Int.
J.
Electron.
Commun.
(AEÜ)
67 (2013) 491–
499
Contents
lists
available
at
SciVerse
ScienceDirect
International
Journal
of
Electronics
and
Communications
(AEÜ)
jo
ur
n
al
homepage:
www.elsevier.com/lo
cate/aeue
A
novel
DOA
estimation
algorithm
of
cyclostationary
signal
based
on
UCA
in
impulsive
noise
Guo-hong
You, Tian-shuang
Qiu
∗
,
Jiao
Yang
Faculty
of
Electronic
Information
and
Electrical
Engineering,
Dalian
University
of
Technology,
Dalian
116024,
PR
China
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
29
May
2012
Accepted
15
November
2012
Keywords:
Co-channel
interference
Directions-of-arrival
estimation
Fractional
lower
order
cyclic
MUSIC
Uniform
circular
array
Impulsive-noise
a
b
s
t
r
a
c
t
Co-channel
interference
is
a
common
existence
in
communication
system,
which
will
result
in
the
deteri-
oration
for
directions-of-arrival
(DOAs)
estimation.
To
solve
this
problem,
a
fractional
lower
order
cyclic
MUSIC
method
suitable
for
uniform
circular
array
(UCA)
is
presented.
The
proposed
method
can
eliminate
both
impulsive-noise
and
interference
in
the
same
frequency
band,
and
it
also
performs
well
in
Gaussian
noise.
Especially,
the
algorithm
can
be
applied
to
either
narrow
or
broad-band
signal
models.
Simulation
results
strongly
verify
the
high
performance
of
this
method.
© 2012 Elsevier GmbH. All rights reserved.
1.
Introduction
The
uniform
circular
array
(UCA)
plays
a
significant
role
in
many
array
processing
applications
due
to
its
full
360
◦
azimuth
and
90
◦
elevation
angle
coverage.
Compared
with
uniform
lin-
ear
array
(ULA)
[1],
UCA
has
much
more
advantages
since
it
is
almost
omni-directional
in
azimuthal
direction
because
of
its
sym-
metric
geometric
configuration.
Many
direction-of-arrival
(DOA)
estimation
methods
have
been
proposed
that
can
be
applied
to
UCAs
composed
of
omni-directional
elements
[2–4].
These
clas-
sical
methods
exploit
the
eigenstructure
of
the
sample
correlation
matrix
of
received
signals
by
which
the
observation
space
can
be
partitioned
into
a
signal
subspace
and
a
noise
subspace.
Then
DOA
is
determined
based
on
the
orthogonality
of
the
signal
subspace
and
the
noise
subspace.
However,
all
of
these
methods
have
good
per-
formance
only
when
the
signals
are
independent
or
in
a
different
frequency
band
with
each
other
under
Gaussian
noise.
If
the
carrier
frequencies
are
close
in
co-channel
or
the
noise
is
much
impul-
sive,
the
resolution
performance
will
be
deteriorated
seriously
or
even
invalidate
completely.
This
is
because
noise
vector
will
fall
into
the
signal
subspace,
which
will
result
in
the
rank
loss
of
the
spatial
covariance
matrix.
Recently,
it
has
been
shown
that
noise
often
takes
on
apparent
nonstationarities
and
non-Guassianness
or
even
accompanied
with
short
time
amplitude
pulse.
This
kind
of
noise
is
not
adequately
described
in
term
of
Gaussian
models
but
␣-stable
distribution
[5].
The
estimates
based
on
the
second-order
∗
Corresponding
author.
Tel.:
+86
411
84709573;
fax:
+86
411
84709573.
E-mail
address:
qiutsh@dlut.edu.cn (T.-s.
Qiu).
statistics
may
be
highly
unreliable
because
␣-stable
noise
has
no
finite
second-order
statistics.
Instead,
properties
of
fractional
lower
order
moments
(FLOMs)
[6]
should
be
used.
In
this
paper,
a
novel
DOA
estimation
algorithm
for
cyclosta-
tionary
signal
based
on
the
concept
of
fractional
lower
order
cyclic
correlation
[7,8]
is
presented.
Both
[7]
and
[8]
are
based
on
the
ULA
which
can
give
accurate
estimations
of
azimuth
angle
in
impul-
sive
noise
for
only
narrowband
signals.
However,
if
one
needs
to
estimate
both
azimuth
and
elevation
angles
for
wideband
cyclo-
stationary
signals,
the
estimation
algorithms
based
on
ULA
are
powerless.
So
we
propose
a
fractional
lower
order
cyclic
MUSIC
algorithm
based
on
the
UCA,
which
can
estimate
both
azimuth
and
elevation
angles
accurately
even
in
impulsive
noise.
It
allows
to
select
the
desired
signals
and
suppress
interference
and
noise
by
exploiting
cyclostationarity
[9–11].
The
advantage
of
the
proposed
method
is
that
both
temporal
properties
and
frequency
domain
characteristics
of
the
signals
are
considered
so
as
to
improve
the
signal
detection
ability
and
the
resolution
performance
of
the
DOA
estimation
[12,13].
It
is
a
robust
method
since
it
transforms
the
DOA
estimation
of
the
broad-band
cyclostationary
signals
into
the
narrow-band
problem
with
the
“center
frequency”
ε,
by
using
the
phase
shift
characteristics
of
the
fractional
lower
order
cyclic
cor-
relation.
That
is
to
say
the
proposed
method
is
available
for
both
narrow-band
and
broad-band
signal
models.
2.
The
proposed
signal
model
and
noise
model
Let
us
have
a
uniform
circular
array
of
M
antenna
elements
with
radius
r.
Assume
that
L
narrow-band
source
signals
are
S(t)
=
s
1
(t),
.
.
.,
s
L
(t),
impinging
on
the
array
from
the
unknown
1434-8411/$
–
see
front
matter ©
2012 Elsevier GmbH. All rights reserved.
http://dx.doi.org/10.1016/j.aeue.2012.11.006