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.
8
,
No.
6
,
D
ece
m
ber
201
8
, pp.
4448
~
44
55
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
8
i
6
.
pp
4448
-
44
55
4448
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Wireles
s
Techn
ology
for Mon
itori
ng Site
-
s
pecifi
c Landsl
ide
in Vietn
am
Gian Qu
oc
-
A
nh
1
, Ng
uy
e
n
Dinh
-
Ch
in
h
2
, T
ran Duc
-
N
ghi
a
3
, Tr
an
Duc
-
Tan
4
,
Kieu
Thi
N
guy
en
5
,
Kumbes
an S
andrase
gar
an
6
1
,2,4
El
e
ct
roni
cs
a
nd
Telec
om
m
unic
ation
Facu
lty
,
VN
U,
Hanoi
-
Univer
sit
y
of E
ngin
ee
ring
and
Tech
nolog
y
,
Vi
et
n
am
1
Depa
rtment of
El
e
ct
roni
cs,
Na
m
Dinh
Univer
si
t
y
of Te
chnol
og
y
Educat
ion
,
Vi
e
t
n
am
3
Instit
ute of Info
rm
at
ion
T
ec
hno
l
og
y
,
Vie
tna
m
ese
Aca
dem
y
of
Sci
enc
e
and
T
ec
hno
log
y
,
Vi
et
n
am
5
Facul
t
y
of
Me
c
hani
c
al
,
E
le
c
tric
al
,
and
Elec
tronic
Eng
ine
e
ring
,
N
gu
y
en
T
at T
han
h
Univer
sit
y
,
Vi
et
nam
6
Facul
t
y
of Engi
nee
ring
and
IT
a
nd
CRIN,
Univ
e
rsit
y
of Te
chnolog
y
S
y
dn
e
y
,
Aus
tra
lia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
2
, 2
01
8
Re
vised
Jun
1
,
201
8
Accepte
d
J
un
14
, 201
8
Cli
m
at
e
ch
ange
has
ca
used
an
in
cre
asing
num
ber
of
la
ndslide
s
,
e
spec
iall
y
in
the
m
ounta
inous
provinc
es
of
V
ie
tn
am,
result
in
g
in
the
dest
ruc
t
ion
of
vit
a
l
tra
nsport
and
othe
r
infra
stru
c
ture
.
Curre
n
t
m
onit
oring
and
fore
ca
stin
g
s
y
stems
of
the
m
et
eor
olog
y
d
e
par
tment
ca
nnot
del
iv
er
ac
cu
rate
and
r
el
i
abl
e
fore
ca
sts
for
we
at
her
ev
ent
s
and
issue
ti
m
ely
wa
rnings.
Thi
s
pap
er
desc
rib
es
the
developm
ent
of
a
sim
ple
,
low
cost,
and
eff
i
ci
en
t
s
y
stem
for
m
onit
oring
and
warni
ng
la
n
dslide
in
rea
l
-
tim
e.
The
aut
hors
foc
us
on
the
us
e
of
wire
le
ss
and
re
la
t
ed
t
ec
h
nologi
es
in
th
e
implementa
t
ion
of
a
te
chn
ical
s
olut
ion
and
som
e
of
the
prob
le
m
s
of
the
wire
le
ss
sensor
net
wo
rk
(W
S
N)
rel
at
e
d
to
power
consum
pti
on.
Prom
ising
comp
ressed
sensing
(CS)
base
d
s
olut
ion
for
la
ndslide m
onit
o
ring
is d
iscussed
and evaluated in
the pa
p
er.
Ke
yw
or
d:
Com
pr
essed se
ns
in
g
Lan
ds
li
de
m
onit
or
in
g
Power c
ons
umpti
on
Re
al
-
tim
e w
arn
in
g
syst
em
Sensors
W
i
reless
Senso
r
N
et
w
ork
Copyright
©
201
8
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
:
Tran D
uc
-
Tan
,
Ele
ct
ro
nics
and Telec
omm
un
ic
at
ion
Fac
ulty
,
VNU, Ha
noi
-
Un
i
ver
sit
y o
f En
gin
eeri
ng a
nd Tec
hnol
og
y
,
E3 b
uildin
g,
144 X
u
a
n
T
huy,
Cau
Giay
, Ha
No
i,
V
ie
t
nam
.
Em
a
il
: t
antd@
vnu.
e
du.
vn
1.
INTROD
U
CTION
Cl
i
m
a
te
chan
ge
has
cause
d
an
increasi
ng
nu
m
ber
of
la
ndsli
des,
es
peci
al
ly
in
the
m
ountain
ous
reg
i
on
s o
f
Viet
nam
.
These
la
nd
sli
des
ca
n
cau
se
a
disastro
us
eff
ect
on
the nei
ghborin
g
co
m
m
un
it
ie
s
as
well
as
the
local
infr
a
structu
re
an
d
econom
y
[
1
]
.
Lan
ds
li
des
can
be
broa
dly
classified
into
f
our
m
ai
n
t
ypes:
pr
e
-
existi
ng,
rain
fa
ll
-
induced
,
eart
hqua
ke
-
i
nduce
d,
a
nd
e
ndoge
nous
la
ndsli
des.
In
Viet
nam
,
m
os
t
la
ndsli
de
e
ven
ts
are trig
ge
red b
y rainfall
a
nd t
heir harm
is serio
us
[
2
]
,
[
3
]
.
Ther
e
are
t
wo
ty
pes
of
m
on
it
or
i
ng
i
n
La
nds
li
de
Syst
e
m
s,
nam
ely
sh
ort
te
rm
and
lo
ng
t
erm
[
4
]
-
[
8
]
.
Lo
ng
-
te
rm
m
o
nitor
i
ng
us
es
a
com
bin
at
ion
of
rem
ote
sensing
data
f
ro
m
sat
el
li
te
s,
glo
ba
l
po
sit
io
ning
s
yst
e
m
,
geog
raphic
in
f
or
m
at
ion
syst
em
s,
and
relat
e
d
m
at
he
m
at
ic
a
l
m
od
el
s
to
predict
la
ndsli
de
s
over
la
rg
e
tim
e
intervals.
S
hor
t
-
te
rm
m
on
it
ori
ng
ide
ntifie
s
the
early
sign
s
of
the
la
nd
sli
des
us
i
ng
a
co
m
bin
at
ion
of
m
any
sens
or
s
su
c
h as
accel
erati
on, soil
, r
ai
n
a
nd tem
per
at
ur
e.
W
i
reless
com
m
un
ic
at
ion
networks
form
a
crit
ic
al
enab
li
ng
te
chn
ol
og
y
of
Lan
ds
li
de
Mo
ni
toring
an
d
Re
al
-
tim
e
W
ar
ning
(LMR
W)
syst
e
m
s
.
W
ire
le
ss
Senso
r
Ne
twork
(
WSN)
and
ot
her
wir
e
le
ss
te
chnolo
gi
es
are
the
m
os
t
appr
opriat
e
te
ch
nolog
y
for
la
ndsli
de
m
on
it
or
i
ng
due
t
o
dif
ficult
to
acc
e
ss
te
rr
ai
n,
eas
e
of
m
ai
ntenan
ce,
c
heap
a
nd
qu
ic
k
instal
la
ti
on
al
ong
with
the
need
to
sat
isfy
the
i
m
po
rtant
real
-
ti
m
e
req
ui
rem
ent
of
LMR
W
sys
tem
[
5
]
,
[
7
]
,
[
9
]
,
[
10
]
.
Most
rural
com
m
un
i
ti
es
in
Viet
nam
hav
e
at
le
as
t
2G
co
ver
a
ge,
wh
ic
h
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Wi
rel
ess
Tech
nolo
gy
fo
r M
on
i
torin
g Sit
e
-
sp
e
ci
fi
c …
(
Tra
n Du
c
-
T
an
)
4449
will
b
e
upgrad
ed
to
3G a
nd
4G
netw
orks
i
n t
he
nea
r fu
t
ur
e
.
Furthe
rm
or
e, t
hese s
yst
em
s w
il
l pro
vid
e
vi
ta
l dat
a
for nat
ion
al
we
at
her
m
on
it
ori
ng syst
em
s.
A
nu
m
ber
of
LMR
W
hav
e
been
re
porte
d
in
the
li
te
rature
toda
y.
I
n
I
dukki,
I
nd
ia
,
a
com
plex
an
d
el
aborate
LM
R
W
de
plo
ym
ent
of
50
se
ns
ors
a
nd
20
W
S
N
nodes
was
desc
rib
ed
in
[
5
]
.
A
no
t
her
dep
l
oym
ent
[
9
]
us
es
se
nsor
node
buried
in
the
slo
pe
t
o
detect
m
ov
e
m
ent
sig
nal
an
d
com
bin
es
wit
h
s
oil
par
am
et
ers
f
or
pr
e
dicti
ng
ti
m
e
of
la
ndsli
de.
A
l
ow
-
c
ost
so
luti
on
to
detect
la
ndsl
ide
in
[
10
]
ut
il
iz
e
s
acce
le
ro
m
et
ers
to
e
valuate
th
e
la
ndsli
de
ris
k,
but
the
ef
fe
ct
of
oth
er
para
m
et
ers
su
c
h
a
s
rai
nf
al
l
has
not
been
consi
der
e
d.
Ot
her
researc
he
r
s
[
11
]
,
[
12
]
ha
ve
us
e
d
sat
urat
ed
hy
dr
a
ulic
cond
uctivit
y
and
i
nf
il
trat
ion
of
rain
water
into
slop
es
al
ong
w
it
h
com
pu
te
r
m
od
el
s
and
sim
ula
ti
on
s
to
predict
la
ndsli
de
risk
but
their
abili
ty
for
real
-
ti
m
e
a
nd
r
obust
pr
e
dicti
o
ns
is
qu
est
ion
a
ble.
More
over,
powe
r
consum
ption
fo
r
WSN
is
al
so
an
i
m
po
rtant iss
ue
.
This
pa
per
des
cribes
the
us
e
of
wireless
te
chnolo
gy
f
or
a
si
m
ple,
low
co
st,
and
e
ff
ic
ie
nt
la
nd
sli
de
m
on
it
or
ing
a
nd
real
-
ti
m
e
war
ni
ng
syst
em
with
an
i
nteg
r
at
ed
rain
ga
uge
wh
ic
h
pr
ov
i
des
the
rain
da
ta
.
W
e
fo
c
us
e
d
on
a
n
inno
vative
de
velo
pm
ent
of
Com
pr
essed
S
ensin
g
(CS)
ba
sed
al
go
rith
m
fo
r
la
ndsli
de
risk
m
on
it
or
ing.
The
al
gorithm
f
ind
s
the
prop
e
rtie
s
of
data
a
cqu
i
red
in
the
tim
e
do
m
ai
n.
Also
,
th
e
al
gor
it
h
m
is
sp
eci
al
ly
dev
el
op
e
d
to
overc
om
e
the
pr
obl
e
m
of
po
wer
consum
ption
.
W
i
reless
data
transm
issi
on
m
od
ule
ZigBee
use
s
th
e
802.1
5.4
sta
ndar
d
for
wirele
ss
com
m
un
ic
at
ion
.
ZigBee
m
odules
al
s
o
pro
vid
e
t
he
sle
e
p
m
od
e
to
save
po
wer
consum
ption
.
Data
is
the
n
tra
ns
m
itted
to
a
s
erv
e
r
us
in
g
a
3G/2
G
m
ob
il
e
ne
twork
.
At
t
he
serv
e
r
inf
or
m
at
ion
re
cei
ved
from
the
rem
ote
equ
ip
m
ent
is
com
par
ed
with
t
he
pr
edeterm
ined
th
reshold
s
est
abl
ished
by
ex
per
ts
i
n
the
fiel
d
t
o
pr
edict
li
kelihoo
d
of
la
ndsli
de
.
The
data
rec
ei
ved
from
sensor
nodes
ca
n
be
m
on
it
or
ed
on t
he pr
oject
w
e
bsi
te
an
d wa
rn
i
ng m
essages ar
e
sen
t t
o
re
gister
ed user
s m
ob
il
e phones
.
2.
SY
STE
M
I
M
PLE
MENT
A
TION
Figure
1
s
hows
the im
ple
m
ent
at
ion
of LMR
W wh
e
re t
he
te
rr
ai
n i
s
div
ide
d i
nto
t
wo areas:
saf
e a
reas
(for rain
g
a
uge
)
a
nd potenti
al
sli
de
areas
whe
re s
e
ns
in
g
a
nd
transm
itti
ng
nodes
a
re
placed
.
Figure
1. Lan
dsl
ide
m
on
it
or
in
g
syst
em
2.1.
Sy
s
tem
Topolo
gy
Figure
2
sho
ws
the
topolo
gy
of
the
syst
em
.
T
his
netw
ork
ha
s
sever
al
se
nso
r
node
s
that
co
m
m
un
ic
at
e
us
in
g
a
Zi
gb
ee
protoc
ol
with
a
sink
node
t
ha
t
is
par
t
of
th
e
data
lo
gg
e
r.
The
rai
n
gaug
e
is
connecte
d
to
th
e
data logger
th
r
ough a w
ire
d
com
m
un
ic
at
ion
li
nk
. T
he
logge
r
transm
it
s d
ata
to a d
at
abase throug
h
the I
nt
ern
et
.
The
i
nfor
m
at
ion
in
the
data
ba
se is use
d t
o u
pdat
e the
we
bs
it
e an
d
se
nd SM
S alerts to
the
com
m
un
it
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
4448
-
4455
4450
Figure
2. Syst
em
top
ol
og
y
2.2.
Sens
or
Im
plem
ent
at
i
on
The
se
ns
or
c
olu
m
n
con
sist
s
of
m
ajo
r
c
om
po
ne
nts:
sensors,
m
ic
ro
pr
ocess
or
s
,
tran
sm
it
te
r
and
rech
a
rg
ea
ble
ba
tt
ery
(see
Fig
ur
e
3(a,
b)).
T
hree
ty
pes
of
se
ns
ors
a
re
us
e
d
in
this
im
ple
m
entat
ion
,
nam
el
y
so
il
m
oistur
e,
te
m
per
at
ur
e
an
d
a
ccel
ero
m
et
er
(
ti
lt
m
e
te
r
an
d
geop
hone
)
se
nsors
.
T
he
m
ic
r
opr
ocess
or
rec
ei
ves
sens
or
data
as
input
an
d
exec
ut
es
a
nu
m
ber
of
processes
,
f
or
exam
ple,
filt
er
ing
noise
,
cal
ib
rati
ng
se
nsors
,
et
c.
Ther
ea
fter
,
the
wireless
m
od
ule,
XBeePR
O
[
13
]
base
d
on
802.1
5.4,
tra
nsm
it
s
the
proce
ssed
da
ta
to
a
central
com
pu
te
r
at
sp
eed
s
250
kbps
on
t
he
2.4
GH
z
ba
nd
with
50m
W
tra
nsm
itted
powe
r.
The
po
wer
s
upply
is
pro
vid
e
d
thr
ou
gh
a
batte
ry
w
it
h
su
f
fici
ent
capaci
ty
to
op
e
rate
for
one
ra
iny
seaso
n.
Fi
gure
3(
b)
s
hows
th
e
photo o
f
a
sens
or col
um
n.
Figure
3. (a
)Bloc
k dia
gr
am
o
f
sen
s
or syst
em
;
(b)
Se
nsor
im
p
lem
entat
ion
The
Weathe
r
Stat
ion
WS
-
3000
was
us
e
d
in
the
pro
j
ec
t
to
colle
ct
w
eat
her
data
s
uc
h
as
wind
directi
on,
wind
sp
ee
d,
a
nd
rai
nf
al
l.
T
he
acc
uracy
of
t
he
W
S
-
30
00
was
te
ste
d
agai
ns
t
ot
her
weathe
r
st
a
ti
on
s
and the
res
ults
sh
ow t
he de
vice to
be
acc
ur
at
e an
d reli
able,
bu
t i
nexpe
ns
iv
e.
2.3.
Em
bedde
d Comp
ut
in
g Modul
e
In
this
pa
per,
the
W
as
pm
ote
us
in
g
ATm
ega1
28
1
m
ic
ro
co
nt
ro
ll
er
was
ch
ose
n
to
co
nn
ect
and
process
inf
or
m
at
ion
re
cei
ved
from
se
ns
ors.
The
At
m
e
ga1
28
1
are
su
it
able
f
or
e
nginee
rin
g
a
se
ns
or
node,
w
hich
ha
s
batte
ry
co
ns
tra
int,
due
to
a
dvantage
ous
cha
r
act
erist
ic
s
su
c
h
as
hi
gh
pe
rfor
m
ance,
lo
w
powe
r,
et
c.
T
he
data
wh
ic
h
se
nsor
nodes
gat
her
e
d
on
the
sl
ope
is
transm
it
ted
wirelessl
y
to
the
sin
k
node,
a
nd
s
ubse
qu
e
ntly
delivere
d
t
o
t
he
central
com
pu
te
r
w
her
e
the
data
is
tran
sf
err
e
d
t
o
a
My
SQ
L
data
base
instal
le
d
on
th
e
we
b
serv
e
r. The
use
rs
m
on
it
or r
em
otely
infor
m
at
i
on th
rou
gh the
web ap
plica
ti
on as
sho
wn in Fi
gure
4.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Wi
rel
ess
Tech
nolo
gy
fo
r M
on
i
torin
g Sit
e
-
sp
e
ci
fi
c …
(
Tra
n Du
c
-
T
an
)
4451
Figure
4.
W
e
b i
nterf
ace
for L
MR
W
2.4.
War
ning
SMS t
o an
y
P
ho
ne
To
broa
dcasti
ng
an
al
ert
m
ess
age
to
a
ny
ph
one,
t
he
cent
ral
com
pu
te
r
is
co
nn
ect
e
d
to
a
G
SM/
GP
R
S
m
od
ule.
In
the
war
ni
ng
sta
te
,
an
al
ert
m
essa
ge
is
issued
an
d
autom
at
ic
al
l
y
sen
t
to
the
respon
si
ble
people
as
sh
ow
n
in
Fi
gur
e 5
.
Figure
5. P
hone
al
erts
3.
SY
STE
M DESIGN
In
the
early
w
ork
[
14
]
,
the
env
i
ron
m
ental
m
easur
em
ents
wer
e
se
nt
in
the
discrete
-
ti
m
e
without
any
com
pr
essio
n.
Data
rec
orde
d
from
senso
rs
in
LMR
W
is
pr
i
m
aril
y
low
f
re
qu
e
ncy
data.
I
n
the
c
urre
nt
s
yst
e
m
,
the
auth
or
s
re
desig
ne
d
the
syst
e
m
to
red
uc
e
the
a
m
ou
nt
of
tra
ns
m
itted
data
and
sa
ve
powe
r
.
I
n
this
pape
r
,
com
pr
essed
se
ns
in
g
(CS
)
te
c
hn
i
qu
e
was
use
d
to
re
duce
the
data
tra
nsm
issi
on
[
15
]
by
us
in
g
the
Four
ie
r
Transf
or
m
to
conve
rt
data
t
o
the
f
reque
nc
y
do
m
ai
n
fro
m
the
tim
e
dom
ai
n
and
se
nding
it
al
on
g
w
it
h
the
corres
p
on
ding Four
ie
r
coe
ff
ic
ie
nts.
By
recei
ving
the
t
ran
s
m
itted
data,
a n
onli
nea
r
al
gor
it
h
m
would
be
app
li
e
d
to r
ec
onstr
uct the
or
i
gin
al
data.
CS
is
an
ef
fic
ie
nt
te
chn
i
qu
e
that
em
plo
ys
a
com
pact
num
ber
of
sam
ples
to
rec
onstr
uct
a
sp
a
rse
sign
al
th
rou
gh
us
in
g
of
nonl
inear
al
gorith
m
s,
su
ch
as
Or
th
ogonal
Ma
tc
hing
Purs
uit
or
l
1
norm
[
16
]
,
[
17
]
.
So
m
e
publish
ed
wor
ks
a
ppli
ed
this
te
c
hniq
ue:
in
form
at
ion
syst
em
s
[
15
]
,
bio
m
ed
ic
al
syst
e
m
s
[
18
-
20
]
,
netw
orke
d
syst
e
m
s
[
21
]
,
c
omm
un
ic
at
ion
syst
e
m
s
[
22
]
-
[
24
]
,
robo
ti
c
syst
em
s
[
25
]
.
T
his
te
chn
i
qu
e
c
onsi
sts
of
two
m
ai
n
pro
cesses:
1)
ra
ndom
or
c
ha
ot
ic
unde
rsam
pl
ing
a
nd
2)
nonlinea
r
recon
structio
n.
I
n
m
any
app
li
cat
io
ns
,
unde
rsam
pling
will
help
to
re
du
ce
a
la
r
ge
num
ber
of
m
easur
em
ents,
an
d
thu
s
,
re
du
ce
t
he
power
consum
ption
.
Nonlinea
r
reconst
r
uction
of
te
n
c
onsu
m
es
m
or
e
tim
es
com
par
e
d
t
o
the
re
const
ru
ct
io
n
w
it
ho
ut
CS. Ho
wever,
it
is n
ot s
uc
h
a
disad
va
ntage
i
n
m
any app
li
ca
ti
on
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
4448
-
4455
4452
Howe
ver,
the
r
equ
i
rem
ent
of
the
sig
nal
of
in
te
rest
x
is
that
it
m
us
t
be
sp
a
r
se
in
the
f
or
m
of
a
li
near
represe
ntati
on
Φ.
Af
te
r
that,
t
he
unde
rsam
pling
proces
s
is
m
ade
w
her
e
th
e
eq
uiv
al
e
nt
m
easur
em
ent
m
a
trix
is
denoted
by
Ψ
.
Con
se
quently
,
the
m
easur
em
e
nts
are
gi
ven
by
y
=
Θ
s
,
w
her
e
Θ
=ΨΦ
.
The
ta
rg
et
of
this
w
ork
i
s
cl
ear that it
is
need
e
d
t
o reco
ns
tr
uct
x
from
y
(
or
s
f
r
om
y
)
.
4.
RESU
LT
S
A
ND D
I
SCUS
S
ION
In
this
w
ork,
a
determ
inist
ic
basis
create
d
by
a
se
qu
e
nce
of
ps
e
udo
-
ra
ndom
is
pr
opos
e
d
to
s
ub
sti
tute
for
pure
ra
ndom
basis.
In
co
m
par
ison
with
CS,
the
stren
gt
h
of
this
s
olu
ti
on
is
that
this
sequ
e
nce
can
be
easi
ly
instal
le
d
into
the
m
ic
ro
co
ntr
ollers
be
fore
assem
bling
the
sens
or
no
des
in
the
fiel
d
sit
e.
Ther
e
f
or
e,
th
e
pap
e
r
consi
ders
a
dy
nam
ic
deter
m
i
nisti
c
syst
e
m
whose
c
har
act
erist
ic
is
determ
inist
ic
no
nlin
ear.
T
he
dete
r
m
inist
i
c
com
pr
essed
sa
m
pl
ing
te
ch
ni
qu
e
is
eq
ui
valent
to
t
he
rando
m
on
e
on
t
he
acc
ur
acy
of
outc
om
e
[
26
]
.
The
auth
or
s
us
e
d
a
log
ist
ic
m
ap
based
dynam
ic
struc
tu
re
wh
ic
h
is
tran
sf
or
m
ed
into
a
seq
ue
nc
e
that
would
hav
e
a
Gau
s
sia
n
-
li
ke beha
vior:
(
1
)
(
)
(
1
(
)
)
q
n
q
n
q
n
(1)
wh
e
re
ρ
is
the
con
tr
ol
pa
ram
et
er
[
27
]
;
the
init
ia
l
con
diti
on
q(0
)
seri
ou
sl
y
eff
ect
s
the
dy
nam
ic
of
Eq
uation
1.
q(0)
cha
nges
a
s
m
al
l
value
will
quic
kly
re
su
lt
a
big
c
hange
i
n
the
value
of
q(n).
T
he
s
pa
rse
sign
a
l
can
be reco
ns
tr
u
ct
ed by
us
in
g t
he
l
1
-
re
gula
riz
ed
le
ast
s
qu
a
re
s m
et
ho
d
[
20]
. T
he
s
olv
i
ng pr
ob
le
m
is
(2)
wh
e
re
λ
is
a
c
on
sist
e
ncy
tu
nin
g
co
ns
ta
nt,and
F
u
is
t
he
unde
r
sam
pled
F
ourier
ope
rator.
Data
of
te
m
per
at
ur
e
,
acce
le
rati
on
,
a
nd
m
oistur
e
a
r
e
reconstr
ucte
d
at
the
recei
ve
r
w
hich
a
re
s
how
n
in
Fi
gur
es
6
-
8.
The
da
ta
is
reduce
d by a fa
ct
or
of 25%
.
It
can
be
see
n
from
Figu
re
6
that
the
te
m
pe
ratur
e
increa
se
s
from
30
o
C
to
33
o
C
a
nd
it
can
be
note
d
that
the
ave
ra
ge
diff
e
re
nce
be
tween
t
he
re
pro
duced
data
a
nd
t
he
or
igi
nal
on
e
is
only
0.58%.
I
n
th
e
se
cond
scenari
o
as
show
n
i
n
Fi
gure
7
,
t
he
data
of
m
oistur
e
can
be
rec
on
st
ru
ct
e
d
with
a
n
a
verage
e
rror
of
1.51%
i
n
com
par
ison wi
th the o
rigin
al
on
e
.
Figure
6. Data
from
te
m
per
atu
re
senso
r,
r
=
0.2
5
Figure
7. Data
from
m
oistur
e
sens
or
f
or
r
=
0.25
The
Fig
ur
e
8
s
hows
the
va
riat
ion
betwee
n
t
he
reconstr
ucte
d
a
nd
or
i
gin
al
data
f
r
om
the
acce
le
rati
on
sens
or
f
or
a
lo
w
com
pr
essio
n
rati
o
of
r
=
0.2
5.
It
can
be
observ
e
d
that
the
re
is
con
si
der
a
ble
error
betwe
en
the
act
ual
and
se
nse
d
data
f
or
this
value
of
com
pr
essio
n
rati
o.
Nex
t,
the
a
uthors
in
vestigat
e
the
best
com
pr
ession
rati
o.
2
21
2
a
r
g
m
in
u
x
u
F
x
y
x
s
u
b
je
c
t
to
F
x
y
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Wi
rel
ess
Tech
nolo
gy
fo
r M
on
i
torin
g Sit
e
-
sp
e
ci
fi
c …
(
Tra
n Du
c
-
T
an
)
44
53
Figure
8. Data
from
the accel
erati
on senso
r for
r
=
0.2
5
Figure
9. Ef
fec
t of com
pr
essi
on r
at
io
on rel
a
ti
ve
error.
To
analy
ze
th
e
per
f
orm
ance
of
the
rec
ons
tructi
on
syst
em
and
the
effe
ct
of
com
pr
ession
rati
o,
a
par
am
et
er call
ed
the
r
el
at
ive
r
econst
ru
ct
e
d
e
r
ror (e) is
prop
ose
d:
1
ˆ
||
1
100%
ˆ
||
L
ii
i
i
xx
e
Lx
(3)
wh
e
re
L
is
t
he
total
nu
m
ber
of
data
us
ed
for
cal
culat
io
n,
x
de
note
s
th
e
or
i
gin
al
data
and
ˆ
x
denotes
the
reconstr
ucted
one.
Figure
9
in
dica
te
s
the
in
flue
nc
e
of
the
c
om
pr
essio
n
rati
o
on
the
relat
ive
e
rror.
It
is
obser
ved
that
the
error
is
la
r
ger
f
or
lo
w
r
an
ge
c
om
pr
essio
n
rat
ios
betwe
en
0.25
a
nd
0.5
.
If
t
he
com
pr
essi
on
rati
o
is
great
er
than
or
e
qual
to
0.5
5,
t
he
er
ror
is
qu
ic
kly
reduce
d
to
ze
r
o.
T
he
se
res
ults
sug
ge
st
that
the
k
-
s
pace
data
is
de
cent
enou
gh
for
rec
on
st
ru
ct
io
n
of
the
or
i
gin
al
da
ta
.
Fo
r
the
fu
t
ure
w
ork,
a
com
pr
essi
on
rati
o
of
r=
0.5
5
is
sel
ect
ed
base
d
on
t
his
resu
lt
.T
he
al
gorithm
find
s
t
he
pro
per
ti
es
of
data
ac
qu
ire
d
i
n
the
ti
m
e
do
m
ai
n.
It
is
s
pe
ci
al
ly
dev
el
op
e
d
to
overc
om
e
the
pr
ob
le
m
of
powe
r
co
ns
um
ption
.
W
i
reless
data
transm
issi
on
m
odule
ZigBee
us
e
s
the
80
2.15.4
s
ta
nd
a
r
d
f
or
wi
reless
com
m
un
ic
at
ion
.
ZigB
ee
m
od
ules
al
so
prov
i
de
the
sle
ep
m
od
e
to
sa
ve
powe
r
c
on
s
umpti
on.
5.
CONCL
US
I
O
N
This
pa
pe
r
de
scribe
d
a
suc
cessf
ully
i
m
plem
ented
LMR
W
with
a
n
oper
at
ion
al
schem
e
fo
r
transm
itti
ng
co
m
pr
essed
data
that
wer
e
acq
ui
red
f
ro
m
three
diff
e
ren
t
sen
s
or
s
.
Prop
e
rtie
s
of
data
acq
uir
ed
in
the
tim
e
do
m
ai
n
are
ex
plo
i
te
d
toa
pp
ly
C
om
pr
essed
Se
ns
in
g
te
c
hn
i
que
for
powe
r
s
avin
g.
A
num
ber
of
wireless
te
chnolo
gies
wer
e
us
e
d
in
the
real
-
tim
e
s
yst
e
m
design
based
on
syst
e
m
,
po
we
r
an
d
data
rate
requirem
ents.
A
fi
nite
num
ber
of
Four
ie
r
c
oeffici
ents
of
t
i
m
e
-
do
m
ai
n
da
ta
wer
e
tran
sm
it
te
d
and
the
a
m
ou
nt
of
data
tra
ns
m
it
te
d
was
halv
ed
thu
s
re
duci
ng
the
po
wer
consum
ption
.
This
prototype
can
fo
rm
the
basis
of
so
lvi
ng r
eal
w
or
l
d prob
le
m
s r
el
at
ed
to
n
at
ural
d
isa
ste
rs
a
nd
to
assist
c
omm
un
it
ie
s thr
ou
ghout t
he glo
be
.
REFERE
NCE
S
[1]
T.
Gl
ade
,
et
a
l.
,
"La
ndslide ha
za
r
d
and
r
isk
"
,
John
W
il
e
y
&
Sons
,
2006.
[2]
D.M.
Duc,
"Rai
nfa
ll
-
tri
gger
ed
l
arg
e
la
ndslid
es
on
15
Dec
ember
2005
in
Van
Canh
Distri
ct
,
B
inh
Dinh
Provin
ce,
Viet
nam
"
,
Lands
li
des
,
vol
.
10(2)
,
pp.
219
-
230,
20
12.
[3]
D.H.
Loi
,
e
t
al
.
,
"The
28
Jul
y
20
15
rap
id
l
andsli
d
e
at
H
a
Long
C
ity
,
Quang
Ninh,
Viet
nam"
,
Land
slide
s
,
vol
.
14(3)
,
pp.
1207
-
1215
,
2017.
[4]
O.
Mons
err
at
,
e
t
al.
,
ed
it
ors.
"Lo
ng
te
rm
la
ndslid
e
m
onit
oring
with
Gr
ound
Based
SA
R
"
,
EGU
Gene
ral
Assembly
Confe
renc
e
A
bstracts
,
2014
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
20
88
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
201
8
:
4448
-
4455
4454
[5]
M.V.
Ramesh,
"
Design,
deve
lop
m
ent
,
and
depl
o
y
m
ent
of
a
wire
l
ess
sensor
net
w
ork
for
det
ec
ti
on
of
la
ndslide
s
"
,
A
d
Hoc
Net
works
,
v
ol.
13
,
pp
.
2
-
18
,
2014.
[6]
F.
Aglia
rd
i
,
et al
.
,
"In
situ and
re
m
ote
long
t
erm r
ea
l
-
ti
m
e
m
onit
or
ing
of a
l
arg
e al
p
ine
ro
ck
slid
e"
,
La
ndslid
e
Sci
en
ce
and
Prac
ti
c
e: Springe
r, 2013. p.
415
-
421.
[7]
Q.A.
Gian
,
e
t
al.
,
"D
esign
an
d
implementati
on
of
site
-
spe
cific
r
ai
nfa
ll
-
indu
ce
d
l
andsli
d
e
e
arly
warn
ing
a
nd
m
onit
oring
s
y
st
e
m
:
a
c
ase
stud
y
at
Nam
Dan
la
n
dslide
(Vi
et
nam)
"
,
Geomatic
s
,
N
atural
Haz
ards
and
Ri
sk
,
vol
.
8(
2),
pp.
1978
-
1996
,
2017.
[8]
Q.A.
Gian
,
e
t
al
.
,
"M
onit
oring
o
f
La
ndslide
s
in
Mountai
nous
Re
gions
base
d
on
FEM
Modell
ing
and
Rai
n
Gaug
e
Mea
surem
ent
s
"
,
Inte
rnational
Jo
urnal
of
El
e
ct
ri
c
al
and
Compute
r
Engi
nee
ring
(
IJE
CE)
,
vol
.
6(5)
,
pp.
210
6
-
2113,
2016.
[9]
A.
Te
r
zi
s
,
e
t
al
.
,
edi
tors.
"S
li
p
sur
fac
e localization
in
wire
l
ess sensor ne
tworks for
l
andsli
de
pr
edi
c
tion
"
,
Proceedi
ng
s
of
th
e
5th
in
te
rn
ati
onal conf
ere
n
ce
on
Informatio
n
proce
ss
ing
in
s
ensor ne
tworks
,
2006:
ACM
.
[10]
H.Z
.
Kot
ta
,
e
t
a
l.
,
"W
ire
le
ss
sen
sor
net
work
for
la
ndslid
e
m
o
nitoring
in
Nus
a
T
engga
ra
Ti
m
ur
"
,
TEL
KOMNIKA
(
Tele
communic
ati
on
Computing
El
e
ct
ronics
and
Control)
,
vol. 9(
1),
pp
.
9
-
18
,
201
1.
[11]
A.
Ali
,
e
t
al
.
,
"
Sim
pli
fie
d
quantita
t
ive
r
isk
assess
m
ent
of
rai
nfa
l
l
-
induced
la
ndsl
i
des
m
odel
le
d
b
y
infi
nite
slopes"
,
Engi
ne
ering
Ge
ology
,
vol
.
179
,
pp.
102
-
116
,
20
14.
[12]
B.
D.
Coll
ins
an
d
D.
Zni
dar
cic,
"S
ta
bil
ity
an
aly
ses
of
rai
nfa
ll
induc
ed
l
andsli
d
es
"
,
Journal
of
Geote
chn
ic
al
an
d
Geoe
nvi
ronm
ental
Eng
ine
ering
,
vol.
130(4)
,
pp
.
362
-
372,
2004
.
[13]
A.H.
Kiou
m
ars
and
L.
Ta
ng,
ed
it
ors.
"A
Tmega
and
XBee
-
base
d
wire
le
ss
sensing
"
,
Inte
rnationa
l
Confe
renc
e
on
Aut
omation
,
Rob
oti
cs
and
Applic
ati
ons (
ICAR
A)
,
2011:
IE
EE.
[14]
D.
C.
Ngu
y
en
,
et
al.
,
"M
ult
i
-
sens
ors
int
egr
ation
for
l
andsli
de
m
onit
oring
applic
ation
"
,
VNU
Journal
o
f
Sci
en
ce
,
v
ol.
30(6S
-
B),
pp.
20
2
-
210,
2014
.
[15]
D
.
L. Donoho,
"
Com
pre
ss
ed
sensing
"
,
IE
EE
Tr
a
nsacti
ons on inf
orm
ati
on
the
ory
,
vol. 52(4), pp. 1
289
-
1306,
2006
.
[16]
Y.
Zha
ng
,
e
t
al.
,
"A
Study
on
I
m
age
Rec
onfigu
rat
ion
Algori
th
m
of
Co
m
pre
sse
d
Sensing"
,
TEL
KOMNIKA
,
vol.
15(1),
pp
.
299
,
2
017.
[17]
J.A.
Tropp
and
A.C.
Gilbe
r
t,
"S
igna
l
r
ec
over
y
fr
om
ran
dom
m
ea
surem
ent
s
via
or
thogona
l
m
atchi
ng
pursuit"
,
IEEE
Tr
ansacti
ons on information
th
eo
ry
,
vol
.
53(12
), p
p.
4655
-
4666
,
2
007.
[18]
S.
Yang
and
M.
Gerl
a
,
edi
to
rs.
"Ene
rg
y
-
eff
ic
i
en
t
a
cce
le
rom
eter
dat
a
tra
nsf
er
for
hum
an
bod
y
m
ovement
studi
es
"
,
Inte
rnational
Co
nfe
renc
e
on
Sens
or Ne
tworks,
Ub
iqui
tous,
and
Tr
ustwor
thy
Computing
,
2010:
IEEE.
[19]
M.
Lusti
g
,
et
al
.
,
"S
par
se
MRI:
The
appl
i
cation
of
compress
ed
sensing
for
rap
id
MR
imaging"
,
Ma
gnet
i
c
resonanc
e
in
medi
ci
ne
,
vo
l. 58(6), pp. 1182
-
1195,
2007
.
[20]
Y.
Zha
ng
,
et
al
.
,
"Expone
nti
a
l
wave
l
et
itera
t
ive
shrinkage
thr
esholdi
ng
al
gori
thm
with
ran
dom
shi
ft
for
compress
ed
sen
sing
m
agne
tic
resona
nc
e
imaging
"
,
IEEJ
Tr
ansacti
ons
on
El
e
ct
rical
and
E
lect
ronic
E
ngineeri
ng
,
vol.
10(1),
p
p.
116
-
117,
2015
.
[21]
H.
Zhe
ng
,
et
al
.
,
"D
ata
ga
the
ri
ng
with
compre
ss
ive
sensing
in
wire
le
ss
sensor
net
works
:
a
r
a
ndom
walk
base
d
appr
oac
h"
,
I
EE
E
Tr
ansacti
ons on
Parallel
and
Di
stribute
d
S
yste
m
s
,
vol. 26(1)
,
pp
.
35
-
44,
2015.
[22]
K.
Ha
y
ashi
,
et
al.
,
"A
user'
s
guide
to
compress
ed
sensing
for
comm
unic
at
ions
sy
st
ems
"
,
IEI
CE
transacti
ons
on
communic
ati
ons
,
vol. 96(3), pp. 6
85
-
712,
2013
.
[23]
C.
Cai
on
e
,
et
al.
,
ed
it
ors.
"Com
pre
ss
ive
sensing
opti
m
iz
ation
ov
e
r
Zi
gBe
e
n
et
wor
k
s
"
,
Industrial
E
mbedde
d
Syst
e
ms
(
SIES)
,
2010
Inte
rnational
Symp
osium on
,
2010:
IEE
E
.
[24]
S.U.
Khan
,
et
al.
,
"D
ia
gnosis
of
Faulty
Senso
rs
in
Antenna
Arra
y
using
H
y
brid
Diffe
ren
t
ia
l
Evol
uti
on
bas
e
d
Com
pre
ss
ed
Sensing
Te
ch
nique
"
,
Inte
rnationa
l
Jo
urnal
of
E
le
c
trical
and
Computer
Engi
ne
ering
(
IJ
ECE
)
,
vol.
7(2),
pp.
961
-
966
,
20
17.
[25]
S.
Qiu
,
e
t
al
.
,
"Brai
n
–
Mac
hin
e
Inte
rf
ac
e
and
Visual
Com
pre
ss
ive
Sensing
-
Ba
sed
Te
l
eope
r
at
i
on
Control
of
an
Exoskel
e
ton
Ro
bot
"
,
I
EE
E
Tr
ansacti
ons on Fuzz
y
Syst
ems
,
vol
.
2
5(1),
pp
.
58
-
69
,
2017.
[26]
J.A.
Tropp
,
et
a
l.
,
ed
it
ors
.
"Ran
dom
fil
te
rs
for
compress
ive
sam
pli
ng
and
rec
o
nstruct
ion
"
,
A
co
ustic
s,
Speech
a
nd
Signal
Proce
ss
in
g,
2006
ICASSP
2006
Proceedi
n
gs 2006
IEEE
In
te
rnational
Conf
ere
nce on
,
2006:
IEEE.
[27]
J.C.
Sprott
and
J.C.
Sprot
t,
"Cha
os a
nd
t
ime
-
seri
es
anal
y
s
is
"
,
Oxf
ord
Univer
sit
y
P
ress Oxford,
200
3.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Gian
Quoc
-
Anh
was
born
in
1981.
He
recei
v
ed
t
he
B.
S.
degr
ee
i
n
Ph
y
sics
from
VN
U,
Hanoi
-
Univer
sit
y
of
Scie
n
ce
in
2003
and
M
.
S.
d
eg
ree
in
El
e
ct
roni
cs
and
Telec
o
m
m
unic
at
ion
te
chno
log
y
from
VN
U,
Hanoi
-
Univer
sit
y
of
Enginee
ring
and
T
echnolog
y
(UE
T)
i
n
2010.
He
is
cur
ren
t
l
y
wor
king
towar
ds
th
e
Ph.D.
deg
ree
i
n
El
e
ct
roni
c
En
gine
er
ing
a
t
VN
U
-
UET.
His
rese
arc
h
in
te
rest
s a
re
application
s of
digital signa
l
proc
essing
and
embedde
d
s
y
s
tem
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N:
20
88
-
8708
Wi
rel
ess
Tech
nolo
gy
fo
r M
on
i
torin
g Sit
e
-
sp
e
ci
fi
c …
(
Tra
n Du
c
-
T
an
)
4455
Ngu
y
en
Dinh
-
C
hinh
recei
ved
t
he
B.
S.
degr
ee
in
El
e
ct
ron
ic
and
Telec
om
m
unic
a
ti
on
from
Viet
nam
Nat
ion
al
Unive
rsit
y
,
Hanoi
–
Univ
ersity
of
Engi
n
ee
r
in
g
and
T
ec
hno
lo
g
y
in
2014
and
M.S.
degr
ee
in
El
e
ct
roni
c
and
Com
m
unic
at
ion
from
the
sam
e
unive
rsit
y
in
201
7.
His
rese
ar
ch
int
er
ests
ar
e
dig
i
ta
l
signal process
ing,
m
ac
hin
e learning and em
bedde
d
s
y
s
te
m
s
Tra
n
Duc
-
Nghi
a
was
born
in
19
86.
He
is
a
s
ci
e
nti
st
a
t
Insti
tute
of
Inform
at
ion
Te
chno
log
y
(IOIT),
Vi
et
nam
Aca
dem
y
of
Sci
enc
e
and
Techn
olog
y
(VA
ST).
He
is
cur
r
entl
y
a
PH
D
student
of
'
Drug,
Toxico
log
y
,
Chemistr
y
,
Im
age
rie
s'
(MT
CI
ED
563)
doct
ora
l
school
of
Sorbonne
Pari
s
Cit
é
(Franc
e
).
His
rese
ar
ch
i
nte
rests
are
m
at
hemat
ic
s
and
signal
proc
essing,
El
e
ct
ron
Para
m
agne
tic
R
esona
nce
(EPR),
par
amet
er
est
i
m
at
ion,
d
at
a
anal
y
sis.
In
his
th
esis,
he
fo
cuse
s
on
signal
pro
ce
s
sing
of
EPR
spe
ct
ra
fo
r
in
vi
vo
expe
riments.
He
did
his
m
aste
r
i
n
'
Sc
ie
nc
e
i
n
Inform
at
ion
T
echnolog
y
'
a
t
Univ
ersity
of
Eng
ineeri
ng
and Te
chn
olog
y
,
VN
U.
Tra
n
Duc
-
Tan
w
as
born
in
1980
.
He
recei
ved
his
B.
Sc,
M.Sc
,
a
nd
PhD
.
degr
ee
s
r
espe
ctively
in
2002,
2005,
an
d
2010
at
th
e
Univer
sit
y
of
E
ngine
er
ing
and
Te
chno
log
y
(UET
),
Vi
et
n
am
Nati
ona
l
Univer
sit
y
–
Hano
i,
Vi
et
nam
(VN
UH
),
where
he
h
as
bee
n
a
lectur
er
si
nce
2006.
H
e
was
the
r
ecipie
n
t
of
th
e
Vi
et
nam
Nati
ona
l
Univer
sit
y
,
Hano
i,
Vietnam
Young
Scientifi
c
Aw
ard
in
2008.
He
is
cur
ren
t
l
y
an
As
socia
te
Profe
ss
or
w
it
h
the
Facul
t
y
of
El
e
ct
roni
cs
and
Te
l
ec
om
m
unic
ations,
Univer
sit
y
of
Engi
n
ee
r
ing
and
Te
ch
nolog
y
,
Viet
n
a
m
Nati
onal
Univer
sit
y
,
Han
oi,
Viet
n
am.
He
is
the
aut
hor
a
nd
coa
uthor
of
30
pape
rs
on
MEMS
base
d
sensors
and
th
ei
r
application
.
His
pre
sent
rese
arc
h
inter
est
is i
n
DS
P a
pplications.
Kieu
Thi
Ngu
yen
was
born
in
1983.
She
is
a
hea
d
depa
r
tme
nt
of
m
ec
h
ani
c
a
l,
Fa
cul
t
y
of
Mec
hanica
l
,
Ele
ct
ri
ca
l
,
and
Elec
troni
c
Engi
n
ee
r
i
ng,
Ngu
y
en
T
at
Tha
nh
Univer
si
t
y
,
Viet
n
am
.
She
did
her
m
as
te
r
at
HCM
C
U
nive
rsit
y
of
Te
c
hnolog
y
and
Ed
uca
t
ion
,
Viet
n
a
m
.
Her
re
sea
rch
int
er
ests
ar
e
ne
t
work a
nd
sign
al
proc
essing.
Kum
besa
n
Sandra
sega
ran
is
a
n
As
socia
te
Pr
ofe
ss
or
at
UTS
and
Cen
tre
fo
r
Rea
l
-
Ti
m
e
Inform
at
ion
Net
works
(CRIN).
He
holds
a
Ph.D.
in
Elec
t
rical
Eng
ineeri
ng
from
McGill
Univer
sit
y
(Can
ada
)(1994)
,
a
M
aste
r
of
Sc
ie
nc
e
Degre
e
in
T
el
e
comm
unic
at
ion
Engi
ne
eri
ng
from
Essex
Univer
sit
y
(198
8)
and
a
B
ac
h
el
or
of
Scie
n
ce
(Honours
)
Degre
e
in
E
lectr
i
ca
l
Engi
ne
eri
ng
(Fir
st Cl
ass) (1985).
His c
urre
nt
rese
arc
h
work foc
us
es
on
two
m
ai
n
are
as
(a)
rad
io
resourc
e
m
ana
g
ement
in
m
obil
e
net
works
,
(b)
e
ngine
er
ing
of
re
m
ote
m
onit
oring
s
y
stems
for
novel
appl
i
ca
t
i
ons
with
indus
tr
y
through
th
e
use
of
embedde
d
s
y
s
te
m
s,
sensors
and
comm
unic
at
ions
s
y
stems
.
He
has
publi
shed
over
100
ref
er
ee
d
pu
bli
c
at
ions
and
2
0
consultanc
y
rep
orts
spanning
telec
om
m
unic
a
t
ion
and
computi
ng
s
y
st
ems
.
Evaluation Warning : The document was created with Spire.PDF for Python.