Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
1
,
Jan
uar
y
201
9
,
pp.
7
2
~
7
6
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
1
.pp
7
2
-
76
72
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Universi
ty cours
e tim
etabli
ng mod
el using
ant c
olony
optimiz
ation al
go
rithm ap
proach
Mun
ir
ah
M
az
lan, M
ok
h
airi
Makhtar
, Ah
mad
Fir
daus
Kha
ir
Ahm
ad
Kh
airi,
Moham
ad A
fe
ndee
Moham
e
d
Facul
t
y
of
Infor
m
at
ic
s a
nd
Com
puti
ng,
Univer
sit
i
Sulta
n
Z
ai
n
al
Abidin,
Te
r
engg
anu,
M
a
lay
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
O
ct
1,
2018
Re
vised
N
ov
2
1
, 2
018
Accepte
d
D
ec
2
, 2
018
Due
to
the
in
cr
ea
sed
num
ber
o
f
student
s
and
r
egul
a
ti
ons,
a
ll
e
duca
t
iona
l
insti
tutions
have re
newe
d
their
in
te
rest to appe
ar
i
n
the
num
ber
o
f com
ple
x
i
t
y
and
fle
x
ibi
l
ity
si
nce
th
e
resour
ces
and
eve
n
ts
are
bec
om
ing
m
ore
diffi
cu
lt
to
be
sche
dule
d
.
Tim
et
abl
ing
is
the
t
y
pe
of
problem
s
where
the
eve
nts
nee
d
to
be
orga
nized
into
a
num
ber
of
ti
m
eslot
s
to
pre
vent
the
con
flicts
in
using
a
give
n
set
of
res
ourc
es.
Thus
in
the
in
te
rv
eni
ng
dec
ad
es,
signif
icant
progre
ss
has
bee
n
m
ade
in
the
cour
se
tim
et
abl
ing
probl
em
m
onit
oring
with
m
et
a
-
heur
isti
c
adj
ust
m
ent
.
In
thi
s
stud
y
,
ant
co
lon
y
o
pti
m
iz
ation
(ACO
)
al
gorit
hm
appr
oac
h
has
b
ee
n
develope
d
for
unive
rsit
y
c
ourse
ti
m
et
ablin
g
proble
m
.
ACO
is
bel
ie
ved
to
be
a
powerful
soluti
on
appr
oa
ch
for
var
ious
combinat
ori
al
o
pti
m
iz
ation
prob
le
m
s.
Thi
s
app
ro
ac
h
is
used
acco
rding
to
t
he
dat
a
set insta
nc
e
s tha
t
have
b
ee
n
col
l
ec
t
ed.
Its
per
form
anc
e
is pr
es
ent
ed
using
the
appr
opri
at
e
al
gorit
hm
.
Th
e
result
s
are
arg
u
ably
wi
thi
n
the
best
result
s
ran
ge
from
th
e
literature
.
Th
e
pe
r
form
anc
e
assessment
and
r
esult
s
are
used
to
det
ermine
whet
her
t
h
e
y
a
re
r
el
i
abl
e
in
pr
ep
ari
ng
a
qua
li
f
ying
cour
s
e
ti
m
et
abling
pro
c
ess.
Ke
yw
or
d
s
:
An
t C
olony
O
pti
m
iz
at
ion
Course Tim
et
a
bling
Me
ta
-
heurist
ic
s
Op
ti
m
iz
ation
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
:
Mun
ira
h
Ma
zl
an
,
Facult
y o
f
I
nfo
rm
atics and
C
om
pu
ti
ng
,
Un
i
ver
sit
i S
ultan Zai
nal Abid
in,
Tereng
ganu,
Ma
la
ysi
a
.
Em
a
il
:
m
un
i
m
azl
an@gm
ail.co
m
1.
INTROD
U
CTION
Ti
m
et
abling
in
edu
cat
io
nal
in
sti
tuti
on
s
is
an
i
m
po
rtant
act
ivit
y
to
sche
du
l
e
courses
or
e
xam
inati
on
,
wh
ic
h
m
us
t
be
entirel
y
assi
gn
e
d
i
nto
a
pp
ropr
ia
te
ti
m
eslo
ts
f
or
stu
de
nt
s,
le
ct
ur
e
rs
a
nd
r
oo
m
s
subje
ct
to
const
raints.
C
ourse
ti
m
et
abling
is
a
com
m
on
prob
le
m
fo
r
e
ver
y
aca
dem
ic
com
m
un
it
y
an
d
is
oft
en
at
te
m
pted
by
researc
hers
[1]
.
Alth
ough
var
io
us
sci
en
ti
fic
and
com
m
ercial
wo
rks
wer
e
m
ade
i
n
this
area,
in
m
an
y
edu
cat
io
nal i
nst
it
ution
s,
ti
m
etab
le
is sti
ll
sche
du
le
d
m
anu
al
ly
an
d eve
n
if
the c
om
pu
te
r
is
fr
e
qu
e
ntl
y use
d,
it
is
on
ly
necessa
ry
to
pr
ese
nt
dat
a
or
chec
king
const
raints
vali
dation
[2]
.
It
is
extrem
ely
diff
ic
ult
to
so
lve
a
la
rg
e
course
ti
m
et
ab
li
ng
prob
le
m
s
with
m
anu
al
a
ppr
o
ach
w
hich
m
a
y
need
a
gro
up
of
aca
de
m
ic
sta
ff
to
w
ork
for
sever
al
days
[3]
.
Th
us
,
un
i
ver
si
ty
cou
rse
ti
m
e
ta
bling
prob
le
m
is
con
sidere
d
as
no
n
-
deter
m
inist
ic
po
ly
no
m
ial
(N
P
)
hard
pro
blem
,
w
hich
m
eans
the
c
om
plexity
of
c
om
pu
ta
ti
on
al
am
ou
n
t
on
ti
m
e
re
qu
i
red
inc
reases
expo
nen
ti
al
ly
with
prob
le
m
siz
e
[4]
.
T
he
un
i
ver
sit
y
tim
et
abling
pro
ble
m
s
are
act
ual
ly
diff
ere
nt
f
r
om
each
oth
e
r
dep
e
ndin
g
on
th
e
ty
pe
of
ed
ucati
onal
insti
tuti
on,
t
he
sche
dule
d
e
nt
it
ie
s,
and
t
he
const
raints
i
nvolv
e
d.
The
m
ai
n
idea
of
this pro
blem
is
to
assign
a
s
et
of
eve
nts
int
o
a
lim
it
ed
num
ber
of
tim
eslo
ts
sub
j
ect
sat
isfyi
ng
a
nu
m
ber
of
c
on
st
raints.
T
he
set
of
con
st
raints
are
us
ually
div
ide
d
into
two
par
ti
cular
t
ypes
wh
ic
h
ar
e
hard
and
s
oft
const
r
ai
nts
[6]
.
T
herefo
re,
the
pro
bl
e
m
ob
j
ect
i
ve
i
s
to
sat
isfy
the
hard
co
ns
trai
nt
s
and
m
ini
m
ize
the
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Un
iv
ersit
y c
ourse
ti
meta
bling m
od
el
us
i
ng ant col
on
y
opti
miza
ti
on
algori
thm a
pp
r
oac
h
(
Mun
ir
ah Mazl
an
)
73
vio
la
ti
on
of
t
he
soft
co
ns
trai
nts.
It
is
the
refor
e
neces
sary
t
o
us
e
e
ff
ic
ie
nt
appr
oach
to
pr
oduce
ti
m
et
abl
e
that
sat
isfie
s the c
onstrai
nts
.
Ther
e
ha
ve
be
en
nu
m
ero
us
a
m
ou
nts
of
re
search
on
s
olvi
ng
c
our
se
an
d
e
xam
inati
on
tim
et
abling
pro
blem
s
[7
-
10]
.
Nowa
days
with
bette
r
co
m
pu
ti
ng
te
chnolo
gy,
var
i
ous
m
et
ho
ds
are
pro
po
se
d
to
pr
oduce
a
bette
r
so
luti
on
of
the
tim
et
abling
proces
s
[11]
.
This
pa
per
con
ce
ntrates
on
th
e
co
ur
se
ti
m
et
abling
pro
blem
s.
Me
ta
heu
risti
c
or
a
ppr
oxim
a
t
ion
optim
iz
at
i
on
al
gorithm
s
are
beco
m
ing
increasi
ng
ly
eff
ect
ive
f
or
s
olv
in
g
course
tim
et
abling
pro
blem
su
c
h
as
Gen
et
ic
Algo
rithm
(GA)
[12
,
13]
,
Tabu
Searc
h
[14
,
15]
,
Sim
ulate
d
a
nn
eal
in
g
[16]
and
An
t
Colo
ny
Op
ti
m
iz
a
tio
n
(
AC
O)
[
17]
.
These
al
gorithm
s
can
pr
od
uce
high
-
qual
it
y
so
luti
ons
but
do
no
t
gua
ran
te
e
opti
m
a
li
t
y
[6]
.
In
re
cent
ye
a
rs,
AC
O
has
be
en
e
xtensi
vel
y
stud
ie
d
for
ta
ckling
this
pro
blem
.
It
has
been
ta
ken
up
by
s
om
e
sci
entist
s
and
m
at
he
m
at
i
ci
ans
an
d
ha
s
been
e
xploit
ed
an
d
wides
pr
ea
d
i
n
the
early
90'
s
[18
,
19]
.
AC
O
is
insp
ire
d
by
ant
sti
gm
erg
ic
be
hav
i
or
w
hich
sim
ulate
s
a
set
of
agen
ts
that
w
ork
to
gethe
r
to
fin
d
so
l
ution
s
t
o
opti
m
i
zat
ion
pro
ble
m
s
b
y
rely
ing
on
unc
om
plica
te
d
com
m
un
ic
at
ion
.
In
li
gh
t
of
the
above,
this
pa
per
disc
us
ses
t
he
us
e
d
of
a
nt
colo
ny
op
ti
m
i
zat
ion
in
deali
ng
with
the
course
tim
et
abling
pro
blem
base
d
on
a
ppr
opriat
e
al
gorit
hm
s
app
li
ed.
This
pa
per
is
orga
nized
a
s
f
ollows:
Sect
ion
2
will
su
m
m
arize
the
prob
le
m
s
def
init
ion
of
co
urs
e
tim
et
abling
and
Sect
io
n
3
el
aborates
the
ACO
appr
oach
a
nd
al
gorithm
us
ed.
Sect
io
n
4
pr
ese
nts
the
r
esults,
w
hile
Sect
ion
5
will
highli
gh
t
the
ov
e
rall
con
cl
us
io
n of t
he pape
r.
2.
COUR
SE TI
METABL
IN
G P
ROBLE
M
D
ES
CRIPT
I
ON
The
un
i
ver
sit
y course
ti
m
et
ab
li
ng
p
r
ob
le
m
is
an
op
ti
m
iz
ation
p
r
ob
le
m
wh
e
re
a
set
o
f
e
ve
nt
s
needs
to
be
sche
du
le
d
in
tim
esl
ots
fo
r
stud
ents
,
le
ct
ur
e
rs
an
d
locat
ed
in
the
ap
propriat
e
room
wh
il
e
m
a
intai
nin
g
the
const
raints.
T
he
pro
blem
pr
es
ents
asset
of
N
co
ur
se
s
to
be
sche
du
le
d
in
5
days
of
9
per
i
od
s
eac
h
wh
ic
h
ti
m
e
T
is
eq
ual
to
45
ti
m
es
lots,
a
s
et
of
R
room
s
wh
e
re
eac
h
r
oom
has
a
set
of
F
featu
res
a
nd
ca
pacit
y,
a
s
et
of
M
stud
e
nts
a
nd
a
set
of
feat
ur
es
require
d
by
the
co
ur
ses
ass
oci
at
ed
[
20]
.
T
he
pro
blem
ob
j
ect
ive
is
to
sat
isf
y
the
hard c
onstrai
nt
s and t
o
m
ini
m
iz
e the v
i
olati
on
of the s
oft
constraints
.
The
ge
ner
al
c
on
st
raints
f
or
course
tim
et
abling
pro
blem
can
be
cl
assifi
ed
into
tw
o
ty
pes
wh
ic
h
a
re
hard
co
ns
trai
nt
s
and
s
of
t
co
ns
trai
nts
[6]
[21]
.
A
pr
act
ic
al
and
fe
asi
ble
ti
m
et
able
m
us
t
sat
isfy
al
l
the
ha
r
d
const
raints
wit
h
no
co
ns
ide
ra
ti
on
w
hile
the
so
ft
co
ns
trai
nt
s
are
no
t
ab
s
olu
te
ly
essenti
al
bu
t
the
am
o
un
t
of
relat
ed violat
io
n
s
houl
d be m
i
nim
iz
ed
to m
axim
iz
e the p
e
rfec
ti
on
th
e ti
m
e
ta
ble.
The har
d
c
onst
raints c
on
sider
ed
in
this
pro
ble
m
are:
H1:
N
o
stu
de
nt ca
n be assi
gn
e
d
t
o
m
or
e tha
n o
ne
cours
e
at the
sam
e t
i
m
e.
H2:
The
roo
m
sho
uld
sat
isfy t
he feat
ur
e
s r
e
quir
ed by t
he
c
ours
e.
H3:
The
num
ber
of stude
nts att
en
ding the
cou
rse
shou
l
d be less
than o
r
e
qu
al
t
o
the
capa
ci
ty
o
f
the
room
.
H4:
N
o
m
or
e tha
n on
e
cou
rse
is a
ll
ow
ed
at a ti
m
esl
ot in
eac
h r
oom
.
Be
sides
sat
isfyi
ng
the
hard
con
strai
nts,
the
vio
la
ti
on
of
so
ft
co
ns
trai
nts
can
be
co
ns
ide
red
a
s
pr
e
fer
e
nces
t
ha
t
will
fu
lfil
l
the
use
r
requir
e
m
ents
to
i
m
pr
ove
the
qual
it
y
of
ti
m
e
ta
ble.
So
ft
c
onstrai
nt
s
are
of
te
n
c
onfro
nted wh
ic
h
incl
ude the
fo
ll
owin
g:
A
stu
de
nt
has
t
o
at
te
nd
on
ly
one c
ourse
i
n
a
day.
A
stu
de
nt
has
t
o
at
te
nd m
or
e t
han tw
o
c
ourse
s consec
utively
.
A
stu
de
nt
has
t
o
at
te
nd a
co
urse in t
he
la
st
pe
rio
d
in
an
y
da
y.
3.
AN
T
COL
ONY O
PTIMIZ
A
TION
ALG
O
RITH
M
An
t
c
olony
op
t
i
m
iz
ation
(
ACO)
is
a
te
ch
nique
that
m
i
m
ic
s
the
nu
m
ber
of
arti
fici
al
ants
m
ov
ing
on
a
gr
a
ph
that
enc
od
e
t
he
pro
ble
m
it
se
lf
propo
sed
by
D
ori
go
[
22
,
23]
.
T
he
or
i
gin
al
m
e
m
b
er
of
ACO
Algorith
m
can
be
s
pecifi
cal
ly
cl
assifi
ed
as
A
nt
Col
ony
(A
S
).
Af
te
r
few
ye
a
rs,
AC
O
is
e
nh
a
nce
d
with
tw
o
s
uc
cessf
ul
var
ia
nts
w
hich
are
An
t
C
olony
Syst
em
(A
CS)
a
nd
Ma
x
-
Mi
x
A
nt
Syst
e
m
(MM
AS
)
with
bette
r
se
archi
ng
perform
ance.
Nowa
days,
the
se
appro
ac
hes
hav
e
bee
n
wi
dely
app
li
ed
i
n
so
lvi
ng
the
discrete
opti
m
iz
at
ion
pro
blem
and
ot
her
c
om
bin
at
ori
al
pro
blem
s
f
or
e
xam
ple
travell
ing
sal
esm
en
pr
ob
le
m
(TSP
),
grap
h
c
ol
or
i
ng,
sche
d
ulin
g pro
blem
s an
d veh
i
cl
e routin
g pro
blem
s
[24
,
25]
.
ACO
al
gorith
m
is
insp
ire
d
from
the
fora
ging
be
hav
i
or
of
real
ant
c
olonies.
I
n
A
CO
al
gorit
h
m
arti
fici
al
ants
su
ccess
fu
ll
y
con
st
ru
ct
s
olu
ti
on
base
d
on
the
global
inf
or
m
at
ion
(phe
ro
m
on
es
)
an
d
local
inf
or
m
at
ion
.
T
he
pher
om
on
e
the
n
act
s
a
s
a
pro
ba
bili
sti
c
m
od
el
f
or
th
e
co
ns
tr
uctio
n
of
s
olu
ti
ons
and
is
const
antly
a
m
plifie
d
by
a
nts
bu
il
t
with
hi
gh
qu
al
it
y
sol
ution
s
.
H
e
nce
,
pher
om
on
e
evapo
rati
on
s
urpasse
s
pr
em
at
ur
e c
onverge
nce to a
poor l
ocal opti
m
um
.
In
ACO
,
the
prob
le
m
is
act
ual
ly
dealt
with
by
si
m
ulati
ng
som
e
arti
fici
al
ants
that
m
ov
e
on
the
gr
a
ph
that
issues
t
he
pro
blem
it
sel
f.
The
tra
velin
g
sal
es
m
an
pro
ble
m
(TSP
)
play
s
a
m
ajo
r
r
ole
in
AC
O
as
it
is
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
1
,
Ja
nu
a
ry
201
9
:
7
2
–
7
6
74
first
pro
blem
t
o
be
at
ta
cke
d
by
ACO
.
Th
us,
TSP
has
be
en
im
ple
m
ented
us
i
ng
A
nt
Syst
e
m
a
lgo
rith
m
s
as
sh
ow
n
in
Fi
gur
e 1
.
T
he
fi
gure
exp
la
in
s the
b
a
sic
p
ri
nciples
of the
A
C
O
al
go
rithm
s.
3.1
.
The
St
r
ateg
y of A
nt’s
Movem
e
nt
s
The
basic
m
ater
ia
l
of
AC
O
i
s
the
us
e
of
a
pro
bab
il
ist
ic
so
luti
on
c
onstruc
ti
on
m
echan
is
m
based
on
the
sti
gm
erg
y.
All
l
ants
are
i
niti
al
ly
placed
rand
om
l
y
at
node
i.
Eac
h
a
nt
bu
il
ds
a
com
plete
tim
e
ta
ble
in
eac
h
it
erati
on
al
gori
thm
s
in
wh
ic
h
al
l
the
har
d
const
raints
are
sat
isfie
d
at
th
e
hig
hest
le
ve
l
in
ever
y
resu
lt
in
g
so
luti
on.
The
a
nt
k
travels
f
rom
no
de
to
no
de
sta
rting
f
ro
m
the
node
i.
Th
e
transiti
on
prob
a
bili
ty
P
ij
k
that
the
ant
k,
c
urren
tl
y
at
node
i
will
choose
an
d
m
ov
e
to
the
ne
xt
node
j
is
cal
c
ul
at
ed
usi
ng
the
rand
om
pr
opor
ti
on
al
ru
le
giv
e
n by,
P
ij
k
=
[
τ
ij
]
α
[
η
ij
]
β
∑
[
τ
il
]
α
[
η
il
]
β
l
∈
N
i
k
,
j
∈
N
i
k
(1)
W
he
re
τ
ij
is
th
e
pher
om
on
e
tr
ai
l
value
on
t
he
e
dge
c
onne
ct
ing
t
o
node
i
to
node
j
m
eanwhil
e
η
ij
is
the
heurist
ic
val
ue
of
t
hat
e
dg
e
a
nd
η
ij
=
1
/
d
ij
,
α
an
d
β
are
t
he
par
am
et
ers
that
dete
rm
ine
the
relat
ive
in
f
luence
of
pher
om
on
e
trai
l
and
the
he
ur
ist
ic
in
form
a
ti
on
a
nd
N
i
k
is
a
s
et
of
al
l
no
des
that
rem
ai
ns
to
be
visit
ed
w
he
n
the an
t
k
is
at n
od
e
i.
Af
te
r
it
erati
ng the
proc
ess, eac
h
a
nt c
om
plete
s the tour
.
3.2
.
P
herom
one U
pd
at
e
The
pher
om
one
trai
l
value
is
update
d
at
t
he
en
d
of
each
it
erati
on
al
gorit
hm
.
The
a
nts
with
s
horter
route
will
le
ave
m
or
e
pher
om
on
e
trai
l
tha
n
those
with
l
onge
r
r
ou
te
.
T
he
pher
om
on
e
update
r
ule
de
fines
the
way
in
wh
ic
h
good
s
olu
ti
ons
are
rei
nfor
c
ed
in
the
p
he
r
om
on
e
trai
l
by
a
dding
a
qua
ntit
y
∆
τ
ij
wh
ic
h
is
high
on
the
be
st
so
l
ution
arcs
.
T
her
e
f
or
e
,
the
am
ou
nt
of
pher
om
one
in
eac
h
route
will
be
ad
justed
by
the
pher
om
on
e
trai
l
decay
equat
ion
(
2).
T
he
trai
l
le
vels
are
updated
a
s
on
ever
y
r
oute
each
ant
le
ave
s
ph
e
r
om
on
e
qu
antit
y
giv
e
n
by
Q
L
k
⁄
,
w
here
Q
is
co
ns
ta
nt
an
d
L
k
is
the
le
ngth
of
it
s
best
tour
re
sp
ect
ive
ly
.
On
the
ot
he
r
hand,
th
e
evapo
rati
on
of
ph
er
om
on
e
tr
ai
l
is
app
li
ed
at
the
end
of
ea
ch
it
erati
on
al
gorithm
.
Thu
s,
t
he
ru
le
f
or
upda
ti
ng
trai
ls cou
l
d be
def
i
ned as
fo
ll
ow
s:
τ
ij
(
t
+
1
)
←
(
ρ
)
τ
ij
(
t
)
+
∆
τ
ij
(
t
)
(2)
∆
τ
ij
=
∑
∆
τ
ij
k
l
k
=
1
(3)
∆
τ
ij
k
(
t
)
=
{
Q
L
k
,
if
a
nt
k
tr
a
ve
ls
on
edg
e
(
i
,
j
)
0
ot
he
rwi
se
(4)
W
he
re t is t
he
i
te
rati
on
c
ounte
r,
p
is t
he ph
e
r
om
on
e eva
pora
ti
on
rate,
∆
τ
ij
is t
he
am
ou
nt of i
nc
rease in
the t
r
ail
l
evel on the e
dge (i,
j) and
∆
τ
ij
k
is t
he
inc
rease
d
le
vels of trail
at the ed
ge (i,j
)
ca
us
e
d
by the
ant
s k
r
e
sp
ect
ivel
y.
The
al
gorithm
n
eed
s to
be
a
ppli
ed
with t
he ph
e
r
om
on
e eva
porati
on pro
ce
ss to
av
oid
t
he un
li
m
i
te
d
pher
om
on
e
colle
ct
ion
a
nd
init
ia
l
conve
r
ge
nce
to
ward
a
sub
op
ti
m
al
s
olu
ti
on
reg
i
on.
The
ne
xt
it
er
at
ion
t
+
1
will
sta
rt
after
updatin
g
t
he ph
e
r
om
on
e
trai
l process.
F
igure
1
is t
he
basi
c p
ri
nciple
of the
A
C
O
al
go
rithm
.
ACO
Al
gori
thms
-
I
nitiali
ze
phero
mone
trail
s
-
Do whi
l
e
(
Stop
condition/if crit
eria
are
no
t
sat
isfi
ed)
-
loop
Gene
rate
solu
ti
o
n
Local
Searc
h
sol
uti
on
Update
pheromo
ne
trai
ls
-
End
Do
-
End
Figure
1.
The
basic P
rinciple
s A
C
O Alg
or
it
hm
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Un
iv
ersit
y c
ourse
ti
meta
bling m
od
el
us
i
ng ant col
on
y
opti
miza
ti
on
algori
thm a
pp
r
oac
h
(
Mun
ir
ah Mazl
an
)
75
4.
E
X
PERI
MEN
TAL RES
UL
T
The
ACO
a
pp
ro
ac
h
is
c
on
st
ru
ct
e
d
an
d
te
s
te
d
in
a
web
-
base
d
com
pu
t
er
syst
em
of
the
U
niSZ
A
Course
Tim
et
a
bling
Syst
e
m
su
pp
or
te
d
by
the
CPU
In
te
l
Core
i7
-
36
10QM
with
2.3
GH
z
a
nd
R
A
M
6G
B
unde
r
wi
ndow
7.
T
o
e
valuate
the
pro
posed
a
ppr
oach,
we
c
onduct
ed
se
veral
exp
e
rim
ents
.
A
pr
io
rity
is
give
n
on
seve
ral
te
sts
to
dete
rm
ine
the
al
locat
ion
of
c
ourse
ti
m
etab
li
ng
f
or
a
di
ff
e
ren
t
la
r
ge
num
ber
of
stu
de
nts
to
be
sc
he
du
le
d
i
nto
t
he presc
ri
bed tim
e.
The
ex
pe
rim
e
nt
te
st
has
be
en
car
ried
out
on
the
c
ours
e
tim
e
ta
bling
wh
e
re
ap
plied
us
in
g
ACO
appr
oach
t
o
com
pu
te
the
co
ur
se
ti
m
et
ablin
g
res
ults
with
the
te
st
cas
e
from
Faculty
of
I
nfo
rm
ati
cs
an
d
Com
pu
ti
ng
(FI
C)
dataset
s
as
sh
own
in
Ta
ble
1.
The
res
ult
of
the
te
st
case
in
Table
2
ind
ic
at
ed
that
the
FI
C_
A
1
w
hich
is
witho
ut
pr
i
or
it
y
h
as
pro
duced
0.2
9
of
st
and
a
r
d
de
viati
on.
O
n
the
ot
he
r
ha
nd,
the
re
su
lt
of
the
te
st
case
in
Ta
ble
3
i
ndic
at
ed
that
FIC
_A2
wh
ic
h
is
pr
i
or
it
y
giv
e
n
ha
s
pro
duce
d
0.3
8
of
sta
nda
r
d
dev
ia
ti
on.
T
he
resu
lt
s
ind
ic
a
te
d
that
the
cou
rse
ti
m
e
ta
bling
was
ef
fecti
ve
ly
sched
ule
d
and
ob
ta
ine
d
bette
r
resu
lt
rather t
ha
n
the
o
t
her te
s
t case
r
es
ults.
Table
1.
T
he
F
IC D
at
aset
s
Test Case
Su
b
jects
Enro
lled
Stud
en
ts
Ti
m
eslo
ts
Priority
(
P
)
/
No
Pr
io
rit
y
FIC_
A1
32
1650
10
P
FIC_
B1
32
1415
10
P
FIC_
A2
32
1650
10
NP
FIC_
B2
32
1415
10
NP
4.1
.
E
xp
eri
m
ent test
wi
tho
ut
th
e
priorit
y
The
T
able
2
sh
ows
the
A
CO
appr
oach
perform
ance
without
pr
i
or
it
y
on
tim
e
ta
bling
prob
le
m
instances.
Table
2.
ACO
Approac
h per
f
or
m
ance w
it
ho
ut prio
rity
Test Case
ACO Ap
p
roach
Den
sity
Co
n
f
lict (
%)
m
e
an
v
ar.
st.d
ev
(
)
FIC_
A1
1
.93
0
.74
0
.08
0
.29
FIC_
B1
2
.26
0
.71
0
.15
0
.39
4.2
.
E
xp
eri
m
ent test
wi
th
t
he priori
ty
The
T
a
ble
3
s
hows
the
A
C
O
a
ppr
oach pe
rform
ance w
it
h p
rior
it
y o
n
ti
m
et
a
bling p
r
ob
le
m
instances.
Table
3.ACO
Approac
h per
f
or
m
ance
with
pr
i
or
it
y
Test Case
ACO Ap
p
roach
Den
sity
Co
n
f
lict (
%)
m
e
an
v
ar.
st.d
ev
(
)
FIC_
A2
1
.93
0
.39
0
.14
0
.38
FIC_
B2
2
.26
0
.45
0
.20
0
.45
5.
CONCL
US
I
O
N
This
pa
pe
r
has
introd
uced
t
he
ant
colo
ny
optim
iz
at
ion
al
gorithm
in
so
lvin
g
the
un
i
ve
rsity
course
tim
e
ta
bling
prob
le
m
.
This
stud
y
has
res
ulte
d
in
a
f
easi
ble
appr
oach
for
c
ourse
ti
m
e
ta
ble
us
in
g
the
ge
ner
at
e
d
ACO
a
ppr
oach.
The
im
ple
m
e
nted
ACO
a
pp
ro
ac
h
gen
e
rat
e
d
com
par
a
ble
r
esults
li
ke
oth
e
r
hi
gh
perf
or
m
ance
resu
lt
s
pr
ese
nt
ed
in
the
li
te
r
at
ur
e.
The
pre
sented
e
xperi
m
ent
resu
lt
s
a
re
go
od
an
d
i
nd
ic
at
ed
a
r
el
ia
ble
appr
oach
f
or
unive
rsity
co
urs
e tim
et
abling
pro
blem
s.
W
e
bel
ie
ved
that
bet
te
r
res
ults can
be ob
ta
ine
d b
a
sed o
n
how
dataset
pr
ob
le
m
s
are
fixed
.
I
n
ad
diti
on,
the
al
gorithm
s
i
m
ple
m
ented
can
be
en
hanc
ed
de
pendin
g
on
the
adjustm
ent
res
ults.
F
uture
w
ork
will
aim
t
o
im
pr
ove
the
ACO
a
ppr
oac
h
an
d
a
pp
li
ed
the
pro
po
se
d
a
ppr
oach
on r
eal
-
w
or
l
d
c
ourses
ti
m
et
abl
ing
.
AC
KNOWL
E
GEMENT
This
wor
k
is
pa
rtia
ll
y supported
by UniSZ
A (G
ran
t R
R
008 C
RIM/
2016).
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
1
,
Ja
nu
a
ry
201
9
:
7
2
–
7
6
76
REFERE
NCE
S
[1]
H.
Babaei
,
J.
K
ari
m
pour,
and
A.
Hadidi,
“
A
surve
y
of
appr
oa
che
s
for
univ
ers
ity
cour
se
ti
m
etabli
ng
prob
le
m
,
”
Comput.
Ind
.
En
g.
,
vol. 86, pp. 4
3
–
59,
2015
.
[2]
W
.
Le
g
ie
rski,
“
Sy
stem for
solvin
g
ti
m
etabli
ng
pr
oble
m
s,”
Proc. S
ymp.
M
et
hods A
rtif
.
Intell
.
,
pp.
7
6
–
77,
2003
.
[3]
S.
a.
MirHass
an
i,
“
A
computat
i
onal
appr
o
ac
h
t
o
enha
nci
ng
cou
rse
ti
m
et
abling
with
int
eg
er
pro
gra
m
m
ing,
”
App
l.
Math.
Comput
.
,
vol.
175
,
pp
.
814
–
822,
2006
.
[4]
K.
Socha,
M.
S
ampels,
and
M.
Manfrin,
“
Ant
Algorit
hm
s
for
the
Univer
sit
y
C
ourse
Ti
m
et
ab
ling
Problem
with
Rega
rd to
the
St
at
e
-
of
-
the
-
Art
,
”
Appl
.
E
vol
.
Comput.
,
pp.
334
–
34
5,
2003
.
[5]
S.
Abdulla
h
and
H.
Tur
abi
eh
,
“
Gene
ra
ti
ng
un
ive
r
sit
y
cour
se
ti
m
e
t
abl
e
using
Gen
etic
Algor
it
hm
s
an
d
local
s
ea
rch
,
”
in
Proceedi
ngs
-
3rd
Inte
rnation
al
Confe
ren
ce
o
n
Conve
rgen
ce
and
Hybrid
Info
rm
ati
on
Technology,
ICCIT
200
8
,
2008,
vol
.
1
,
pp
.
254
–
260.
[6]
R.
Le
wis,
“
A
su
rve
y
of
m
et
ah
eu
risti
c
-
b
ase
d
te
ch
nique
s
for
Univer
sit
y
T
imeta
bl
in
g
proble
m
s,”
OR
Spec
tr.
,
vol.
3
0,
no.
1
,
pp
.
167
–
1
90,
2008
.
[7]
M.
W
.
C.
I
an
d
G.
La
porte
,
“
Rec
en
t
Deve
lop
m
ent
s
in
Prac
ti
ca
l
Course
Ti
m
et
ab
li
ng,
”
Pract.
Theory
Aut
om
.
Timetabl
ing
II
,
v
ol.
1408
,
pp
.
3
–
1
9,
1998
.
[8]
M.
W
.
Carter
a
nd
G.
La
po
rte,
“
Rec
ent
d
eve
lop
m
ent
s
in
pra
c
tic
al
ex
amination
t
imeta
bli
ng
,
”
in
Lect
ure
Not
es
i
n
Computer
Sci
e
nce
(
inc
ludi
ng
subs
erie
s
Lect
ure
Note
s
in
Arti
ficial
In
telli
gen
ce
and
L
ec
ture
No
te
s
i
n
Bi
oinf
orm
atics)
,
1996,
vol
.
1153
,
pp.
3
–
21
.
[9]
A.
Scha
erf
,
“
Surve
y
of automat
e
d
ti
m
etabli
ng
,
”
Arti
f. Intell. Re
v.
,
vol
.
13
,
no
.
2
,
p
p.
87
–
127
,
1999
.
[10]
E.
K.
Burke
and
S.
Petrovi
c,
“
Rec
en
t
rese
ar
ch
d
ire
c
ti
ons
in
aut
o
m
at
ed
ti
m
et
ab
ling,”
Eur.
J
.
Ope
r.
Re
s.
,
vol
.
140
,
no.
2
,
pp
.
266
–
2
80,
2002
.
[11]
S.
A.
Mirhassani
and
F.
H
abi
bi
,
“
Soluti
on
appr
o
aches t
o
th
e course
ti
m
et
ab
li
ng
p
ro
ble
m
,
”
Arti
f
.
In
t
el
l
.
R
ev.
,
vo
l. 39,
no.
2
,
pp
.
133
–
1
49,
2013
.
[12]
S.
Yang
and
S.
N.
Jat,
“
Gene
tic
Algorit
hm
s
W
it
h
Guided
an
d
Loc
a
l
Sear
ch
Strat
eg
ie
s
for
Univer
sit
y
Cour
se
Ti
m
et
ab
li
ng,
”
I
E
EE
Tr
an
s.
S
yst.
Man,
Cyb
ern.
P
art C
(
Appl
ic
a
tions
Re
v
.
,
vol
.
41
,
no
.
1
,
pp
.
93
–
1
06,
2011
.
[13]
S.
Gy
o
ri,
Z
.
Pet
res,
and
A.
Várkon
y
i
-
Kóc
z
y
,
“
Gene
tic
Algorithm
s
in
Ti
m
et
abl
ing.
A
New
Ap
proa
ch.,”
Budap
est
Univ.
Te
chnol. E
con.
,
2001.
[14]
E.
P.
Fod,
“
The
o
r
y
and
M
et
hodol
og
y
Ta
b
u
sea
r
ch
for
l
arg
e
sca
le t
i
m
et
abl
ing
problem
s,”
vol. 54, pp. 39
–
47,
1991.
[15]
A.
Hert
z, “Ta
bu
sea
rch
for
l
arg
e
sca
le t
imet
abling
proble
m
s,”
Eur.
J. Ope
r.
Re
s
.
,
v
ol.
54
,
no
.
1
,
pp
.
39
–
47,
1991
.
[16]
S.
Abdulla
h,
K
.
Shaker
,
B
.
Mc
col
lum,
and
P
.
Mcm
ull
an,
“
Dual
seque
n
c
e
sim
ula
t
ed
anneal
ing
with
round
-
rob
in
appr
oac
h
for
un
i
ver
sit
y
cour
se
tim
et
abl
ing
,
”
Eur.
Conf. Evol. Co
mput.
Comb.
Optim.
,
pp.
1
–
10,
2
010.
[17]
G.
Molnar
and
M.
Cupi,
“
Univ
ersity
Course
Tim
et
abl
ing
Us
ing
AG
O:
A
Case
Stud
y
on
La
bor
at
or
y
Ex
ercises,
”
Knowle
dge
-
Bas
e
d
Intell. Inf. Eng
.
Syst
.
Pt I
,
vol.
6276,
no
.
36
,
pp
.
100
–
110,
2010.
[18]
M.
Dorigo,
V.
Manie
z
zo,
A.
C
olorni
,
and
M.
D
origo,
“
Pos
it
ive
Feedba
ck
as
a
S
ea
rch
Str
ateg
y
,
”
Tech.
R
ep.
91
-
0
16
,
no.
June
,
pp
.
1
–
20,
1991
.
[19]
M.
Dorigo,
V
.
Manie
z
zo,
and
A.
Colorni,
“
Ant
s
y
st
em:
Optimization
b
y
a
colon
y
of
coop
era
t
i
ng
age
n
ts,”
I
EEE
Tr
ans.
Syst.
Man
,
Cyb
ern.
Part B
Cybe
rn.
,
vo
l. 26
,
no
.
1
,
pp
.
29
–
4
1,
1996
.
[20]
M.
Chia
ran
din
i,
M.
Bira
ttari
,
K.
Socha,
and
O.
R
oss
i
-
Doria
,
“
An
eff
ective
h
y
br
id
al
gorit
hm
for
unive
rsit
y
cour
s
e
ti
m
et
abl
ing,”
J.
Sche
d.
,
vo
l. 9, n
o.
5
,
pp
.
403
–
43
2,
2006
.
[21]
E.
K.
Burke
,
B.
McColl
um
,
A.
Meisel
s,
S.
Petr
ovic
,
and
R.
Qu,
“
A
gra
ph
-
base
d
h
y
per
-
h
eur
isti
c
for
educ
a
ti
on
a
l
ti
m
et
abling
prob
le
m
s,”
Eur.
J. Ope
r.
R
es.
,
vol
.
17
6,
no
.
1
,
pp
.
177
–
192,
2007
.
[22]
A.
Colorni
,
M.
Do
rigo,
and
V.
Manie
z
zo,
“
Distribut
ed
opti
m
iza
ti
on
b
y
an
t
col
o
nie
s,”
in
Proc
eedings
of
the
Fi
rs
t
European
Conf
e
renc
e
o
f Artificial
Life
,
19
91,
pp
.
134
–
142
.
[23]
A.
Colorni,
M.
Dorigo,
and
V
.
Manie
z
zo,
“
An
I
nvesti
gation
of
s
om
e
Properti
es
o
f
an`
`Ant
Algori
thm’’.,
”
Ppsn
,
n
o.
Pps
n
92,
pp.
2
–
7
,
1992
.
[24]
Z.
Chi
,
S.
Su,
and
M.
A.
Khi
ne,
“
An
Ant
Colon
y
Opt
imizat
ion
Algorit
hm
f
or
Solving
Travel
ing
Sa
le
sm
an
Problem,”
vol
.
1
6,
pp
.
54
–
59
,
20
11.
[25]
M.
Dorigo
and
C.
Blum
,
“
Ant
c
olon
y
opti
m
i
za
t
i
on
the
or
y
:
A
sur
ve
y
,
”
Theor.
Co
mput.
Sc
i.
,
vol
.
344,
no.
2
–
3,
pp
.
243
–
278,
2005
.
Evaluation Warning : The document was created with Spire.PDF for Python.