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.
9
, No
.
6
,
Decem
ber
201
9
, p
p.
5586~
5595
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
9
i
6
.
pp5586
-
55
95
5586
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Price
effect
an
alysis
and
p
re
-
res
eravtion
s
ch
eme on
electric
ve
hi
cle
ch
arging
network
s
Ju
n
gho
on
Lee
,
G
yu
n
g
-
Le
en
Park
De
p
ar
t
m
ent
o
f
C
om
pute
r
Scie
n
ce a
nd
St
a
ti
st
ic
s,
Jeju
Na
ti
ona
l
Uni
ver
sit
y
,
Republi
c
of
Kor
ea
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r
1
, 2
01
9
Re
vised
Ju
l
2
1
,
201
9
Accepte
d
J
ul
28
, 2
01
9
Thi
s
pap
er
inv
es
ti
gates
th
e
pr
ic
e
eff
ect
to
th
e
charging
demand
co
m
ing
from
el
e
ct
ri
c
vehicl
es
and
the
n
eva
lu
at
es
th
e
per
for
m
anc
e
of
a
pre
-
rese
rva
t
io
n
m
ec
hani
sm
usin
g
th
e
re
al
-
l
ife
d
emand
p
at
t
ern
s.
O
n
the
cha
rg
ing
ne
twork
in
Jeju
cit
y
,
the
o
ccupancy
ra
te
s
for
3
pric
e
g
rou
ps,
n
amel
y
,
fr
ee,
m
edi
um
-
pric
e
,
and
expe
nsive
c
har
ger
s,
ar
e
s
ep
ara
t
ed
al
m
o
st
ev
enly
b
y
about
9.
0
%,
whil
e
a
set
of
ch
arg
ers
dom
ina
t
es
the
cha
rg
ing
d
emand
during
hot
hours.
The
virt
u
al
pre
-
r
ese
rva
t
ion
sch
e
m
e
m
at
che
s
elec
tri
c
veh
ic
l
es
to
a
ti
m
e
slot
of
a
cha
rg
er
so
as
n
ot
onl
y
t
o
avoi
d
int
ole
r
able
w
ai
t
i
ng
ti
m
e
in
cha
rg
ing
sta
ti
ons
s
y
stematica
lly
b
ut
al
so
to
in
crea
se
the
r
eve
nu
e
of
service
provi
der
s,
t
aki
ng
int
o
ac
coun
t
bo
t
h
biddi
ng
le
v
el
s
spec
ified
b
y
elec
tri
c
vehicl
es
and
pre
fer
ence
cri
t
eri
a
d
efi
n
ed
b
y
cha
rg
ers.
The
per
fo
rm
ance
anal
y
s
is
r
esul
ts
obta
in
ed
b
y
prototy
p
e
i
m
ple
m
ent
at
ion
show
tha
t
t
he
proposed
p
re
-
rese
rva
t
ion
m
ec
hani
sm
improves
the
rev
en
ue
of
s
erv
ice
pr
ovide
rs
b
y
up
t
o
9.
5
%
an
d
42.
9
%
,
compa
red
wi
th
the
l
ega
c
y
FC
FS
a
nd
rese
rv
at
ion
-
l
ess
walk
-
in
sche
m
es
for
th
e give
n
p
erf
orm
an
ce
p
ara
m
eter
se
t
s.
Ke
yw
or
d
s
:
C
hargin
g netw
ork
E
le
ct
ric
ve
hicle
P
re
-
rese
rv
at
io
n
P
rice ef
fect
R
evenue i
ncr
e
ase
Copyright
©
201
9
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
:
Gyu
ng
-
Lee
n
P
ark
,
De
p
ar
t
m
ent
o
f
C
om
pute
r
Scie
n
ce a
nd
St
a
ti
st
ic
s,
Jej
u
Nati
on
al
Un
i
ver
sit
y
,
Jej
ud
ae
ha
kno 102, R
ep
ubli
c
of Korea
.
Em
a
il
: jh
le
e
@jejun
u.
ac.
kr
1.
INTROD
U
CTION
Ele
ct
ric
ve
hicle
s,
or
E
Vs
i
n
s
hort,
a
re
r
eplaci
ng
le
gac
y
gas
oline
-
power
e
d
ve
hicle
s,
bri
ngi
ng
a
po
ll
utio
n
-
f
re
e
trans
port
era
[1
]
.
A
s
they
get
ene
rg
y
f
rom
the
el
ect
ricity
netwo
r
k,
t
r
ans
port
syst
em
s
are
beco
m
ing
a
pa
rt
of
the
el
ect
r
ic
it
y
gr
id,
sp
e
ci
fical
ly
,
the
s
m
art
gr
id
[
2].
Def
i
nitel
y,
the
wide
ex
pa
ns
i
on
of
chargin
g
in
fr
as
tructu
res,
i
nclu
ding
cha
r
ging
sta
ti
on
s,
e
nerg
y
pr
ovisi
onin
g,
and
t
he
li
ke,
i
s
the
m
os
t
i
m
p
or
ta
nt
pr
e
requisi
te
f
or
the
fast
pe
ne
trat
ion
of
E
Vs.
H
ow
e
ver,
we
ca
nnot
c
om
plete
ly
avo
id
the
in
suffic
ie
ncy
in
the
c
hargin
g
c
apacit
y
or
the
dem
and
-
sup
ply
m
is
m
a
tc
h
in
early
sta
ge
s.
Hen
ce
,
it
i
s
ne
cessary
to
e
m
plo
y
an
e
ff
ic
ie
nt
m
anag
em
ent
syst
e
m
fo
r
c
hargin
g
facil
it
ie
s
to
m
on
it
or
their
r
eal
-
tim
e
beh
av
iors,
not
on
ly
t
o
detect
fail
ur
es
im
m
e
diate
ly
bu
t
al
so
t
o
ob
ta
in
operati
on
rec
ords
f
or
bette
r
plann
i
ng.
M
or
e
ov
er,
t
he
m
anage
m
ent
syst
e
m
can
su
ppor
t
a
reserv
at
ion
m
echan
ism
and
al
so
i
m
ple
m
ent
an
appr
opriat
e
pri
ce
po
li
cy
bas
ed
on
the an
al
ysi
s
of
the ope
rati
on hi
story d
at
a
[
3].
Ma
nag
em
ent
capab
il
it
ie
s,
com
bin
ed
wit
h
a
rese
rv
at
ion
strat
e
gy,
ca
n
al
le
via
te
in
her
e
nt
inco
nv
e
nien
ce
stemm
ed
fr
om
long
c
hargi
n
g
tim
e
by
fr
eei
ng
dri
ve
rs
fro
m
wasti
ng
thei
r
ti
m
e
in
the
wait
ing
qu
e
ue
[4
]
.
M
oreo
ver,
a
s
ophi
sti
cat
ed
m
at
c
hing
m
echan
ism
between
E
Vs
a
nd
c
ha
rg
e
rs
can
be
desi
gn
e
d
t
o
distrib
ute
e
nergy
loa
d
ov
e
r
the
ti
m
e
and
s
pace
axis,
as
well
as
to
inc
r
ease
the
prof
it
of
c
ha
r
ging
s
erv
ic
e
pro
vid
er
s
[5
]
.
I
t
is
tr
ue
t
hat
t
he
cha
r
ging
c
os
t
ha
s
le
ss
ef
fect
on
the
cha
r
ging
dem
and
,
co
m
par
ed
with
ga
so
li
ne
fu
el
in
g,
du
e
to
m
uch
c
hea
pe
r
el
ect
rici
ty
pr
ic
e.
H
oweve
r,
local
dri
ve
rs,
w
ho
a
re
fam
i
li
ar
with
the
c
urrent
chargin
g pr
ic
es
arou
nd
t
heir
vi
ci
nity
, h
ave no
reas
on to disrega
rd
t
he
pri
c
e d
iffe
re
nce.
T
his all
ow
s
us
t
o
sh
a
pe
the
overall
de
m
and
by
lowe
rin
g
the
c
harg
ing
fee
on
no
n
-
busy
cha
rg
e
rs
du
rin
g
the
hours
of
lo
w
traff
ic
.
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
Pric
e eff
ect
a
naly
sis a
nd
pr
e
-
reseravti
on
sc
he
me on el
ect
ric
vehicl
e c
ha
r
gi
ng n
et
w
or
ks (J
unghoo
n
Lee)
5587
In
a
dd
it
io
n,
if
the
c
ha
rg
i
ng
pr
i
ce
is
set
hi
gher
f
or
the
ho
t
s
ta
ti
on
s
duri
ng
pe
ak
hours,
the
r
evenue
of
cha
r
ging
serv
ic
e
pro
vid
e
rs wil
l i
ncr
ease
[6].
The
first
ste
p
for
pr
ic
e
po
li
c
y
plan
ning
is
t
he
a
naly
sis
of
the
c
os
t
e
ff
ect
to
the
cha
r
ging
dem
and
.
It
can
be
ca
rr
ie
d
out
m
aking
use
of
real
-
li
fe
op
e
rati
on
rec
ords
accum
ulate
d
in
the
ta
r
get
EV
cha
r
ging
s
yst
e
m
.
Our
ci
ty
,
nam
el
y,
Jej
u,
Re
public
of
K
or
ea
,
c
urren
tl
y
run
s
a
ci
ty
wide
c
hargin
g
net
work
c
on
sist
in
g
of
a
bout
245
DC
c
harg
ers
i
ns
ta
ll
ed
f
or
public
us
e
[7
]
.
T
her
e
are
4
c
hargin
g
se
rv
ic
e
pro
vi
ders
f
r
om
bo
th
ci
vil
an
d
gove
rn
m
ental
do
m
ai
ns
,
w
hile
the
na
ti
on
al
auth
or
it
y
m
a
kes
them
repo
rt
their
real
-
ti
m
e
wo
r
king
st
at
us
t
o
the
cent
ral
ser
ver
via
the
av
ai
la
ble
com
m
u
nicat
ion
c
ha
nnel
.
Af
te
r
ac
qu
i
rin
g
the
pe
rm
i
t
to
ex
plo
it
thi
s
data
,
our
a
ppli
cat
ion
retrieves
the
operati
on
rec
ords
of
c
harge
rs
belo
ngin
g
t
o
our
ci
ty
fro
m
the
central
databas
e
ever
y
5
m
inu
te
s
an
d
st
or
es
i
n
the
local
data
base,
c
reati
ng
an
e
vergro
wing
ti
m
e
-
series.
I
t
m
us
t
be
m
entione
d
that
cha
rg
e
rs
are
ad
de
d
or
detache
d
to
or
f
ro
m
the
m
on
it
ori
ng
syst
e
m
fo
r
m
any
reasons
su
c
h
as
ne
w
instal
la
ti
on
s,
c
om
po
ne
nt
fail
ures,
operati
on
s
trat
egy
cha
nge
s,
a
nd
the
li
ke.
Ther
e
f
or
e,
the
nu
m
ber
of
c
hargers,
even n
ot s
o fr
e
qu
e
ntly
, k
ee
ps
changin
g
a
nd t
he data
set
un
de
rgoes
i
ns
ta
bili
ty
f
ro
m
tim
e to ti
m
e.
In
t
he
m
eantim
e,
we
can
as
su
m
e
that
a
pre
-
rese
rv
at
io
n
a
pp
li
cat
io
n
f
or
the
ne
xt
day
c
hargin
g
will
app
ea
r
with
th
e
support
of
ha
rdwar
e
im
plem
entat
ion
w
hi
ch
di
rectl
y
co
ntr
ols
the
c
harger
operati
on
unde
r
the
co
ordi
nation
of
upp
e
r
la
ye
r
ap
plica
tio
ns.
In
ste
a
d
of
proce
ssin
g
r
ese
rv
at
io
n
r
equ
e
sts
one
by
on
e
,
it
al
locat
es
(c
ha
rg
e
r,
tim
e
slo
t)
pairs
to
m
ult
iple
E
Vs
sim
ul
ta
neously
to
ac
hieve
the
gi
ve
n
pe
rfor
m
ance
goal
.
The
n,
it
is
po
ssible
to
desig
n
a
bi
dd
i
ng
m
echan
ism
fo
r
hot
cha
r
ger
s
durin
g
peak
hours
s
o
as
to
avo
i
d
unpredict
able
energy
dem
and
im
balance.
Actuall
y,
our
r
esearch
te
am
ha
s
pr
opos
e
d
an
EV
-
c
harger
m
at
ching
schem
e
wh
ic
h
no
t
on
ly
incre
ases
the
num
ber
of
m
at
ches
bu
t
al
so
e
nhan
ces
the
re
venu
e
of
c
hargin
g
serv
ic
e
pro
vid
er
s,
ta
il
o
ri
ng
the
sta
bl
e
m
arr
ia
ge
prob
le
m
to
ta
ke
as
m
any
high
-
bi
dd
e
rs
a
s
possible
[
8].
H
oweve
r
,
the
pe
rfor
m
ance
has
been
m
e
asur
e
d
us
in
g
a
rather
unreali
st
ic
rando
m
pro
cess
m
od
el
.
H
ence,
t
his
pa
pe
r
is
to
first
s
umm
ariz
e
the
pri
ce
e
ffec
t
on
c
ha
rg
i
ng
dem
and
us
i
ng
t
h
e
Jej
u
data
set
[
9]
a
nd
the
n
fee
d
t
he
res
ult
to
the E
V
-
c
harger
pre
-
reservat
io
n
m
echan
ism
t
o
asses
s it
s
perform
ance b
ase
d on a
real
-
li
fe
scenari
o.
This
pa
per
is
orga
nized
as
f
ollows:
Af
te
r
descr
i
bing
the
m
ai
n
top
ic
a
nd
it
s
c
orrelat
ion
with
our
pr
e
vious
rese
a
rch
es
in
Sect
ion
1,
Se
ct
ion
2
s
hows
the
m
ai
n
featu
re
of
the
m
on
it
or
ing
data
a
rch
i
ve
a
nd
the
analy
sis
re
su
lt
in
te
rm
s
of
the
cha
rg
i
ng
dem
and
dy
nam
i
cs.
Sect
io
n
3
c
aptu
res
the
rea
li
sti
c
de
m
and
patte
rn
for
the
e
valu
at
ion
of
E
V
-
r
el
at
ed
app
li
ca
ti
on
s,
be
ginning
with
the
exp
la
natio
n
of
our
pre
-
rese
r
vation
m
echan
ism
.
Sect
ion
4
c
onduct
s
e
xp
e
rim
e
nts
to
m
easure
the
pe
rfor
m
ance
of
the
E
V
-
c
hargin
g
m
at
ching
schem
e
with
t
he
a
naly
zed
ch
arg
i
ng
loa
d.
Fi
nally
,
Sect
io
n
5
c
oncl
udes
t
hi
s
pa
per
with
a
bri
ef
intr
oduct
ion
of
fu
t
ur
e
w
ork.
2.
PRI
CE EFFE
CT AN
ALY
SI
S
2.1.
Data ar
chive
To
beg
i
n
with,
we
plo
t
the
l
oc
at
ion
of
eac
h
charge
r
on
t
he
ro
a
d
netw
ork
of
Je
ju
ci
ty
as
show
n
i
n
Figure
1.
T
his
m
ap
is
do
wnloade
d
from
an
open
data
si
te
in
a
n
ESRI
sh
a
pe
file
for
m
at
and
pl
otted
on
the
R
pac
kag
e
workspace
[
10]
.
The
coa
stl
ine
of
t
his
isl
and
stret
ches
a
bout
200
km
l
ong,
wh
il
e
the
t
wo
m
ost
popula
te
d
to
w
ns
are
l
ocated
in
the
nort
he
rn
m
os
t
and
s
outhe
rn
m
os
t
re
gions.
Je
ju
ci
ty
is
on
e
of
th
e
m
os
t
fam
ou
s
t
ourist
at
tract
ion
s
in
t
he
East
Asia,
s
o
t
ourists
occ
upy
a
la
r
ge
po
rtion
of
the
who
le
ci
ty
traf
fic
a
l
m
os
t
al
l
ye
ar
r
ound.
Ther
e
f
or
e,
the
chargin
g
l
oad
is
not
re
stric
te
d
to
local
resi
de
nts
bu
t
c
om
es
fr
om
tou
rists
dr
iving
EV
re
nt
-
a
-
cars
.
The
fig
ur
e
s
hows
t
hat
cha
rgers
are
i
ns
ta
ll
ed
la
r
gely
in
pro
po
rtion
t
o
the
popula
ti
on
de
ns
it
y,
wh
ic
h
is
de
fin
it
el
y
interrelat
ed
with
the
c
urren
t
E
V
pe
ne
trat
ion
[
7].
T
hu
s
,
we
can
s
ee
tw
o
da
rk
a
reas
i
n
the
up
per
a
nd
lowe
r
par
ts
of
the
m
ap.
In
add
it
io
n,
t
o
m
eet
the
dem
and
f
r
om
tou
rists,
s
om
e
char
ge
rs
a
r
e
instal
le
d
in
fa
m
ou
s
tourist
at
tract
ion
s.
T
hose
c
ha
rg
e
rs
are
scat
te
re
d
ov
e
r
t
he
ci
ty
area.
I
n
the
f
igure,
wh
e
n
m
or
e t
ha
n on
e
ch
a
r
ger
i
s instal
le
d
i
n
th
e sam
e b
uildin
g,
m
ulti
ple let
t
ers ov
e
rlap
and
just
on
e
appea
rs.
On
the
m
ap,
c
harger
s
a
re
m
a
rk
e
d
by
F
(
Fr
e
e),
M
(Me
diu
m
-
pr
ic
e),
a
nd
E
(E
xpen
sive)
,
res
pecti
vely
.
Ther
e
a
re
c
urr
ently
3
pri
ce
l
evels.
First,
it
is
fr
ee
(
F)
on
60
cha
r
ger
s
op
e
rated
by
th
e
local
go
vern
m
ent.
Seco
nd,
it
cost
s
about
3
USD
(M)
to f
ully
charge
a 2
0
kw
h
EV
batte
ry o
n
49
c
ha
rg
e
rs
possesse
d
by
a
nationa
l
gove
rn
m
ent
m
inist
ry.
Last
,
6
USD
(E
)
is
cha
r
ged
f
or
t
he
s
am
e
el
ectr
i
ci
ty
a
m
ou
nt
at
136
c
harge
rs
run
by
pri
vate
se
r
vi
ce
pro
vid
er
s
[
7].
E
ven
th
ough
the
di
ver
sit
y
of
pri
ce
le
vel
s
is
quit
e
lim
ited,
we
ca
n
est
i
m
at
e
how
m
uch
the
pri
ce
will
in
f
luence
the
dri
ver
si
de
be
hav
i
or
f
or
f
uture
c
harger
dep
l
oym
ent.
As
m
entione
d
earli
er,
the
nu
m
ber
of
c
harg
ers
unde
r
the
c
on
t
ro
l
of
the
m
on
it
or
ing
sys
tem
has
c
ha
nged
from
tim
e
t
o
ti
m
e.
Durin
g
A
ugus
t
2017,
the
cha
rg
i
ng
net
work
m
os
t
reli
ably
op
e
rated
,
a
nd
li
ng
the
la
rg
e
st
nu
m
ber
of
c
ha
rg
i
ng
transacti
ons.
H
ence,
t
his
re
se
arch
f
ocuses
on
the
data
a
rc
hi
ve
obta
ine
d
duri
ng
this
per
i
od.
Be
side
s,
sl
ow
AC
charger
s,
m
ai
n
ly
instal
le
d
i
n
ind
ivi
du
al
hom
es
f
or
ove
rn
i
gh
t
c
hargin
g,
are
not
i
nclu
de
d
i
n
the
m
onit
or
in
g
do
m
ai
n,
as t
he
y hav
e
no reas
o
n t
o be
global
ly
co
ordi
nated.
W
e a
re
only
intereste
d i
n
t
he
pub
li
c
fast c
ha
rg
e
rs.
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.
9
, N
o.
6
,
Dece
m
ber
201
9
:
5586
-
5595
5588
Figure
1
.
EV c
harger
d
ist
ri
bu
t
ion
This
data
arc
hiv
e
is
quit
e
diff
e
re
nt
from
the
le
ga
cy
ones
w
hich
ta
ke
c
hargin
g
re
cords
f
r
om
resp
ect
ive
E
vs
[1
1].
So,
w
e
s
el
ect
the
occ
up
ancy
rate
as
th
e
m
a
in
m
e
tric
for
the
pri
ce
ef
fect
analy
sis
i
n
our
exp
e
rim
ent.
H
ere,
th
e
occ
up
ancy
rate
is
t
he
pro
bab
il
it
y
that
a
c
harger
is
occ
up
ie
d
at
a
sp
eci
fic
ti
m
e
instant
and
cal
culat
e
d
by
di
vid
in
g
the
num
ber
of
repor
ts
in
di
cat
ing
a
c
harge
r
is
ser
vicin
g
a
n
EV
by
that
of
total
repor
ts
.
I
n
ca
s
e
a
ne
w
E
V
re
places
t
he
pre
vious
one
wit
hi
n
a
5
-
m
inu
te
repor
t
inter
val
in
t
he
s
am
e
char
ge
r,
it
is
no
t
possib
le
to
know
whet
her
a
ne
w
tra
ns
act
io
n
has
st
arted,
as
t
he
re
port
do
es
no
t
i
nclu
de
E
V
i
de
ntifie
rs
for
the
sa
ke
of
pri
vacy
protec
ti
on
.
He
nce,
th
e
occ
upancy
ra
te
is
the
m
os
t
m
eaningfu
l
i
nfor
m
at
ion
we
c
an
get
with
this
data
arch
i
ve.
It
giv
e
s
us
a
n
insi
gh
t
of
c
harge
r
util
iz
at
ion
a
nd
c
ha
rg
i
ng
dem
and
.
Ba
sed
on
t
he
a
bove
def
i
niti
on
s,
on
e
of
our
pr
e
vio
us
w
orks
cl
ust
ers
the
da
il
y
patte
rn
on
ea
ch
cha
r
ger
to
check
i
f
there
exists
a
com
m
on
ser
vic
e b
e
hav
i
or sh
a
r
ed by a
gro
up
of cha
rg
e
rs, b
ut
w
it
hout th
e c
on
si
der
at
io
n o
f
the
pr
ic
e le
vel
[
7]
.
2.2.
Dem
an
d
beh
aviors
The
first
ex
pe
rim
ent
m
easur
es
the
o
ccu
pa
ncy
rate
f
or
t
hose
3
gr
oups
of
c
hargers
durin
g
A
ugus
t
2017,
a
nd
the
resu
lt
s
are
s
ho
wn
i
n
Fig
ur
e
2.
Fi
gure
2(a)
plo
ts
the
occupan
cy
rate
for
a
whole
day,
wh
il
e
Figure
2(b)
only
f
or
t
he peri
od fro
m
1
2 PM
to
6
PM
. Ch
a
r
ging d
em
and
is
co
nce
ntrate
d
i
n
this
inte
rv
al
, duri
ng
wh
ic
h
the
ci
ty
wide
e
ne
rg
y
dem
and
al
s
o
rea
ches
t
he
peak.
Figure
2(
a
)
s
hows
that
t
he
oc
cup
a
ncy
rate
of
free
charger
s
is
hi
gh
e
r
t
han
the
oth
e
rs
by
8.9
and
9.3
%
on
ave
rag
e
,
respec
ti
vely
,
w
hile
the
occ
up
a
nc
y
rate
diff
e
re
nce
bet
ween
m
edium
-
pr
ic
e
an
d
ex
pe
ns
ive
c
hargers
is
ve
ry
sm
al
l.
On
the
c
on
t
rary
,
the
ga
ps
betwee
n
3
curves
bec
ome
s
m
or
e
viv
i
d
in
Fig
ure
2(
b).
F
ree
cha
r
ge
rs
are
m
or
e
fr
e
qu
e
ntly
us
e
d
tha
n
m
ediu
m
-
pr
ic
e
charger
s
by
up
to
13.0
%,
bu
t
the
gap
has
be
en
once
re
ve
r
sed.
T
his
sit
ua
ti
on
can
possi
bly
happe
n,
as
m
or
e
m
edium
-
pr
ic
e
charger
s
a
re
i
nst
al
le
d
in
t
ouri
st
at
tract
ion
s
,
wh
e
re
t
he
dayt
i
m
e
dem
and
c
an
e
xplo
de.
A
nyway,
the
e
nlar
ged
ga
p
i
nd
ic
at
es
th
at
so
m
e
dr
i
vers
c
hange
cha
r
ger
s
wh
e
n
fi
ndin
g
t
hat
fr
ee
on
e
s
t
hey
want
are
al
read
y
occupi
ed
in
bu
sy
hours.
H
ow
e
ver, e
ven in t
hat cas
e, they
rar
el
y c
hoos
e
expe
ns
iv
e ones.
(a)
Whol
e d
ay
occupa
nc
y rate
(b)
Hot
-
hour
occ
up
a
ncy
rate
Figure
2
.
D
ai
ly
ch
a
rg
i
ng b
e
ha
vior
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
Pric
e eff
ect
a
naly
sis a
nd
pr
e
-
reseravti
on
sc
he
me on el
ect
ric
vehicl
e c
ha
r
gi
ng n
et
w
or
ks (J
unghoo
n
Lee)
5589
Nex
t,
Fig
ur
e
3(a)
a
nd
Fig
ur
e
3(b)
plo
t
the
hourl
y
occ
up
a
nc
y
rates
for
al
l
charger
s
an
d
t
ho
s
e
f
or
ho
t
20
c
hargers,
re
sp
ect
ively
,
t
o
c
heck
the
loa
d
im
balance
betw
een
t
he
c
harge
rs.
O
ur
ob
se
r
va
ti
on
has
disc
overe
d
that
a
bout
50
c
harger
s
dom
in
at
e
the
ove
rall
cha
r
ging
act
ivit
ie
s.
Am
ong
t
he
se,
23
fr
ee
,
12
m
edium
-
pr
ic
e,
a
nd
15
ex
pe
ns
ive
charger
s
a
re
i
nclu
ded
a
nd
they
are
us
uall
y
instal
le
d
in
high
-
popula
ti
on
a
reas.
Most
of
al
l,
the
tw
o
fig
ur
es
sho
w
a
n
a
bsolute
di
ff
e
ren
ce
i
n
the
occupa
nc
y
rate.
In
te
r
est
ing
ly
,
e
xpe
ns
i
ve
cha
rg
e
rs
a
re
m
or
e
fr
e
qu
e
ntly
us
e
d
tha
n
the
ot
he
rs
over
t
he
tim
e
interval
f
ro
m
2
AM
to
4
A
M.
Dr
i
ver
s
see
m
to
char
ge
t
he
ir
EVs
without
c
on
si
de
rin
g
the
pri
ce
at
dawn.
O
n
t
he
c
on
tra
ry,
duri
ng
dayt
im
e,
the
di
ff
e
ren
ce
betwee
n
t
he
three
i
s
m
agn
ifie
d.
Th
e
peak
dem
and
ta
kes
place
be
tween
16
PM
and
18
PM
in
Jeju,
w
he
re
to
ur
ist
s
occ
upy
a
la
rge
portio
n
of
ci
ty
wide
tra
ff
ic
on
the
ro
a
d.
In
t
ui
ti
vely
,
daily
t
ours
us
ually
ne
ed
one
m
or
e
chargin
g
ar
ound
this
tim
e,
pr
ovide
d
that
EVs
a
re
fu
ll
y
charge
d
ov
e
r
night
an
d
sta
rt
the
ir
t
rips
in
t
he
m
ornin
g.
In
a
ddit
ion
,
Figure
3(b
)
tra
ces
in
div
id
ual
dem
and
patte
r
ns
.
He
re,
20
out
of
hot
50
cha
rg
e
rs
a
re
rand
om
ly
sel
ect
ed
to
av
oi
d
too
m
uch
cu
rve
com
plexity
.
The
fi
gure
s
hows
that
m
os
t
of
high
-
dem
and
chargers
sh
a
r
e
a
com
m
on
de
m
an
d
patte
rn
,
in
dicat
ing
that
a
pri
ce
po
li
cy
can
be
ap
plied
unif
or
m
ly
.
Af
te
r
al
l,
a
reserv
at
ion
m
echan
ism
is
necessa
ry dur
i
ng the i
nter
val
from
1
2
to
6 P
M.
(a)
H
ourly
o
cc
up
a
ncy
rate
(b)
Hot
-
cha
rg
e
r hour
ly
rate
Figure
3
.
Ho
url
y char
ging
be
hav
i
or
As
a
n
e
xten
de
d
ver
si
on
of
[9]
,
this
w
ork
f
oc
us
es
on
the
pr
ic
e
eff
ect
t
o
th
e
dem
and
patte
rn
to
obser
ve
the
pe
rfo
rm
ance
of
E
V
c
hargin
g
a
pp
li
cat
io
ns
.
For
m
or
e
de
ta
il
ed
ex
per
i
m
ent
resu
lt
s,
inclu
ding
the
e
ff
ect
of
charger
de
ns
it
y
and
the
dista
nce
to
a
nearb
y
fr
ee
c
ha
rg
e
r
to
the
occ
up
a
nc
y
rate,
re
fer
to
[
9].
In
ad
diti
on,
f
r
om
the
te
m
po
ral
on
-
off
stream
,
we
ca
n
detect
the
sta
rt
an
d
e
nd
ti
m
e
of
c
ha
rg
i
ng
tra
ns
act
ion
s
.
It
al
lo
ws
us
to
est
i
m
at
e
the
len
gt
h
of
cha
rg
i
ng
tim
e,
the
a
m
ou
nt
of
e
nergy,
a
nd
the
li
k
e,
m
uch
re
duci
ng
t
he
am
ount
of
m
e
m
or
y t
o
sto
r
e the
data arc
hi
ve.
3.
DEM
AND P
A
TT
ERN
F
OR
THE
PR
E
-
RE
SERVATIO
N
M
E
CHANI
S
M
3.1.
Reser
vation
mecha
nism
W
it
hout
a
res
erv
at
io
n
m
echan
ism
,
an
EV
dr
i
ver
visit
s
a
chargin
g
sta
ti
on
on
the
r
oute
an
d
ta
kes
a
cha
rg
e
r
i
f
a
vaila
ble.
Othe
rw
ise
,
the
dri
ver
will
ei
ther
wait
or
m
ov
e
to
a
nothe
r
s
ta
ti
on
.
E
ve
n
t
ho
ug
h
a
releva
nt
ap
pl
ic
at
ion
sho
ws
t
hat
a
cha
rg
e
r
i
s
fr
ee
now
,
it
can
be
ta
ke
n
by
oth
e
r
E
Vs,
whil
e
an
EV
is
he
adin
g
for
the
c
harge
r
.
As
t
he
ser
vic
e
tim
e,
na
m
e
ly,
the
cha
r
ging
transacti
on
le
ngth,
f
or
eac
h
E
V
can
e
xte
nd
t
o
te
ns
of
m
inu
te
s,
th
e
wait
in
g
ti
m
e
can
grow
s
oon.
W
it
h
the
s
uppo
rt
of
an
int
erf
ace
ci
rc
uit
capab
le
of
i
nject
ing
el
ect
rici
ty
to
an
EV
duri
ng
t
he
sp
eci
fic
ti
m
e
interval
under
the
co
ntr
ol
of
the
co
ordinati
on
entit
y,
a
cha
r
ging
sta
ti
on
ca
n
pr
ov
i
de
a
rese
r
va
ti
on
m
echan
i
sm
.
Gen
erall
y,
a
reservat
io
n
re
qu
e
st
is
processe
d
one
by
one
accor
ding
to
t
he
FCFS
(F
irs
t
Com
e
First
S
erv
ic
e)
poli
cy
.
He
re,
th
e
request
s
ubm
itted
fir
st
has
an
ul
tim
a
te
pr
i
or
it
y
ov
e
r
t
he
la
tt
er
ones
,
wh
il
e
dri
ve
rs
s
el
ect
charge
rs
(alo
ng
with
ti
m
e
slots)
out
of
a
vaila
ble
on
es
with
the
help
of
an
app
li
cat
io
n
whic
h
s
hows
the
up
-
to
-
date
rese
rv
at
io
n
sta
tus
[
12
]
.
A
n
E
V
ca
n
iss
ue
a
reservat
i
on
request
un
ti
l
j
ust
b
ef
ore the
ti
m
e slot it
w
ant
s,
as
l
ong
a
s it
can a
rr
ive
.
Eve
n
th
ough
t
he
FCF
S
po
li
c
y
is
so
c
ommon,
easy
to
im
plem
ent,
and
f
ai
r
to
requeste
rs,
it
ca
nnot
coor
din
at
e
requests
t
o
ac
hieve
oth
e
r
syst
e
m
go
al
s
su
c
h
as
gl
ob
al
pea
k
sh
a
ping
an
d
per
s
onal
iz
ed
pri
ci
ng
.
Me
anwhil
e,
th
ere
are
E
Vs
w
ho
s
e
ne
xt
day
tour
sc
hedules
are
fixe
d,
f
or
exam
ple,
deliv
ery
ser
vices,
t
ourists
veh
ic
le
s,
a
nd
s
cheduled
busin
ess
tri
ps
.
Mo
re
ov
e
r,
E
Vs
ca
n
po
s
sibly
try
t
o
reserve
a
c
harger
a
nd
m
ake
a
to
ur
plan
acc
ordin
g
to
the
reservat
ion
re
su
lt
,
w
he
n
they
ha
ve
m
ulti
ple
op
ti
ons
.
I
n
this
case,
a
set
of
re
quest
s
from
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.
9
, N
o.
6
,
Dece
m
ber
201
9
:
5586
-
5595
5590
EVs
a
re
ha
ndle
d
sim
ultaneou
sly
.
That
is,
t
he
co
ordi
nato
r
colle
ct
s
reque
sts
by
the
e
nd
of
a
ce
rtai
n
de
adline
and
pairs
EVs
and
c
ha
rg
e
rs.
Her
e
,
an
E
V
m
ay
su
bm
it
m
or
e
than
on
e
cha
rg
e
r
it
wan
ts
,
wh
il
e
at
m
os
t
on
e
c
a
n
be
a
d
m
itted.
The
n,
m
any
-
to
-
m
any
coo
r
dina
ti
on
is
ca
rr
ie
d
ou
t
i
n
a
bat
ch
m
od
e
as
a
va
riant
of
bi
par
ti
te
m
at
ching
.
T
his
is
the
m
ai
n
idea
of
pr
e
-
rese
r
vation.
W
e
as
s
um
e
that
the
tim
e
axis
is
div
ided
i
nto
a
seri
es
of
fixe
d
-
siz
e
slot
s
a
nd
re
ser
vations
a
re
m
ade
by
the
sl
ot.
E
V
c
ha
rg
i
ng
ca
n
sta
rt
on
ly
from
the
be
ginn
ing
of
the tim
e slot th
e EV
h
as
r
ese
r
ved.
Fr
om
the
view
po
i
nt
of
EV
s,
the
rese
rv
at
io
n
is
fu
lfil
le
d
in
two
ste
ps
,
na
m
el
y,
pr
e
-
rese
rv
at
io
n
a
nd
ordina
ry
FCF
S
-
ba
sed
rese
rvat
ion
,
resp
ect
i
vely
.
I
n
the
pre
-
reservat
i
on
ste
p,
E
Vs
ca
n
basical
ly
nom
inate
m
ul
ti
ple
(ch
ar
ger,
slot)
pair
s
they
wa
nt.
T
he
reservat
io
n
re
su
lt
s
are
not
cr
eat
ed
unti
l
the
end
of
a
n
ap
plica
ti
on
dead
li
ne
.
T
hen,
slots
are
m
atch
ed
t
o
EVs
a
ccordin
g
to
th
e
pr
e
fer
e
nce
at
the
EV
side
a
s
well
as
the
pri
or
it
y
functi
on
set
by
cha
rg
e
rs.
A
n
EV
m
ay
fail
in
rese
rv
i
ng
a
sl
ot
if
al
l
re
quest
entries
a
re
bea
te
n
by
oth
e
rs.
A
slot
cannot
be
assi
gn
e
d
in
case
no
EV
has
a
ppli
ed.
T
ho
se
E
Vs
fail
ed
in
the
pre
-
reservat
io
n
sta
ge
a
re
re
qu
i
red
t
o
fo
ll
ow
t
he
next
reservat
io
n
proce
s
s.
Her
e
,
ne
wco
m
ers
can
j
oi
n
the
pr
oces
s
an
d
eac
h
request
is
handled
one
by
one.
He
re,
par
ti
ci
pa
nts
do
no
t
ha
ve
to
nam
e
m
ulti
pl
e
can
did
at
es
s
i
m
u
lt
aneo
us
ly
.
In
the
sec
on
d
ste
p,
the
dri
ve
r
m
us
t
searc
h
slots
le
ft
after
the
pr
e
-
rese
rv
at
io
n
proces
s
a
nd
sel
e
ct
t
he
on
e
w
hi
ch
best
m
eet
s
his
or
her
dri
ving
sc
hedule.
We
t
hi
nk
t
hat
so
m
e
In
te
r
net
portal
s
pro
vid
e
we
b
ser
vices
in
dicat
ing
the
l
ocat
ion
of
charger
s
on
th
e
m
ap
an
d
can
possibly
im
pl
e
m
ent
a
reserv
at
ion
m
echan
ism
j
us
t
li
ke
ho
t
el
book
i
ng
ser
vices
.
Our
desi
gn
put
s
em
ph
asi
s
on
the
pr
e
-
re
ser
va
ti
on
ste
p,
as
t
he
ordina
ry
pro
cess
is
no
t
pec
uliar
a
nd
has
nothin
g
to im
pr
ov
e
.
Figure
4
s
umm
arizes
this
m
ec
han
ism
,
assumi
ng
that
n
c
ha
r
ger
s
a
re
pa
rtic
ipati
ng
for
the
t
i
m
e
per
iod
from
12
PM
to
18
PM.
Only
d
uri
ng
the
h
ot
h
ou
rs,
t
he
r
ese
rv
at
io
n
ser
vice
is p
rovide
d, w
hile no
t
eve
ry charger
par
ti
ci
pates
i
n
the
ser
vice.
Th
e
tim
e
slot
is
30
m
inu
te
s,
m
a
king
the
total
num
ber
of
slots
to
m
at
ch
is
n
x
12
if
each
cha
rger
op
e
ns
12
sl
ots
for
rese
rv
at
io
n.
T
he
slot
le
ng
t
h
will
be
r
oughly
set
to
the
aver
a
ge
c
ha
rg
i
ng
transacti
on
tim
e
a
nd
m
us
t
inc
lud
e
the
ov
e
r
he
ad
to
m
anipu
l
at
e
the
re
gistra
ti
on
de
vice,
pl
ug
t
he
EV
c
onnecto
r
into
a
cha
rg
e
r,
pay
the
fee,
an
d
the
li
ke.
W
e
assum
e
that
EVs
will
le
ave
be
fore
the
e
nd
of
re
ser
ved
ti
m
e
slots,
po
s
sibly
du
e
t
o
str
ong
pe
nalty
poli
ci
es.
In
th
e
fig
ure,
slots
are
nu
m
ber
e
d
seq
uen
ti
al
ly
from
Charg
e
r
0.
An
E
V
sp
eci
fies
the
li
sts
of
(
c
harger
,
slot) p
ai
rs,
w
hi
ch
are
m
app
e
d
to
res
pecti
ve
i
den
ti
fier
s
f
or
a
ll
ocati
on
. A
s
at
m
os
t
on
e
of t
he
re
qu
est
s can be ac
c
epted pe
r
E
V,
t
he
sl
ot assig
nm
ent is reduce
d
to
a
bip
a
rtit
e m
at
ching
pro
ble
m
.
W
it
hout
any
oth
er
requirem
ent,
the
goal
of
t
he
m
at
ch
pr
oc
edure
will
be
t
o
m
axi
m
iz
e
the
nu
m
ber
of
(EV,
sl
ot)
pair
s.
Howe
ver,
c
ha
rg
e
rs
prefe
r
hi
gh
est
bi
dd
e
rs
to
im
pr
ove
the
ir
re
ve
nu
e
.
T
he
n,
the
r
e
a
re
w
ei
gh
t
s
on
E
V
-
sl
ot
li
nks
a
nd
th
e
pro
blem
gets
com
plex.
Her
e,
the
prefe
ren
ce
f
r
om
each
par
ty
i
s
ex
plici
tl
y
defi
ned
.
On
t
he
EV
side,
eac
h
E
V
sim
ply
li
s
ts
(ch
ar
ge
r,
s
lot)
pair
s
acc
ordin
g
to
it
s
pr
e
fer
e
nce
order.
Un
li
ke
the
ordina
ry
m
at
chi
ng
sc
hem
e,
th
e
num
ber
of
app
li
cat
io
n
e
nt
ries
is
m
uch
s
m
al
le
r
than
th
e
total
nu
m
ber
of
sl
ots
an
d
can
be
diff
e
re
nt
EV
by
EV.
On
t
he
oth
e
r
ha
nd,
at
the
cha
rg
e
r
si
de,
a
c
harge
r
giv
es
pr
i
or
it
y
to
t
hos
e
EVs
wh
ic
h
wan
t
t
o
pay
m
or
e
to
i
ncr
ea
se
it
s
re
venue.
T
o
this
en
d,
our
ser
v
ic
e
m
akes
EVs
su
bm
it
the
pr
i
ce
they
w
ould
li
ke
to
pay,
m
aking
a
rese
r
vation
ent
ry
(c
harger
,
slot
,
bi
d).
I
n
t
he
e
xa
m
ple
of
Figure
4,
E
V
0
app
li
es
for
2
s
lots
w
hile
EV
1
just
on
e
.
T
he
n,
the
rese
rv
at
i
on
m
echan
ism
chooses
first
t
ho
s
e
requests
biddin
g
hi
gher.
I
n
ca
se
two
or
m
or
e
biddin
gs
a
re
ti
ed,
t
he
re
quest
hav
i
ng
a
s
horter
prefe
re
nce
l
eng
t
h
will
be
sel
ect
ed
as
it
has
le
ss
po
s
sibil
it
y
to
be
pic
ked.
I
f
ti
ed
agai
n,
the
r
equ
e
sts
w
hich
arr
ive
d
first
wi
ll
be
pr
e
ferred
.
Af
te
r
al
l,
t
he
pr
e
fe
ren
ce
f
unct
ion
at
the
cha
r
ger
side
is
c
om
pl
et
ed.
T
his
pro
cedure
is
cu
stom
iz
ed
from
the S
MP
(S
ta
ble Ma
rr
ia
ge
P
r
ob
le
m
)
sol
ver
.
Figure
4
.
Mat
chin
g
-
base
d pr
e
-
al
locat
ion
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
Pric
e eff
ect
a
naly
sis a
nd
pr
e
-
reseravti
on
sc
he
me on el
ect
ric
vehicl
e c
ha
r
gi
ng n
et
w
or
ks (J
unghoo
n
Lee)
5591
The
bip
arti
te
m
at
ching
pr
oc
edure
r
uns
with
the
cl
ass
def
i
niti
on
s
how
n
i
n
Fig
ure
5
[8
]
.
The
Ch
arger
cl
ass
is
insta
ntiat
ed
f
or
each
(c
harger
,
sl
ot)
pair.
Eac
h
E
V
obj
ect
c
onta
ins
a
prefe
re
nce
li
st
c
on
sis
ti
ng
of
m
ul
ti
ple
reque
st
entries,
w
hi
le
pos
fiel
d
points
t
o
the
re
qu
e
st
entry
t
o
proces
s
nex
t.
It
ca
nnot
go
beyo
nd
the
num
ber
of
entries
a
nd
if
it
do
es
,
the
EV
ca
nnot
be
assig
ned
to
a
slot
an
d
will
be
el
im
inate
d
from
the
m
at
ching
process
.
I
n
ad
diti
on
,
the
m
at
ch
fiel
d
i
n
the
cha
rg
e
r
cl
ass
kee
ps
t
rack
of
the
c
u
r
ren
t
m
at
ch
.
The
m
at
ching
schem
e
proces
ses
from
the
fi
rst
E
V,
wh
ic
h
arr
ive
s
first.
H
ow
e
ve
r,
this
orde
r
does
not
aff
ec
t
the
final
m
at
c
hing
re
su
lt
.
If
the
(c
harger
,
s
lot)
is
not
ye
t
m
at
ched
,
it
w
il
l
be
assigne
d
to
the
E
V,
t
ha
t
is,
the
m
atch
fiel
d
in
the
c
h
ar
ge
r
cl
ass
will
be
se
t
to
that
EV
.
O
ther
wise,
the
c
urren
t
E
V
a
nd
the
chall
en
ging
E
V
fig
ht
acc
ordin
g
t
o
th
e
c
rite
ria
desc
ribe
d
previ
ou
sly
.
Th
e
wi
nn
e
r
will
ta
ke
t
he
place
wh
il
e
t
he
l
os
e
r
will
con
ti
nue
the
c
on
te
ntion
with
the
ne
xt
requ
est
entry.
He
re
,
the
po
s
point
er
will
procee
d
by
on
e
.
We
do
no
t
consi
der the ca
se a sin
gle E
V t
ries to r
e
ser
ve t
wo
or m
or
e slots sim
ultaneou
sly
, as o
ne
f
ul
l char
ging ca
n cove
r
daily
trips
i
n
m
os
t case
s
.
Figure
5
.
EV a
nd ch
a
r
ger
pref
eren
ce
3.2.
Requ
es
t
gener
at
i
on
sce
na
ri
o
As
s
how
n
in
the
pr
e
vious
se
ct
ion
,
each
c
ha
rg
e
r
has
it
s
ow
n
dif
fer
e
nt
occ
up
a
ncy
rate,
a
nd
we
thi
nk
that
this
dem
and
distrib
utio
n
will
be
ke
pt
as
li
fe
sty
le
s
or
dr
i
ving
patte
rn
s
of
ci
ti
zens
will
no
t
c
ha
ng
e
i
n
a
short
ti
m
e.
The
num
ber
of
ch
ar
ger
s
w
hi
ch
will
par
ti
ci
pate
in
t
he
res
erv
at
io
n
se
r
vice
is
not
kn
own
ye
t.
Howe
ver,
f
or
tho
s
e
in
t
he
ser
vice
gro
up,
E
Vs
will
send
r
eserv
at
io
n
requests
m
uch
si
m
il
arly
as
the
curren
t
dem
and
patte
r
n.
Hen
ce
,
f
or
n
cha
r
ger
s
part
ic
ipati
ng
in
th
e
reservat
io
n
s
e
rv
ic
e,
our
lo
a
d
ge
ne
rati
on
m
od
ule
su
m
s
up
the
c
urren
t
occupa
ncy
rates
f
or
t
hem
and
the
n
cal
culat
es
the
pro
bab
il
it
y
that
a
charger
will
be
sel
ect
ed
just
li
ke
the
well
-
known
R
ou
le
tt
e
wh
eel
m
et
ho
d.
Her
e
,
n
is
gi
ven
as
a
n
ex
pe
rim
ent
par
am
et
er.
As
for
t
he
nu
m
ber
of
E
Vs,
EVs
will
be
m
ai
nly
boug
ht
by
local
resi
de
nt
s
in
t
he
dow
nto
w
n
area
due
t
o
easy
acce
ssibil
it
y
to
cha
r
ging
facil
it
ie
s.
A
ddit
ion
a
ll
y,
EVs
will
be
pu
rch
ase
d
by
re
nt
-
a
-
car
c
ompanies
se
rv
ic
in
g
f
or
tourist
s,
who
ge
t
m
or
e
inte
re
ste
d
in
eco
-
f
ri
end
ly
d
ri
ving
with
E
Vs.
E
ve
n
th
ough
t
he
num
ber
of
E
Vs
ta
ken
by
to
ur
ist
s
gets
highe
r
f
ro
m
tim
e
to
tim
e,
especial
ly
during
t
he
vacati
on
seas
ons,
th
e
m
ajo
rity
of
E
Vs
on
the
ro
a
d
is
owne
d
by
loca
ls.
Hen
ce
,
the
chargin
g
de
m
and
will
no
t
change
in
s
pite
of
forthc
om
ing
p
enet
rati
on
s
.
Nex
t,
t
he
pro
bab
il
it
y
that
an
E
V
sel
ect
s
low
-
pri
ce
cha
r
ger
s
c
an
be
i
nf
e
rr
e
d
from
the
cu
rr
e
nt
occupa
ncy
rat
e
acco
rd
i
ng
t
o
the
pr
ic
e
le
ve
l.
The
dri
vers
act
ively
pr
efe
r
rin
g
f
ree
c
hargers
are
s
ensit
ive
to
the
cha
r
ging
c
os
t.
R
oughly
sp
ea
king,
t
he
dem
and
rati
o
for
3
groups
i
s
2:3
:
4,
eac
h
of
w
hich
re
presents
exp
e
ns
i
ve,
m
e
diu
m
-
pr
ic
e,
a
nd
free
c
ha
rg
e
r
s,
res
pecti
vely
.
As
the
pr
ic
e
le
vel
will
be
di
ff
e
ren
t
a
nd
ke
ep
changin
g,
es
pe
ci
al
ly
wh
en
c
om
bin
ed
with
t
he
biddin
g
pro
cess
in
t
he
nea
r
f
uture,
t
her
e
exist
s
a
dr
ive
r
gro
up
wh
ic
h
will
put
the
to
p
pri
ori
ty
on
the
cha
rg
i
ng
co
st
an
d
wi
ll
bid
as
lo
w
a
s
possible
.
Mo
reover
,
loc
al
s
te
nd
t
o
m
or
e
li
ke
cheap
er
c
harger
s.
Hen
ce
,
we
set
the
rati
o
of
the
gro
up
as
a
perform
ance
par
am
et
er.
In
th
e
data
arch
i
ve
a
naly
sis,
the
d
iffe
re
nc
e
in
util
iz
at
ion
betwee
n
m
edium
-
pr
ic
e
a
nd
e
xpensi
ve
c
ha
rg
e
rs
gets
le
s
s
vi
vid
al
ong
the
tim
e
axis.
The
sel
ect
ion
see
m
s
to
de
pe
nd
m
or
e
on
th
e
cha
rg
e
r
loc
at
ion
tha
n
t
he
cost
.
Hen
ce
,
f
or
the
non
-
sen
sit
ive
gro
up,
the
pro
bab
il
it
y
to
sel
ect
ou
t
of
t
he
res
t
of
c
harger
s,
will
be
t
he
sam
e,
nam
ely, d
ist
rib
utes r
a
ndom
ly
.
4.
PERFO
R
MANC
E
A
NA
L
Y
SIS RES
ULTS
This
sect
ion
m
easur
e
s
t
he
pe
r
form
ance
of
t
he
pr
opos
e
d
sc
hem
e,
com
par
ing
with
the
le
gacy
FC
FS
strat
egy
a
nd
the
reservat
io
n
-
le
ss
walk
-
in
poli
cy
.
The
pe
r
form
ance
m
e
t
rics
inclu
de
th
e
num
ber
of
m
at
ched
charges
an
d
t
he
gai
ned
prof
it
.
T
he
e
xperim
ent
ta
kes
nu
m
ber
of
EVs
,
num
ber
of
c
ha
r
ger
s
(
slots)
,
a
ver
a
ge
pr
e
fer
e
nce
le
ngth,
a
nd
pri
ce
sensiti
vity
as
pe
rfor
m
ance
pa
ram
et
ers.
By
def
ault,
they
a
re
set
to
800,
50
,
3.0,
and
0.3
3
,
res
pe
ct
ively
.
Perform
ance
is
m
eas
ur
e
d
cha
ngin
g
on
e
par
am
et
er
with
the
oth
er
s
rem
ai
nin
g
at
def
ault
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.
9
, N
o.
6
,
Dece
m
ber
201
9
:
5586
-
5595
5592
values
.
O
ur
prot
otype
is
i
m
plem
ented
with
the
Ja
va
program
m
ing
la
ngua
ge
an
d
it
s
exec
ution
ti
m
e
is
ob
s
er
ved
to
be
le
ss
t
han
0.1
sec
onds
for
al
l
slot
assig
nm
e
nts,
m
aking
the
rese
rv
at
i
on
process
a
re
al
-
li
fe
app
li
cat
io
n.
T
ha
t i
s,
an
a
verag
e p
er
f
or
m
ance PC can
run
t
he
p
re
-
re
ser
vatio
n process
for
t
hous
a
nds
of
E
Vs
a
nd
te
n
th
ou
s
an
ds
of
slots
within
an
acce
ptable
tim
e
bo
un
d.
The
request
e
ntries,
nam
el
y,
(ch
a
r
ger,
slot
,
bi
d)
tup
le
s,
will
be
gen
e
rated
acc
ordin
g
to
the
l
oad
a
naly
sis
re
su
lt
ob
ta
i
ned
i
n
the
pr
e
vious
sect
ion
.
In
a
ddit
ion
,
for
al
l pa
ram
eter
sett
ings,
20
set
s ar
e
ge
ner
a
te
d
an
d
t
he o
btained res
ults ar
e ave
rag
e
d.
The
first
e
xp
e
rim
ent
m
easur
es
the
e
ff
ect
of
t
he
n
um
ber
of
EV
s,
a
nd
the
re
su
lt
s
a
r
e
show
n
i
n
Figure
6.
First,
Fig
ur
e
6(a)
pl
ots
the
num
ber
of
acce
pted
re
qu
e
sts,
wh
e
n
t
he
nu
m
ber
of
EVs
pa
rtic
ipati
ng
in
the
m
at
ching
proces
s
ra
nges
f
ro
m
100
t
o
1,0
00.
Her
e
,
the
le
ng
t
h
of
a
pr
e
f
eren
ce
li
st
ranges
rand
om
l
y
fr
o
m
1
to
5,
m
aking
th
e
aver
a
ge
c
onve
rg
e
t
o
3.0
.
As
can
be
see
n
in
the
fig
ure,
the
curve
of
t
he
pro
posed
SMP
-
base
d
schem
e
ou
t
perform
s
the
ot
he
r
tw
o.
Howe
ve
r,
t
he
ga
p
from
the
FCFS
sc
hem
e
is
not
s
o
l
arg
e
,
at
m
os
t
6.5
%,
for
500
EVs
.
This
e
nh
a
nce
m
ent
is
achie
ve
d
as
t
he
S
MP
-
base
d
sch
e
m
e
giv
es
pre
ceden
ce
to
E
Vs
ha
ving
sh
ort
er
pr
e
fere
nce
le
ngth.
At
the
tw
o
r
ang
e
boun
dar
i
es,
the
ga
ps
are
bel
ow
1.0
%.
E
ve
n
thou
gh
the
i
m
pr
ovem
ent
is
qu
it
e
s
m
al
l,
the
pr
op
os
e
d
schem
e
works
bette
r
f
or
th
e
w
ho
le
exp
e
rim
ent
setting
s.
The
perform
ance
gap
f
ro
m
the
walk
-
in
sc
he
m
e
reaches
27.
8
%
wh
e
n
the
nu
m
ber
of
E
vs
is
600.
T
he
num
ber
of
avail
a
ble
sl
ots
is
al
s
o
600,
so
f
or
the
c
urr
ent
dem
and
distribu
ti
on,
a
bout
1:1
E
V
-
c
harg
er
rati
o
ca
n
e
nhance
the r
ese
r
vation effici
en
cy
.
Nex
t,
Fig
ure
6(b
)
s
hows
t
he
gaine
d
prof
it
f
or
3
schem
es.
Her
e
,
we
ass
um
e
that
there
a
re
3
bi
dd
i
ng
cl
asses,
each
of
w
hich
costs
6,
5,
an
d
4
USD
,
res
pecti
vely
,
for
a
fu
ll
slot
oc
cup
at
io
n.
T
his
sel
ect
ion
c
on
siders
the
cu
rr
e
nt
c
ha
rg
i
ng
pri
ce
pol
ic
y
in
Jej
u.
W
e
assum
e
that
the
pr
ic
e
is
in
de
pende
nt
of
th
e
am
ou
nt
of
c
harge
d
el
ect
rici
ty
,
as
m
os
t
Evs
m
a
king
re
ser
vations
hi
gh
ly
li
ke
ly
to
wan
t
t
o
c
harge
al
m
os
t
to
t
he
f
ull
le
vel.
The
pr
opos
e
d
schem
e
giv
es
pr
ece
de
nce
to
tho
se
requests
biddin
g
hi
gh
e
r
.
The
nu
m
ber
of
biddin
g
le
ve
ls
can
get
la
rg
e
r
an
d
so
does
th
e
pr
ic
e
gap
bet
ween
biddin
g
cl
asses.
H
ow
e
ver,
how
t
o
s
et
tho
se
pa
ra
m
et
ers
is
ano
t
her
prob
le
m
.
The
fi
gure
sh
ows
that
the
gai
ned
prof
it
increases
by
32.
3
%
a
nd
9.5
%,
c
om
par
ed
with
the
walk
-
in
a
nd
FCFS
schem
es,
res
pecti
vel
y.
The
ear
ne
d
prof
it
gets
higher
w
hen
t
he
nu
m
ber
of
E
V
s
gets
cl
os
er
to
t
he
num
ber
of
(c
ha
rg
e
r,
slot)
pai
r
s.
Actuall
y,
in
the
real
-
li
fe
r
equ
e
st
m
od
el
,
chargin
g
dem
and
is
m
or
e
co
ncen
t
r
at
ed
t
o
a
sm
al
l
set
of
c
hargers.
The
perf
or
m
ance
e
nha
ncem
ent
is
r
el
at
ively
sm
alle
r
t
han
the case
of r
a
ndom
ch
ar
ger se
le
ct
ion
s as i
n [
8].
(a)
Ma
tc
hed ch
arg
e
rs
(b)
Gaine
d pro
fit
Figure
6
.
Effec
t of the
num
ber
of E
Vs
Figure
7
plo
ts
t
he
ef
fect
of
t
he
num
ber
of
cha
rg
e
rs,
a
nd
the
exp
e
rim
ent
m
a
kes
it
ra
nge
f
rom
20
to
80
.
This
m
eans
th
at
the
num
ber
of
slots
ra
ng
e
s
f
ro
m
240
to
960.
Her
e
,
E
Vs
sel
ect
a
(c
harg
er,
slot)
acco
r
di
ng
to
the
c
urren
t
l
oad
distrib
ution.
If
the
nu
m
ber
o
f
cha
r
ger
s
is
20,
it
m
eans
that
top
20
c
ha
rgers
i
n
the
occ
up
a
ncy
rate
are
i
nclud
e
d
in
the
s
erv
ic
e
gr
oup.
Hen
ce
,
w
he
n
the
num
ber
of
c
hargers
i
ncrea
ses
,
the
dem
and
im
balance
al
s
o
ge
ts
higher
.
As
sh
ow
n
in
Fig
ure
7(a)
plo
tt
in
g
the
num
ber
of
m
at
ched
re
qu
e
sts,
the
SM
P
-
base
d
schem
e
outp
erfor
m
s
the
ot
her
s
e
xce
pt
one
pa
ram
et
er
se
t.
T
he
be
nef
it
gaine
d
f
ro
m
gi
vin
g
pr
ece
de
nce
to
a
shorter
pr
efere
nce
li
st
dim
inishes
in
this
sit
uation.
The
pro
pos
ed
sc
hem
e
increases
the
num
ber
of
m
at
ched
pairs
by
up
to
24.
6
%,
c
om
par
e
d
with
t
he
wal
k
-
in
sc
hem
e.
I
n
add
it
io
n,
the
ga
ined
prof
it
cu
rv
e
is
qu
it
e
sim
il
ar
to
the
case
of
Fi
gure
6(
b)
.
T
he
SMP
-
based
sc
hem
e
i
m
pr
ove
s
the
rev
e
nue
by
up
to
6.5
%
a
nd
31.
8
%,
com
par
e
d
with
the
FCFS
a
nd
t
he
wa
lk
-
in
schem
es.
The
perf
or
m
ance
gap
te
nds
to
g
et
enlar
ged acco
r
ding to
the i
ncrea
se in t
he num
ber
o
f
c
ha
rg
e
rs
as m
or
e
opti
on
s
are
av
ai
la
bl
e in m
at
ching
slots.
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
Pric
e eff
ect
a
naly
sis a
nd
pr
e
-
reseravti
on
sc
he
me on el
ect
ric
vehicl
e c
ha
r
gi
ng n
et
w
or
ks (J
unghoo
n
Lee)
5593
(a)
Ma
tc
hed ch
arg
e
rs
(b)
Gaine
d pro
fit
Figure
7
.
Effec
t of the
num
ber
of c
hargers
Now,
t
he
ex
pe
rim
ent
m
easure
s
the
eff
ect
of
the
aver
a
ge
le
ng
t
h
of
the
pr
e
fer
e
nce
li
st,
an
d
the
res
ults
are
s
how
n
i
n
Figure
8.
T
he
longer
a
prefe
r
ence
li
st,
t
he
m
or
e
op
ti
ons
t
he
reservat
io
n
m
echan
ism
can
ha
ve.
Figure
8(a)
s
hows
t
he
num
ber
of
m
at
ched
charges
w
he
n
t
he
a
ve
rag
e
num
ber
of
pr
e
fere
nce
li
st
le
ng
t
h
ra
ng
e
s
from
1.
5
to
5.
0.
T
o
be
gin
w
it
h,
the
num
ber
of
m
a
tc
hed
r
equ
e
sts
is
almost
sam
e
fo
r
S
MP
a
nd
FC
FS
-
base
d
schem
es,
and
there
is
no
s
upe
rior
it
y
bet
wee
n
them
.
Even
th
ough
the
SMP
-
base
d
s
chem
e
m
or
e
m
a
tc
hes
EVs
and
c
ha
rg
e
rs
i
n
total
,
the
gap
is
le
ss
than
2
%,
com
par
ed
with
the
FCF
S
schem
e.
The
FCFS
sc
hem
e
m
at
ches
m
or
e
in
3
out
of
7
par
am
et
er
sel
ect
ion
s
.
It
t
el
ls
that
bo
t
h
hav
e
the
sam
e
cha
nce
f
or
th
e
num
ber
of
m
at
ched
pairs.
When
th
e
pre
fer
e
nce
li
st
is
s
uffici
entl
y
la
rg
e
a
nd
t
he
FCF
S
sche
m
e
ca
n
al
s
o
ta
ke
adv
a
ntage
of
pl
enty
sel
ect
able
op
ti
on
s
.
T
he
rese
r
vationless
s
ch
e
m
e
is
no
t
infl
uen
ce
d
by
the
li
st
le
ng
th,
as
it
ta
kes
a
char
ge
r
or
wait
s
unti
l
the
charger
bec
ome
s
avail
able.
T
he
SMP
-
ba
sed
schem
e
ou
tpe
r
form
s
the
walk
-
in
sc
hem
e
by
up
t
o
33.8
%.
Th
e
p
e
rfor
m
ance g
a
p gets la
r
ger acc
ordin
g
t
o
the
in
crease in
the
num
ber
of c
harg
er tuple
s in
the
li
st.
In
a
ddit
ion
,
Fi
gure
8(b)
sho
w
s
the
gai
ne
d
pr
of
it
acco
r
ding
to
the
pr
e
fer
e
nc
e
li
st
le
ng
th.
Eve
n
th
ough
the
num
ber
of
m
at
ched
cha
r
ge
s
is
al
m
os
t
sam
e,
th
is
res
ult
rev
eal
s
t
hat
ou
r
sc
hem
e
i
m
pr
ov
e
s
the
re
venue
of
chargin
g
ser
vi
ce
prov
i
der
s
by
7.2
%,
c
ompare
d
with
t
he
FCF
S
schem
e.
As
ex
pecte
d,
the
gap
gets
la
rg
er
accor
ding
to
th
e
increase
in
t
he
li
st
le
ng
th,
be
nef
it
in
g
f
ro
m
the
la
rg
e
op
ti
on
sp
ace
.
Be
sid
es,
ou
r
sc
hem
e
ear
ns
42.9
%
m
or
e
rev
e
nue
tha
n
the
walk
-
in
sc
hem
e.
W
hile
t
he
FC
FS
sche
m
e
can
al
s
o
be
nef
it
from
a
la
rg
e
r
pr
e
fer
e
nce
li
st
in
t
he
nu
m
ber
of
m
at
ched
pairs,
the
SM
P
-
ba
sed
sc
he
m
e
can
sel
ect
highe
r
bidder
s
m
or
e
eff
ic
ie
ntly
.
Howev
e
r,
it
is
quit
e
unreali
st
ic
f
or
EV
s
to
li
st
m
or
e
than
5
ca
nd
i
dates
f
or
the
ne
xt
da
y
tour
sche
du
le
.
It
is
desira
ble
to
giv
e
an
ad
diti
on
a
l
ben
e
fit,
su
c
h
as
higher
disc
ount
rate,
to
E
V
s
ha
ving
a
la
r
ge
r
li
st,
as
they
can
e
nh
a
nce
t
he
po
ssibil
it
y
of
suc
cessf
ul
m
a
tc
hin
g.
Act
ually
,
the
pr
e
-
rese
r
vation
nee
ds
t
o
offe
r
an
ea
rly
b
ir
d ra
te
, ch
ea
per tha
n
the
r
e
gula
r re
serv
at
io
n o
r w
al
k
-
in
ch
a
rg
i
ng
.
(a)
Ma
tc
hed ch
arg
e
rs
(b)
Gaine
d pro
fit
Figure
8
.
Effec
t of the
pr
e
fere
nce len
gth
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.
9
, N
o.
6
,
Dece
m
ber
201
9
:
5586
-
5595
5594
Finall
y,
Fig
ur
e
9
s
how
s
the
e
ff
ect
of
the
pr
i
ce
sensiti
vity
.
If
it
is
0,
al
l
E
Vs,
ha
ving
no
con
ce
r
n
on
the
pri
ce,
bid
f
or
the
highest
l
evel.
On
the
c
on
t
rar
y,
i
f
it
is
1,
al
l
E
Vs
bi
d
lowest.
I
n
t
hes
e
two
e
xtrem
e
cases,
there
is
al
m
os
t
no
possi
bili
ty
for
th
e
pro
po
sed
sc
hem
e
to
i
m
pr
ove
the
r
evenue.
The
im
pr
ov
em
ent
can
be
ob
ta
ine
d
only
by
the
inc
rease
d
num
ber
of
m
at
ched
pairs
.
F
or
p
se
ns
it
ivit
y,
the
tw
o
highe
st
biddin
g
le
ve
ls
are
sel
ect
ed
with
pro
ba
bili
ty
of
(1
-
p
).
Bot
h
ha
ve
the
sam
e
chan
ce
to
be
sel
ect
ed
with
i
n
this
r
ang
e
.
I
n
Fig
ure
9(
a
),
the
SMP
-
base
d
and
FCFS
sc
hem
es
sh
ow
the
alm
os
t
sa
m
e
nu
m
ber
of
m
at
c
hed
pairs
,
outp
erfor
m
ing
the
walk
-
in
sch
em
e
by
about
23
%
over
t
he
whole
range
.
Ne
xt,
as
sho
wn
in
F
igure
9(b),
the
SMP
-
base
d
s
chem
e
i
m
pr
oves
the
r
evenue
by
up
t
o
4.7
%
c
om
par
ed
with
t
he
FCFS
sc
hem
e,
wh
e
n
t
he
se
nsi
ti
vity
is
0.
5.
T
hat
is,
EVs
bi
d
unif
orm
ly
ov
er
t
he
a
vaila
ble
biddin
g
ra
ng
e
.
I
n
ad
di
ti
on
,
at
bo
t
h
ti
ps
,
the
SMP
-
ba
sed
sc
hem
e
m
at
ches
m
or
e p
ai
rs, sh
owin
g
a
bette
r r
evenue
f
or
t
he whole
e
xp
e
rim
ent r
a
nge.
(a)
Ma
tc
he
d
c
harger
s
(
b)
Gai
ne
d prof
it
Figure
9
.
Effec
t of the
pr
ic
e
s
ensiti
vity
5.
CONCL
UDI
NG RE
MAR
KS
In
this
pap
e
r,
we
ha
ve
an
al
yz
ed
the
pr
ic
e
e
ff
ect
to
the
c
ha
rg
i
ng
dem
and
ta
kin
g
a
dvant
age
of
data
arch
i
ves
obta
in
ed
from
a
cha
r
ging
netw
ork
c
urren
tl
y
in
op
e
rati
on.
T
he
occ
up
a
ncy
rates
f
or
3
pr
ic
e
le
ve
ls
are
even
ly
se
par
at
ed
by
9
%,
s
howing
t
he
li
near
pri
ce
se
ns
it
ivit
y.
H
our
ly
trace
revea
ls
the
necessi
ty
of
an
ef
fici
ent
res
erv
at
io
n
m
echan
ism
fo
r
hot
charger
s
duri
ng
peak
l
oad
per
i
od.
The
SMP
-
base
d
pre
-
reservat
i
on
schem
e
giv
es
pr
ece
de
nce
to
tho
se
E
Vs
biddin
g
hi
gh
e
r,
ha
ving
a
s
horter
prefe
ren
ce
li
st,
an
d
ar
rivi
ng
fir
st
,
seq
uen
ti
al
ly
.
Accor
ding
t
o
the
pe
rfor
m
ance
m
easur
e
m
ent
res
ult
usi
ng
the
prac
ti
cal
dem
and
m
od
el
,
the
pro
po
s
ed
pr
e
-
rese
r
vation
m
echan
ism
i
m
pr
ov
es
the
r
evenue
of
ser
vice
pro
vid
e
rs
by
up
to
9.5
%
a
nd
42.9
%
,
c
om
par
ed
with
t
he
le
gacy
FCF
S
and
reserv
at
io
n
-
le
ss
walk
-
i
n
sc
hem
es
for
t
he
give
n
perform
ance
par
am
et
er
set
s.
E
V
pen
et
rati
on
is
no
t
m
atu
re
ye
t
an
d
c
ha
rg
i
ng
in
fr
ast
r
uctu
res
a
re
sti
ll
under
c
on
st
r
uction
in
m
any
co
un
t
ries.
I
ntell
igent
inf
orm
a
ti
on
te
ch
no
l
ogie
s,
es
pecial
ly
soph
ist
ic
at
ed
al
gorithm
s
an
d
arti
fici
al
intel
lig
ence
te
ch
niqu
es,
m
ake
it
possible
to
overc
om
e
the
curre
nt
ins
uffici
enc
y
and
inc
onve
nienc
e
in E
V dr
ivi
ng.
As
f
uture
w
ork,
we
will
re
f
ine
our
dem
and
m
od
el
with
the
acc
um
ulatio
n
of
c
hargin
g
operati
on
arch
i
ves
a
nd
i
nvest
igate
t
he
e
f
fect
o
f
ne
w
fac
il
ity
instal
la
ti
on
a
nd
E
V
de
pl
oym
ent.
This
a
naly
sis
will
help
us
to
de
sig
n
dive
rse
E
V
a
ppli
cat
ion
s
s
uc
h
as
new
ch
ar
ger
sit
e
sel
ec
ti
on
[13],
V
2G
(V
e
hicle
-
to
-
G
rid
)
coor
din
at
io
n
[
14
]
,
a
nd
the
li
ke.
M
or
e
over,
ren
e
wa
ble
ene
rg
y,
su
c
h
as
w
ind
a
nd
s
un
li
gh
t,
has
been
dr
awin
g
m
uch
at
te
ntion
in
t
he
ta
rg
et
a
rea.
O
ur
re
sear
ch
te
am
is
pla
nn
i
ng
to
dev
el
op
a
ci
ty
wide
chargin
g
co
ord
inati
on
integrati
ng
t
he
ren
e
wa
ble
ene
rg
y
gen
e
rati
on.
Her
e
,
a
var
ie
t
y
of
dataset
s
a
re
colle
ct
ed
re
gardin
g
s
olar
e
nergy
gen
e
rati
on a
nd cl
i
m
at
e h
ist
ory
r
eco
rds [1
5].
ACKN
OWLE
DGE
MENTS
This
r
esearc
h
was
s
upporte
d
by
the
20
19
sci
entifi
c
pro
m
ot
ion
pro
gr
a
m
fu
nd
e
d
by
Jej
u
N
at
ion
al
Un
i
ver
sit
y
, So
uth
K
or
ea
.
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
Pric
e eff
ect
a
naly
sis a
nd
pr
e
-
reseravti
on
sc
he
me on el
ect
ric
vehicl
e c
ha
r
gi
ng n
et
w
or
ks (J
unghoo
n
Lee)
5595
REFERE
NCE
S
[1]
S.
Ramchrun,
et
al
.
,
“
Putti
ng
th
e
‘Sm
art
s’
in
to
the
Sm
art
Grid:
A
Grand
Cha
llenge
for
Art
ifi
c
i
al
In
telli
g
enc
e
,
”
Comm
unic
ati
on
of
th
e
ACM
,
vo
l. 55, pp. 89
-
97,
2
012.
[2]
X.
He
,
et
al
.
,
“
A
big
da
ta
ar
chite
ct
ure
design
for
sm
art
grids
b
ase
d
on
ran
dom
m
a
tri
x
the
or
y
,
”
I
EEE
Tr
ansacti
on
s
on
Smar
t
Gr
id
,
v
ol.
8
,
pp
.
674
-
68
4,
2017
.
[3]
J.
Mukherjee
an
d
J.
A.
Gupta
,
“
A
rev
ie
w
of
charge
sche
du
li
ng
o
f
el
e
ct
r
ic
v
ehi
c
l
es
in
sm
art
grid
,
”
IEEE
Syst
ems
Journal
,
vo
l. 8,
pp.
1541
-
1553
,
2014.
[4]
S.
Orcio
ni
,
e
t
a
l
.
,
“
El
e
ct
ri
c
veh
ic
l
es
cha
rg
in
g
r
ese
rva
t
ion
bas
ed
on
OCP
P,”
IE
EE
Int
ernati
ona
l
Confe
ren
ce
on
Env
ironment
an
d
Elec
tri
cal E
ng
ine
ering
,
2018
.
[5]
Z.
C
ao,
et
al
.
,
“An
autonomo
us
el
e
ct
ric
ve
h
i
cl
e
bas
ed
char
ging
system:
M
atc
hing
and
ch
arging
strategy,
”
IEE
E
Pow
er
&
Ene
rg
y
Soc
ie
t
y
I
nnovat
iv
e
Sm
art Grid
T
ec
hno
log
ie
s Confe
r
ence,
2018.
[6]
I.
B
a
y
ram,
e
t
al
.
,
“
A
Surve
y
on
Ene
rgy
Tr
ading
in
Smar
t
Gr
id
,”
IEEE
G
loba
l
Confer
ence
on
Signa
l
an
d
Inform
at
ion
Pro
ce
ss
ing
,
pp.
258
-
262
,
2014
.
[7]
J.
Lee
and
G.
Park,
“
Te
m
pora
l
data
str
ea
m
a
naly
s
is
for
EV
cha
rging
infr
astruc
ture
in
Je
ju,”
ACM
Re
searc
h
in
App
li
ed
Computati
on
S
ymposi
um
,
pp.
36
-
39
,
2
017.
[8]
J.
L
ee
and
G.
Park,
“C
harger
reservat
ion
s
c
heme
for
el
e
ct
r
ic
ve
h
icles
bas
ed
on
stabl
e
mar
riage
proble
m
adapta
ti
on
,
”
Int
ern
ational Conference on
Inform
at
ion
Technol
og
y
:
IoT
and
Sm
art
Cit
y
,
2018
.
[9]
J.
L
ee
and
G.
Park,
“Price
ef
f
ec
t
anal
ysis
on
e
le
c
tric
v
ehicle
charging
ne
tw
orks
in
Jeju
cit
y,
”
Inte
rn
at
ion
a
l
Confer
ence
on
C
ontrol
,
Mec
h
at
ro
nic
s a
nd
Autom
a
ti
on
,
pp.
175
-
17
9
,
2018
.
[10]
C.
Brunsdon
and
L. Comber,
An
I
ntroduct
ion
to R for Spati
al
Ana
l
ysis
&
Mapping
,
SA
GE
Public
a
tion L
td
.
,
2015.
[11]
J.
Q
.
Torr
os,
et
al
.
,
“
A
statistica
l
anal
y
s
is
of
EV
cha
rging
b
eha
v
i
or
in
the
UK
,
”
I
EE
E
P
ES
Innov
ati
v
e
Smar
t
Gr
i
d
Technol
ogi
es
L
ati
n
Ame
rica
,
pp
.
445
-
449
,
2015
.
[12]
A.
Ghos
h
and
Y.
Aggrawal
,
“
Menu
-
base
d
pr
i
ci
ng
for
ch
arg
in
g
of
el
e
ct
ri
c
v
eh
ic
l
es
with
vehicl
e
-
to
-
grid
servi
ce,”
IEE
E
Tr
ansacti
o
ns on
Ve
h
ic
u
lar
Technol
ogy
,
vo
l. 67,
pp.
10268
-
1
0280,
2018
.
[13]
N.
Shahra
ki
,
e
t
al
.
,
“
Optimal
lo
ca
t
io
n
of
e
lectr
i
c
public
ch
arg
in
g
stat
ions
using
rea
l
world
vehic
le
tr
avel
patter
ns
,
”
Tr
anspo
rtati
on
Re
search Part D
,
vol
.
41
,
pp
.
165
-
176,
2015
.
[14]
G.
Buj
a,
e
t
al
.
,
“
Rea
ct
iv
e
pow
er
compensat
ion
c
apa
bi
li
ties
of
V2
G
-
ena
ble
d
e
lectr
ic
v
ehicle
s
,
”
IE
EE
Tr
ansacti
ons
on
Powe
r
Elec
tr
onic
s
,
vo
l. 32, p
p.
9447
-
9459
,
2
017.
[15]
B.
B
anht
hasi
t,
e
t
al
.
,
“
Optimal
g
ene
ra
ti
on
sche
du
li
ng
of
power
s
ystem
for
m
axi
m
um
ren
ewa
bl
e
e
ner
g
y
h
arv
esti
ng
and
power
losse
s
m
ini
m
iz
at
ion
,
”
Inte
rnat
ional
Jo
urnal
of
E
le
c
trical
and
Computer
Engi
ne
ering
(
IJ
ECE
)
,
vol
.
8
(4)
,
pp.
1954
-
1966
,
2018.
Evaluation Warning : The document was created with Spire.PDF for Python.