TELKOM
NIKA
, Vol. 11, No. 6, June 20
13, pp. 3028
~
3
035
e-ISSN: 2087
-278X
3028
Re
cei
v
ed
Jan
uary 9, 2013;
Re
vised Ma
rch 16, 2013; A
c
cepted Ap
ril 12, 2013
Eliminating Noise of Mud Pressure Phase Shif
t Keying
Signals with A Self-Adaptive Filter
Yue Shen*
1
, Lingtan Zh
a
n
g
2
, Heng Zh
ang
3
, Yinao Su
4
, Limin
S
h
eng
5
, Lin Li
6
1,2,
3
School of S
c
ienc
e, Chin
a Univers
i
t
y
of
Petrole
u
m, Qing
dao, 26
65
80, P
.
R. China
4,5,
6
Drillin
g T
e
chno
log
y
R
e
se
a
r
ch Institute,
CNPC, Beiji
ng, 1
001
95, P. R. Chin
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: shen
y1
96
1@
ya
ho
o.com.cn*
1
, zhanglt@u
p
c
.
edu.cn
2
,
zhan
g6
6h@
16
3.com
3
, su
y
i
na
o@p
e
trochi
na.
com.cn
4
, slmdri@cnpc.com.cn
5
, lilin5
507
03@
ya
ho
o.com.cn
6
A
b
st
r
a
ct
T
he feasi
b
il
ity of app
lyin
g a s
e
lf
-ad
aptive fi
lter to el
imin
ate
nois
e
in th
e d
o
w
nhol
e
mu
d
pressur
e
phas
e sh
ift ke
ying (PSK)
sig
nals
is stud
ie
d
.
The self-a
da
ptive filter
w
i
th carrier
w
a
v
e
as the fi
lter i
n
put
sign
al
an
d
mu
d pr
essure
PS
K sig
nal
i
n
clu
d
i
ng
no
ise
as
the fi
lter
expect
ed
inp
u
t si
gn
al
in
structure
w
a
s
adopted t
o
process the
mud pr
essu
r
e
PSK signals
with the br
oadba
nd s
i
gnal charac
t
e
ristic
in
communic
a
tio
n
.
Mathe
m
atic
al
mod
e
l
of the
filter w
a
s
b
u
ilt to
reco
nstruct the mu
d press
u
re
PSK
sig
n
a
l
s
base
d
on th
e eval
uatio
n crit
erio
n of least
me
an sq
uar
e error (LMS) an
d the mathe
m
atical
mo
del
of mu
d
pressur
e
PSK
signals. Accor
d
ing to t
he
filter m
a
them
atic
al m
o
del, a special self-ad
aptive control algor
it
hm
w
a
s ado
pted
to re
ali
z
e
the
filt
er by
ad
justin
g
the fi
lter
w
e
ig
h
t
coefficie
n
ts s
e
lf-ad
aptive
l
y
a
nd th
e i
m
pacts
of
the filter ste
p
-
s
i
z
e
factor
on
sign
al to
nois
e
ratio
(SN
R
) a
nd d
i
stortio
n
factors of the r
e
constructe
d
mu
d
pressur
e
PSK signals were analy
z
ed
. Numerical c
a
lculation
and sim
u
lation show t
hat
the self-adapt
ive
filter can efficie
n
tly eli
m
i
nate r
and
o
m
nois
e
i
n
the
sign
al fre
que
ncy ba
nd a
nd reco
nstruct the mu
d press
u
re
PSK signals. I
n
addition, low distor
tion fact
ors of the rec
onstructed
m
u
d press
u
re PS
K signals c
a
n
be
obtai
ne
d by re
ason
abl
e sel
e
c
t
ing
the filter step-si
z
e
factor.
Ke
y
w
ords
: sel
f
-adaptiv
e filter
, mud pr
essure
phase sh
ift keying si
gna
l, noi
se, carrier w
a
ve
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Measurement
Whil
e Dri
lling (M
WD) co
nsi
s
ts
of maki
ng
variou
s
d
o
wn
hole
measurement
s and the
n
tran
smitting this inform
ati
on to the su
rface fo
r display and furt
her
interp
retation
or imme
diat
e u
s
e. O
ne
o
f
the
mo
st co
mmon metho
d
s of
pa
ssing
the
info
rmati
on
from the do
wnh
o
le se
nsors to the surface is
through p
r
e
s
sure pulse
s in
the mud flow, a
techni
que
kn
own
a
s
m
u
d
pulse tele
met
r
y. The
ne
we
r mu
d
pul
se
telemetry u
s
es
a m
u
d
si
ren
type of modulator to ge
n
e
rate mu
d continuo
us
p
r
essure wave
signal
s an
d
allows co
m
p
lex
modulatio
n method
s to be use
d
to produ
ce high
er
dat
a rates by accurate co
ntrol
of the phase o
r
freqen
cy of the mud siren, the mod
u
lation
meth
od is call
ed
phase
shift keying (PS
K
)
modulation. Modulation type su
ch as differential
phase
shift keying (DPSK) and
even
more
compl
e
x modulation method such as
quadrature
phase shift keying (QPSK) can be used to
gene
rate m
u
d pressu
re P
SK signal
s to
tran
smitting i
n
mud
with a
mud
sire
n mo
dulator. In
M
W
D
system, the p
r
inci
pal noi
se
sou
r
ce is the
pre
s
sure
flu
c
tuations
cau
s
ed by bit vibration, downh
ole
motor stalli
ng
or drill stri
ng
bucklin
g and
the noi
se pre
s
ent
s a ban
d-limited white
Gau
ssi
an noi
se
due to
the l
o
wer noi
se
fre
quen
cy
spe
c
t
r
um [1].
T
h
o
ugh f
r
equ
en
cy of the n
o
ise source
is n
o
t
high, there is still some n
o
i
se into the
si
gnal
fre
quen
cy
band, cau
s
ing relativel
y
larger ran
d
o
m
pre
s
sure fluct
uating in am
plitude an
d makin
g
sig
n
a
l
to noise ratio (SNR) of th
e downhol
e mud
pre
s
sure
sig
nal
seve
rely
redu
ce.
Due
to
spe
c
tru
m
alia
sing
of
noise a
nd th
e mu
d p
r
e
ssure
sign
al, conve
n
tional sig
nal
processin
g
method
s
ca
n
not effectively eliminate or sup
p
re
ss the
noise. Some
resea
r
chers
p
u
t forward th
e mat
c
hed
f
ilter m
e
thod
to
eliminate th
e
noise effe
cts
by
cal
c
ulatin
g th
e self-correlat
i
on
coefficie
n
t
s of
si
gnal
m
i
xed with
noi
se ba
sed
on
the differen
c
e
of
noise an
d sig
nal in
correla
t
ion [2], but this meth
od i
s
only suita
b
le
for limited
si
ngle fre
que
n
cy
sign
al mo
dul
ated by th
e freque
ncy
shift
keyin
g
(FSK) method
havi
ng lo
w tran
smissi
on
effici
ency
and not for the frequency band si
gnal modulated by mud pressure PSK method. An adaptive
comp
en
satio
n
metho
d
[3]
,
prop
osed b
y
Brando
n et
c., ca
n elimi
n
ate theo
retically noise in
the
MWD sign
al by
extra
c
ting
app
rop
r
iate
prop
ortio
n
of
the
sign
al m
i
xed with
noi
se
as refere
n
c
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Elim
inating Noise of Mud P
r
essure Phase Shi
ft Keying Signals
with Self-Adapti
v
e... (Yue Shen)
3029
input
sign
al
of the a
dapti
v
e co
mpen
sator
and
aut
omatical
adj
usting
the
n
o
ise
inten
s
ity by
feedba
ck of
o
u
tput si
gnal
to bal
an
ce
noi
se
of M
W
D
si
gnal, b
u
t it is
difficult to im
plement
and
t
he
effect is limited. Acco
rdin
g to the f
r
eq
u
ency
band
transmi
ssion
chara
c
te
risti
c
s of mud
pressure
PSK signal
s and mathem
atical theory
of self-adapti
v
e filter, a mat
hematical
model of the
self-
adaptive filter with a carrier wave a
s
refe
ren
c
e
in
put si
gnal an
d M
W
D si
gnal mixe
d with noi
se
as
the expected
input
signal i
s
built for proc
essi
ng
the mud pressure
PSK si
gnal
s
with broadband
signal characteristic,
and t
he feasibility of elim
inating noise in the mud pressure PSK signal
s
based on
self
-ada
ptive filtering meth
od i
s
also studi
ed
in this pape
r.
2. The Math
e
m
atical Mod
e
l
of Self-adaptiv
e Filter
2.1. The Stru
cture o
f
Self-adap
t
iv
e F
ilter
Self-ada
ptive filter is a
kin
d
of digital
f
ilter b
a
sed
on
mode
rn
adap
tive control th
eory [4,
5], it can be u
s
ed to
reali
z
e
dynamic t
r
acking
and n
o
ise eliminatio
n
of the
sign
al by
self-a
dapti
v
e
adju
s
ting the
filter para
m
et
ers
acco
rdin
g
to the si
gnal
feature
s
. Self-adaptive filter is
comm
on
ly
use
d
in
pro
c
essing
na
rro
w ba
nd
sig
n
a
l in radio
communi
catio
n
sy
stem, in
whi
c
h th
e ratio
betwe
en sig
n
a
l freque
ncy band an
d ca
rrier wave fre
quen
cy is greatly less tha
n
1 and sign
al
freque
ncy in frequ
en
cy ban
d has little ch
ange
comp
ar
i
ng with carrie
r wave fre
q
u
ency. Figu
re
1
is the gen
eral stru
cture of
adaptive filter, in which
()
x
n
is input si
gnal with noi
se,
()
dn
is
expecte
d sig
nal
inp
u
t,
()
y
n
is the filter
out
put, and
()
en
is
error
sig
nal o
u
tput. The
expecte
d
sign
al is
sp
ecial sig
nal
refl
ecting th
e fea
t
ure of
extracted sig
nal. Under the
effe
ct of erro
r
sig
nal,
the self-a
da
ptive filter adjust
s
the filter coeffi
cien
ts self
-adapti
v
ely and the output sig
nal
contin
uou
sly approa
che
s
to the expect
ed sig
nal
to make th
e error minimal ev
entually, and
the
effective ch
a
r
acte
ri
stic in
cluded i
n
the
input si
gnal i
s
dyna
micall
y extracted,
then the
use
f
ul
signal reconstruction and noise elimi
nati
on or suppression
will reali
z
e.
Acco
rdi
ng to
the linear system theory
,
t
he self-ad
aptive filter output matrix
()
Yn
is
convol
ution o
f
the input matrix
()
X
n
and unit impulse re
spo
n
se matrix
()
H
n
a
nd can b
e
sh
own
as
follows
.
(
)
()
()
Yn
X
n
H
n
(1)
Comp
ari
ng with conventi
onal digital
f
ilter
st
ru
cture
with fixed
para
m
eters, the
self-
adaptive filter paramete
r
s
form
a
weigh
t
coeffici
ent
matrix
()
Wn
with
1N
dimen
s
io
n. If the
input mat
r
ix
()
X
n
is
N1
dimen
s
io
n, then th
e o
u
tput matrix
()
Yn
of
the
self-ada
ptive filter
ca
n
be expre
s
sed
as:
N1
0
()
()
()
()
()
(
)
i
i
YW
X
w
x
nn
n
y
n
n
n
i
(2)
Whe
r
e,
n
and
i
are di
screte
variable
s
,
()
Wn
is wei
ght co
efficient matrix
of the self-a
daptive
filter,
()
i
wn
is
matrix c
oeffic
i
ent of
()
Wn
,
()
x
ni
is matrix coefficient of the matrix
()
X
n
and can
be expre
s
sed
by the unit delay sampli
ng
value of inpu
t signal.
we
i
g
ht coeffici
ent matrix
()
Wn
+
()
dn
()
x
n
()
en
()
y
n
Fi
g
ure 1. The
stru
cture of self-ada
p
tive filter
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No. 6, June 20
13 : 3028 – 3
035
3030
In
th
e
filte
r
ing
pr
oc
e
s
s
,
ba
s
e
d
on
th
e
s
p
ec
ia
l
control alg
o
rithm
[6, 7], the
self
-ada
ptive
filter obtai
ns the filter weig
ht co
e
fficient
s a
c
co
rdin
g t
o
the
erro
r
si
gnal
()
en
a
nd th
e
input
matrix
()
X
n
and ite
r
ativel
y update
s
we
ight co
efficie
n
t matrix
()
Wn
. Throug
h finite it
eration, th
e o
u
tpu
t
signal will approach to
the
expected signal.
In MWD sy
stem, transmitt
ed MWD signal or
mud pressure PSK signal are a kind of
mech
ani
cal
modulatio
n si
gnal [8, 9]. Beca
use of me
cha
n
ical syst
em ine
r
tia an
d pre
s
su
re si
gnal
transmitting i
n
the m
ud, th
e carrier wave freq
uen
cy
i
s
limited
in th
e lo
w fre
que
ncy a
bout
a f
e
w
tens
Hert
z. T
he ratio
of m
odulatio
n si
g
nal fr
eq
uen
cy
band
and
ca
rrie
r
wave fre
quen
cy is
usually
clo
s
e to
1
an
d is a
ki
nd
of typical
broa
dban
d
sign
al
[10, 1
1
]. Th
erefo
r
e, h
o
w
to co
nst
r
u
c
t the
expecte
d sig
nal is a
key to get the tra
n
smitted
M
W
D sig
nal by self-ada
pt
ive filtering. Fo
r the
mud pressure PSK signals, the si
gnal
spectrum is
related to
data encodi
ng. If the signal
averag
e
power
sp
ectrum i
s
u
s
e
d
as th
e effective
fe
ature
to
con
s
truct th
e exp
e
cted
si
gnal,
it is
not sufficie
n
t to represent all the signal
s in diffe
rential
enco
d
ing sta
t
us and only the ca
rrie
r
wa
ve
can b
e
used
to represent variou
s en
co
ding mod
u
la
ti
on sig
nal cha
r
acte
ri
stics. Howeve
r, for the
broa
dba
nd
si
gnal, si
gnal f
r
equ
en
cy in f
r
equ
en
cy ba
nd chan
ge
s
greatly relative to the
ca
rri
er
wave frequ
e
n
cy a
nd the
en
codin
g
in
formation
si
g
nal
cann
ot b
e
extra
c
ted
prop
erly if u
s
in
g
carrie
r wave
as the
expecte
d si
gn
al. Theref
ore, the self
-adaptive
filte
r
structu
r
e
and
mathemati
c
al
modeli
ng fo
r pro
c
e
s
sing
b
r
oad
ban
d
sig
nal
sho
u
ld b
e
adju
s
ted
ba
sed o
n
the
ba
sic
mathemati
c
al
princi
ple of a
daptive filter.
2.2. The Stru
cture o
f
Self-adap
t
iv
e F
ilter
The cha
r
a
c
te
ristics ch
ang
e
of self-a
da
ptive
f
ilter is implem
ente
d
by a
d
justi
ng filter
weig
ht co
efficient
s with
self-ada
ptive
algorith
m
an
d all filter
weight coeffici
ent adju
s
tme
n
t
algorith
m
s
are trying to m
a
ke
output
si
gnal
()
y
n
approa
ch exp
e
cte
d
sign
al
()
dn
. The least
mean
-squa
re
error (LMS)
algorith
m
a
d
j
u
sts weight
coeffic
i
ent
s
matrix to mak
e
the mean-s
quare
value
of erro
r sig
nal
()
(
)
()
en
d
n
y
n
minimize, an
d wh
en
is minimu
m, the optima
l
weig
ht
coeffici
ent m
a
trix
()
Wn
can
be
obtaine
d to a
dapt the
statisti
cal ch
ara
c
t
e
risti
cs of
un
kno
w
n or
time-varying
sign
al and n
o
i
se an
d the optimal
filterin
g effect will b
e
achi
eved [1
2, 13].
Suppo
se that
the discrete
downhole
sig
nal is the
su
m of MWD
si
gnal
()
s
n
and wh
ite
Gau
ssi
an no
ise
w
()
nn
. Accord
ing to th
e
communi
catio
n
theo
ry, the noi
se
introdu
ced
in
broa
dba
nd d
o
wn
hole M
W
D sig
nal is a
d
d
itive rando
m
interfere
n
ce and its me
an
-sq
u
a
r
e value
is
not zero.
Wh
en
is mi
nimu
m, it must
ap
proa
ch
the
m
ean-sq
ua
re v
a
lue
of ra
ndo
m noi
se, a
nd
()
y
n
will approa
ch
the MWD
sig
nal
()
s
n
.
Whe
n
taki
n
g
the do
wn
hole M
W
D
sign
al with
noise a
s
the expe
cted
sign
al
w
()
()
()
dn
s
n
n
n
and ca
rrie
r
wave
cc
()
s
i
n
(
)
x
nA
n
as in
p
u
t sign
al, mu
d pre
s
su
re P
SK signal
sc
()
s
i
n
(
)
s
nA
n
f
n
as MWD signal. Amon
g those formulas,
c
A
is the ca
rri
er wave
amplitude,
cc
2
f
is the an
gula
r
f
r
equ
en
cy of carri
er
wave,
c
f
is the
ca
rri
er
wave frequ
en
cy,
s
A
is the signal
amplitude,
()
f
n
is the phase
-
shift function. Then,
the out
put signal of the self-
adaptive filter can be exp
r
e
s
sed a
s
:
N1
N
1
00
cc
()
(
)
(
)
()
s
i
n
[
(
)
]
ii
ii
wx
w
yn
n
n
i
n
A
n
i
(3)
The mea
n
-sq
uare valu
e of error si
gnal
can be de
scrib
ed as:
2
2
2
ww
()
()
()
()
2
(
)
(
)
(
)
Ee
n
E
s
ny
n
E
n
n
E
n
n
s
ny
n
(4)
Considering
()
s
n
and
()
y
n
are n
o
t respe
c
tively relevant wi
th rand
om n
o
ise
w
()
nn
,
there are
w
2(
)
(
)
(
)
0
En
n
s
n
y
n
and
2
w
()
0
En
n
, then the
minimum m
e
an-squ
a
re val
ue of
error si
gnal
can be expressed a
s
:
2
N1
2
sc
c
c
w
0
mi
n
=
m
i
n
s
i
n
[
(
)
]
(
)
s
i
n
[
(
)
]
+
(
)
i
i
EA
n
f
n
w
n
A
n
i
E
n
n
(5)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Elim
inating Noise of Mud P
r
essure Phase Shi
ft Keying Signals
with Self-Adapti
v
e... (Yue Shen)
3031
Whe
n
appro
a
ch
es the mi
nimum value
we can get th
e result as fol
l
ows:
N1
cc
s
c
0
()
s
i
n
[
(
)
]
s
i
n
[
(
)
]
i
i
wn
A
n
i
A
n
f
n
(6)
The physi
cal
meaning of Equation (6
) is that
whe
n
the mean-squ
a
re valu
e
of error
sign
al is mi
ni
mum, the lin
ear
sum
of the carrie
r
wa
ve value .. at one time a
n
d
N-1 pa
st time
values of
cc
sin
(
)
A
ni
with weig
ht from
the weig
ht coefficient mat
r
ix can
be u
s
ed to app
roa
c
h
the informatio
n signal valu
e of
sc
sin
(
)
A
nf
n
and re
al
ize the re
co
n
s
tru
c
tion of mud pre
s
sure
PSK signal. Where,
N
is the matrix dime
nsio
n numb
e
r or ord
e
r of th
e digital filter.
Therefore, the mud pressure PSK
si
gnal
can be reconstructed
by
the filter output
signal
as:
N1
cc
0
()
()
()
s
i
n
[
(
)
]
i
i
yn
s
n
w
n
A
n
i
(7)
Whe
r
e,
()
i
wn
is matrix coefficie
n
t of the optimal weig
ht co
efficients mat
r
ix
()
Wn
.
The
weig
ht
coeffici
ents
matrix can
be o
b
tained
by Wi
dro
w
-Hoft ra
ndo
m
gra
d
ient
algorith
m
[14] and the matri
x
coefficient
s can b
e
expre
s
sed a
s
:
(1
)
(
)
2
(
)
()
wn
w
n
e
n
x
n
(8)
Whe
n
both the co
ndition
s of min
and
*
()
()
0
Wn
W
Wn
are satisfied, the op
timal weight
coeffici
ents m
a
trix
()
Wn
can be
obtaine
d.
In the Eq
uati
on
(8),
is the
self
-ad
aptive step
-size fa
ctor
whi
c
h
det
ermin
e
s the
system
stability
and convergence rate;
if
is
ove
r
si
ze, the
con
v
ergen
ce
rate is hig
her bu
t the tra
c
king
pre
c
isi
on of sign
al will be
worse and t
he
system will be diverge
n
t when seri
ously, if
is
unde
rsi
z
e, th
e conve
r
g
e
n
c
e rate is u
n
s
atisfying
a
n
d
the trackin
g
perfo
rma
n
ce of sign
al will
become worse.
3. The Numerical Simulation Anal
y
s
is
of Self-adaptiv
e F
iltering Effect
Takin
g
mu
d
pre
s
sure P
SK signal
a
s
the tran
smitted MWD sign
al, accordin
g to
mathemati
c
al
model
of mud pr
essure
DPSK signal
and QPSK si
g
nal [15, 16], t
he M
W
D
signa
l
can
be exp
r
e
s
sed a
s
sc
()
s
i
n
[
2
(
)
]
s
tA
f
f
t
. Among th
e formula, carrier wave frequ
e
n
cy is
c
20
H
z
f
, signal a
m
plitude i
s
s
1P
a
A
, data cod
e
of the DPSK
signal i
s
D
PSK
=[1 1
1 1 1
1
1 1 1
1]
C
, data code of the QPSK signal is
Q
PSK
=[
0 0 0
1
1 0 1
1 0 0]
C
, both
maximum fre
quen
cie
s
of t
he two ki
nd
s of coding
si
gnal
spe
c
tru
m
are
ma
x
30
Hz
f
, the signal
power i
s
22
ss
(
/
2
)
0.
5P
a
PA
; the mean-sq
ua
re value of the
introdu
ce
d white Gau
ssi
a
n
noise
is
22
w
(
)
0.5P
a
nt
, the signal
-to-n
o
ise rati
o is
2
sw
/(
)
1
SN
R
P
n
t
, the ord
e
r of filter is
K1
0
1
, the
self-a
daptive
step-si
ze fa
ctor is
0.0
0
1
, the initial weight coefficient is
(0
)
0
w
, th
e sam
p
ling
freque
ncy is
s
400
0
H
z
f
. The effect of self-ada
ptive filter
takes
the improve
m
ent of sign
al-to-
noise ratio (S
NR) a
nd the
distortio
n
fa
ctor of
si
gn
al
waveform a
s
the evaluati
on criteri
on.
The
sign
al-to
-
noi
se ratio of sign
al can b
e
defi
ned a
s
:
M
2
1
M
2
1
()
[(
)
(
)
]
k
k
yk
SN
R
yk
y
k
(9)
The wavefo
rm distortio
n
factor of
signa
l can be d
e
fin
ed as:
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No. 6, June 20
13 : 3028 – 3
035
3032
2
M
ma
x
1
ma
x
M
2
1
()
(
)
()
()
=
()
k
k
yk
s
k
sk
yk
D
sk
(10)
In Equation
(9) a
nd Eq
uat
ion (10),
()
s
k
is the mud press
u
re PSK original
s
i
gnal and
ma
x
()
sk
is
its maximum value;
()
y
k
is
the outp
u
t sig
nal of
self-ad
aptive filter;
()
y
k
is the
output
sign
al of digit
a
l low-pa
ss filt
er after self-adaptive filtering;
max
()
yk
is the m
a
ximum outp
u
t signal
value of the d
i
gital low-pa
ss filter;
k
is sa
mpling seq
u
e
n
ce numb
e
r;
M
is sa
mplin
g numbe
r in a
codi
ng pe
riod
.
3.1. The Nu
merical Simulation and
Analy
s
is of Self-ad
a
ptiv
e Filtering o
n
Mud Pres
sure
DPSK Signal
Figure 2 sho
w
s the mu
d pressu
re DPS
K
signal
with
out noise, Fig
u
re 3 sho
w
s the mu
d
pre
s
sure
DP
SK sign
al mi
xed with
white Ga
ussi
an
no
is
e w
i
th SNR
=
1
an
d
F
i
gu
r
e
4
s
h
o
w
s
th
e
recons
truc
ted mud press
u
re DPSK s
i
gnal waveform
after s
e
lf
-adaptive filtering. In Figure
4, the
noise in re
co
nstru
c
ted
sig
nal de
crea
se
s su
bsta
ntiall
y and SNR i
s
25.5, raised
nearly 2
5
times.
Thro
ugh freq
uen
cy spe
c
trum analysi
s
,
the noise in
reco
nstructe
d
sign
al is hig
h
-
freq
uen
cy n
o
ise
outside the frequency
band of m
u
d pressure
DPSK
si
gnal, and
can be eliminated by
an
ordi
nary
digital lo
w-pa
ss filter. Fi
gure 5
sho
w
s th
e si
gnal
wave
form of
re
co
n
s
tru
c
ted
mud
pre
s
sure
DP
SK
sign
al pa
ssin
g a di
gital lo
w-p
a
ss filter
with
cut-off freque
ncy 4
0
Hz a
nd the
no
ise o
u
tsid
e t
he
freque
ncy b
a
nd is alm
o
st
eliminated. In
Figure
5,
the re
con
s
tru
c
t
ed sig
nal ha
s some extent
o
f
waveform di
stortion
comp
aring
with th
e ori
g
inal
sig
nal in Fi
gu
re
2 an
d the
di
stortion
facto
r
is
about 10.9%.
The rea
s
o
n
is that the
filter step
-si
z
e fa
ctor i
s
too
sm
all for improving the ability of
tracking
noi
se, whi
c
h
re
su
lts in lo
we
r
converg
e
n
c
e
rate an
d the i
n
crea
sing
re
constructio
n
e
r
ror
of low frequenc
y c
o
mponent in mud pres
sure
D
PSK s
i
gnal frequenc
y s
p
ec
t
r
um. Inc
r
eas
i
ng the
step-si
ze fact
or will b
r
ing d
o
wn the a
b
ility of tra
ckin
g
noise and the
SNR of re
co
nstru
c
ted
sig
nal,
but the
ability of tracking low frequency compo
nent in mud pressure
DPSK signal
frequency
spe
c
tru
m
will
be improve
d
and the signal di
st
orti
on facto
r
wil
l
be decre
a
s
ed. Th
erefo
r
e,
approp
riate in
cre
a
si
ng the
step-si
ze fact
or ca
n im
prove the re
con
s
tructed
sign
al quality, but the
signal distortion factor will
rais
e when the step-si
z
e fa
ctor reaches
the crit
i
c
al value because
o
f
highe
r
conve
r
gen
ce
rate
a
nd lo
w t
r
a
cki
ng a
c
cura
cy.
Table
1
sh
ows the
nu
meri
cal computatio
n.
results amo
ng the value of step-si
ze facto
r
, the SNR and
the distorti
on factor of
the
recons
truc
tion s
i
gnal.
T
i
me
(se
c
on
d
s
)
Figure 2. Mud Pressure
DPSK Signal Without Noise
T
i
me
(se
c
on
d
s
)
Figure 3. Mud Pressure DPSK Signal
Mixed with white Gaussi
an
T
i
me
(se
c
on
d
s
)
Figure 4. Reconstruc
ted M
ud Pressure
DPSK Signal a
fter self- adaptive Filtering
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Elim
inating Noise of Mud P
r
essure Phase Shi
ft Keying Signals
with Self-Adapti
v
e... (Yue Shen)
3033
T
i
me
(se
c
on
d
s
)
Figure 5. Waveform of the Recons
tructed Mud Pressure DPSK Signal
Table 1. Impact of filter
step-si
ze factor
on reconstruc
ted mud pressure DPSK si
gnal
Step-size factor
SNR
Distortion factor(
%
)
0.0005
67.1
14.8
0.001
25.5
10.9
0.002
11.4
7.1
0.003
6.4
6.4
0.004
4.2
4.4
0.005
3.1
6.6
The num
eri
c
al simulatio
n
and filtering
effects
sh
o
w
that the self-ada
ptive filter can
eliminate the
noise in sig
n
a
l freque
ncy
band a
nd ch
oosi
ng an a
p
p
rop
r
iate
ste
p
-si
z
e fa
ctor
can
minimize dist
ortion facto
r
of reco
nst
r
u
c
ted si
gnal. T
houg
h there
are certain n
o
ise
s
left in the
recon
s
tru
c
ted
sig
nal, the
n
o
ise
s
are
out
side
the
sig
n
a
l fre
quen
cy
band
an
d
can
be
eliminate
d
by
an ordi
na
ry digital low-pa
ss filter, then
the sign
al SNR
will be imp
r
ov
ed gre
a
tly.
3.2 The Num
e
rical Simulation and
An
aly
s
is of Self-ada
ptiv
e Filtering on Mu
d Pressure
QPSK Signal
Figure 6
sh
o
w
s the m
ud p
r
essu
re
QPS
K
sign
al with
out noi
se, Fi
g
u
re
7
sho
w
s the mu
d
pre
s
sure QP
SK signal
mi
xed with
whi
t
e Gau
s
sian
noise with
S
N
R=1
and
F
i
gure
8
sho
w
s
reconstructedmud pressure QPSK signal
wavefo
rm aft
e
r self
-adapti
v
e filtering.
T
i
me (s
ec
on
ds
)
Figure 6. Mud Pressure QPSK Signal without Noi
s
e
T
i
me (s
ec
on
ds
)
Figure 7. Mud Pressure QPSK Signal
Mixed with white Gaussi
an Noise
T
i
me (s
ec
on
ds
)
Figure 8. Reconstructed M
ud Pressure
QPSK Signal afte
r Self-Adaptive Filtering
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No. 6, June 20
13 : 3028 – 3
035
3034
T
i
me (s
ec
on
ds
)
Figure 9. Waveform of the Re
constructed Mud Pres
sure QPSK Aignal Passing a Digital Low-
Pass Filte
r
wi
th 40Hz cut
-
o
ff Frequen
cy
In Figure 8, the noise in reco
nstructe
d sign
al decre
a
s
e
s
sub
s
tanti
a
lly and SNR is 24.4,
raised ne
arly
23 times. Beca
use of the noise
in re
con
s
tru
c
ted
signal bein
g
outside the si
gnal
freque
ncy
ba
nd, it can
be
eliminate
d
b
y
an o
r
din
a
ry
digital l
o
w-p
a
ss filter. Fi
g
u
re
9
sho
w
s
the
s
i
gnal
wavef
o
rm
of recons
truc
ted mud press
u
re
QPSK s
i
gnal pass
ing a
digit
a
l low-pass filter
with cut-off f
r
equ
en
cy 40
Hz and
noi
se out
side th
e freq
uen
cy
band i
s
almo
st elimin
ated.
In
Figure 9, the
dis
t
ortion fac
t
or
of rec
o
ns
t
r
uc
ted mud press
u
re
Q
PSK s
i
gnal is
about 7.3%. Table
2 shows th
e
nume
r
ical results am
ong
th
e value
of
th
e ste
p
-size fa
ctor, the
SNR and th
e
sign
al
distortio
n
fa
ctor. Thi
s
i
ndi
cates th
at the
distor
tio
n
fa
ctor of
re
co
nstructed
mud
p
r
essure QPS
K
sign
al is
gen
erally smalle
r than th
at of re
co
n
s
tru
c
ted mud
pre
s
sure
DPSK
sign
al an
d
the
recons
truc
tion quality of
mud pres
sure QPSK s
i
gnal after
s
e
lf
-ada
ptive filtering is
relatively
better than that of mud pressure
DPSK signal,
but choosi
ng a
reas
onabl
e
step-size factor i
s
t
he
key to g
e
t l
o
we
r
sign
al
distortio
n
fa
ctor. Becau
s
e
of more
co
mplex dem
o
dulation
of
mu
d
pressure QP
SK signal than that
of m
ud pressure
DPSK signal,
the low di
st
ortion factor
of
reconstructed mud pre
ssure QPSK signal provi
des a g
ood condition for the correct
demod
ulation
of the signal.
Table 2. Impact of filter
step-si
ze factor
on reconstruc
ted mud pressure QPSK si
gnal
Step-size factor
SNR
Distortion factor(
%
)
0.0005
68.9
8.8
0.001
24.4
7.3
0.002
9.8
5.6
0.003
6.7
5.6
0.004
4.7
4.2
0.005
3.2
4.4
4. Conclusio
n
s
Theo
retical a
nalysi
s
and
nume
r
ical si
mulation
sho
w
that the self-ada
ptive filter usin
g
transmitted M
W
D
sign
al mi
xed with noi
se as the
exp
e
cted
sign
al and carrier
wave as the in
put
sign
al can re
alize th
e noi
se elimi
natio
n of broadb
a
nd si
gnal,
which i
s
suitab
le for elimi
n
a
t
ing
random noi
se introduced in
mud pressure PSK signal in transmission process.
Self-ada
ptive filter can eli
m
inate the
ra
ndom
noi
se i
n
si
gnal f
r
equ
ency
band.
T
he noi
se
in re
con
s
tructed sig
nal i
s
outsid
e
the
sign
al
freq
u
ency b
and
a
nd can b
e
e
liminated by
an
ordin
a
ry digit
a
l low-pa
ss fil
t
er, a highe
r SNR will b
e
o
b
tained.
The q
uality of
re
con
s
tructe
d si
gnal
de
pe
nds on
the
si
gnal
disto
r
tio
n
facto
r
being
relate
d
to the filter step-si
ze fa
ct
or, therefore
the lo
we
r
si
gnal di
stortio
n
factor
ca
n
be obtain
e
d
by
cho
o
si
ng a reasona
ble filter step
-size factor. In
addi
tion, numeri
c
al cal
c
ulation
sho
w
s that the
distortio
n
fa
ctor of
re
con
s
t
r
ucte
d mu
d
pre
s
sure
QP
SK signal
is
smalle
r tha
n
that of the m
ud
pre
s
sure DP
SK signal un
der conditio
n
of the same fi
lter step
-si
z
e
factor.
Ackn
o
w
l
e
dg
ements
This
wo
rk
was finan
cially
sup
porte
d
by the Proje
c
t of Nation
al Natu
ral S
c
ien
c
e
Found
ation
of Chin
a un
der
Gra
n
t 5
1274
236
an
d
the Proje
c
t of High
-tech Re
se
arch
and
Develo
pment
Prog
ram
of
Chin
a u
nde
r
Grant
200
6A
A06A101. T
h
e auth
o
rs
wo
uld li
ke to
ex
pre
ss
their than
ks f
o
r the spon
so
ri
ng of publi
s
hing this p
a
p
e
r.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Elim
inating Noise of Mud P
r
essure Phase Shi
ft Keying Signals
with Self-Adapti
v
e... (Yue Shen)
3035
Referen
ces
[1]
Hutin R, T
ennet
RW, Kashik
ar SV.
N
e
w
mud
puls
e
te
le
metry techn
i
q
u
e
s
for d
e
e
p
-w
ater a
ppl
icati
o
n
s
and i
m
prov
ed real-ti
m
e data capa
bil
i
ties
. S
PE/IADC Drill
i
ng C
onfere
n
ce
. Amsterdam, Nether
lan
d
s
.
200
1. SPE/IADC 677
62: 1-8.
[2]
Marsh JL, Fraser EC, Holt AL.
Me
asur
e
m
e
n
t-w
h
ile-
d
ril
l
i
ng
mud
p
u
ls
e d
e
tectio
n
p
r
ocess: a
n
investi
gatio
n
of matche
d fi
lter resp
ons
es
to si
mul
a
ted
and
rea
l
mud pr
essur
e
puls
e
s
. SPE
S
y
mp
osi
u
m on
Petrole
u
m Ind
u
str
y
A
ppl
icati
ons of
Micr
oco
m
puters. San
Jose, Ca
liforn
i
a. 198
8. SP
E
177
87: 1-8.
[3]
Brand
on
T
L
, Mintchev
MP, T
abler H.
A
daptiv
e c
o
mp
ensati
o
n
of th
e mu
d
pump
no
ise
in
a
measur
ement-
w
h
i
l
e
-dri
lli
ng s
ystem.
SPE Journal
. 19
99; 4(2)
: 128-13
3.
[4]
Kuan H
e
, T
ao Huan
g. Ada
p
t
ive filter des
i
gn bas
ed
on
Matlab.
Jo
urn
a
l of W
uha
n Univers
i
ty of
T
e
chno
logy
. 2
008; 30(
1): 70-
72. (in Ch
ines
e
)
[5]
Cho
ng
Z
h
u, Xi
aomi
ng Lia
ng. Performanc
e
s
t
ud
y
of a
d
a
p
tiv
e
spe
e
ch
no
is
e canc
el
ing
ba
sed o
n
LMS
algorithm.
Jour
nal of Guil
in U
n
iversity of Ele
c
tronic T
e
chn
o
l
ogy
. 200
8; 28(
4): 298-3
01. (in
Chin
ese)
[6]
Suish
eng
Ye,
Ron
ggu
i Y
u
,
Xiao
W
u
, eta
l
.
T
he
researc
h
and
a
ppl
icatio
n
of a ne
w
a
d
aptive
LMS.
Electrical M
eas
ure
m
e
n
t & Instrumentati
o
n
. 2
008; 45(
7): 19-
22. (in Ch
ines
e
)
[7]
Yan Geng, D
uan
jin Z
h
a
ng.
Surve
y
of a
daptiv
e filterin
g alg
o
rithms.
Information and Electronic
Engi
neer
in
g
. 2008; 6(4):3
15-
320. (in C
h
in
es
e)
[8]
Moriart
y
K A.
Pressure
puls
e
gen
erator for
me
asur
e
m
ent-
w
hile-dri
lli
ng s
ystems w
h
ic
h
prod
uces
hig
h
sign
al strengt
h and ex
hi
bits hi
gh resista
n
ce to ja
mmin
g
. US
621
93
01 (Pate
n
t). 2001.
[9]
Malone
D.
Sin
u
soi
dal
pressu
re puls
e
ge
ner
ator for me
as
ure
m
e
n
t w
h
ile
drilli
ng to
ols
. US48
478
1
5
(Patent). 198
9.
[10]
Yue Sh
en, Yi
n
ao Su, Ge
nsh
eng
Li, eta
l
. N
u
merica
l mo
de
ling
of DPSK
pressur
e
sig
n
a
l
s an
d the
i
r
transmissio
n
c
haracter
i
stics i
n
mud cha
n
n
e
l
s
.
Petroleu
m S
c
ienc
e
. 200
9; 6(3): 266-2
70.
[11]
Yue S, Yinao S, G
ensheng L, etal.
T
r
ansmission
characteristics of DPSK m
ud pressur
e
signals in
a
straight w
e
ll.
Petrole
u
m
Scien
c
e and tec
hno
l
ogy
. 201
1; 29(
12): 124
9-1
256
.
[12]
Yongfa
ng
Xi
e, Hongj
un W
u
, Yanni De
ng,
et
al. Harmon
i
c detectio
n
method b
a
se
d on modifi
e
d
ada
ptive filter.
Journ
a
l of W
u
Han U
n
ivers
i
ty of T
e
chnol
ogy
.
2008; 3
0
(7): 1
23-1
26. (in Ch
i
nese)
[13]
Ron
g
Mei, Sha
nhu
a Yao. Ada
p
tive filt
er app
l
i
ed in a
d
a
p
tive
noise canc
ell
e
r sy
st
em.
Instrumentati
o
n
T
e
chno
logy
. 2
008; 8: 13-
15. (in Chi
nes
e)
[14]
Li
ye
Son
g
, Ji
n
g
she
ng W
a
ng,
Jishe
n
Pe
ng. A
l
gorit
hm R
e
sea
r
ch of A
daptiv
e
F
ilter a
n
d
DSP
Simul
a
tio
n
Real
izati
on.
Moder
n Electro
n
i
cs T
e
chniq
u
e
. 200
9; 5: 112-1
14. (in Ch
ines
e
)
[15]
Yue S
hen, Y
i
n
ao S
u
, Li
n Li,
etal. An
al
ysis
on tr
a
n
smissi
o
n
char
acteristi
cs of differe
nti
a
l p
has
e sh
ift
ke
yin
g
sign
al o
f
continuo
us pr
essure
w
a
ve in
drilli
ng flu
i
d ch
ann
el.
Acta Petrolei Sinica
. 2
009; 30(
4):
593-
597. (i
n Chin
ese)
[16]
Yue Sh
en, Ju
n Z
hu, Yina
o
Su, etal. T
r
a
n
smissi
o
n
ch
a
r
acteristics of the dril
lin
g flui
d pressur
e
qua
dratur
e ph
ase sh
ift ke
yin
g
sig
nal
al
ong
a dir
e
ction
a
l
wellb
ore.
Acta Petrolei Sinica
. 201
1;
32(
2):
340-
345. (i
n Chin
ese)
Evaluation Warning : The document was created with Spire.PDF for Python.