi proximity
• RightSPOT FM radio triangulation
• TempIO Inside/outside from temperature

8
Location from 802.11 with LOCADIO
with Eric Horvitz
Wi-Fi (802.11) access point

• Mobile device measures signal strengths from
  Wi-Fi access points
• Computes its own location

Location from Radio
9
LOCADIO Radio Survey
Radio survey to get signal strength as a function
of position
10
LOCADIO - Constraints
Make the client as smart as possible to reduce
calibration effort
No passing through walls
No speeding
We know when you move
11
LOCADIO - Results
Hidden Markov model gives median error of 1.53
meters
12
Outline

• Introduction
• LOCADIO Wi-Fi triangulation
• NearMe Wi-Fi proximity
• RightSPOT FM radio triangulation
• TempIO Inside/outside from temperature

13
NearMe
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
with Ken Hinckley
Find people and things nearby
printers
people
reception desk
bathroom
conferencerooms
14
The Basic Idea
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
802.11 Wi-Fi access point
NearMe Proximity Server
15
Location vs. Proximity
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
x1 (x,y) location
x2 (x,y) location
d12 f(x1, x2)
s1 measured signals
s2 measured signals
16
NearMe Client
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
PocketPC 2003
Windows XP

• Requirements
• Windows XP
• WWW access
• Microsoft .NET Framework

17
NearMe Client Test Connections
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
18
NearMe Client Register
Download from http//research.microsoft.com/jckru
mm/NearMe.htm

• Register with
• Name
• Email (optional)
• URL (optional)
• Expiration interval

19
NearMe Client Report Wi-Fi
Download from http//research.microsoft.com/jckru
mm/NearMe.htm

• List of detectable Wi-Fi access points
• Access points used only as beacons
• Periodic reports for mobility

20
NearMe Client -- Query
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
Adjustable Look back time to filter outdated
reports
21
NearMe Client Nearby Things
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
22
Simple Distance Function
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
d -2.53nn 2.90?s - 22.31
rms error 14.04 meters ?s 0.39
23
Access Point Layout
Download from http//research.microsoft.com/jckru
mm/NearMe.htm
1
2
A
F
3
24
NearMe Demo
25
Outline

• Introduction
• LOCADIO Wi-Fi triangulation
• NearMe Wi-Fi proximity
• RightSPOT FM radio triangulation
• TempIO Inside/outside from temperature

26
SPOT Watch Location
with Adel Youssef, Ed Miller, Gerry Cermak, Eric
Horvitz
traffic
weather
dining
movies
Commercial FM transmit new data every 2 minutes
Filter on watch to take what it wants
Watch displays personalized data
27
Location-Sensitive Features

• Nice to have
• Local traffic
• Nearby movie times
• Nearby restaurants

Need to know location of device
28
Use FM Radio Signal Strengths
Scan signal strengths of 32 FM radio stations at
1 Hz
29
Ranking Approach
Redmond KPLU lt KMTT lt KMPS Bellevue KMTT lt KPLU
lt KMPS Issaquah KMTT lt KMPS lt KPLU
Any monotonically increasing function of signal
strength preserves ranking
Measured Power
N radio stations ? N! possible rankings

• A B C
• A C B
• B A C
• B C A
• C A B
• C B A

A
B
C
Input Power

• Each watch scales signal strengths differently
• Impractical to calibrate every watch

30
Test
Six suburbs and six radio stations
81.7 correct from 8 radio stations
31
Avoid Manual Training
Seattle
KMPS 94.1 MHz
KSER 90.7 MHz
32
Classify Into Grid Cell

• Find location in grid
• Use predicted signal strengths to avoid manual
  training

8 kilometers average error
Summer intern Adel Youssef, U. Maryland
33
Outline

• Introduction
• LOCADIO Wi-Fi triangulation
• NearMe Wi-Fi proximity
• RightSPOT FM radio triangulation
• TempIO Inside/outside from temperature

34
TempIO Inside/Outside Classification
with Ramaswamy Hariharan
Suunto X9 GPS, altimeter, thermometer

• Are you inside or outside?
• Turn off GPS if inside to save batteries
• Metadata for digital photos
• Higher-level context reasoning

Suunto N3 SPOT watch, knows outside
temperature, location
Bayes Net
35
World Weather Stations
6509 weather stations ? http//weather.noaa.gov/we
ather/metar.shtml ? our web service
36
Inside/Outside from Temperature
Kyoto

• From hourly temperature data in five US cities,
  2003
• Average correct 81

37
The End