Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
3
,
June
2020
,
pp. 3
235
~
3243
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
3
.
pp3235
-
32
43
3235
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
An enha
nced l
ossless c
om
pression
with cr
yp
tograph
y hybrid
mech
anism for E
CG bi
om
edical
signal m
onitorin
g
Musta
fa Em
ad H
ameed
1
,
M
as
rull
iz
am
M
at Ibr
ah
im
2
,
Nu
rulf
ajar
Abd
Manap
3
, A
li
A
.
Moham
med
4
1,2,3
Cent
re
for Te
le
comm
unic
atio
n
Resea
r
ch and
I
nnovat
ion
(CeTRI),
Facu
lty
of E
le
c
troni
c
and
Co
m
pute
r
Engi
n
ee
r
ing
,
Univer
siti
Te
kn
i
kal
Ma
lay
sia
Me
la
ka
(UTe
M),
M
al
a
y
si
a
1
Depa
rtment of
Com
pute
r
Techn
ique
s E
ng
ine
e
rin
g,
Bi
la
d
Al
-
raf
id
an
Univer
si
t
y
C
o
ll
eg
e, I
raq
4
Facul
t
y
of
Infor
m
at
ion
and
Com
m
unic
at
ions
Tec
hnolog
y
(FTMK
),
Univer
siti
Te
kn
i
kal
Ma
lay
sia
Me
la
ka
(UTe
M)
,
M
al
a
y
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
6
, 2
019
Re
vised
Oct
13
,
2019
Accepte
d
Oct
25
, 201
9
Due
to
the
ir
use
in
dai
l
y
l
ife
situ
a
ti
on,
demand
for
remote
hea
l
th
a
ppli
c
at
ions
and
e
-
he
al
th
m
onit
oring
equ
ipment
is
growing
quic
kl
y
.
In
thi
s
phase
,
for
fas
t
dia
gnosis
and
t
her
ap
y
,
informa
ti
on
ca
n
b
e
tra
n
sferre
d
from
th
e
pa
ti
en
t
t
o
the
dista
nt
c
li
ni
c.
Now
aday
s
,
t
he
m
ost
chr
onic
disea
se
is
ca
r
diova
scul
ar
disea
ses
(CVD
s).
How
eve
r
,
th
e
s
tora
ge
and
tra
ns
m
ission
of
the
E
CG
signal
,
consum
es
m
ore
ene
rg
y
,
bandwi
dth
and
da
ta
se
cur
ity
whi
ch
is
fac
ed
m
a
n
y
cha
l
le
nges.
Hen
ce
,
in
th
is
work,
we
pre
sent
a
c
om
bine
d
appr
oa
c
h
for
ECG
dat
a
compress
ion
and
cr
y
ptogr
a
ph
y
.
The
compress
ion
is
per
form
ed
using
ada
pt
ive
Huffm
an
en
codi
ng
an
d
enc
r
y
p
ti
ng
is
done
using
AES
(CBC)
sche
m
e
with
a
256
-
bit
ke
y
.
To
inc
re
ase
the
se
cur
ity
,
we
inclu
de
Diffie
-
Hell
m
an
Ke
y
e
xcha
nge
to
aut
h
ent
i
ca
t
e
th
e
r
ecei
ver
,
RS
A
ke
y
gene
r
at
io
n
for
enc
r
y
pti
ng
and
dec
r
y
pt
ing
the
dat
a
.
Exp
eri
m
ent
a
l
results
show
tha
t
the
proposed
ap
proa
ch
a
chieve
s
bet
t
er
per
form
an
ce
in
te
rm
s
of
c
om
pre
ss
ion
and
en
cr
y
pti
o
n
o
n
MIT
-
BIH E
C
G da
ta
se
t.
Ke
yw
or
d
s
:
Ad
a
ptive
H
uff
m
an
co
ding
AES
CB
C
m
od
e
Diff
ie
-
Hell
m
a
n
ECG
si
gn
al
RSA
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Mustafa Em
ad Ham
eed
,
Dep
a
rtm
ent o
f Te
le
com
m
un
icati
on
E
nginee
ring, F
ac
ulty
of
Ele
ct
ro
nic
and
Com
pu
te
r
E
ng
i
neer
i
ng,
Un
i
ver
sit
i Te
knikal M
al
ay
sia
Mel
aka, Ha
ng Tu
a
h
Jay
a,
76
100 D
ur
ia
n
T
unggal
, Mel
a
ka, M
al
ay
sia
.
Dep
a
rtm
ent o
f C
om
pu
te
r
Tec
hn
i
qu
e
s E
ng
i
ne
erin
g,
Bi
la
d
Al
-
raf
i
da
n Un
i
ver
sit
y C
ollege,
Diya
la
, B
aquba
h,
Ir
a
q
.
Em
a
il
:
m
ihh
h2
21@
gm
ail.co
m
1.
INTROD
U
CTION
Du
e
t
o
se
veral
prob
le
m
s
su
ch
as
un
heal
thy
li
festy
le
a
nd
ge
netic
in
her
it
ance
of
the
pa
ti
ent,
the
m
edical
healt
h
-
relat
ed
pro
blem
s
are
grow
i
ng
daily
.
These
pro
blem
s
co
ntribute
to
the
de
velo
pme
nt
of
sever
al
c
hro
nic
diseases
su
c
h
as
cance
r,
dia
be
te
s,
Alz
heim
e
r'
s
and
car
diov
ascular
disease
s,
et
c.
A
s
urve
y
was
su
bm
it
te
d
by
the
Wo
rld
He
al
th
Organ
iz
at
ion
re
veali
ng
that
card
i
ov
as
cular
diseas
e
is
the
m
os
t
chro
nic
disease
[
1].
T
hu
s
,
ef
fici
ent
diag
nosis
of
c
ard
i
ov
asc
ular
disease
ha
s
be
com
e
a
pr
om
i
nen
t
researc
h
t
op
ic
i
n
the
fiel
d
of
bi
om
edical
app
li
cat
ion
s.
Ele
c
trocardio
gram
(ECG
)
sig
nals
are
co
ns
ide
r
ed
as
the
pro
m
isi
ng
non
-
in
vasive
te
chn
i
qu
e
to
m
easur
e
the
He
art
-
relat
ed
act
i
viti
es,
functi
onal
it
y
and
al
so
pro
vid
es
str
uctu
ral
inf
or
m
at
ion
of
the
hea
rt
w
hi
ch
helps
t
o
an
al
yz
e
the
heart
-
relat
ed
disea
ses
[
2].
Fig
ure
1
s
hows
at
rial
and
ven
t
ricular
de
po
la
rizat
ion
an
d
repolariz
at
io
n
of
hear
t
by
represe
nting
th
e
ECG
sign
al
as
a
series
of
wave
s
wh
ic
h
are
kn
own
as
P
,
Q
,
R,
S
an
d
T
wa
ve
s.
T
he
ti
m
e
interval
betwee
n
these
pea
ks
pr
ov
i
des
se
ver
al
crit
ic
al
inf
or
m
at
ion
wh
ic
h
is
us
ef
ul
f
or
Hea
rt
analy
sis.
Ther
e
e
xist
sever
al
ty
pes
of
inter
vals
w
hi
ch
are
i
m
po
rta
nt
for
m
edical
analy
sis.
These
inte
r
vals
are
known
as
PQ
inte
rv
al
,
RR
interval,
QRS
du
rati
on,
ST
segm
ent,
and
QT
interval
wh
ic
h
ca
n
be
ide
ntifie
d
ba
sed
on
th
e
P,
QRS
c
omplex
a
nd
T
wa
ves
[
3].
C
on
se
qu
e
ntly
,
ECG
sign
al
s
are
widely
a
dopte
d
for
rem
ote
he
al
th
m
on
it
or
i
ng
syst
em
s
us
ing
a
dva
nced
te
le
m
edi
cal
ap
plica
ti
ons
w
her
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
32
3
5
-
32
4
3
3236
patie
nts
can
be
diagnose
d
a
nd
treat
e
d
fro
m
re
m
ote
assis
ta
nce
[4
]
.
I
n
this
ty
pe
of
proces
s,
the
sig
nal
is
transm
itted ov
er a wirel
ess c
hannel
wh
e
re t
he ori
gin
al
si
gnal
g
et
s
conta
m
inate
d.
More
ov
e
r,
t
he
chall
eng
e
s
of
ECG
si
gn
al
are
re
qu
i
re
m
or
e
sp
ace
f
or
st
or
a
ge
a
nd
ba
ndwi
dth
requirem
ent
for
data
tra
ns
m
is
sion
a
nd
al
so
t
hese
sig
nals
a
r
e
transm
it
te
d
ov
e
r
the
unsec
ur
e
d
c
hannel
he
nce
secur
it
y
al
so
be
com
es
a
cru
ci
al
ta
sk
f
or
t
he
r
esearch
com
m
un
it
y
[
5].
T
o
m
itigate
the
da
ta
stora
ge
re
qu
irem
en
t
and
sec
ur
it
y,
ECG
com
pr
es
sion,
an
d
cryp
tograp
hy
-
base
d
schem
es.
As
well
as,
the
pr
i
vacy
is
su
e
without
lim
it
ing
tradin
g
feat
ur
es
t
o
a
vo
i
d
data
c
ollec
ti
on
at
ta
cks
and
t
o
ke
ep
c
orrect
ope
n
tr
ansacti
on
rec
ord
[6
]
.
Durin
g
sig
nal
transm
issi
on
f
or
r
em
ote
m
o
nitor
i
ng
syst
e
m
s,
the
chann
el
conditi
on
s
aff
ect
the
EC
G
sig
nal
qu
al
it
y.
Th
ough,
a
few
c
om
plexiti
es
arise
her
e
due
to
the
i
m
pr
oper
tra
nsm
issi
on
rate
or
low
ba
ndwi
dt
h
in
com
m
un
ic
at
ion
cha
nnel
.
Oft
en,
co
ntin
uous
flo
w
of
sig
nal
m
ay
le
ad
to
an
i
ncr
ease
in
sh
ee
r
vo
l
um
e
of
data
[7
]
.
The
r
efore,
the
siz
e
of
data
in
the
com
m
un
ic
ation
cha
nnel
su
bs
ta
ntial
ly
af
f
ect
s
the
un
s
ui
ta
ble
transm
issi
on
r
at
e
or
low
bandw
i
dth
[
8,
9].
Con
ti
nu
ou
s
E
CG
sign
al
fl
ow
can
oft
en
le
ad
to
pac
ket
f
al
l
and
i
m
po
rtant
i
nform
at
ion
loss.
Howe
ver,
the
com
pr
essio
n
w
it
h
the
el
im
inati
on
of
re
dund
ancy
in
data,
wh
ic
h
i
s
necessa
ry
to
optim
iz
e
sto
rage
sp
ace
an
d
re
du
ce
t
he
tim
e
need
e
d
f
or
dat
a
transm
issi
on
[10,
11
]
.
F
ur
t
he
rm
or
e,
the i
m
pr
ovem
e
nt o
f
c
om
pr
ess
ion
m
echan
ism
w
hic
h
is de
pe
nd
on a p
e
rform
ance ev
al
uati
on
c
rite
rio
n.
T
hough,
the
m
os
t
i
m
po
rtant
thin
g
is
t
he
ne
w
pro
pos
ed
com
pr
essio
n
al
gorithm
w
ork
a
high
co
m
pr
ession
rati
o
(CR)
,
wh
ic
h
usual
ly
needs
a
low
pe
rcen
ta
ge
root
m
ean
sq
ua
re
di
ff
e
ren
ce
(
PRD
).
Als
o,
the
EC
G
bio
m
edical
sign
al
s
con
ta
in
se
ns
it
ive
pr
i
vate
heal
th
in
form
at
ion
as
well
as
detai
ls
that
ser
ve
to
in
div
i
du
al
ly
disti
nguis
h
pa
ti
ents,
hen
ce
m
us
t
be
encr
y
pted
pr
i
or
to
t
ran
sm
issi
on
ac
ro
ss
publ
ic
m
edia
so
as
to
preve
nt
unaut
horized
ac
cess
by
adv
e
rsa
ries f
rom
c
yber
-
at
ta
ck
s [12,
13
]
.
Figure
1.
ECG
sign
al
represe
nt
at
ion
-
P, Q,
R,
S,
T
Ther
e
f
or
e,
t
he
com
bin
at
ion
betwee
n
the
c
om
pr
essio
n
an
d
crypt
ogra
phy
m
echan
ism
s
it
s
at
tract
ed
researc
hers'
atten
ti
on
i
n
rece
nt
tim
es. I
n
the
pap
e
r
[
14]
p
r
opose
d
the c
onc
ept o
f
c
om
pr
es
sed
sen
sin
g
(C
S)
, on
e
of
the
ne
west
sign
al
proce
ssin
g
te
ch
niques
w
hich
a
pp
li
ed
to
com
pr
ess
the
ECG
sig
nals
w
it
h
the
adv
a
nta
ge
of
ver
y
l
ow
com
pu
ta
ti
onal
c
omplexit
y
at
com
pr
essi
on
sta
ge
to
enc
rypt
sig
nals
f
or
tra
nsm
issi
on
ove
r
i
ns
ec
ur
e
channels.
As
well
as,
a
nove
l
el
ect
ro
card
i
ogram
(ECG)
com
pr
essio
n
m
et
ho
d
f
or
e
-
healt
h
ap
plica
ti
o
ns
by
adap
ti
ng
ada
pt
ive
Fou
rier
deco
m
po
sit
io
n
(
AF
D
)
al
gor
it
h
m
hybr
i
dized
with
sym
bo
l
substi
tuti
on
(S
S
)
te
chn
iq
ue
ha
ve
been
presente
d
in
[15].
This
m
et
ho
d
as
achieves
ave
ra
ge
d
CR
of
17.6
t
o
44.
5
and
PR
D
of
0.8%
t
o
2.0%
with
a
hi
ghly
li
near
a
nd
robu
st
PRD
-
CR
relat
ion
s
hip.
A
no
t
her
resea
rc
h
[16]
ha
ve
pr
opos
e
a
schem
e
that
can
com
pr
ess
t
he
data
withou
t
any
loss
of
im
po
rtant
inf
orm
at
ion
and
al
s
o
ap
ply
crypto
gr
a
phic
schem
e
to
pre
serv
e
co
nfi
de
nt
ia
li
t
y
fr
om
unaut
horized
ac
cess.
In
their
work,
t
hey
use
m
ob
il
e
com
pu
t
in
g
dev
ise
t
o
el
im
inate
us
a
ge
of
com
pu
te
rs.
T
hey
pe
rfor
m
pr
e
processi
ng
su
c
h
as
rem
ov
al
of
gaussia
n
a
nd
baseli
ne
noise
,
detect
peak
s
,
do
hear
t
rate
a
naly
sis
and
c
om
pr
ess
the
ECG
sig
nal.
At
the
com
pr
essio
n
they
app
ly
3
le
vel
wav
el
et
tra
ns
f
or
m
at
ion
(db
04)
a
nd
us
e
t
hr
e
sh
ol
d
m
echan
ism
s.
Nex
t,
H
uffm
an
cod
in
g
te
chn
i
que
is
us
e
d
t
o
c
ompress
a
nd
enc
r
ypt
the
si
gn
al
.
They
get
the
c
om
pr
essio
n
rat
e
of
97.72%
w
hich
is
quit
e
de
cent
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
e
nhance
d
l
os
sle
ss c
ompre
ssion wi
th cry
pt
og
r
aphy
h
y
br
i
d mech
anism
for
…
(
M
us
taf
a Em
ad
H
ameed
)
3237
for
any
com
pressi
on
sc
hem
e.
Fu
rt
her,
the
E
CG
sig
nals
are
transm
itted
over
TCP/
IP
to
te
lem
edi
ci
ne
cl
i
nic
for
sp
eci
al
ist
’
ass
essm
ent.
Other
wise,
t
he
exi
sti
ng
of
e
ncry
ption
-
the
n
-
c
om
pr
ession
(
E
TC)
ap
proac
he
s
for
processi
ng
EC
G
data
pr
opose
d
in
[
17]
,
w
hi
ch
ca
n
protect
data
pri
vacy
a
nd
al
so
pro
vide
the
sam
e
qu
a
li
ty
of
the
reconstr
uct
ed
sig
nals
with
ou
t
sac
rifici
ng
the
com
pr
essio
n
eff
ic
ie
ncy.
F
ur
t
her
m
or
e,
an
oth
e
r
resea
rch
[18]
pro
po
se
d
sc
he
m
es
fo
r
com
pr
essi
on
-
the
n
-
encr
y
ption
of
the
MIT
-
BI
H
Arr
hythm
ia
ECG
sign
al
s
.
Their
schem
e
inv
olve
s
com
plexity
so
rtin
g
(Beat
detect
ion
,
2D
ECG
ar
ray
fo
rm
at
ion
,
P
erio
d
N
or
m
al
i
zat
ion
,
Dc
E
qual
iz
at
ion
,
C
om
plexity
Sorti
ng,
C
od
e
c
Q
uan
ti
zat
ion
an
d
JP
EG
2000
c
od
ec
)
a
nd
c
oupled
cha
otic
m
ap
m
uta
ti
on
.
I
n
their
sc
hem
e,
t
hey
assum
e
wireless
transm
i
ssion
(u
si
ng
Ra
yl
ei
gh
fad
i
ng
wireless
cha
nnel
)
of
ECG
th
rou
gh
OFDM
(
Or
t
hogo
nal
Fr
e
que
ncy
Di
vision
Mult
iplexin
g)
and
en
ha
nce
i
t
to
co
rr
ect
i
m
pair
sam
ples
us
ing
MM
F
(Mo
vin
g
Me
dian
Fil
te
ring
)
.
T
hey
show
st
or
a
ge
sp
ace
m
ini
m
iz
at
ion
thr
ough
2D
com
pr
essio
n
m
echan
ism
an
d
com
bin
ed
it
with
cha
otic
ba
sed
on
m
utatio
n
sc
hem
e
to
r
andom
iz
e
EC
G
vecto
r
for
m
a
intai
nin
g
sh
ie
ld
to
da
ta
con
fi
den
ti
al
it
y
thu
s
preve
nt
from
eaves
dro
pp
i
ng.
Conseq
uen
tl
y,
the
study
sh
ows
th
at
,
m
os
t
of
the
prev
iou
s
wor
ks
co
ncen
t
rate
m
or
e
on
qu
al
it
y
of
sign
al
ba
sed
on
pea
k
detect
i
on
but
la
cks
secu
rity
aspect.
T
houg
h
fe
w
w
orks
a
re
fou
nd
wit
h
secur
it
y
m
od
el
al
on
g
with
co
m
pr
ession
sch
e
m
es,
bu
t
t
ho
se
sche
m
es
are
not
so
eff
ic
ie
nt
in
te
r
m
s
of
lossless
com
pr
essio
n
or
hy
br
i
d
of
the
encr
y
ption
sch
e
m
e
is
inco
rpor
at
e
d
to
prov
i
de
the
se
cur
it
y.
W
e
pro
po
s
e
a
ne
w
ap
proac
h
base
d
on
lossless
sc
he
m
e
su
ch
as
A
da
ptive
Huff
m
an
f
or
c
om
pr
essio
n
s
o
that
there
is
no
inf
or
m
at
ion
lo
ss
up
on
rec
ons
tructi
on.
Als
o,
to
av
oid
the
ch
ance
s
of
data
tam
peri
ng
we
m
ake
us
e
hy
br
i
d
bet
ween
sym
m
et
r
ic
key
and
asy
m
m
e
tric
key
of
ci
ph
e
rin
g
ba
sed
on
the AES
-
CB
C,
RSA
a
nd
Dif
fie
-
Hell
m
a al
go
r
it
h
m
s w
it
h 256
-
bit
key size
.
2.
RESEA
R
CH MET
HO
D
This
sect
io
n
presents
the
pro
po
s
ed
s
olu
ti
on
for
ECG
sig
na
l
encodin
g
&
decodin
g
us
in
g
pro
pose
d
weig
hted
a
da
pt
ive
H
uffm
an
appr
oach
with
AES
-
C
BC
,
RSA
f
or
key
gen
e
rati
on
a
nd
Diff
ie
-
Hell
m
a
key
exch
a
nge
base
d
ECG
data
en
crypti
on.
The
com
plete
pr
oc
ess
is
div
ided
into
the
f
ollow
i
ng
ph
ase
s
wh
i
ch
are
as sho
wn in Fi
gure
2
.
Figure
2.
Bl
oc
k diag
ram
o
f
pro
posed
syst
em
An
EC
G
sign
a
l
is
con
sidere
d
fr
om
a
pu
blicl
y
avail
able
data
kn
ow
n
as
“M
IT
-
BI
H”.
I
n
this
sign
al
,
we
co
ns
ide
r
that
ECG
sign
al
div
ide
int
o
fou
r
blo
c
ks
with
the
sam
e
siz
e
t
o
m
a
intai
n
the
pr
oce
ssin
g
ti
m
e
and
secur
it
y
i.e.
th
e
com
plete
sign
al
will
be
processe
d
into
di
ff
ere
nt
pa
rts
and
finall
y,
it
can
be
c
om
bin
ed
at
the
receive
r
e
nd.
T
his
process
helps
t
o
m
ai
nt
ai
n
the
a
nony
m
ity
in
the
sig
nal
an
d
pr
e
ve
nt
s
the
outsi
der
a
tt
ack
on
t
he
sig
nal.
Howe
ver,
in
the
ne
xt
phase
,
we
c
onside
r
each
blo
c
k
an
d
pe
rfo
rm
the
weig
hted
A
da
ptive
Huff
m
an
encodin
g
to
achiev
e
the
com
pr
essed
ECG
sig
nal.
Af
te
r
c
om
pr
ession,
we
ap
pl
y
AES
-
CB
C,
256
-
bit
key
inc
orp
or
at
e
RSA
al
gorithm
fo
r
t
he
ke
y
gen
e
rati
on
with
Diff
ie
-
He
llm
a
key
exchang
e
-
based
en
crypti
on
schem
e
to
encr
ypt
of
eac
h
co
m
pr
essed
blo
c
k.
At
this
sta
ge
,
this
data
is
re
ady
for
tra
ns
m
i
ssion
ov
e
r
uns
ecur
e
d
channels.
Lat
e
r,
we
a
pply
A
ES
base
d
de
cr
ypti
on
sc
hem
e
to
each
recei
ve
d
bl
ock
t
o
de
crypt
the
data
into
it
s
com
pr
esse
d
or
encode
d
f
or
m
.
Ther
e
f
or
e,
i
n
the
ne
xt
phase,
we
ap
ply
an
A
dap
ti
ve
H
uffm
an
dec
odin
g
sc
hem
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
32
3
5
-
32
4
3
3238
to
reconst
ru
ct
the
data
into
it
s
or
i
gin
al
f
or
m
for
each
blo
c
k
an
d
al
l
blo
c
ks
are
c
om
bin
ed
to
f
orm
the
act
ual
sign
al
vect
or.
Finall
y,
we
ev
al
uate
the
perf
or
m
ance
o
f
c
om
pr
ession
an
d
creypto
gr
a
ph
y
in
te
r
m
s
of
PRD,
com
pr
essio
n
ra
ti
o
an
d
sec
ur
it
y analy
sis.
2.1.
Sign
al
bloc
k c
reat
i
on
Let
noisy
sig
na
l be
NS
,
len
gth o
f
it
is L,
num
ber o
f
blo
c
ks
r
equ
i
red to
be NB, t
hu
s
, th
e
bl
ock
siz
e
ℬ
=
ℒ
/
ℬ
(1
)
Now,
ℬ
(
)
is t
he
se
t of sig
nal
|
|
, th
a
t can
be
c
om
pute
d
as:
ℬ
(
)
=
|
|
=
(
(
(
−
1
)
∗
ℬ
)
+
1
,
(
(
−
1
)
∗
ℬ
)
+
2
,
.
.
,
ℬ
∗
i
)
(2
)
wh
e
re
ℬ
,
is t
he E
CG sig
nal
blo
c
k.
2.2
.
Adapt
i
ve
h
uff
man
enc
od
in
g
Accor
ding
to
the
pro
posed
flow,
in
this
sta
ge,
we
perfor
m
ECG
data
c
om
pr
essio
n
usi
ng
ada
ptiv
e
Huff
m
an
encodin
g
schem
e.
The
H
uffm
an
cod
i
ng
is
wi
de
ly
ado
pte
d
f
or
data
com
pr
ession
or
e
nc
od
i
ng,
accor
ding
to
t
his
proces
s
a
set
of
val
ues
of
fixe
d
co
der
word
s
a
re
re
placed
by
the
o
ptim
al
se
t
of
co
de
words
[
19
]
.
T
he
H
uffm
an
cod
i
ng
c
onside
rs
any
in
put
data
as
an
i
nput
stream
and
pe
rfo
rm
s
fr
equ
e
nc
y
distrib
ution
of
the d
at
a. Th
is
f
reque
ncy d
ist
ribu
ti
on is lat
er co
nsi
der
e
d
to c
om
pr
ess the d
a
ta
. I
n
the f
ir
st ph
a
se,
data
is
com
pr
e
ssed
with
the
he
lp
of
dy
nam
ic
bit
reduct
io
n
te
chn
i
qu
e
a
nd
i
n
sec
ond
ph
ase
Huff
m
an
co
din
g
is
us
e
d
to
c
om
pr
ess
the
data
f
urt
her
t
o
produc
e
the
final
outpu
t.
I
n
the
first
ph
a
se,
wh
e
n
a
us
er
enter
s
an
input
te
xt
data,
the
syst
e
m
wil
l
find
ou
t
the
occ
urr
ence
of
the
nu
m
ber
of
un
i
qu
e
sy
m
bo
ls
in
the
inp
ut
te
xt
strin
g
an
d
then
num
eric
cod
e
will
be
assigne
d
to
these
un
iq
ue
sym
bo
ls.
Fo
r
eac
h
num
eric
cod
e,
c
orres
pondin
g
bi
nar
y
cod
e
s
will
be
gen
e
rated
dyna
m
ic
al
l
y
to
ob
ta
in
the
(c
om
pr
essed)
bin
a
ry
outp
ut.
The
n
th
e
AS
CI
I
co
de
will
be
gen
e
rated
f
r
om
the
bin
ary
o
utput
ob
ta
in
ed
wh
ic
h
will
serv
e
as
the
in
put
to
the
second
ph
ase
of
the
s
yst
e
m
.
In
th
e
seco
nd
ph
a
se,
H
uffm
an
Co
ding
will
be
ap
plied
to
t
he
outp
ut
of
the
first
ph
a
se
to
furthe
r
c
ompress
the d
at
a
an
d
im
pro
ve
the
p
e
rfo
rm
ance o
f d
yn
a
m
ic
b
it
r
ed
uction
al
gorit
hm
.
Huff
m
an
co
ding
f
ollo
ws
a
t
op
-
dow
n
a
ppr
oac
h
m
eans
the
bin
ary
tree
is
bui
lt
fr
om
the
top
to
do
wn
t
o
gen
e
rate
a
n
op
tim
a
l
resu
lt
.
In
H
uffm
an
Codi
ng
t
he
c
har
act
ers
in
a
data
file
are
c
onve
rted
to
bi
nar
y
c
ode
a
nd
the
m
os
t
co
m
m
on
char
act
er
s
in
the
file
hav
e
t
he
shorte
st
bin
ary
code
s,
and
the
c
ha
racters
w
hich
are
le
as
t
com
m
on
will
hav
e
t
he
lo
ngest
bin
ary
c
ode.
In
a
sim
il
ar
way
m
e
tho
d
of
dec
om
pr
ession
works
in
r
eve
r
s
e
order.
Com
pr
e
ssed
data
is
fir
st
deco
m
pr
ess
ed
by
Huff
m
an
Dec
oder
a
nd
then
by
dy
na
m
ic
te
xt
co
m
pr
e
ssio
n
decode
r
to
get
back
t
he
ori
gi
nal
data.
F
ollow
i
ng
a
re
the
ste
ps
to
c
ompress
t
he
data
with
the
help
of
ou
r
pro
po
se
d
syst
e
m
.
Accor
ding
to
the
pro
pose
d
ap
proac
h,
init
ia
ll
y,
the
occ
urren
ce
of
uniq
ue
sym
bo
ls
is
identifie
d,
a
nd
a
uniq
ue
nu
m
eric
co
de
value
i
s
assig
ned
to
t
he
ide
ntifie
d
gro
up
of
sym
bo
ls.
I
n
the
ne
xt
sta
ge,
a
bin
ary
co
de
word
is
ge
nerat
ed
for
each
sy
m
bo
l
assigned
init
ia
ll
y.
This
bin
a
ry
cod
e
w
ord
re
pr
e
sents
the
com
pr
esse
d
data.
Huff
m
an
c
od
i
ng
is
bas
ed
on
the
t
op
-
dow
n
c
om
pu
ta
ti
on
ap
proac
h
w
her
e
a
bin
ary
t
ree
is
const
ru
ct
e
d
t
o
ge
ner
at
e
the
op
ti
m
al
so
luti
on
f
or
data
co
m
pr
ession.
Ac
cordin
g
t
o
t
he
co
nce
pt
of
H
uffm
an
cod
i
ng,
the
da
ta
wh
ic
h
is
conve
rted
int
o
bin
ary
c
od
e
is
con
si
der
e
d
for
analy
sis
wh
e
re
m
os
t
c
omm
on
char
act
e
rs
of
input
stream
as
assigne
d
s
hortest
bin
a
ry
cod
e
an
d
le
ast
co
m
m
on
char
act
ers
are
a
ssigne
d
longest
bin
a
ry
co
des.
Sim
il
a
rly
,
the
data
de
com
pr
essio
n
is
pe
rfor
m
ed
by
a
pp
ly
in
g
th
e
re
ver
se
proc
ess
of
Huff
m
an
enc
odin
g.
2.3.
Hy
brid
cry
pt
ogra
ph
y
f
or
ECG
si
gnal
In
this
sect
io
n,
we
present
th
e
crypto
gr
a
phy
m
od
el
to
secure
the
ECG
sig
nal
w
hile
trans
m
itti
ng
ove
r
the
unsecu
red
channel.
I
n
or
der
to
perform
this
ta
sk
,
we
us
e
AE
S
256
-
bit
key,
base
d
encr
y
ption
m
od
el
to
gen
e
rate
the
ci
ph
e
rs
w
hic
h
ar
e
ob
ta
ine
d
usi
ng
the
Ci
pher
B
lock
Chai
ning (
CB
C)
m
e
th
od.
The
A
ES
ap
proac
h
is
a
sym
m
et
ric
key
al
go
rithm
w
hich
ta
kes
128
-
bit
blo
c
k
i
nput
data
for
th
e
giv
e
n
sp
eci
fi
ed
ti
m
es
[2
0].
In
t
his
appr
oach,
enc
r
ypti
on
a
nd
de
c
ryptio
n
m
od
ul
es
us
e
t
he
sam
e
keys w
hic
h
a
re
assig
ne
d
by
the
us
e
r.
In
thi
s
w
ork,
we
inc
orp
or
at
e
R
SA
based
key
ge
ner
at
i
on
m
od
el
w
hi
ch
is
desc
ribe
d
la
te
r
in
this
sect
ion
.
The
siz
e
of
an
enc
rypt
-
dec
rypt
key
m
a
y
var
y
as
128,
192
a
nd
256
bits
f
or
the
input
blo
c
k
of
128
bi
t.
Encr
ypti
on
m
od
ule
gen
e
rates
the
c
iph
e
r
data fro
m
the
input
ECG
data b
loc
k
a
nd
d
ec
r
ypti
on p
r
ocess
recons
tructs
the o
ri
gin
al
dat
a
fro
m
ci
ph
er
bl
ock
s
of
ECG
sign
al
.
H
owe
ver,
enc
rypt
–
de
crypt
key
pla
ys
an
im
po
rtant
r
ole
in
this
fiel
d
because
the
num
ber
of
r
ou
nd
s
t
o
be
pe
r
form
ed
on
th
e
inp
ut
bl
ock
d
epe
nds
on
the
siz
e
of
th
e
key.
The
init
ia
l
rou
nd
of
AES
e
nc
ryptio
n
sche
m
e
per
f
or
m
s
“Add
Ro
und
K
ey
”
in
wh
ic
h
the
in
pu
t
EC
G
data
is
XO
R
’ed
with
RSA
ge
ner
at
e
d
ci
phe
r
key.
The
Fig
ur
e
3
sh
ows
a
flo
w
char
t
of
AES
base
d
e
ncr
y
ption
an
d
decr
y
ption m
eth
od
ology.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
e
nhance
d
l
os
sle
ss c
ompre
ssion wi
th cry
pt
og
r
aphy
h
y
br
i
d mech
anism
for
…
(
M
us
taf
a Em
ad
H
ameed
)
3239
Figure
3
.
Bl
oc
k diag
ram
s o
f
t
he AES
alg
or
it
hm
In
orde
r
to
pe
r
form
the
encr
y
ption,
we
us
e
Ci
ph
er
Bl
oc
k
Chainin
g
(CBC
)
Mo
de
of
A
ES
al
gorithm
.
Accor
ding
to
t
his
sc
hem
e
chaining
of
in
put
data
is
ap
plied
with
the
pre
vio
us
ci
pherte
xt
blo
c
ks
[21].
I
n
orde
r
to
do
this,
it
re
qu
i
res
an
i
niti
al
iz
at
ion
vecto
r
(IV)
wh
ic
h
i
s
no
t
sec
ur
e
but
unpredict
able.
Accor
ding
to
CB
C,
init
ia
ll
y,
the
first
blo
c
k
of
an
inp
ut
sig
nal
is
XO
R
-
ed
with
the
init
ia
li
za
tio
n
vecto
r
w
hich
ge
ner
at
es
th
e
first
ci
ph
e
r
blo
c
k.
T
he
gen
e
rated
ci
ph
e
r
blo
c
k
is
then
X
OR
-
e
d
w
it
h
the
ne
xt
pla
in
te
xt
or
in
put
ECG
sig
nal.
T
hu
s
,
the
ci
ph
e
r
bloc
ks
are
c
reate
d
by
perform
i
ng
t
he
X
OR
operati
on
betwe
en
the
ci
phe
r
blo
c
k
an
d
plaint
te
xt.
Figure
4
de
pi
ct
s
the
CB
C
m
od
e
of
e
ncry
ption
.
The
de
crypti
on
of
CB
C
m
od
e
can
s
how
in
Figure
5.
The
E
nc
ryptio
n process
can
be p
resen
te
d
as:
(
)
=
ℰ
(
,
[
(
)
⊕
(
)
−
1
]
)
,
=
1
,
2
,
.
.
,
(3
)
wh
e
re
ℰ
de
note
s
the
e
ncr
y
ption
functi
on
w
hich
c
ons
ider
s
as
the
sec
ret
ke
y,
X
OR
ope
rati
on
betwee
n
plai
n
sign
al
bl
ock
(
)
with
the
ci
phe
r
of
t
he
previ
ou
s
blo
c
k,
t
his
process
is
r
epeate
d
f
or
al
l
bl
ock
s
[22]
.
Si
m
il
arly
, th
e d
ecry
ption p
rocess i
s
perform
ed
as:
Figure
4.
Enc
r
ypti
on
i
n
th
e C
BC
m
od
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
32
3
5
-
32
4
3
3240
Figure
5.
Dec
r
ypti
on
i
n
th
e C
BC
m
od
e
(
,
(
)
)
=
(
,
ℰ
(
,
[
(
)
⊕
(
)
−
1
]
)
(
,
(
)
)
=
(
)
⊕
(
)
−
1
}
(4
)
(
,
(
)
)
⊕
(
)
−
1
=
(
)
−
1
⊕
(
)
−
1
⊕
(
)
=
(
)
Wh
e
re
D
re
pr
e
sents
the
de
cry
ption
f
unct
io
n
secret
key
an
d
ci
ph
e
r
bl
ock
s
a
nd
r
et
urns
t
he
pl
ai
n
bl
ock.
In
order
t
o
ge
ner
at
e
the
public
key,
we
us
e
the
RS
A
al
gorithm
a
nd
t
he
ge
ne
r
at
ed
key
is
use
d
in
the
aforem
entio
ne
d
process
of
e
ncr
y
ption
and
decr
y
ptio
n
[
23
]
.
T
he
R
SA
key
gen
e
r
at
ion
al
gorith
m
is
as
fo
ll
ows:
RSA
KeyGe
n()
Inp
ut: Pr
ov
i
de t
wo
la
rg
e
prim
e num
ber
s as
p an
d q
Ou
t
pu
t:
P
ubli
c and private
ke
ys for e
ncr
y
ption an
d dec
rypt
ion
.
Step
1: co
m
pu
t
e n
a
s
←
∗
Step
2: co
m
pu
t
e Eu
le
r
phi
val
ue
as
(
)
←
(
−
1
)
∗
(
−
1
)
Step
3: selec
t a
r
a
ndom
n
um
ber
wh
ic
h
sat
isfi
es the c
onditi
on
1
<
<
(
)
an
d
gcd
(
,
(
)
)
=
1
Step
4: co
m
pu
t
e d
a
s
←
−
1
(
(
)
)
In
t
his
w
ork,
we
use
the
D
H
key
exc
ha
ng
e
protoc
ol.
Be
lo
w
gi
ven
Fig
ur
e
6
s
hows
a
sa
m
ple
rep
resen
t
at
ion
of
Dif
fie
-
Hell
m
a
key exc
ha
ng
.
Figure
6.
Dif
fi
e
-
hellm
a k
ey
ex
cha
nge
m
od
el
3.
RESU
LT
S
A
ND
DI
SCUS
S
ION
S
In
this
sect
ion,
we
pr
ese
nt
the
exp
erim
ental
resu
lt
and
are
of
the
propose
d
appr
oach
to
pe
rfor
m
ECG
sign
al
filt
erin
g,
c
om
pr
essio
n,
a
nd
e
nc
ryp
ti
on
.
The
c
om
ple
te
exp
e
ri
m
ental
stud
y
is
carrie
d
out
us
i
ng
MATLAB
to
ol
and
vali
dated
on
publicl
y
avail
able
MIT
-
BIH
ar
r
hythm
ia
dataset
.
This
da
ta
bas
e
co
ntains
48
sign
al
rec
ords
(36
0H
z
)
of
47
un
iq
ue
patie
nt
s
and
with
a
le
ng
t
h
of
30
m
i
nu
te
s
eac
h.
To
com
par
e
the
resu
lt
s
with
the
m
et
ho
ds
i
n,
t
he
ex
pe
rim
ental
per
fo
rm
ance
us
in
g
the
sig
nals
nu
m
ber
ed
as
101,
102,
10
8,
202,
205,
209, 21
9
a
nd 223 are
r
e
ported
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
e
nhance
d
l
os
sle
ss c
ompre
ssion wi
th cry
pt
og
r
aphy
h
y
br
i
d mech
anism
for
…
(
M
us
taf
a Em
ad
H
ameed
)
3241
3.1.
Co
m
pressi
on
perf
ormance
The
n,
in
the
ne
xt
phase,
the
pur
pose
of
the
ECG
sig
nal
co
m
pr
ession
is
to
achieve
hi
gh
com
pr
essio
n
rates
with
out
changin
g
t
he
qu
al
it
y
of
the
sign
al
.
The
c
om
pr
ession
rate
sho
uld
be
c
he
cked
with
th
e
oth
e
r
par
am
et
ers
to
evaluate
t
he
exp
erim
ental
resu
lt
s
of
th
e
qu
al
it
y
of
the
reconstr
uct
ed
sign
al
.
T
he
refor
e
,
the
c
om
pr
ession
pa
ram
et
ers
are
sel
ect
ed
to
va
li
date
th
e
dyna
m
ic
com
pr
ession
m
echan
ism
,
we
evaluate
the
pe
rfor
m
ance
in
te
rm
s
of
com
pr
essio
n
rati
o
an
d
PR
D
.
The
c
om
pr
es
sion
rati
o
(CR
)
is
the
m
easur
e
of
com
pr
essio
n
a
chieve
d
in
sig
na
l
thr
ough
e
nc
od
i
ng
m
echan
ism
s.
It
do
es
n’t
pro
vid
e
inf
or
m
at
ion
on
c
ompress
ed
sign
al
qual
it
y bu
t m
easur
es
the e
ff
ic
ie
ncy
of a
n
al
gorithm
in
re
duci
ng sto
rag
e
sp
ace
[1
7].
=
∑
(
(
[
]
̅
̅
̅
̅
̅
̅
)
)
=
1
∑
(
(
[
]
)
)
=
1
(5
)
Her
e
,
the
e
ncoded
si
gnal
(S
[
n]
)
̅
a
nd
or
i
gin
al
sig
nal
S[n]
is
conver
te
d
to
bits
an
d
the
n
the
le
ng
t
h
o
f
encode
d
bits
is
div
i
ded
with
ori
gi
nal
sig
nal
le
ngth
to
fin
CR
val
ue.
The
lo
we
r
the
CR
,
the
be
tt
er
is
the
ef
fici
ency.
Howe
ver,
the
Perce
ntage
R
oot
-
m
ean
-
squar
e
Dif
fer
e
nce
(
PRD)
is
a
m
easur
e
to
eval
uate
erro
r
or
dif
fer
e
nce
be
tween
t
he
or
i
gin
al
S
[
n]
an
d
rec
on
st
ru
ct
e
d
sign
al
(S
̂
[
n]),
wh
e
re
N
is
t
he
le
ng
t
h
of
sig
na
l
S.
The pr
ocess
ca
n be
def
i
ned as
[16
]
:
=
√
∑
(
[
]
−
̂
[
]
)
2
=
1
∑
[
]
2
=
1
×
100
(
6
)
T
he
ab
ove
pro
cess
m
ay
no
t
giv
e
an
acc
ur
a
te
m
easur
e
m
ent
of
perform
ance
as
PRD
reli
es
m
uch
on
the
m
ean
valu
e
of
the
ori
gina
l
sign
al
.
Re
se
arch
e
rs
pr
e
viously
est
ablishe
d
that
t
her
e
shou
l
d
be
e
xtra
effort
s
ta
ken
t
o
cl
ean
the
bas
el
ine
or
redu
ce
t
he
D
C
le
vel.
T
o
overco
m
e
this
issue
a
m
od
ifie
d
and
ad
va
nced
ver
si
on
of PRD
equati
on is
descr
i
bed in the
p
a
st w
hi
ch
is i
nd
e
pe
ndent of m
ean v
a
lue:
_
=
√
∑
(
[
]
−
̂
[
]
)
2
=
1
∑
(
[
]
−
́
[
]
)
2
=
1
×
100
(7
)
Her
e
,
S ́[n]
i
ndic
at
e
the
m
ea
n
valu
e
of
t
he
ori
gi
nal
si
gn
a
l
S[n]
.
Since
,
the
disto
rtion
i
ncr
ease
s
a
s
the
CR
rati
o
increases,
a
com
pr
essi
on
rati
o
s
uitable
for
the
ty
pe
of
ser
vice
and
wi
reless
channel
en
vir
onm
ent
sh
oul
d
be
sel
ec
te
d.
I
n
or
der
to
com
pr
ess
an
d
deco
m
pr
ess
t
he
sign
al
,
we
ha
ve
de
velo
pe
d
a
n
ada
ptive
Huf
fm
an
appr
oach
t
o
ac
hieve
a
bette
r
com
pr
essio
n
ra
ti
o.
A
ny
increa
se
in
CR
m
a
y
resu
lt
in
a
decl
ine
in
the
qu
al
i
ty
of
the
signa
ls
rec
on
st
ru
ct
e
d.
T
he
qu
al
it
y
scor
e
(Q
S
)
is
us
e
d
t
o
assess
c
om
pr
essio
n
ef
fic
ie
ncy
w
hile
ta
kin
g
i
nto
consi
der
at
io
n t
he reco
ns
tr
uction m
ist
akes th
at
h
ave
b
e
en
c
om
pr
om
ise
d
[24]. T
he defi
niti
on of t
he QS is
:
=
(8
)
A
high
Q
S
va
lue
co
rr
es
pond
s
to
a
high
qual
it
y
of
com
pressi
on
in
a
l
ossy
com
pr
ession
m
et
ho
d.
Ther
e
f
or
e,
Ta
ble
1
il
lustrate
s
CR
,
PRD
and
QS
par
a
m
et
ers
fo
r
dif
fer
e
nt
ECG
re
cords
of
the
MIT
-
B
I
H
arrhyt
hm
ia
database.
All
depi
ct
ed
cu
rv
e
s
ar
e
of
sim
il
ar
natur
e
(as
CR
i
nc
reases
PRD
a
lso
inc
reases
)
wh
ic
h
assures the
wo
rk
i
ng ab
il
it
y of t
he pr
opos
e
d m
et
ho
d o
n diffe
ren
t
EC
G rec
ords.
Table
1.
Re
s
ults
obta
ine
d wit
h p
rop
os
ed
ad
a
pt
ive huffm
an
cod
i
ng for
8
re
c
ords
Dataset
CR
PRD (%
)
QS
100
2
1
.68
0
.22
9
8
.54
102
3
7
.6
0
.15
2
3
7
.76
108
1
8
.25
0
.20
9
1
.25
117
3
2
.13
0
.24
1
2
9
.58
118
2
3
.8
0
.18
1
2
9
.32
209
2
4
.45
0
.18
1
2
8
.
8
8
219
2
1
.69
0
.19
1
0
9
.09
223
2
9
.06
0
.16
1
7
5
.45
The
pro
posed
co
m
pr
essio
n
s
chem
e
achieve
s
im
po
rtant
c
om
pr
ession
rati
os
w
it
h
a
ve
ry low
PRD
rate.
It
reaches
a
m
ean
com
pr
ession
rati
o
of
23.91
f
or
the
adap
ti
ve
Huffm
an
cod
in
g.
t
hu
s
,
the
m
ean
PRD
is
the
sam
e
fo
r
th
e
diff
e
re
nt
dat
a
set
and
is
eq
ual
to
0.2%
.
T
able
2
pr
ese
nts
a
com
par
ison
with
a
dif
fer
e
nt
sta
te
of
the
art
te
chni
qu
es
.
It’s
obvi
ou
s
that
our
c
om
pr
ession
al
gorithm
ou
tpe
rfo
rm
s
the
sta
te
-
of
-
the
-
art
te
ch
niq
ue
s
in term
o
f
acc
uracy
and e
ff
ic
i
ency
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
32
3
5
-
32
4
3
3242
Table
2.
Per
for
m
ance
of
diff
e
ren
t
EC
G
si
gnal
co
m
pr
essio
n schem
es f
or
re
cord
n
o.
100
Alg
o
rith
m
CR
PRD (%
)
QS
Prop
o
sed
Algo
rith
m
2
1
.68
0
.22
9
8
.54
CS
[
2
5
]
1
7
.76
5
.60
-
Static L
o
w
cod
in
g
Dy
n
a
m
i
c L
o
w
cod
in
g
Dy
n
a
m
i
c with
m
in
i
m
u
m
win
d
o
w size
[
2
6
]
3
7
.5
4
0
.82
4
2
.13
0
.50
-
CAE
[
2
7
]
3
2
.25
1
.47
-
SVD
[
1
7
]
4
0
.1
0
.42
9
5
.47
QLV
-
Sk
eleto
Haf
f
m
an
[
2
8
]
1
6
.91
0
.64
2
9
.36
Howe
ver,
we
noti
ce
that
no
sig
nificant
disto
rtion
is
introd
uce
d
betwee
n
the
or
i
gin
al
a
nd
reconstr
ucted
s
ign
al
s.
T
he
pro
po
s
ed
ECG
c
om
pr
ession/d
ec
om
pr
essio
n
al
go
rithm
is
ver
y
low
com
plexity
and
ver
y
highly
ef
f
ic
ie
nt co
m
par
ed wit
h othe
r
a
ppr
oac
hes.
I
n
a
ddit
ion
,
the
re
no n
ee
d for
huge m
e
m
or
y resour
ces.
3.2.
ECG
si
gnal se
curi
ty anal
ys
is
In
this
sect
i
on,
we
desc
ribe
the
var
i
ous
ty
pe
s
of
at
ta
ck
s
w
hich
can
be
m
i
ti
gated
us
in
g
t
he
pro
po
s
ed
m
echan
ism
.
Hen
ce,
t
he
sec
uri
ty
of
the
data
pro
duced
m
ain
ly
de
pends
on
the
sec
ur
it
y
of
t
he
key
generato
r
.
Fu
rt
her
m
or
e,
t
he
only
at
ta
ck
m
od
el
that
can
be
ap
plied
to
the
pro
posed
e
ncr
y
ption
sc
he
m
e
is
the
atta
cks
on
pr
i
vacy,
as
des
cribe
d
in
[
29]
.
Additi
on
al
ly
,
a
ccordin
g
t
o
t
he
secu
rity
m
ana
gem
ent
aspect,
we
ha
ve
us
e
d
D
H
key
exc
hange
,
AES
c
rypto
gr
a
phy
an
d
pa
rtit
ion
in
g
the
entire
ECG
sign
al
int
o
4
boc
ks
.
T
he
refor
e
,
the
pro
posed
appr
oach
ca
n
be
us
e
d
for
a
vo
i
ding
the
fol
lowing
at
ta
ck
s
su
c
h
as
us
e
r
data
pr
i
vacy
in
this
appr
oach,
we
do
not
tra
ns
m
i
t
the
data
in
t
he
plain
-
te
xt
for
m
.
The
ori
gin
a
l
data
is
e
ncr
y
pted
us
i
ng
A
E
S
-
CBC
m
echan
ism
wi
th
a
256
-
bit
ke
y.
Hen
ce
,
pri
va
te
data
inf
orm
at
ion
can
be
secur
e
d.
The
Mutual
aut
h
en
ti
cat
ion
befor
e
tra
ns
m
itti
ng
the
data
over
an
unsecu
r
ed
wireless
ch
ann
el
,
we
us
e
the
D
H
key
exc
hange
to
aut
he
ntica
te
the
recei
ver
m
od
ule.
T
his
authe
ntica
ti
on
process
hel
ps
t
o
a
vo
i
d
e
aves
droppe
rs.
Howe
ver,
the
m
utu
al
authe
ntica
ti
on
us
in
g
DH
key
exch
a
nge
pro
cess
help
s
to
avo
i
d
the
m
an
-
in
-
t
he
-
m
idd
le
at
ta
cks.
T
he
session
Key
m
anag
em
ent
duri
ng
tra
ns
m
issi
on
of
the
data,
we
pe
rfor
m
DH
ke
y
exch
a
nge
w
hich
is
li
m
i
ted
f
or
the
cu
rr
e
nt
ses
sion.
He
nce
,
t
he
f
al
se
data
c
annot
be
tra
nsm
itted
i.e.
us
er
1
data
ca
n
only
be
tra
ns
m
i
tted
f
or
us
er
1’
s
ses
sio
n
by
us
i
ng
th
e
par
ti
cula
r
session
key.
O
n
an
oth
e
r
ha
nd,
the
ph
is
hing
at
ta
ck:
the
authe
ntica
ti
on
process
help
s
to
av
oid
the
f
al
se
receive
r
m
od
ule
he
nce,
the
ph
is
hing
at
ta
cks
can
be
av
oid
e
d
us
in
g
this
appr
oach.
Co
nse
quently
,
t
he
AES
-
CB
C
with
a
256
-
bit
ke
y
based
enc
ryp
ti
on
sc
hem
e
helps
to
sec
ure
the
dat
a
wh
il
e
se
nd
i
ng
the
data
int
o
the
e
ncr
y
pted
f
or
m
[2
1].
More
ov
e
r,
w
e
div
ide
t
he
or
i
gin
al
sig
nal
into
m
ul
ti
ple
blo
ck
s
w
hich
inc
re
ases
the
sec
uri
ty
to
var
io
us
at
ta
cks.
T
he
pro
po
se
d
a
ppr
oac
h
al
so
r
est
rict
s
the eave
sdro
pp
ing
att
ac
ks
by
us
in
g
a
n
e
ncr
y
pted key e
xcha
ng
e
m
echan
ism
.
4.
CONCL
US
I
O
N
In
this
work,
we
fo
c
us
on
the
dev
el
opm
ent
of
a
secur
e
d
an
d
ef
f
ic
ie
nt
m
echan
is
m
fo
r
EC
G
processi
ng
w
he
re
EC
G
si
gn
a
l
com
pr
essio
n,
an
d
e
ncr
y
ptio
n
a
re
the
m
ai
n
ta
sk
s.
Howe
ve
r,
T
he
ECG
s
ign
al
com
pr
essio
n
ta
sk
is
done
us
ing
a
n
ada
ptive
Huff
m
an
cod
i
ng
sc
hem
e
and
data
encr
ypti
on
is
achieve
d
us
in
g
the
AES
enc
ryption
process.
In
or
der
to
inc
rease
secu
rity
,
we
inco
rpo
rate
RSA
key
generati
on
an
d
D
H
key
exch
a
nge
prot
oco
ls
.
T
he
perform
ance
of
the
pro
po
se
d
a
ppr
oach
is
m
e
asur
e
d
i
n
te
rm
s
of
PR
D,
CR
a
nd
secur
it
y
eff
ic
a
cy
.
the
obta
in
ed
pe
rfo
rm
ance
is
com
par
ed
with
the
e
xis
ti
ng
te
ch
nique
s,
the
c
om
par
at
ive
analy
sis
sh
ow
s
that
the
pro
po
s
ed
a
ppr
oac
h
achie
ves
bet
te
r
perform
ance.
Th
us
,
the
r
esults
of
t
his
pap
e
r
appr
ov
e
the
a
pp
li
cabil
it
y
of
the
propose
d
m
et
ho
d,
f
or
the
com
pr
ession
m
echan
ism
,
and
c
rypto
gr
aph
ic
te
chn
iq
ue.
Thi
s
m
et
ho
d
is
the
n
belie
ve
d
to
s
erv
e
as
a
novel
syst
e
m
m
od
el
for
the
high
co
m
pr
ession
rati
o
wit
h
m
ini
m
u
m
r
e
con
str
uction er
ror
a
nd a
high le
vel of sec
uri
ty
.
REFERE
NCE
S
[1]
E.
Yuli
ant
o
,
A.
Sus
ant
o,
T.
S.
W
idodo,
and
S.
W
ibowo,
“
Cla
ssi
f
y
ing
th
e
EE
G
Signal
through
Stim
ulus
of
Motor
Movem
ent
Us
ing
New
T
y
pe
of
W
ave
le
t
,
”
IAE
S
Inte
rnational
Jo
urnal
of
Artifi
ci
a
l
Intelli
g
ence
(
IJ
-
AI)
,
vol.
1
,
no
.
3,
pp.
139
–
148
,
Se
p
.
2012.
[2]
J.
S.
Sahambi,
S.
N.
Ta
ndon
,
an
d
R.
K.
P.
Bhat
t
,
“
Us
ing
wave
le
t
tra
nsform
s
for
ECG
cha
ra
ct
er
izati
on
.
An
on
-
li
n
e
digi
tal
signal
pr
oce
ss
ing
s
y
stem
,
”
in
I
EEE
Eng
ine
ering
in
Me
dic
in
e
and
Bi
o
l
ogy
Magaz
ine
,
vol.
16,
no
.
1,
pp.
77
-
83
,
Jan
.
-
Feb.
1997
.
[3]
A.
Alesa
nco
and
J.
Garc
´
ıa
,
“
Autom
at
ic
Real
-
T
i
m
e
ECG
Coding
Methodol
og
y
G
uar
antee
ing
Sign
al
Inte
rpr
etati
on
Quali
t
y
,
”
in
I
EEE
Tr
ansacti
ons
on
Bi
omed
ic
al
E
ngine
ering
,
vol
.
55,
no
.
11
,
pp
.
2
519
-
2527,
Nov.
2008.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
e
nhance
d
l
os
sle
ss c
ompre
ssion wi
th cry
pt
og
r
aphy
h
y
br
i
d mech
anism
for
…
(
M
us
taf
a Em
ad
H
ameed
)
3243
[4]
D.
M.
Mirris
and
A.
L.
Goldber
ger
,
“
Braunwa
ld
:
Hea
rt
Disea
se:
A
Te
xtb
ook
of
Cardi
ovasc
u
la
r
Medic
in
e,
”
6th
e
d.
,
Cop
y
righ
t
©
200
1
W
.
B.
Saunder
s Com
pan
y
,
200
1.
[5]
H.
Kupw
ade
Pat
il
and
R
.
Seshad
ri,
“
Big
Dat
a
Se
cur
ity
and
Priva
c
y
Iss
ues
in
Hea
lt
hc
are
,
”
2014
I
EE
E
In
te
rnation
al
Congress
on
Bi
g
Data,
Anchora
g
e
,
AK
,
pp.
762
–
765,
2014
.
[6]
K.
Gai,
Y.
W
u,
L.
Zhu
,
and
M.
Qiu,
“
Privacy
-
pr
ese
rving
En
erg
y
Tra
ding
Us
ing
Consortium
Bloc
kchain
in
Sm
art
Grid,
”
in
I
EEE
T
rans
act
ions o
n
I
ndustrial
Informatic
s
,
vol
.
15
,
no
.
6
,
pp
.
354
8
-
35
58,
Jun.
2019
.
[7]
J.
Andreu
-
Pere
z
,
C.
C.
Y.
Poon,
R.
D.
Me
rrif
ie
ld
,
S.
T
.
C.
W
ong,
and
G.
Z
.
Yang
,
“
Big
Data
fo
r
Hea
lt
h
,
”
in
IEEE
Journal
of
Bi
om
edi
ca
l
and
Health Inf
orm
ati
cs
,
v
ol
.
19
,
no
.
4
,
pp
.
1193
-
1208,
Jul
.
2015.
[8]
S.
K.
Mukhopa
dh
y
a
y
,
S.
Mi
tra,
and
M.
Mit
ra,
“
An
ECG
signal
compress
ion
t
ec
hniqu
e
using
AS
CII
cha
racte
r
enc
oding
,
”
Me
as
urement
,
vo
l. 45, no. 6, pp. 1651
–
1660,
Jul.
2012.
[9]
M.
K.
Abdulha
m
ee
d,
Z.
Za
k
aria
,
I.
Ibr
ahi
m
,
an
d
M.
K.
Mohs
en,
“
Novel
design
of
tri
ple
-
b
and
E
BG
Novel
design
of
tri
ple
-
band
EB
G,”
TEL
KO
MNIKA
(
Tele
communic
ati
on
Comp
uti
ng
Elec
tron
i
c
s
and
Control
)
,
vol.
17,
no.
4
,
pp.
1683
–
1691
,
Aug
.
2019.
[10]
D.
Gurve,
B.
S.
Saini
,
and
I
.
Saini,
“
An
impro
ve
d
loss
y
and
l
oss
le
ss
combined
ECG
dat
a
co
m
pre
ss
ion
using
AS
CII
cha
rac
t
e
r
enc
oding
,
”
In
te
rnational
Jou
rnal
of
Me
di
ca
l
Engi
ne
ering
and
Informatic
s
,
vol.
8
,
no.
4
,
pp.
758
–
764,
20
16.
[11]
A.
F.
Hus
sein,
S.
J.
Hashim
,
A.F.A.
Aziz,
F.Z
.
R
okhani
,
and
W
.
A.
W
.
Adnan,
“
A
rea
l
t
ime
EC
G
dat
a
compres
sion
sche
m
e
for
enhance
d
bluetooth
l
ow
ene
rg
y
ECG
s
y
stem
power
co
nsum
pti
on,
”
Jou
rnal
of
Ambie
nt
Inte
lligen
ce
and
Hum
anized
Computing
,
p
p
.
1
–
14,
Aug
.
2017
.
[12]
K.
Gai,
M
.
Qiu,
Y.
Li,
and
X
.
Li
u,
“
Advanc
ed
Full
y
Hom
om
o
rphic
En
cr
y
pti
o
n
Scheme
Over
Rea
l
Num
ber
s,
”
2017
IEE
E
4
th
Inte
rnational
C
onfe
renc
e
on
C
ybe
r
Sec
uri
ty
a
nd
Cloud
Computing
(
CSCloud)
,
New
York,
NY
,
pp.
64
-
69
.
2017
.
[13]
H.
Al
-
Ham
adi
,
A.
Gawanm
eh,
J
.
Ba
ek,
and
M.
Al
-
Quta
y
ri,
“
Li
g
htwei
ght
Se
cur
ity
Proto
col
for
E
CG
Bio
-
Sensors
,
”
in
Wire
le
ss
P
ersonal
Comm
unications
,
vo
l. 95, n
o.
3
,
pp
.
5097
–
5
120,
Apr
.
2017.
[14]
M.
Fira,
“
Appli
ca
t
ions
of
comp
ressed
sensing:
Com
pre
ss
ion
and
enc
r
y
p
ti
on,
”
2
015
E
-
Heal
th
a
nd
Bi
o
engi
ne
eri
ng
Confe
renc
e
(
EHB)
,
Iasi,
pp.
1
-
4
,
2015
.
[15]
J.
Ma,
T.
Zha
ng
,
and
M.
Dong,
“A
novel
ECG
dat
a
compress
ion
m
et
hod
using
ada
pti
v
e
fourie
r
de
compos
it
ion
with
sec
urity
guar
antee
in
e
-
he
al
th
ap
pli
c
at
ions,
”
in
IEE
E
Journal
of
Bi
omedi
cal
and
Healt
h
Informati
cs
,
vol.
19,
no.
3
,
pp.
986
-
994
,
Ma
y
2015.
[16]
M.
Rae
iatiba
n
a
dkooki
and
S.
R.
Qucha
ni
,
“
Com
pre
ss
ion
and
Enc
r
y
p
ti
on
of
ECG
Signal
Usi
ng
W
ave
le
t
an
d
Chaot
ical
l
y
Huff
m
an
Code
in
T
elem
edi
ci
n
e
Appl
i
ca
t
ion,
”
Mobi
le
Syste
ms
,
vol
.
40
,
no.
3,
pp.
1
-
8,
Jan
.
2016
.
[17]
T.
Y.
L
iu,
K.
J.
L
in,
and
H.
C
.
W
u,
“
ECG
dat
a
e
ncr
y
p
ti
on
th
en
c
om
pre
ss
ion
usin
g
singula
r
val
u
e
dec
om
positi
on,
”
in
IE
EE J
ournal
of
B
iomedi
ca
l
a
nd
Healt
h
Infor
matic
s
,
vo
l. 22,
no.
3
,
pp
.
707
-
7
13,
Ma
y
2018.
[18]
A.
Pande
y
,
B.
S.
Saini
,
B.
Singh
,
and
N.
Sood,
“
Com
ple
xity
sort
ing
and
coupl
ed
c
haot
i
c
m
ap
base
d
on
2D
ECG
dat
a
compress
ion
-
the
n
-
en
cr
y
pti
o
n
and
i
ts
OF
DM
tra
nsm
ission
with
impair
s
ample
cor
recti
on
,
”
Mult
imedi
a
Tool
s
and
Applications
,
vol
.
78
,
no
.
9
,
p
p.
11223
–
11261
,
Ma
y
2019
.
[19]
N.
A.
M.
Mus
tafa
Emad
Ham
eed,
Masrull
izam
Mat
Ibra
h
im,
“
Com
p
ression
and
Enc
r
y
pt
ion
for
ECG
Biom
edi
c
a
l
Signal
in
He
al
th
ca
re
S
y
stem,
”
TEL
KOMNIKA
(Tele
communic
a
t
ion
Computing
El
e
ct
ronics
and
Control)
,
vol.
1
7,
no.
6
,
pp
.
2826
-
2833,
De
c
.
2019
.
[20]
M.
E.
Ham
ee
d,
M.
M.
Ibra
him,
and
N.
A.
Mana
p,
“
Revi
ew
on
Im
prove
m
ent
of
Adva
nce
d
En
cr
ypti
o
n
Standa
rd
(AES
)
Alg
orit
h
m
base
d
on
Tim
e
Exe
cu
ti
on
,
Diffe
ren
t
ia
l
Cr
ypta
na
l
y
sis
and
Le
ve
l
of
Secur
i
t
y
,
”
Journal
of
Tele
communic
a
t
ion,
El
e
ct
roni
c
a
nd
Computer
En
gine
ering
,
vol
.
1
0,
no
.
1
,
pp
.
139
–
145,
2018
.
[21]
M.
L.
A.
Mus
taf
a
Emad
H
amee
d
,
Masrull
iz
am
M
at
Ibr
ahi
m
,
Nuru
lfa
j
ar
Abd
Man
a
p,
“
Com
par
at
iv
e
stud
y
of
sev
era
l
oper
ation
m
odes
of
AES
a
lgori
th
m
for
enc
r
y
p
t
ion
ECG
biomedi
cal
signal,”
Int
ernati
onal
Journal
of
Elec
tric
al
an
d
Computer
Engi
n
ee
ring (
IJE
C
E)
,
vol.
9
,
no
.
6
,
pp
.
4850
-
4859,
De
c
.
2019.
[22]
M.
Vaide
h
i
and
B.
J.
R
abi
,
“
Design
and
an
aly
s
is
of
AES
-
CBC
m
ode
for
high
sec
uri
t
y
app
li
c
at
ions,
”
Se
con
d
Inte
rnationa
l
C
onfe
renc
e
on
Curr
ent
Tr
ends
i
n
Engi
n
ee
rin
g
and
Technology
-
ICCT
ET
2014
,
Coim
bat
or
e,
pp.
499
-
502
,
20
14.
[23]
E.
R.
Arboleda
,
J.L
.
B
al
ab
a,
and
J.C.
L.
Espin
el
i
,
“
Chaot
ic
r
ive
st
-
sham
ir
-
adl
erma
n
al
gori
thm
wit
h
dat
a
en
cr
y
pti
o
n
standa
rd
sch
eduling,
”
Bulle
ti
n
o
f El
e
ct
rica
l
Eng
in
ee
ring a
nd
Infor
matic
s
,
vo
l. 6, n
o.
3
,
pp
.
219
–
22
7,
Sep
.
2017
.
[24]
S.
Le
e
,
J.
Kim
,
and
M.
L
ee,
“
A
rea
l
-
t
ime
EC
G
dat
a
c
om
pre
ss
ion
and
tra
nsm
ission
al
gorit
hm
for
an
e
-
h
ealth
devi
c
e,”
in
IE
EE Tr
ansacti
ons on
Bi
omed
ic
al
Eng
ine
ering
,
vol
.
58
,
no
.
9
,
pp
.
2448
-
2455,
Sep
.
201
1.
[25]
A.
Singh,
L.
N.
Sharm
a,
and
S.
Danda
pat
,
“
Multi
-
ch
annel
ECG
dat
a
compress
ion
using
compr
essed
sensing
in
ei
genspa
ce,”
Computers i
n
B
iol
o
gy
and
M
edicine
,
vol
.
73
,
pp
.
24
–
37,
Jun
.
2016.
[26]
H.
Anas,
R.
La
t
if,
and
M.
Ario
ua,
“
Eff
icient
elec
tro
ca
rdiog
ram
(ECG)
lossy
c
om
pre
ss
ion
sche
m
e
for
rea
l
ti
m
e
e
-
Hea
l
th
m
onit
o
ring,
”
Int
ernati
o
nal
Journal
of
Biology And Biom
edi
ca
l
Eng
ineeri
ng
,
vol
.
11
,
pp
.
1
01
–
114,
2017
.
[27]
O.
Yildi
r
im,
R.
S.
Ta
n
,
and
U.
R.
Acha
r
y
a
,
“
An
eff
i
ci
en
t
compress
ion
of
ECG
signal
s
using
d
e
ep
convo
lut
ion
al
aut
oen
code
rs,
”
Cognit
ive
S
ystem
s R
ese
arch
,
vo
l.
52
,
pp
.
198
–
21
1,
De
c
.
2018.
[28]
H.
Kim
,
R.
F.
Yaz
icioglu,
P.
Merke
n,
C.
V
an
Hoof,
and
H.
J.
Yoo,
“
ECG
sig
nal
compress
ion
and
cl
assificat
i
on
al
gorit
hm
with
quad
le
vel
ve
ctor
for
ECG
holt
er
s
y
stem,”
in
I
EE
E
Tr
ansacti
o
ns
on
Infor
mation
Technol
ogy
i
n
Bi
omedi
ci
ne
,
vo
l
.
14
,
no
.
1
,
pp
.
9
3
-
100,
Jan
.
2010
.
[29]
M.
Burhanuddi
n,
A.
Moham
m
ed
,
R
.
Ism
ai
l,
M.
E.
Ham
ee
d,
A.
N.
Kare
em,
and
H.
Basiron,
“
A
Re
vie
w
on
Secur
i
t
y
Chal
le
ng
es
and
Feat
ure
s in
W
ire
le
ss
Sensor
Networks:
IoT
Pers
pec
t
ive
,
”
Journal
of
Tele
communic
ati
on
,
Elec
tron
i
c
and
Computer
E
ngine
ering
,
vol
.
10,
no
.
1
–
7
,
pp
.
17
–
21,
2018
.
Evaluation Warning : The document was created with Spire.PDF for Python.