Sie sind auf Seite 1von 14

1

1
Shinsuke Hara, Ph. D.
Gradate School of Engineering, Osaka City University
11/August/2009
hara@info.eng.osaka-cu.ac.jp
Localization-1
2
Contents of Shins seminar
Whats localization?
Why localization now?
Context-aware services
Location-based services
Importance of localization
Category
Principle
Performance bounds
3
Whats localization?
To estimate the 2-dimensional or 3-dimensional
location of a target, such as a mobile user and node
Location estimation
Positioning
Localization (technical term in robotics)
Ranging is to estimate (measure) the distance
between a pair of transceivers
Usually, localization is based on multiple ranging
with different known locations, such as base
stations and anchor nodes
4
Assumptions for localization
Some nodes are available whose locations are
known in advance, which are called reference
nodes or anchor nodes. In this seminar, we call
them anchor nodes.
From a target node, anchor nodes can get some
information related to the distance to the target
node
Examples are: TOA (time of arrival), TDOA (time
difference of arrival), AOA (angle of arrival), RSSI
(received signal strength indication, received
power) and so on
2
5
Anchor
node #3
Anchor
node #1
Anchor
node #4
Target node
Anchor
node #2
t
1
=g(d
1
)
TOA (time of arrival)
t
2
=g(d
2
)
t
3
=g(d
3
)
t
4
=g(d
4
)
6
How can we measure TOA? (1)
One-Way Ranging (synchronization, the same
clock among nodes)
time
time
0
0
Anchor node
Target node
t
d=ct
c: speed of light (3.0x10
8
m/s)
Establishment of the synchronization is very difficult!
7
How can we measure TOA? (2)
Round trip time measurement
time
time
0
Anchor node
Target node
t t
T
d
(known in advance)
T
R
t =(T
R
-T
d
)/2
d=ct
8 t t
At
0
1
0
1

Anchor
node
(A)
Target
node
(B)
T
T
A
T
T
A
t
T
R
B
T
T
B
T
T
B
T
R
B
T
R
A
How can we measure TOA? (3)
Two-Way Ranging
3
9
T
T
A
+t=T
R
B
+At
T
R
A
-At=T
T
B
+t
2
) ( ) (
B
R
B
T
A
T
A
R
p
T T T T
t

=
TWR does not need synchronization !
How can we measure TOA? (4)
10
Anchor
node #3
Anchor
node #1
Anchor
node #4
Target node
Anchor
node #2
TDOA (time difference of arrival)
A
21
=t
2
-t
1 A
31
=t
3
-t
1
A
41
=t
4
-t
1
11
How can we measure TDOA?
Anchor nodes are synchronized and can talk
with each other
time
time
0
0
Anchor node 1
Anchor node 2
time
t
1
t
2
A
21
=t
2
-t
1
Target node
12
Anchor
node #3
Anchor
node #1 Anchor
node #4
Target node
Anchor
node #2
AOA (angle of arrival)
u
2
u
1
u
3
u
4
4
13
How can we measure AOA?
A target node or anchor nodes need to have
array antenna
14
Pros and cons of TOA, TDOA and AOA
Their performance has been long believed to be
much better (than RSSI)
They do not work in non-line-of-sight (NLOS)
environments (tall buildings and people walking
around)
TWR or synchronization is required for TOA
Array antenna is required for AOA
15
Anchor
node #3
Anchor
node #1
Anchor
node #4
Target node
Anchor
node #2
P
1
=f(d
1
)
RSSI (received signal strength indication)
P
2
=f(d
2
)
P
3
=f(d
3
)
P
4
=f(d
4
)
16
RSSI is easily measurable
Cellular
phones
Wireless LAN
terminals
Many current wireless communication standards
support RSSI measurement
IEEE 802.15.4
LQI (link quality indicator)
8bit resolution,
-173dBm to 82dBm
(1dB step)
5
17
Pros and cons of RSSI
Its performance has been long believed to be
much worse (than TOA), because of fading
and shadowing
They can work even in NLOS environments
It is simply implementable
OWR is easily possible
18
Why localization now?
What are the paradigms of next generation mobile
communication systems?
1G system: analogue (1980s)
2G system: digital (1990s)
3G system: multimedia (2000s)
4G system: heterogeneous network & system
and context-aware services
5G system: green, learning,
The word of context means the overall situation in which an event
occurs (The American Heritage Dictionary)
19
NGN
ADSL
WLAN
WiMAX
CPU capability
Traffic load condition
Channel bandwidth
Indoor
Outdoor
In motion
Stationary
3G
Context in communications means the overall situation
surrounding any parts of the entities in a communication system
Context-aware services
20
What is an essential difference of wireless from wired?
End device (terminal) can move around
The location is worth estimating
Services with the location information are unique
only for wireless
Location-based services
6
21
Location-based services (1/3)
Switch your cell
phones to sleep
mode
My cell phone is
automatically
switched to sleep
mode
My cell phone is
automatically switched
to vibration mode
when getting on a
silence cart
(a) School (b) Transportation
22
You are approaching to
a good French restaurant
My cell phone is
automatically
switched off when
entering a theater
(c) Navigation and
advertisement
(d) Theater
Location-based services (2/3)
23
(e) Resource management
My current base station is busy with
handling a lot of traffic, but my next
base station which I can reach in a few
seconds is not busy. So the system
has just decided not to assign a channel
with lower data rate to me in the current
cell and to assign channel with much
higher data rate in the next cell
Location-based services (3/3)
(f) Advice
24
Importance of localization (1/4)
GPS (Global Positioning System)
Cellular (3G, 4G)
WMAN (WiMAX)
WLAN (WiFi)
WPAN (Bluetooth)
WSN (Zigbee, UWB)
RF-ID (Passive, Active)
Providing a variety of location-based services depends on
the accuracy of the estimated location of a target (node, cell
phone and so on)
To localize a target more accurately, we can use any kinds
of wireless systems
WiMAX
GPS
RF-ID
WiFi
Zigbee
Bluetooth
7
25
[1] G. Y. Delisle, Location Awareness and Positioning Methods for Location
Based Services in Wireless Communication Networks, IEEE VTC 2006-Fall.
From always best connected to always best
located [1]
Importance of localization (2/4)
References
26
Importance of localization (3/4)
Wireless Enhanced 911 (FCC Third Order)
Base station-based localization: error<100m (call>67%)
error<300m (call>95%)
GPS-based localization: error<50m (call>67%)
error<150m (call>95%)
All terminals are not equipped with GPS receivers
Cellular base stations and GPS satellites are not available
in indoor environments
27
Importance of localization (4/4)
Increasing number of sessions in international conferences
Increasing number of special issues in international journals
Specialized conferences
Specialized projects
MELT 2008 (The 1st ACM International Workshop on
Mobile Entity Localization and Tracking in GPS-less
Environments), San Francisco, CA, USA, 19 Sep. 2008
WPNC 2009 (The 6th Workshop on Positioning,
Navigation and Communication, Leibniz, Germany, 19
Mar. 2009
WHERE (Wireless Hybrid Enhanced Mobile Radio
Estimators), FP7-ICT-2007-1, 01.01.2008-30.06.2010,
5.551 Million Euros
28
Category
Range-free
APS [1]
Range-based
Deterministic (fingerprinting)
Probabilistic
RADAR [2]
Parametric
Non-parametric
ML, LS, MAP
BP-iterative
BP [3]
[1] D.Niculescu and B.Nath, Ad hoc positioning system (APS), IEEE Globecom
2001, pp.2926 2931, Nov. 2001.
[2] P.Bahl and V.Padmanabhan, RADAR: An in-building RF-based user location
and tracking system, IEEE Infocom 2000, pp.775 - 784 Mar. 2000.
[3] A.T.Ihler, et al., Nonparametric belief propagation for self-localization of sensor
networks, IEEE JSAC, vol.23, no. 4, pp.809-819, Apr. 2005.
References
8
29
Range-free localization (APS)
Anchor #1
Anchor #2
Anchor #3
Anchor #4
Anchor #5
3 hops
3 hops
3 hops
4 hops
5 hops
Target
The location of a target is estimated with the known
locations of anchor nodes and the numbers of hops to
them
30
Problems of APS
Hop count does not correspond to physical
distance
The localization performance is very bad
They believe ranging is difficult (ranging function is
realized in current wireless communication
standards)
31
Deterministic localization (RADAR)
a b c d
Anchor #1
Anchor #2
Anchor #3
RSSI (received signal strength indication)
=received power
Database
[a, -20, -34, -50]
[b, -21, -32, -65]
[c, -34, -11, -45]

RSSI at Anchor #1
[?, -33, -15, -22]
Find the
best
matching
32
Problems of RADAR
Exhausted pre-measurement for making database
is required
Once the room layout is changed, the current
database becomes useless, so a new measurement
is required
To include the effect of moving objects into
database is difficult
9
33
Principle of localization
Target
node
Target
node
Mobile
terminal
Unknown
location
(x, y, z)
Anchor
node
Access
point
Base
station
Known
location
(x
n
, y
n
, z
n
)
WSN WLAN Cellular
N>k for the k-dimensional localization problem
n=1, , N
M
n
: measurement vector at the n-th anchor node for a target
M
n
=[M
n1
, , M
nL
], L: the number of measurements for the target
34
What is the measurement?
The measurement is related to the location of the target node,
(x, y, z), such as
TOA (Time Of Arrival)
TDOA (Time Difference Of Arrival) [1]
AOA (Angle Of Arrival) [2]
RSSI (Received Signal Strength Indication, received power)
Localization: to estimate the location of the target node,
|=(x, y, z)
Ranging: to estimate the distance to the target node
Note:
35
An example of TOA-based method (1/2)
2-dimensional localization: z=0, z
n
=0
(parameter) vector to be estimated: |=(x, y)
One-shot measurement: L=1
Measurement TOA
n
: M
n
=t
n
(the propagation time
between the mobile terminal and
the n-th base station
Distance:
n n
ct d =
c=3.0x10
8
m/sec
[1] N. Priyantha, et al., CRICKET location-support system, Proc. ACM MOBICOM,
Aug. 2003.
[2] H. Kawano and M. Kawano, Development of a pedestrian navigation system
working on the 2.4 GHz band, 11th World Congress on Intelligent Transport
Systems, Oct. 2004 (http://www.radio-com.jp/rcs_technology.html).
References
36
Base
station 4
Base
station 1
Base
station 2
Mobile terminal |=(x, y)
Base
station 3
d
1
d
3
d
2
d
4
NLOS
An example of TOA-based method (2/2)
Localization is based on several ranging from different
places (base stations)
NLOS: non-line-of-sight
10
37
How can we localize the target?
p
42
p
14
p
12
The target must be
located at the
weight of the
triangle p
42
-p
12
-p
14
38
ct
n
LS (Least Squares) estimation (1/2)
find which minimizes
2
1
| ) (
~
| ) (

=
| = |
N
n
n n
d d e
) , ( y x = |
LS estimation problem [1]
2 2
) ( ) (
n n
y x x x +
n-th true distance:
2 2
) ( ) ( )) , ( (
~
n n n
y y x x y x d + = = |
n-th square (ranging) error:
2
| ) (
~
| ) ( | = |
n n n
d d e
Sum of the square errors: ) ( ) (
1
| = |

=
N
n
n
e e
Note:
LS estimation is a nonlinear minimization problem
LS estimation does not care about the distribution of d
n
39
LS (Least Squares) estimation (2/2)
d
4
) (
~
4
| d
Base
station 4
2
4 4 4
| ) (
~
| ) ( | = | d d e
40
ML (Maximum Likelihood) estimation (1/2)
Assume that d
n
has a known conditional probability density
function (pdf) for a given :
) (
~
|
n
d
) | ( )) (
~
| ( | = |
n n n
d p d d p
The likelihood function on |=(x, y) is given by
[
=
| = |
N
n
n
d p l
1
) | ( ) (
11
41
find which maximizes ) , ( y x = |
ML estimation problem [1]
ML (Maximum Likelihood) estimation (2/2)

=
| = |
N
n
n
d p L
1
) | ( log ) (
Note:
ML estimation is a nonlinear minimization problem
ML estimation does not care about the distribution of |
Log-likelihood function on |
42
MAP (Maximum A Posteriori) estimation
Assume that | has a known pdf: p(|)
By Bayes theorem
|) |)
|) |)
= | ( | (
) (
( | (
) | ( p d p
d p
p d p
d p
n
n
n
n
find which maximizes ) , ( y x = |
MAP estimation problem [2]

=
, | = |
N
n
n
d p MAP
1
) ( log ) (
Note:
ML estimation is a nonlinear minimization problem
43
A Gaussian case
Assume that t
n
contains a Gaussian distributed error thus d
n
also contains a Gaussian distributed error [3]:
2
2
2
)) (
~
(
2
1
) (
~
| (
d
n n
d d
d
n n
e d d p
o
o t
|

= |)
find which maximizes ) , ( y x = |
ML estimation problem
ML
N
n d
n n
C
d d
L +
|
= |

=1
2
2
2
)) (
~
(
) (
o
Therefore,
find which minimizes ) , ( y x = |
) ( )) (
~
(
2
1
| = |

=
e d d
N
n
n n
LS estimation
problem !

=
=
|
|
.
|

\
|
= | = |
[
N
n
d
n n
d d N
d
N
n
n n
e d d p l
1
2
2
2
)) (
~
(
1
2
1
)) (
~
| ( ) (
o
o t
44
Concluding remarks
LS and ML estimations are classical frequentist approaches
whereas MAP is a Bayesian approach
The concept of MAP estimation is totally different from that of
ML estimation (The Bayesian approach gives a related but
conceptually different treatment of the parameter estimation
problem (LS and ML) [1])
Only for a Gaussian case, LS estimation becomes equivalent
to ML estimation
The localization performance depends on how accurately we
can know p(d
n
||) and p(|), that is, the channel propagation
characteristic and the system layout
[1] L. Ljung, System Identification: Theory for the User, Prentice-Hall, 1999.
[2] C.M. Bishop, Pattern Recognition and Machine Learning, Springer, 2006.
[3] N. Patwari, et al., Relative location estimation in wireless sensor networks,
IEEE Trans. Signal Process., vol.51, no.8, pp.2137-2148, Aug. 2003.
References
12
45
Performance bounds
Target node location: (x, y, 0) (unknown)
Unknown parameter: |=(x, y)
Anchor node location: (x
n
, y
n
, z
n
) (known) (n=1, , N)
Measurement (RSSI, TOA and so on): M
n
(L=1)
True distance between the target and n-th anchor node:
probability density function (pdf) of M
n
given :
Log-likelihood function on | for the n-th anchor node: L
n
(|)
2 2 2
) 0 ( ) ( ) ( ) (
~
n n n n
z y y x x d + + = |
)) (
~
| ( log ) ( | = |
n n n
d M p L
n
d
~
)) (
~
| ( |
n n
d M p
) (
~
|
n
d
46
Log-likelihood function on |: L(|)
Fishers information matrix is defined as [1]

= =
|) = | = |
N
n
n n
N
n
n
d M p L L
1 1
) (
~
| ( log ) ( ) (
(

=
(
(
(
(

|
c
c
|
c c
c
|
c c
c
|
c
c
= |
22 21
12 11
2
2 2
2
2
2
) ( ) (
) ( ) (
) (
J J
J J
L
y
L
x y
L
y x
L
x
E J
F
Taking into consideration:
0
) (
~
) (
=
(

| c
| c
n
n
d
L
E
Cramer-Rao lower bound for ML estimation (1/6)
47
We have
y
d
x
d
J J
y
d
J
x
d
J
n
N
n
n
n
N
n
n
n
N
n
n
n
c
| c
c
| c
| = =
|
|
.
|

\
|
c
| c
| =
|
|
.
|

\
|
c
| c
| =

=
=
=
) (
~
) (
~
) (
) (
~
) (
) (
~
) (
1
21 12
1
2
22
1
2
11

where
(

| c
| c
= |
2
2
) (
~
) (
) (
n
n
n
d
L
E
Cramer-Rao lower bound for ML estimation (2/6)
48
Defining the inverse matrix of J
F
as I
F
(|):
(

=
(

=
| = |

11 21
12 22
21 12 22 11 22 21
12 11
1
1
) ( ) (
J J
J J
J J J J I I
I I
J I
F F
the Cramer-Rao lower bound on the ML estimation error
variance is given by
2
1 1
2
1
2
1 1
2 2
22 11
2
) (
~
) (
~
) (
) (
~
) (
) (
~
) (
) (
~
) (
) (
~
) (
)] var( ) min[var( ) (

|
|
.
|

\
|
|

|
|
.
|

\
|
|

|
|
.
|

\
|
|

|
|
.
|

\
|
|

|
|
|
.
|

\
|
|

|
|
|
.
|

\
|
|

|
=
+ = + = |


= = =
= =
N
n
n
n
n
n
n
N
n
n
n
n
N
n
n
n
n
N
n
N
n
n
n
n
n
n
n
CRLB
d
y y
d
x x
d
y y
d
x x
d
y y
d
x x
I I y x


o
Cramer-Rao lower bound for ML estimation (3/6)
13
49
0
x
z
y
(x, y, 0)
(x
n
, y
n
, z
n
)
(x
n
, y
n
, 0)
q
n
u
n
From the target node to the n-th
anchor node, define the elevation
angle and the direction angle on the
x-y plane as q
n
and u
n
, respectively
n
d
~
Cramer-Rao lower bound for ML estimation (4/6)
50
The Cramer-Rao lower bound can be re-written as


= + =
=

| |
|
=
|
N
i
N
i j
j i
j i
j i
N
n
n
n
CRLB
1 1 2 2
2
1 2
2
cos cos
) ( sin
) ( ) (
cos
1
) (
) (
q q
u u

q

o
Note:
when the target and anchor nodes are on a line
(u
1
=u
2
=, , =u
N
)
+ = |) (
2
CRLB
o
Cramer-Rao lower bound for ML estimation (5/6)
51
For L measurements case, the Cramer-Rao lower bound
can be re-written as
1
1 1 2 2
2
1 2
2
cos cos
) ( sin
) ( ) (
cos
1
) (
) (

= + =
=

| |
|
=
|


L
L
N
i
N
i j
j i
j i
j i
N
n
n
n
CRLB
q q
u u

q

o
Cramer-Rao lower bound for ML estimation (6/6)
52
Assuming we know a priori knowledge on the location of
the target node as p(|), the priori information matrix is
calculated as
(
(
(
(

|
c
c
|
c c
c
|
c c
c
|
c
c
= |
) ( log ) ( log
) ( log ) ( log
) (
2
2 2
2
2
2
p
y
p
x y
p
y x
p
x
E J
P
Bayesian Cramer-Rao lower bound for MAP estimation (1/2)
14
53
the Bayesian Cramer-Rao lower bound on the MAP
estimation error variance is given by
Defining J
T
(|)=J
F
(|)+J
P
(|) and its inverse matrix as I
T
(|):
(

= | = |

22 21
12 11 1
' '
' '
) ( ) (
I I
I I
J I
T T
22 11
2
' ' )] var( ) min[var( ) ( I I y x
BCRLB
+ = + = | o
Note:
When no priori knowledge on the targets location is available,
the BCRLB (MAP estimation) becomes equivalent to the
CRLB (ML estimation), because J
P
(|)=0
[1] H.L.Van Trees, Detection, Estimation, and Modulation Theory, Part I, John
Wiley & Sons, 1968.
References
Bayesian Cramer-Rao lower bound for MAP estimation (2/2)

Das könnte Ihnen auch gefallen