Sensors
and
Actuators
A
201 (2013) 86–
92
Contents
lists
available
at
ScienceDirect
Sensors
and
Actuators
A:
Physical
jo
u
r
n
al
homep
age:
www.elsevier.com/locate/sna
Signal
processing
for
a
positioning
system
with
binary
sensory
outputs
Long
Cheng
a,∗
,
Yan
Wang
a
,
Chengdong
Wu
a
,
Hao
Wu
b
,
Yunzhou
Zhang
a
a
College
of
Information
Science
and
Engineering,
Northeastern
University,
Shenyang
110819,
China
b
Engineering
Faculty,
University
of
Sydney,
NSW,
Australia
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
20
February
2013
Received
in
revised
form
19
June
2013
Accepted
19
June
2013
Available online 12 July 2013
Keywords:
Binary
sensor
Acoustic
Multiple
sources
Localization
Fuzzy
c-mean
a
b
s
t
r
a
c
t
Energy
based
localization
method
for
source
localization
has
attracted
considerable
attention
in
recent
years.
The
binary
sensor
is
a
low-power
and
bandwidth-efficient
solution
for
positioning
system
since
the
node
provides
only
binary
information
about
the
sources.
Most
of
the
existing
acoustic
source
localization
methods
assume
there
only
exists
a
single
source
and
may
fail
with
the
presence
of
multiple
sources.
In
contrast,
we
directly
consider
the
case
when
the
node
is
influenced
by
multiple
sources.
A
maximum
likelihood
estimation
method
is
applicable
to
the
multiple
sources
localization
problems,
but
is
at
the
cost
of
high
computational
complexity.
To
tackle
these
problems,
this
paper
proposes
a
novel
likelihood
matrix
based
multiple
acoustic
sources
localization
algorithm
for
binary
sensor.
The
fuzzy
c-means
algo-
rithm
is
firstly
employed
to
calculate
the
membership
degrees
of
the
alarmed
nodes
associated
with
the
sources,
and
then
a
likelihood
matrix
is
proposed
for
multiple
acoustic
sources
localization.
Simulation
and
experiment
results
show
that
the
proposed
algorithm
provides
accurate
location
estimations.
© 2013 Elsevier B.V. All rights reserved.
1.
Introduction
Due
to
the
availability
of
low
cost
and
energy
efficient
sen-
sors,
micro-processors
and
radio
frequency
circuits
for
information
transmission,
there
is
a
rapid
development
of
wireless
sensor
net-
work
(WSN).
WSN
is
composed
of
a
large
number
of
inexpensive
nodes
which
are
densely
deployed
in
a
region
of
interests
to
mea-
sure
certain
quantity.
WSN
have
received
significant
attention
due
to
its
potential
applications
such
as
health
surveillance,
battle
field
surveillance
and
environmental
monitoring
[1].
The
source
localization
is
one
of
the
key
techniques
in
WSN.
There
are
several
ways
to
estimate
the
source
location:
energy-
based
[2],
angle
of
arrival
(AOA)
[3],
time
difference
of
arrival
(TDOA)
[4].
As
an
inexpensive
approach,
energy-based
method
is
an
attractive
method
because
it
requires
low
hardware
config-
uration.
In
this
paper,
we
investigate
the
energy-based
sources
localization
method.
The
source
localization
methods
have
a
wide
range
of
possible
applications.
Applications
include
vehicle
or
air-
craft
localization
in
outdoor
environments,
indoor
human
speaker
localization
and
sea
animal
or
ship
localization
in
underwater
envi-
ronments.
The
source
localization
can
be
categorized
as:
single
source
localization
and
multiple
sources
localization
[5].
For
single
source
localization:
since
the
objective
function
of
single
source
∗
Corresponding
author.
Tel.:
+86
15940207611;
fax:
+86
02483687761.
E-mail
address:
chenglong2000
0@yahoo.com.cn
(L.
Cheng).
localization
method
has
multiple
local
optima
and
saddle
points
[6],
the
authors
formulate
the
problem
as
a
convex
feasibility
problem
and
propose
a
distributed
version
of
the
projection
onto
convex
sets
method.
A
weighted
direct/one-step
least
squares
based
algorithm
is
investigated
in
[7]
to
reduce
the
computational
complexity.
In
comparison
with
quadratic
elimination
method,
these
methods
are
amenable
to
a
correction
technique
which
incorporates
the
dependence
of
unknown
parameters
leading
to
further
performance
gains.
In
[8],
the
authors
propose
normalized
incremental
sub-gradient
algorithm
to
solve
the
energy
based
sensor
network
source
localization
problem
while
the
selection
of
the
decay
factor
in
this
method
is
still
an
unsolved
problem.
For
multiple
sources
localization:
A
maximum
likelihood
esti-
mator
[9]
is
used
for
the
multiple
source
localization.
An
efficient
expectation
maximization
algorithm
[10]
is
proposed
to
improve
the
estimation
accuracy
and
to
avoid
trapping
into
local
optima
through
the
effective
sequential
dominant-source
initialization
and
incremental
search
schemes.
An
alternating
projection
[11]
algorithm
is
proposed
to
decompose
the
multiple
source
localiza-
tion
into
a
number
of
simpler,
yet
also
non-convex,
optimization
steps.
This
method
decreases
the
computation
complexity.
And
an
optimal
parametric
maximum
likelihood
solution
[12]
to
locate
wideband
sources
in
the
near
field
is
proposed.
However,
this
method
turns
to
the
solution
of
a
nonlinear
optimization
problem
and
thus
is
still
hard
to
solve.
A
robust
expectation
maximiza-
tion
algorithm
[13]
based
on
the
assumption
that
the
sources
are
corrupted
by
the
noises
with
non-uniform
variances
is
pro-
posed.
This
algorithm
is
much
more
computa
tionally
efficient
0924-4247/$
–
see
front
matter ©
2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.sna.2013.06.020