Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
13
,
No.
1
,
Jan
uar
y
201
9
,
pp.
191
~
198
IS
S
N: 25
02
-
4752, DO
I:
10
.11
591/ijeecs
.v1
3
.i
1
.pp
191
-
198
191
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
An ant c
olony
algorith
m
for uni
versiti
sul
tan zain
al abid
in
examina
tion time
tabling p
ro
bl
em
Ah
m
ad
Fir
daus Kh
air
,
M
okhairi
Makh
t
ar
,
Muni
rah
Ma
z
lan,
Moh
am
ad
Af
e
ndee
Moh
amed
,
Mohd
N
ordin
Ab
d
ul R
ah
m
an
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,
Ma
lay
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
20
, 201
8
Re
vised Oct
30
, 2
018
Accepte
d Nov
19
, 201
8
The
re
al
-
li
fe
co
nstruct
ion
of
ex
aminat
ion
ti
m
etabli
ng
prob
le
m
i
s
conside
red
as
a
comm
on
proble
m
that
al
wa
y
s
en
cou
nte
red
and
ex
per
ie
n
ce
d
in
educ
a
ti
ona
l
insti
tution
whether
in
school,
col
le
g
e,
and
unive
rsi
t
y
.
Thi
s
probl
em
is
usually
exp
eri
en
ce
d
b
y
th
e
a
ca
d
emic
m
ana
gemen
t
depa
rtment
where
they
h
ave
trouble
to
ha
ndle
complexi
t
y
for
assig
n
exa
m
ina
t
ion
into
a
suita
b
le
t
imeslot
m
anua
l
l
y
.
In
thi
s
pape
r
,
a
n
al
gori
thm
appr
oac
h
of
an
t
col
on
y
op
ti
m
isation
(ACO
)
is
pr
ese
nte
d
to
find
a
n
eff
ec
t
iv
e
soluti
on
for
dea
l
i
ng
with
Univer
siti
Sult
an
Za
inal
Abidi
n
(UniSZA)
exa
m
ina
t
ion
ti
m
et
ab
li
ng
probl
e
m
s.
A
combinat
ion
of
heur
ist
ic
with
ACO
al
gorit
hm
cont
r
i
bute
s
the
deve
l
opm
ent
soluti
on
in
orde
r
to
si
m
pli
f
y
an
d
opti
m
iz
e
the
ph
ero
m
one
occ
urr
enc
e
of
m
a
tri
x
updat
es
which
i
ncl
ude
the
constra
in
ts
proble
m
.
Th
e
imple
m
ent
at
ion
o
f
re
al
da
ta
se
t
inst
a
nce
s
from
ac
ad
emic
m
ana
g
ement
is
appl
i
ed
to
the
appr
oa
ch
for
gen
erati
ng
t
he
resul
t
of
exa
m
ina
t
ion
ti
m
et
ab
le
.
Th
e
resul
t
and
per
form
anc
e
that
obta
in
ed
will
be
used
for
furthe
r
use
t
o
evalua
t
e
th
e
q
ual
ity
and
obser
ve
th
e
solut
ion
whethe
r
our
exa
m
ina
t
ion
ti
m
et
ab
li
ng
s
y
s
te
m
is
reliab
l
e
and
eff
i
ci
en
t
th
an
the
m
anual
m
ana
gement that can
d
ea
l
th
e co
nstrai
nts pr
ob
lem
.
Ke
yw
or
ds:
An
t c
olony
op
t
i
m
isa
t
ion
Con
st
raints
Exam
inati
on
ti
m
et
abling
Heurist
ic
Ph
e
ro
m
on
e
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
:
Ah
m
ad
Fir
da
us K
hair
,
Faculty
of In
form
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
:
fird
aus
kh
ai
r@
ya
ho
o.
c
om
1.
INTROD
U
CTION
In
t
he
la
st
dec
ade,
se
ve
ral
of
researc
h
reg
a
r
ding
ex
am
inatio
n
ti
m
et
abling
has
been
co
nducte
d
ov
e
r
the
world
wide
by
pr
e
vious
researc
hers.
G
ener
al
ly
,
the
def
i
niti
on
of
tim
et
able
can
descr
i
be
as
schedu
l
e
plan
ning
on
pa
rtic
ular
e
ven
ts
that
ta
ke
place
in
certai
n
plac
es.
I
n
aca
dem
i
c,
besi
de
ti
m
e
t
able
f
or
e
xam
i
nation,
it
al
so
can
be
us
e
d
for
orga
ni
zi
ng
a
par
ti
cu
la
r
su
bject
or
course.
F
or
this
stud
y,
exam
i
nation
ti
m
et
abl
ing
i
s
descr
i
bing
a
li
st
that
sh
ow
s
the
tim
es
in
the
week
at
wh
ic
h
s
pecif
ic
exa
m
inati
on
is
held.
Ty
pical
ly
,
the
distrib
utio
n
of
exam
inati
on
ti
m
e
ta
bles
is
dep
e
nd
e
d
on
how
t
he
sta
ff
m
anag
em
e
nt
is
m
anag
in
g
th
e
inf
or
m
at
ion
an
d
set
of
data
i
n
e
ver
y
e
ducat
ion
al
in
sti
tuti
on
resp
ect
iv
el
y.
Re
gardin
g
t
he
tim
et
ables
pro
blem
,
the
su
r
vey
on
the
pr
evi
ou
s
researc
h
co
ncl
ud
e
d
as
non
-
de
te
rm
inist
ic
p
olyno
m
ia
l
-
time
har
d
(
NP
-
ha
rd)
[1
]
.
The
e
xam
inatio
n
tim
et
abling
pro
blem
is
kn
ow
n
as
(ET
P)
that
a
rises
arou
nd
the
world
w
hich
i
nvolv
e
s
edu
cat
io
nal
ins
ti
tuti
on
s
wh
et
he
r
sch
oo
ls,
c
olleges
or
un
i
versi
ty
.
Ba
sed
on
the
real
-
li
fe
ex
a
m
inati
on
tim
e
ta
ble
so
lvi
ng
the
sc
hedulin
g
pro
ble
m
fo
r
the
hi
gh
e
du
cat
i
on
al
i
ns
ti
tuti
on
s
are
m
or
e
com
plicated
tha
n
t
he
sc
hool.
It
app
e
ars
that
the
pro
blem
o
f
exam
inati
on
tim
e
ta
bling
is
to
al
locat
e
exa
m
inati
on
into
a
lim
it
ed
nu
m
ber
of
tim
esl
ot
and
wh
il
e
the
sat
isfyi
ng
ad
diti
onal
co
ns
trai
nts
.
Ty
pical
ly
,
the
e
xam
inati
o
n
ti
m
e
ta
ble
pro
ble
m
involves
tw
o
ty
pes
of
ha
rd
and
soft
const
raints
that
need
to
be
sat
isfie
d
in
orde
r
to
pr
od
uce
f
easi
ble
tim
e
ta
ble
[2
]
.
A
c
onflic
t
bet
ween
tw
o
or
m
or
e
exam
inati
on
s
assi
gns
in
t
o
lim
i
te
d
tim
eslo
t
cau
sin
g
diff
i
culti
es
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.
13
, N
o.
1
,
Ja
nu
a
ry 20
19
:
1
9
1
–
198
192
a
m
on
g
the
st
udents
.
P
rev
i
ous
resea
rch
e
rs
ha
ve
s
uggeste
d
a
so
luti
on
w
he
re
a
pen
al
ty
is
giv
e
n
wh
e
n
t
he
re
are
ci
rcu
m
sta
nces
that
vio
la
ti
on
of
c
onflic
t
on
two
or
m
or
e
exam
inati
on
s
is
assigne
d
a
t
the
sam
e
t
i
m
e
an
d
day
[3]
.
To
m
e
asur
e
the
gap
a
nd
ide
ntify
the
avail
abili
ty
of
fr
ee
am
ou
nt
of
tim
e,
sp
read
i
ng
the
pa
per
m
uch
as
po
s
sible
ov
e
r
exam
inati
on
durati
on
c
ould
pro
vid
e
e
noug
h
ti
m
e
fo
r
preparati
on
a
nd
stud
y
t
hat
le
ads
t
o
increase t
he
s
uc
cess of t
he
stu
den
t.
Su
r
vey
on
the
li
te
ratur
e,
m
os
t
of
the
c
onstrai
nts
an
d
pr
ob
le
m
s
wer
e
stu
died
an
d
s
olv
e
d
by
pr
evi
ous
researc
hers
us
i
ng
var
i
ous
ty
pes
of
the
a
ppr
oa
ch
pro
pose
d.
So
m
e
Gen
et
ic
al
gorithm
s
hav
e
been
us
e
d
to
so
lve
popula
t
ion
of
a
chrom
os
om
e
wh
ic
h
is
re
pr
e
sented
as
a
so
l
ution
f
or
fin
din
g
t
he
feasi
ble
tim
et
able.
The
cro
s
s
-
ov
e
r
operat
or
t
ends
to
fi
nd
sui
ta
ble
an
d
pro
duce
a
n
a
ppr
opr
ia
te
resu
lt
for
nex
t
popula
ti
on
wh
ic
h
de
pends
on
bo
t
h
par
e
nts
a
nd
c
hild
[4
–
6]
.
Heurist
ic
so
l
ution
pa
rtic
ularl
y
pro
ved
that
t
his
kind
of
m
eth
od
pro
vid
e
s
s
olu
ti
on
exam
inati
on
ti
m
e
ta
bling
by
deali
ng
the
c
onstrai
nts
e
ve
n
t
hough
not
gua
r
antee
not
op
ti
m
al
and
perfec
t
[7
,
8].
A
he
uri
sti
c
order
i
ng
is
us
e
d
to
exam
inati
on
tim
et
abling
prob
le
m
by
com
par
i
ng
five
dif
fer
e
nt
strat
egie
s
an
d
evaluate
t
he
be
st
res
ult
[
9].
Com
bin
at
ion
of
Genet
ic
and
He
uri
sti
c
al
so
a
re
gr
eat
strat
egies
wh
i
ch
c
a
n
ov
e
rc
om
e
diffi
culti
es
by
m
a
xim
iz
es
the
al
locat
ion
a
nd
m
ini
m
iz
e
as
m
uch
as
possi
ble
the
vio
la
ti
on
s
of
const
raint
in
de
te
rm
ining
t
he
best
s
olu
ti
on for
ti
m
et
abling
pro
blem
[10]
. A
Co
ns
tr
uctive
he
ur
ist
ic
is prese
nted
for
so
l
ving
U
niv
e
rsiti
Ma
laysia
Pahang
exam
inati
on
ti
m
et
abling
by
produce
a
good
qu
al
it
y
so
luti
on
com
par
ed
t
o
e
xisti
ng
softwa
r
e
syst
e
m
[11]
.
The
rece
nt
s
urvey
al
s
o
fou
nd
that
m
e
m
et
i
c
al
gorithm
wh
ic
h
is
m
et
aheu
risti
c
appr
oach
es
ha
ve
be
en
c
onstructe
d
f
or
dea
ls
exclusi
vely
with
fi
xed
le
ng
t
h
ti
m
e
ta
bles
[12]
.
Othe
r
exam
inati
on
tim
et
abling
pro
blem
a
lso
ha
ve
bee
n
s
olv
e
d
by
the
pr
e
vious
w
ork
with
m
et
aheurist
i
c
appr
oach
es
suc
h
as
Ta
bu
Se
arch
(TS),
GR
AS
P
(G),
G
re
at
Delu
ge
(
G
D)
a
nd
et
c.
[13
–
15]
.
Th
e
ev
al
ua
ti
on
perform
ance
of
exam
inati
on
tim
e
ta
ble
has
been
c
om
par
ed
with
ot
her
va
riances
of
A
CO
an
d
ap
pro
ach
that
has
been
pro
po
s
ed
of
Ma
x
-
m
in
ant
syst
e
m
(MM
AS
)
to
dete
rm
ine
the
feasi
ble
so
luti
on
a
nd
bette
r
resu
lt
[16, 1
7]
.
AN
C
OTT
pro
gr
am
us
ed
tw
o
so
luti
ons
wh
i
ch
are
a
nt
ra
nk
-
base
d
syst
em
fr
om
ACO
var
ia
nce
wit
h
heurist
ic
orde
ri
ng
to d
et
erm
ine
the
lo
west nu
m
ber
of
soft
co
ns
trai
nts
an
d
re
du
ci
ng
the
am
ount o
f
no
n
-
fe
asi
ble
tim
e
ta
bles
[9,
17
]
.
A
hybri
d
ant
col
on
y
al
gorithm
and
a
c
om
plete
local
search
with
m
e
m
or
y
he
ur
ist
i
c
wer
e
us
e
d
f
or
s
olv
i
ng
ti
m
esl
ot
pr
ob
le
m
s
on
the
exam
inati
on
tim
e
ta
ble
that
stud
e
nts
assi
gn
e
d
m
or
e
th
an
one
exam
inati
on
si
m
ultaneou
sly
and
m
axim
iz
e
the
a
vaila
ble
tim
e
slot
betwee
n
two
co
ns
e
cutive
exam
inati
on
s
[
18
]
.
In
this
researc
h
pa
per,
the
id
ea
of
the
stu
dy
is
to
sat
isfy
al
l
the
con
st
raint
s
and
pro
vid
e
a
pr
io
rity
to
assign
e
xam
e
ven
ts
i
nto
a
ti
m
et
able.
Ther
e
fore,
the
m
a
in
obj
ect
ive
of
t
his
stud
y
is
to
ba
la
nce
the
distribu
t
i
on
of
ti
m
et
able
sl
ots
an
d
st
ud
e
nt
exam
inati
on
assignm
ent.
We
pr
e
sente
d
wi
th
an
al
gorith
m
so
luti
on
w
hi
ch
is
base
d
on
AC
O
co
nce
pt
of
real
-
li
fe
ant
’
s
behavi
or
to
so
lve
our
e
xam
inati
on
tim
et
abling
for
Un
iS
ZA.
This
al
go
rithm
is
i
m
ple
m
ented
int
o
the
syst
e
m
wh
ere
th
e
res
ult
will
be
com
par
ed
an
d
te
ste
d
with
sever
a
l
dataset
s
to
ana
ly
ze
the
ef
fecti
ven
ess
an
d
fl
exibili
ty
of
ex
a
m
inati
on
ti
m
et
able.
T
his
pa
per
is
prese
nte
d
a
nd
orga
nized
i
nto
sever
al
sect
ions.
I
n
t
he
sec
on
d
sect
io
n,
an
e
xp
la
nation
a
bout
exam
inati
on
tim
et
abling
pr
ob
le
m
and
discussi
on
on
U
niSZ
A
exam
inati
on
tim
e
ta
bling
pr
ob
le
m
with
detai
ls
of
con
s
trai
nts.
A
s
um
m
ar
y
descr
i
bes
the
def
i
niti
on
of
t
he
prob
le
m
in
the
thir
d
sect
ion
.
I
n
the
f
ourt
h
sect
io
n,
t
he
p
r
opos
e
d
m
et
ho
d
appr
oach,
AC
O
is
ex
plained
.
The
ex
pe
rim
e
ntal
resu
lt
s
are includ
e
d
in
t
he
fifth
sect
io
n
a
nd
sect
io
n
si
xth
is
the
final r
e
su
lt
ac
hi
eved
.
T
he
la
st
sect
ion
pr
e
sent
ed ov
e
rall
conc
lusio
ns
a
nd fut
ur
e
work.
2.
AN
T
COL
ONY O
PTIMIZ
A
TION
(ACO
)
The
AC
O
is
a
fam
il
y
of
m
et
aheu
risti
c
that
can
be
use
d
to
so
l
ve
the
discrete
optim
isa
ti
on
pro
blem
[19]
.
An
t
Syst
em
(A
S)
is
t
he
ea
rlie
st
al
gorithm
of
ACO
that
has
bee
n
pro
po
se
d
by
D
ori
go
et
.
a
l
in
1991
for
so
l
vin
g
tra
velin
g
sal
es
m
an
pr
obl
e
m
(TSP
)
[20]
.
Then,
this
A
CO
evo
l
ved
a
f
te
r
a
few
ye
ars
with
e
m
erg
ence
tw
o
va
riants
of
An
t
Col
on
y
S
yst
e
m
(A
CS)
and
the
MM
A
S.
I
n
rece
nt
ye
ars,
m
any
research
e
r
s
hav
e
bee
n
co
ncen
t
rated
on
this
ACO
a
lgorit
hm
that
ap
plied
s
ucc
essfu
l
ly
on
their
var
i
ous
discret
e
op
ti
m
isa
t
ion
pro
blem
and
ot
her
com
bin
at
ori
al
pr
oble
m
s
s
uch
as
the
Q
ua
dr
at
ic
Assign
m
ent
Pr
ob
le
m
(QAP
),
Veh
ic
le
R
ou
ti
ng P
roblem
(
VR
P)
, G
raph C
olorin
g
P
roblem
(
GCP), a
nd Job Sche
du
li
ng
Pro
blem
(
JSP)
[
20]
.
ACO
al
gorith
m
is
insp
ired
by
trai
l
la
yi
ng
for
f
or
a
gi
ng
and
f
ollow
s
th
e
highest
co
nc
entrati
on
of
ph
e
r
om
on
es
in
est
ablishin
g
t
he
s
hortest
pa
th.
He
nce
,
the
pro
bab
il
it
y
of
us
in
g
s
horter
paths
from
nest
to
resou
rce
de
pe
nds
on
t
he
hi
gh
a
m
ou
nt
of
phe
ro
m
on
e.
To
de
scribe
t
he
ste
ps
of
AC
O
al
go
rithm
,
an
exa
m
ple
of
AS
for
T
SP
a
s
sh
owe
d
i
n
Fi
gure
1 an
d
t
hr
ee
levels
op
e
rati
on
for
e
xecu
te
t
he
c
on
ce
pt
of a
nt b
e
ha
vior.
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
An a
nt co
l
on
y
algorit
hm for
universi
ti
su
lt
an
za
i
na
l
abidin
exam
i
na
ti
on…
(
Ahmad Fird
aus K
ha
ir
)
193
Figure
1
.
A
C
O
al
gorithm
To
i
niti
al
iz
e
all
the
phe
ro
m
on
es
trai
ls,
t
he
first
le
vel
oper
at
ion
is
t
o
pla
ce
ant
rand
oml
y
on
node
i
with info
rm
a
t
i
on
th
at
creates t
he
so
luti
on.
T
hen, an
t m
ov
es
thr
ou
gh
th
e no
de
to no
de
fro
m
it
s
starti
ng
point of
node
i
.
A
t
no
de
i,
by
us
i
ng
tr
ansiti
on
pr
o
ba
bili
ty
,
the
ant
v
will
deci
de
th
e
node
w
hic
h
ha
s
not been visi
te
d
ye
t
for next
node
j
.
Th
e
for
m
ula of tra
ns
it
ion p
robab
il
it
y ru
le
s
desc
ribe
a
s foll
ows:
(
)
=
(
(
)
)
.
(
)
∑
(
(
)
)
.
(
)
1
if
j
∈
(1)
-
Set of n
odes
ant
v
wh
ic
h
is
at
the stat
e no
de
i
-
Am
ou
nt
of
pher
om
on
e
-
(
)
Ph
er
om
on
e tr
ai
l i
ntensity
that li
nk
node
i
to
j
.
-
= 1
/
Visibil
it
y(he
ur
ist
ic
in
form
ation
) desi
re
for
c
hoosi
ng c
losest n
ode
j
w
hen at n
ode
i
,
-
α
a
nd β
par
a
m
et
er to
d
et
e
r
m
ine the in
flue
nce
factor t
he
ph
e
r
om
on
e an
d visi
bili
ty
.
The
seco
nd
le
vel
op
e
rati
on
is
search
so
l
ution
that
is
op
t
ion
al
if
the
an
t
find
s
the
be
st
so
luti
on
.
The
phe
ro
m
one
trai
ls
update
is
the
la
st
le
vel
of
oper
at
ion
ACO
al
gorith
m
.
Ther
e
are
ci
rcu
m
sta
nces
in
A
S
wh
e
re
the
co
nst
ant
ph
e
ro
m
one
trai
ls
are
evap
orat
ed
to
le
t
the
ants
la
y
down
ph
e
r
om
on
e
after
com
pletio
n
of
it
s tou
r
. T
he fo
rm
ula is d
efine
d
a
nd d
e
scribe
as foll
ows:
(
t
)
= (1
-
).
(
t
-
1) +
∑
∆
=
1
(2)
(3)
m
nu
m
ber
of a
nts
(
1
−
)
e
vapor
at
io
n r
at
e w
he
re
is c
on
sta
nt
∆
am
ou
nt of
phe
ro
m
on
e lai
d by
an
t a
nt
v
. o
n
t
he
e
dg
e
i
an
d
j
Q
c
on
sta
nt stat
e an
d
is t
he
le
ngth
of a t
our
th
at
an
t
v
(
t
)
t
he
le
ngt
h of a to
ur that a
nt
3.
PROP
OSE
A
CO APP
ROA
CH FO
R
O
U
R
UETP
In
this
sect
io
n,
we
bri
efly
e
xp
la
ine
d
t
he
f
low
proces
s
f
or
m
anag
ing
the
Un
iS
ZA
’s
exam
inati
on
tim
e
ta
bling
pr
ob
le
m
with
the
representat
iv
e
so
luti
on
an
d
pro
pose
d
a
ppr
oach.
The
ACO
al
go
rith
m
and
heurist
ic
is
util
iz
ed
UE
TP
t
o
f
ind
t
he
s
olu
ti
on
to
t
he
heur
ist
ic
inform
at
ion
and
pher
om
one
trai
ls
for
a
fe
asi
ble
so
luti
on.
T
he
go
al
is
to
ac
hieve
best
a
vaila
ble
tim
e
slot
assignm
ent
fo
r
st
ud
e
nts
by
m
axi
m
isi
ng
the
ga
p
betwee
n
tw
o o
r
m
or
e co
ns
ec
ut
ive ex
am
inatio
ns a
nd o
t
her
r
espected
constr
ai
nts.
3.1.
Re
presen
tativ
e
So
lu
tio
n
We
pr
ese
nte
d
the
e
xam
ple
of
tim
esl
ot
ind
ic
es
on
U
niS
ZA
e
xam
inatio
n
wee
k
f
or
our
s
olu
ti
on.
Dep
e
nd
on
ea
ch
sem
est
er,
the
nu
m
ber
of
days
a
nd
ti
m
e
slots
can
be
di
ff
ere
nt
f
or
th
e
exam
inati
on
wee
k.
Fo
r
e
xam
ple,
3
-
wee
k
exam
i
nation
per
i
od
in
this
se
m
est
er
an
d
for
the
nex
t
exam
inati
on
m
a
y
be
diff
ere
nt
wh
et
her
week
or
2
-
weeks.
Sa
m
e
go
es
to
tim
esl
ot,
if
two
ti
m
esl
ots
in
a
day,
m
eans
that
t
wo
in
dices
tim
esl
ots.
So
f
or
exam
ple,
with
3
-
wee
k
exam
inati
on
per
i
od
has
45
tim
esl
ots.
The
total
of
45
tim
esl
ots
ind
ic
es
is
A
C
O
A
l
gori
t
hm
s
-
I
ni
t
i
a
l
i
z
e
phero
m
o
ne tra
i
l
s
-
D
o
wh
i
l
e
(
S
t
op
c
ond
i
t
i
o
n/
i
f
c
ri
t
e
ri
a
a
re
not sa
t
i
s
f
i
e
d)
-
l
oop
Ge
n
e
r
a
t
e
s
o
lut
ion
L
o
c
a
l
S
e
a
r
c
h
s
o
lut
ion
U
p
d
a
t
e
p
h
e
r
o
m
o
n
e
t
r
a
il
s
-
E
nd
D
o
-
E
nd
∆
(t)
=
{Q /
(
t
)
i
f
(
i
,
j
)
∈
(
t
)
0
i
f
(
i
,
j
)
∉
(
t
)
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.
13
, N
o.
1
,
Ja
nu
a
ry 20
19
:
1
9
1
–
198
194
represe
nted
as
15
days
with
out
ta
king
wee
ke
nds
(S
at
urday
and
S
unday
)
because
there
are
no
exam
s
are
hel
d
and in eac
h da
y has
3
ti
m
esl
o
ts (m
or
ning,
af
te
rnoon, eve
ni
ng).
Figure
2
.
Tim
e
slot
in
dices
Ba
sed
on
re
pre
sent
in
dices
ti
m
esl
ot
in
Fig
ure
2,
t
he
ti
m
eslo
t
num
ber
1,
2
an
d
3
a
re
de
note
d
a
s
day
on
e
,
ti
m
esl
ot
nu
m
ber
4,5
an
d
6
i
nd
ic
at
e
as
day
2,
the
n
f
ollow
e
d
by
a
nothe
r
ti
m
esl
ot
nu
m
ber
t
o
de
note
as
ano
t
her
day.
T
his
ind
e
xing
ti
m
esl
ot
is
rep
re
sented
to
pro
vi
de
an
d
fi
nd
a
s
uitable
tim
esl
ot
fo
r
th
e
stu
dent
s
if
a
sit
uation
t
hat they di
d no
t
ha
ve gap
-
f
ree ti
m
e.
In
Table
1,
an
exam
ple
of
r
oo
m
for
the
e
xam
inati
on
ha
s
bee
n
a
pp
li
ed
to
a
data
set
and
with
th
e
detai
ls
of
it
s
ca
pacit
ie
s.
A
pro
cess
of
he
ur
ist
ic
is
searc
hing
f
or
the
pro
per
r
oo
m
to
be
ch
ose
n
for
e
xam
inati
on
base
d
on
the
r
oo
m
detai
ls.
The
exam
inati
on
m
us
t
assign
into
a
room
bu
t
if
it
is
nece
ssary,
m
ulti
ple
of
a
room
(g
r
ouping
)
can
be
util
ized
as
lo
ng
the
exam
inati
on
is
nearby
to
ea
ch
ot
her
a
nd
capaci
ty
sh
oul
d
be
enou
gh
t
o
fit
the
nu
m
ber
of
s
tud
e
nts.
Dista
nc
e
m
a
trix
valu
e
betwee
n
roo
m
is
al
so
giv
e
n
as
sho
wn
i
n
T
able
1.
In
a
pa
rtic
ular
case,
there
is
an
exam
inati
on
that
needs
to
us
e
m
or
e
than
one
r
oo
m
to
facil
it
at
e
capaci
ty
of
stud
e
nts.
T
he
r
efore,
a
r
oo
m
gro
up
i
ng
is
pro
vid
e
d
by
c
om
bin
ing
with
oth
e
r
room
t
o
create
a
ne
w
total
capaci
ty
sh
ow
n
in
Ta
ble
2.
To
facil
it
at
e
the
room
capaci
t
y,
an
ar
rangem
ent
f
or
the
r
oom
gr
ou
ping
is
so
rt
e
d
decr
easi
ng
ly
from
la
rg
e
to
s
m
al
l
siz
e.
If
t
he
r
oo
m
gro
up
i
ng
ha
s
been
ta
ken
by
an
ex
a
m
inati
on
f
r
o
m
the
li
st,
the
oth
e
r
exam
inati
on
with
a
la
rg
e
ca
pacit
y
stud
e
nt
can
be
assigne
d
to
a
ny
su
it
able
roo
m
gr
oup
acc
or
ding
to
the
so
rti
ng
ca
pa
ci
ty
.
It
is
nec
essary
to
orga
nize
the
exam
i
nations
for
the
la
rg
e
stu
den
t
with
a
pr
i
or
it
y
so
that
durin
g
the
assi
gn
at
io
n
f
or
c
hoos
i
ng
a
room
will
be
well
arr
a
ng
e
d
f
or
t
he
ne
xt
exam
i
nation
to
fit
into
the
fo
ll
owin
g
cap
aci
ty
of
the
ro
om
.
No
te
th
at
,
the
value
of
in
form
at
ion
can
be
cha
nged
dep
e
ndin
g
on
the
op
e
rati
ons th
at
involve
d
Table
1
. R
oo
m
D
et
ai
ls
an
d Di
sta
nce
Value
Ma
trix
Ro
o
m
Grou
p
Ro
o
m
Cap
acity
DKU
AC2
1
AC1
9
AC2
0
AC2
3
FBIM
DKU
200
-
-
-
-
-
FIC
AC2
1
120
-
-
2
1
2
FIC
AC1
9
95
-
2
-
1
3
FIC
AC2
0
80
-
1
1
-
3
FIC
AC2
2
70
-
2
2
3
-
Table
2
.
R
oo
m
G
r
oupi
ng
Detai
ls
Ro
o
m
Grou
p
in
g
New
Total Cap
acit
y
Exa
m
in
atio
n
Split
Valu
e
(AC1
9
,AC2
0
,AC2
1
,AC2
2
)
565
3
(AC1
9
,AC2
0
,AC2
1
)
495
2
(AC1
9
,AC2
0
)
200
1
(AC2
1
,AC2
2
)
190
1
(AC1
9
,AC2
0
)
175
1
3.2.
A
nt
St
r
ategies
Moveme
nt
In
ord
er
to
de
m
on
strat
e
the
ACO
oper
at
io
n
for
t
he
e
xam
inati
on
assig
nm
ent,
we
pr
ov
i
de
a
n
e
xam
ple
m
at
rix
so
l
ution
f
or
ti
m
esl
ot
and
e
xam
inati
on
as
sho
wn
in
Table
3.
The
n,
a
sam
ple
of
7
exam
inati
on
da
ta
set
s
and
set
of
num
ber
stud
e
nts
ta
ken
a
n
e
xa
m
is
pr
ovide
d
to
sho
w
the
ant
syst
em
it
e
rati
on
proces
s
wh
ic
h
pr
ese
nted
in T
able 4.
Table
3
.
Ma
tri
x
S
olu
ti
on
of T
i
m
esl
ots an
d E
xam
s
T
i
m
eslo
t
Exa
m
E1
E2
….
….
E30
1
2
……
45
(
1
,2
,3
,4
,5
,6
,7
,8
,9
,1
0
,1
1
,1
2
,1
3
,1
4
,1
5
,2
0
,2
2
,2
3
,2
4
,2
5
,2
6
,2
7
,2
8
,2
9
,3
0
,3
1
,3
2
,3
3
,3
4
,3
5
,3
6
,4
3
,4
4
,
4
5
,4
6
,4
7
,4
8
,4
9
,5
0
,5
1
,5
2
,5
3
,5
4
,5
5
,5
6
,5
7
)
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
An a
nt co
l
on
y
algorit
hm for
universi
ti
su
lt
an
za
i
na
l
abidin
exam
i
na
ti
on…
(
Ahmad Fird
aus K
ha
ir
)
195
Table
4
.
E
xam
ple of
7 Sam
pl
e Ex
am
s an
d S
tud
e
nts
Stu
d
en
t
Grou
p
Exa
m
s
{S1
,S2,S3
,S4,S5
}
E1
{S6
,S7,S8
}
E2
{S6
,S7,S8
}
E14
{S3
,S4}
E17
{S1
,S2}
E16
{S1
,S2}
E19
{S5
}
E30
The
T
a
ble
5
dem
on
strat
ed
the
first
sta
ge
of
ant
proces
s
to
init
ia
li
ze
the
init
ia
l
sta
te
rand
om
l
y
assigne
d
t
he
a
nt
v
is
at
the
firs
t
node
i
in
orde
r
to
pr
ov
i
de
th
e
tim
esl
ot
and
exam
s
into
pr
oper
day
a
nd
se
ssion.
As
ex
plaine
d
on
t
he
re
pr
ese
ntati
ve
so
l
utio
n,
D1
is
ass
um
ed
as
the
fir
st
day
of
the
week
c
onsist
s
of
th
ree
tim
esl
ots (
each
d
ay
has
t
hr
ee t
i
m
esl
ots).
Table
5
.
In
it
ia
l Sta
te
Ran
do
m
A
ssig
n o
n
Ma
t
rix
Date
Sess
io
n
1
2
3
D1
(E
1
,T
1
)
D2
(E
2
,T
5
)
D3
D4
D5
Ba
sed on
t
he p
rev
i
ou
s
secti
on on
ACO
e
xpla
nation,
prob
a
bi
li
ty
so
luti
on
rul
e is used to
de
te
rm
ine the
node
w
hich
is
no
t
visit
ed
ye
t
by
a
nt
v
w
hich
init
ia
ll
y
was
pl
aced
on
node
i
exam
inati
on
will
fin
d
oth
e
r
node
j
exam
inati
on
.
T
he
perf
or
m
ance
of
a
nt
v
play
ed
a
n
im
po
rta
nt
ro
le
on
t
he
he
ur
ist
ic
a
nd
phe
ro
m
on
e
i
nfor
m
at
ion
on
the
e
dge,
w
her
e
the
pro
ba
bili
ty
pr
efe
rence
of
ti
m
esl
ot
exam
inati
on
node
j
f
or
the
c
urren
t
node
i
de
pends
on
it
.
T
able
6
sho
wed
the
t
i
m
esl
ots
and
exam
s
that
ha
ve
been
assi
gned
s
uitable
da
y
aft
er
ta
king
the
pro
bab
il
ist
ic
sol
ution
,
and t
he c
onflic
t co
ns
tr
ai
nts in
eac
h
e
xam
inati
on
are
conside
red.
Table
6
.
Assi
gn
Af
te
r
P
roba
bi
li
sti
c So
luti
on
Date
Sess
io
n
1
2
3
D1
(E
1
,T
1
)
D2
(E
2
,T
5
)
D3
(E
1
9
,T
7
)
(E
3
0
,T
9
)
D4
D5
(E
1
4
,T
1
4
)
3.
3
.
Heuris
tics
The
ACO
a
ppr
oach
for
e
xam
i
nation
ti
m
et
abl
ing
p
r
oble
m
is
su
pp
or
te
d
with
a
heurist
ic
m
e
thod
w
he
re
ta
kin
g
the
Lar
gest
degree
a
nd
La
r
ge
st
E
nroll
m
ent
into
a
ccount.
This
he
ur
ist
ic
is
util
i
zed
to
s
peed
up
th
e
process
of
fi
nding
a
sat
isfact
or
y
so
l
utio
n
e
ven
t
hough
the
possi
bili
ti
es
to
a
chie
ve
opti
m
al
so
luti
on
does
no
t
gu
a
ra
ntee but e
nough t
o p
rovi
de
a
res
ult.
T
he
d
esc
riptio
n o
f
the
m
et
ho
d
is
d
esc
ribing a
s foll
ows:
LD: E
xam
is assign
e
d first i
n t
he
sc
hedule if
the ex
am
h
as
t
he
m
os
t co
nf
li
ct
s w
it
h othe
r
e
xam
s.
LE: Exam
w
it
h l
ar
ge
enrollm
e
nt stu
de
nts is
gi
ven
pri
or
it
y t
o assi
gn ea
rlie
r
.
3.
4
.
Pher
om
one U
pd
at
e
A
f
or
m
ula
is
giv
e
n
t
o
i
nter
pr
et
e
the
be
st
it
erati
on
acc
ordin
g
to
the
up
dated
m
at
r
ix
value
o
f
ph
e
r
om
on
e
trai
ls.
He
nce,
t
o
ge
t
the
str
ong
c
on
ce
ntrati
on
of
the
ph
e
ro
m
one
trai
ls,
ad
diti
onal
so
m
e
a
m
ou
nt
of
value
will
pro
vid
e
res
ult
whet
her
the
best
so
luti
on
ca
n
be
obta
ined
bas
ed
on
t
he
ph
e
r
om
on
e
up
date
ru
le
s
.
A
ne
w
am
ou
nt
of
ph
e
ro
m
on
e
is
updated
when
the
a
nt
has
been
m
ov
e
d
to
al
l
exa
m
that
al
read
y
assig
ne
d
to
each ti
m
esl
ot o
n
the
tim
et
able:
∆
(t) =
{Q /
(
t
)
if (
i, j
)
∈
(
t
)
0
if (
i,
j
)
∉
(
t
)
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.
13
, N
o.
1
,
Ja
nu
a
ry 20
19
:
1
9
1
–
198
196
In
orde
r
t
o
m
ake
s
ure
that
the
un
li
m
i
te
d
inf
or
m
at
ion
pher
om
on
e
at
t
he
e
nd
proces
s
of
t
he
ACO
al
gorithm
, th
e ru
le
s
of
ph
e
rom
on
e evapo
rati
on
proces
s is a
pp
li
ed
to u
pdat
e an
d
c
om
pu
te
the trail
s:
(t)
= (1
-
).
(
t
-
1)
+
∑
∆
=
1
4.
E
X
PERI
MEN
TAL RES
UL
T
The
propose
d
ACO
a
ppr
oach
is
execu
te
d
in
our
sim
ulati
on
ex
pe
rim
ent
fo
r
m
aking
a
n
ob
s
er
vatio
n
and
ide
ntify
m
os
t
op
ti
m
al
resu
lt
pe
rfor
m
ance
pro
duced
by
data
set
.
T
he
al
gorithm
s
were
i
m
ple
m
ented
an
d
cod
e
d
i
n
a
we
b
-
base
d
c
om
pu
te
r
syst
em
on
W
in
dows
10
with
s
upporte
d
by
CP
U
I
ntel
Core
i5
-
5200
U
2.20
GH
z
a
nd
4.0
0GB
RAM.
We
wer
e
al
s
o
usi
ng
e
xam
ples
of
real
dataset
wh
ic
h
are
a
ppli
ed
to
the
sim
ulati
on
pro
gr
am
syst
e
m
.
The
six
of
te
st
case
datase
t
instances
are
pr
ese
nted
ba
se
d
on
fi
ve
detai
ls
of
e
xam
s,
stud
ent
s
,
enroll
ed,
ti
m
es
lots
an
d
pri
ori
ty
as
show
n
in
Table
7.
All
da
ta
set
is
com
p
uted
acc
ordin
g
to
each
num
ber
of
exp
e
rim
ent sim
ula
ti
on
that c
onduct
ed
t
o
c
om
par
e each
res
ult t
hat has
bee
n ob
ta
ine
d by t
he
syst
em
.
Table
7
.
T
he
F
IC D
at
aset
s
Exp
Nu
m
Test Case
Exa
m
s
Stu
d
en
ts
Enro
lled
Ti
m
eslo
ts
Priority
1
FIC3
0
_
A
30
460
1416
45
Y
2
F1
C3
0
_
B
30
460
1416
30
Y
3
FIC3
0
_
C
15
407
1058
30
Y
4
FIC3
0
_
A
30
460
1416
45
N
5
FIC3
0
_
B
30
460
1416
30
N
6
FIC3
0
_
C
15
407
1058
30
N
4
.
1
.
Discussi
on
on Te
st C
ase
A
pr
i
or
it
y
is
gi
ven
on
fe
w
of
the
te
st
case
t
o
determ
ine
exa
m
inati
on
assi
gned
f
or
the
la
r
ge
st
ud
e
nt’s
capaci
ty
or
la
rg
e
e
nrollm
e
nt
of
e
xam
inati
on
(LE
)
a
nd
to
be
sc
he
du
le
d
earli
er
into
the
tim
esl
ots.
The
e
xp
e
rim
e
nt
has
bee
n
ca
rr
ie
d
out
on
the
exam
inati
on
t
i
m
et
abling
that
app
li
e
d
w
it
h
ACO
a
ppr
oach
t
o
com
pu
te
the
e
xam
inati
on
ti
m
et
able
resu
lt
with
t
he
te
st
case
inf
orm
ati
on
that
has
prov
i
ded
in
t
he
pr
e
viou
s
sect
ion
.
The
r
esult
of
the
te
st
case
in
Ta
bl
e
8
i
nd
ic
at
es
t
hat
FI
C
30_A
1
w
hich
is
t
he
pr
i
or
it
y
giv
e
n
has
t
he
lowest
value
0.96
of
sta
nda
rd
de
viati
on
a
nd
F
IC30_B
5
w
hich
is
not
giv
e
n
a
pr
i
ori
ty
pr
od
uced
0.78
of
sta
nd
a
rd
de
via
ti
on
.
B
oth
res
ul
ts
def
ine
d
t
hat
the
e
xam
inatio
n
was
sc
hedu
le
d
ine
ff
ic
ie
ntl
y
and
pro
vide
good
resu
lt
rather t
ha
n othe
r
te
st ca
se r
es
ult f
or
th
e d
ist
rib
utio
ns
of exam
inati
on ev
e
nt in
t
he
ti
m
et
able.
Table
8
.
T
est
Ca
se Result
o
n FIC
Data Set
I
ns
ta
nces
Test Case
Den
sity
Co
n
f
lict (
%)
Mean
Var.
St.dev
FIC3
0
_
A1
2
.12
%
1
.61
0
.92
0
.96
FIC3
0
_
B2
2
.12
%
1
.91
0
.98
0
.99
FIC1
5
_
C3
1
.42
%
2
.43
2
.94
1
.71
FIC3
0
_
A4
2
.12
%
2
.00
1
.17
1
.36
FIC3
0
_
B5
2
.12
%
1
.89
0
.61
0
.78
FIC1
5
_
C6
1
.42
%
2
.62
1
.90
1
.38
4
.
2
.
Result
In
T
a
ble
9,
the
fi
nal
res
ults
of
t
he
FI
C
assi
gnm
ent
exa
m
inati
on
tim
et
ablin
g
wer
e
a
uto
-
gen
e
rated
t
o
su
it
the
rele
va
nt
ti
m
e
slots
and
c
om
pu
te
d
by
co
ns
ide
rin
g
al
l
pro
blem
s
a
nd
s
olu
ti
ons.
T
he
res
ult
al
so
pro
vid
e
s
m
axi
m
u
m
separ
at
ion
for
the
consecuti
ve
e
xam
s
and
exa
m
inati
on
that
has
been
giv
e
n
pr
i
or
it
y
was
a
ssign
e
d
earli
er
int
o
su
it
able
ti
m
esl
ots.
T
o
dep
ic
t
e
xa
m
ples
of
t
he
da
ta
ex
am
inati
on
t
hat h
as b
ee
n
assig
ne
d,
inf
orm
ation
on stu
de
nt take
s ex
am
inati
on
E1, E3
, E
16, a
nd E
19 tim
e slots as s
how
n
i
n
T
able
10.
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
An a
nt co
l
on
y
algorit
hm for
universi
ti
su
lt
an
za
i
na
l
abidin
exam
i
na
ti
on…
(
Ahmad Fird
aus K
ha
ir
)
197
Table
9.
E
xam
i
nation Ti
m
et
ab
le
Result
of FI
C
T
i
m
es
l
o
t
s
E
x
am
s
E1
E2
E3
E4
E5
E6
E7
E8
E9
E
1
0
E
1
1
E
1
2
E
1
3
E
1
4
E
1
5
E
1
6
E
1
7
E
1
8
E
1
9
…
E
3
2
E
3
3
T1
T2
T3
T4
T5
T6
T7
T8
T9
T
1
0
T
1
1
T
1
2
T
1
3
T
1
4
T
1
5
T
2
0
T
2
2
T
2
3
T
2
4
T
2
5
T
2
6
T
2
7
T
2
8
T
2
9
…
…
T
5
7
Table
9
.
E
xam
ples
of
Stu
den
t
Ex
am
inati
on
Assignm
ent
Ti
m
eslo
t
T7
T2
T14
T22
Exa
m
in
atio
n
s
E1
E3
E16
E19
5.
CONCL
US
I
O
N
In
t
his
pa
pe
r,
i
t
can
be
co
ncl
ud
e
d
t
hat
the
i
m
ple
m
entat
ion
of
AC
O
al
go
r
it
h
m
su
ccessf
ul
ly
ob
ta
ined
and
s
olv
e
d
t
he
real
pr
act
ic
al
exam
inati
on
t
i
m
et
abling
pr
oble
m
s
faced
by
the
F
IC,
U
ni
SZA
.
Alth
ough
t
he
pr
ese
nted
res
ult
do
es
no
t
gu
a
ran
te
e
the
best
scenari
o,
at
le
ast
the
auto
-
ge
ner
at
e
d
proce
s
sing
is
m
uch
be
tt
er
than
m
anu
al
ly
process
f
or
the
feasible
so
l
ution
on
our
UTi
m
e
.
It
is
prov
e
d
that
the
a
ppr
oach
ca
n
ob
ta
i
n
good
resu
lt
s d
e
pend
s
on h
ow
t
he
operati
on
as
so
ci
at
ed
with
the proble
m
.
Be
sides,
the p
e
rfor
m
ance
ca
n
be
im
pr
ove
d
and
e
nh
a
nce
d
by
cal
ibrati
ng
on
the
s
olu
ti
on
.
Ther
e
f
or
e,
further
te
sti
ng
an
d
analy
zi
ng
of
this
resear
c
h
w
il
l
be
able
to
ens
ure
the
est
ablis
hme
nt
of
t
he
a
ppr
oach
w
hich
helps
res
olv
e
oth
e
r
var
ia
nts
of
e
ducat
ion
al
insti
tuti
on
exam
inati
on
ti
m
et
abling wit
h t
he diffe
re
nt requ
i
rem
ent an
d
s
pecifica
ti
on
for
the
futu
re.
ACKN
OWLE
DGE
MENTS
This
wor
k
is
pa
rtia
ll
y su
pport
ed by U
niSZ
A (G
ran
t R
R
008 C
RIM/
2016).
REFERE
NCE
S
[1]
M.
Doulaty
,
M.
R.
F.
Dera
khsh
i,
and
M.
Abdi
.
Timetabl
ing:
A
Stat
e
-
of
-
th
e
-
Art
Ev
olu
ti
onary
A
pproach.
Int
.
J.
Mac
h.
Learn
.
C
omput.
2013;
3(
3):
255
–
258.
[2]
W
.
K.
Ho,
A.
Li
m
,
and
W
.
C.
O
on.
Max
imizing
Pape
r
Spread
in
Ex
aminati
on
Ti
metabl
ing
using
a
Ve
hic
l
e
Routin
g
Me
thod.
2001:
3
59
–
366.
[3]
A
y
ob,
M.,
Abdulla
h
,
S.
,
&
Malik,
A.
A
Pract
i
ca
l
E
xaminat
ion
Ti
metabl
ing
Probl
em
at
the
Uni
ver
siti
Kebangsaa
n
Malay
sia
.
Int
ernati
onal Journal of
Computer
S
cienc
e
and
N
et
wor
k
Se
curit
y
.
20
07;
7(9):
198
–
204
.
[4]
M.
Golub
and
D.
Jakobovi.
E
xam
Timetabl
i
ng
Us
i
ng
Gene
t
ic A
lgo
rithm
.
2009:
357
–
362.
[5]
S.
K.
Jha
.
Ex
am
Timetabl
ing
Pro
ble
m using
G
enetic A
lgori
thm
.
In
t.
J. Res. Eng. Te
chnol
.
2014;
3(4
):
649
–
654,
.
[6]
W
.
Erbe
n
and
P.
Y.
Song.
A
Hybrid
Gr
ouping
Gene
tic
Al
gor
it
hm
for
E
xami
nati
on
Timetab
ling
2
The
Hybri
d
Gr
ouping
Gene
tic
A
lgorit
hm
.
200
4.
[7]
Z.
Fu
and
A
.
L
i
m
.
Heuristic
s
for t
he
e
xam
sche
d
uli
ng
probl
em
.
2
000:
172
–
175.
[8]
A.
O.
Adewum
i,
B.
A.
Saw
y
err
,
and
M.
Montaz
Ali.
A
Heurist
ic
Solut
ion
to
the
Unive
rs
it
y
Time
t
abli
ng
Proble
m
.
Eng.
Comput
.
20
09;
26(8):
972
–
9
84
.
[9]
T.
The
pph
akor
n
and
P.
Pongchar
oen,
“
Heuri
sti
c
orde
ring
for
a
nt
col
on
y
base
d
ti
m
et
abling
too
l,
”
2012
,
vol.
4,
pp.
87
–
96
.
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.
13
, N
o.
1
,
Ja
nu
a
ry 20
19
:
1
9
1
–
198
198
[10]
N.
D.
Tha
nh.
Solv
ing
Timeta
bli
ng
Proble
m
Us
ing
Gene
ti
c
and
Heuristic
Al
gorithms
.
AC
IS
Inte
rnational
Confe
renc
e
on
Soft
ware
Eng
in
ee
ring,
Artificia
l
Intelli
g
ence,
Net
working
,
an
d
Parall
e
l/
Distr
ibut
ed
Comput
i
ng
(
SNPD
2007)
,
2007;
3:
472
–
477.
[11]
M.
N.
M.
Kah
ar
and
G.
Ken
dal
l
.
The
Ex
am
inat
ion
Time
tab
li
ng
Probl
em
A
t
Univ
ersiti
Ma
lay
sia
P
ahang:
Compar
ison
of
a
Constr
uct
ive
Heuristic
wit
h
an
Ex
isti
ng
S
oftw
are
Solut
ion
.
Eur.
J.
Oper.
R
es.
2010;
207(2
):
557
–
565.
[12]
E.
K.
Burke,
J.
P.
Newal
l,
and
R.
F.
W
ea
re.
A
Me
metic
Al
go
ri
thm
for
Unive
rs
it
y
Ex
am
Timetabl
ing
Springe
r,
Berl
in
,
He
ide
lb
e
rg,
1996:
241
–
2
50.
[13]
L.
Di
Gaspero
a
nd
A.
Sch
ae
rf
.
T
abu
Searc
h
T
ec
h
nique
s for Exam
inat
ion
Tim
et
abl
ing
.
2001
;
2079
LNCS: 104
–
117.
[14]
S.
Case
y
and
J.
Thomps
on.
GRA
SPi
ng
the
Ex
am
inat
ion
S
che
du
ling
Proble
m
.
Spr
inge
r,
Ber
li
n,
H
ei
de
lbe
rg,
2003:
232
–
244.
[
15]
H.
T.
Ã
and
S.
Abdulla
h.
An
Inte
gra
te
d
H
y
bri
d
Approac
h
To
The
Ex
aminat
io
n
Ti
m
et
ab
li
ng
P
roble
m
.
Om
ega
.
2011;
39(6):
598
–
607.
[16]
A.R.
Mirzaei
and
F.Dja
nnaty
.
En
hanci
ng
Max
-
Mi
n
Ant
Syste
m
for
Ex
aminati
on
Ti
metabl
ing
Probl
ems
.
Int.
J
.
Sof
t
Comput.
2008:
2
30
–
238.
[17]
M.
Ele
y
.
Ant
Algor
it
hms
for
the
Ex
am
Time
tabling
Probl
em
.
Pr
act
.
Theory
Autom
.
Timetabling
VI
.
2006;
3867
:
167
–
180.
[18]
R.
Abounac
er
,
J.
Boukac
hour,
B
.
Dkhiss
i,
A.
E
l
Hila
li
Al
aoui
.
A
Hybrid
Ant
Colony
Al
gorith
m
for
The
Ex
am
Timetabl
ing
Pro
ble
m
.
2016
;
12
: 15
-
42.
[19]
Dorigo,
M.,
&
B
lum,
C.
Ant
Col
ony
Optimizati
o
n
the
ory:
a
Surv
ey
.
Theoretical
Computer
Sci
en
ce
2005;
344
(2
–
3)
:
243
–
278.
[20]
S.
Katiy
a
r
and
A.
Q.
Ans
ari.
A
nt
Colony
Opti
mi
zation :
A
Tut
orial
Revi
ew
An
t
Colony
Optimi
zation :
A
Tutor
i
a
l
Re
v
ie
w
.
2015;
7
no.
AU
GU
ST:
3
5
–
41.
Evaluation Warning : The document was created with Spire.PDF for Python.