TELKOM
NIKA
, Vol.12, No
.4, Dece
mbe
r
2014, pp. 78
7~7
9
4
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i4.497
787
Re
cei
v
ed Au
gust 18, 20
14
; Revi
sed O
c
t
ober 8, 20
14;
Accept
ed O
c
tober 25, 20
1
4
Pre-Timed and Coordinated Traffic Controller Systems
Based on AVR Microcontroller
Fredd
y
Kurnia
w
a
n
1
, Den
n
y
Derm
a
w
a
n
1
, Okto Din
a
ry
anto
2
, Mardiana Ira
w
a
t
i
3
1
Department o
f
Electrical Eng
i
ne
erin
g,
Sekol
ah T
i
nggi T
e
knolo
g
i Adis
utji
pto
2
Department o
f
Mechanic
a
l E
ngi
neer
in
g, Sekola
h T
i
nggi T
e
kno
l
og
i Adis
utjipto
3
Department o
f
Informatics Engi
neer
in
g, Sekola
h T
i
nggi T
e
kno
l
og
i Adis
utjipto
Jln. Janti, Blok
R, Kompleks L
anu
d Adis
utjipt
o
, Yo
g
y
akart
a
, Ph. +
62-274-
4
512
62, F
a
x. +
6
2-27
4-45
12
65
*Co
rre
sp
ondi
ng autho
r
, email: fredd
yk
urni
a
w
a
n
@stta.ac.id
A
b
st
r
a
ct
T
he ma
jor w
e
a
k
nesses of traffic controllers i
n
Indon
esia ar
e una
ble to ac
commod
a
te the variet
y
of traffic volu
me an
d un
abl
e to be co
ord
i
nat
ed. T
o
so
lve th
e pro
b
le
m, a p
r
e-timed a
nd c
oord
i
nate
d
traffi
c
control
l
er system is b
u
ilt. The system
cons
ists of a master and a loc
a
l
control
l
er. Eac
h
control
l
er ha
s a
datab
ase c
ont
aini
ng s
i
gn
al-ti
m
i
ng p
l
a
n
s. T
o
synchro
ni
z
e
th
e sig
nal-ti
m
i
ng,
the master co
ntroll
er sen
d
s th
e
synchro
ni
z
a
ti
o
n
data to the lo
cal contro
ller w
i
reless
ly
, and t
he loc
a
l contro
l
l
er can
mo
dify a cycle le
ngth
by
add
ing or su
btracting the gr
e
en interv
al of a
n
y phas
es
. T
he transitio
n time for synchron
i
z
a
ti
on o
n
ly tak
e
s
one to sev
e
ral
cycles. The al
gorith
m
for co
ntrolli
ng t
he tr
affic inclu
d
in
g
coord
i
nati
on c
an be d
o
n
e
by
an
AVR
microc
on
troller. Me
mor
y
usag
e
of the
micr
oc
ontrol
l
e
r is l
o
w
e
r th
an 1
0
%,
me
a
n
w
h
ile t
he C
P
U
utili
z
a
ti
on is n
o
more th
an 1%,
and thus
the s
ystems co
uld b
e
w
i
dely dev
el
ope
d.
Ke
y
w
ords
: traffic controll
er, pre-timed, coor
d
i
nate
d
, AVR mi
crocontro
ller
1. Introduc
tion
1.1 Bac
k
gro
und of th
e Problem
Traffic j
a
ms i
n
mo
st majo
r citie
s
i
n
In
done
sia
have
re
sulted
in l
o
sin
g
millio
n
rupi
ah
s
every ho
ur. T
r
affic
con
g
e
s
tion is
ca
used
by many fa
ct
ors.
One
of th
e facto
r
s i
s
th
e cu
rrent traff
i
c
controlle
r
can
not a
c
commo
date the
vari
e
t
y of traffic vo
lume. T
r
affic
con
g
e
s
tion
h
appe
ns u
s
ual
ly
at the mai
n
j
unctio
n
s in th
e mo
rning,
b
e
fore th
e
offi
ce
hou
rs (6.0
0 a.m. to 8.0
0
a.m.)
and i
n
the
evening, after
the office h
ours (3.00
p.
m.
to
5.00
p
.
m.). Mean
while, at mid
n
i
ght, som
e
tim
e
s
many vehicle
s
have to wai
t
even thoug
h there a
r
e n
o
others vehi
cle
s
pa
ss at the interse
c
tio
n
.
Bec
a
us
e the traffic
s
i
gnal remains
red for the
fixed
perio
d, the
vehicle
sho
u
l
d
wait until t
h
e
sign
al turn
s to gree
n.
Many re
se
arch
studie
s
h
a
ve bee
n d
o
ne to im
p
r
ov
e the traffic
controlle
r in
orde
r to
accomm
odat
e the t
r
affic v
a
riation.
Fa
zli
obtain
ed
ve
h
i
cle
s
cla
ssifi
cation b
a
se o
n
neu
ral
netwo
rks
for an intelli
gent traffic controlle
r sy
stem
[1]. Khan and Aske
rzad
e implem
ented an im
ag
e
pro
c
e
ssi
ng
a
nd fu
zzy lo
gi
c
control, the
n
sent the
re
sult to
a mi
croco
n
tro
ller to drive the traffic
sign
al in the
desi
r
ed
ma
nner [2]-[3]. Hon
g
jin Zh
u
pre
s
ente
d
a
moving vehi
cle dete
c
tion
and
tracking
syst
em,
which
compri
sin
g
of hori
z
ont
al
ed
ge d
e
tectio
n
method
and
auto
co
rrel
a
tion.
The re
sult sh
ows that it is pos
sible to detect ea
ch i
ndividual veh
i
cle even if the vehicle
s
a
r
e
overlap
p
ing
[4]. Although
vehicle
dete
c
tion ha
s
bee
n imp
r
oved, t
here
is no
gu
arante
e
that
the
pro
c
e
ss
gives the a
c
curat
e
re
sult. To o
v
erco
me
the
probl
em, the traffic co
ntroll
er should
hav
e a
sign
al-timing
plan that
sho
u
ld be
used i
f
there i
s
an i
n
valid re
sult f
r
om d
e
tectio
n
pro
g
re
ss. Th
e
plan can be d
e
fined from traffic volume variation.
Some resea
r
ch
studi
es
ha
ve bee
n do
ne
to imp
r
ove v
ehicl
e flow at
the ro
ad
network. Xi
e
pre
s
ente
d
a sched
ule-driv
en co
ordi
nati
on for real
-ti
m
e traffic net
work [5], and
Shamshi
r
ba
nd
pre
s
ente
d
an
improveme
n
t of control ab
ility by
using
a neural net
work Q
-
lea
r
ni
ng app
roa
c
h
as
on Gam
e
Th
eory [6]. Lei
Wu p
r
e
s
ente
d
a be
e inspir
ed zonal vehi
cle routing
al
gorithm to
provide
a rea
s
on
abl
e and effect
ive optimal route for
th
e Dynami
c
Route Gui
d
a
n
ce System
[7].
Acco
rdi
ng to
Dotoli, one of the effective methods
t
o
improve traffic flow in the network is to
synchro
n
ize t
he traffic
sig
nals at all
int
e
rsecti
on
s esp
e
c
ia
lly a
t
adja
c
en
t in
ter
s
ec
tio
n
s
[8
]. T
h
e
obje
c
tive of the co
ordi
nati
ng is to provi
de co
nti
nuo
u
s
flow of traffic alon
g stre
ets or hig
h
ways.
The
M
anu
al on Uniform
T
r
affic Cont
rol
De
vi
ce
(M
UTCD)
also re
comm
end
s th
at traffic
sign
als
within
800 m
e
ters
must b
e
coo
r
dinated
und
e
r
a
comm
on
cycle l
ength.
O
t
h
e
r
r
e
as
ons
to
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 787
– 794
788
establi
s
h
coo
r
dinatio
n are
when the in
terse
c
ti
on
s a
r
e in clo
s
e p
r
oximity to one anoth
e
r a
n
d
traffic volum
e
s b
e
twe
en th
em a
r
e l
a
rg
e
.
Coo
r
din
a
te
d traffic co
nt
rolle
r
can
pe
rmit continu
o
u
s
movement al
ong an
arte
ri
al or throug
h
out a net
work of majo
r streets with
min
i
mum stop
s
and
delays. By minimizin
g
the amount
of accele
ration a
n
d
brea
kin
g
, the air polluta
nt emitted by the
vehicle
will
b
e
redu
ced
[9
]. It can
be
see
n
from
th
e p
r
eviou
s
rese
arch
that
the b
enefits of
coo
r
din
a
tion
depe
nd
on th
e unifo
rmity o
f
traffic flo
w
.
Ho
wever,
in t
he
scena
rio
that the
flow is no
t
uniform, som
e
improvem
e
n
ts ca
n be do
ne in the future [10].
The mo
dern
traffic controller
system
gene
rally is based o
n
32-bit p
r
o
c
e
s
sor and
coo
r
din
a
ted
b
y
comp
uter systems that a
c
t a
s
a
se
rve
r
[11]. Mea
n
while, present t
r
affic
co
ntroll
ers
in Indone
sia
are b
a
se
d on
MCS-5
1
8-bi
t micro
c
ont
rol
l
er. It is difficult to build the system o
n
the
last microcon
troller du
e to the limitation of
CPU spe
ed and mem
o
ry spa
c
e. T
he com
p
lexity of
the algorith
m
for coo
r
di
nati
ng traffic cont
rolle
rs
coul
d not be don
e b
y
a MCS-51
microcontroll
er.
Several rese
arch
studie
s
have be
en
introdu
ce
d
to
meet the traffic ch
aract
e
risti
c
in
Indone
sia. Primantary pre
s
ente
d
a co
ordin
a
tion
of
the traffic signal mod
e
l [12] and Jatmiko
pre
s
ente
d
th
e archite
c
ture of de
ce
ntra
lized
self
-o
rg
anizi
ng traffic cont
rol in
re
al situatio
n e
v
en
on no
n-stru
ct
ure inte
rse
c
tion like in
Ja
karta
[13
]. Howeve
r, it is found n
o
st
udie
s
ab
out
the
coo
r
din
a
ted
sign
al con
c
e
r
ned with the
traffic c
ontro
ller syste
m
in Indone
sia that is ba
sed
on
microcontroll
er.
The p
u
rpo
s
e
of the
re
sea
r
ch
is to b
u
il
d a
pre
-
time
d an
d
coo
r
di
nated traffic
controlle
r
system
ba
se
d on
an
AVR
microc
ontroll
er. Ba
sed
on
the si
gnal
-tim
ing pl
an, the
new alg
o
rith
m for
controlling traffic can a
ccommod
a
te the traffic
volu
me variation
and coordina
te anothe
r traffic
controlle
r. The algorithm i
s
tested on an
ATmega1
2
8
A
AVR 8-bit microcontroll
er that has m
u
ch
faster o
n
the CPU an
d larg
er memo
ry sp
ace.
1.2 Pre-Time
d and Coor
d
i
nated Tr
affic Con
t
roller
A traffic controller is u
s
ed t
o
swit
ch the traffic sig
nal. The sig
nal se
quen
ce is
red
,
green,
yellow, a
nd
red-clea
ra
nce.
The
main
va
riable
s
i
n
sig
nal-timin
g a
r
e
cycl
e le
ngth
and
gre
e
n
sp
lit.
Cycle length
(
) is the time in secon
d
s that it takes a signal to
compl
e
te on
e full cycle of
indication
s. One cy
cle is th
e sum
of the duratio
n of th
e gre
en inte
rval (
) plus the yellow interval
plus th
e red
-
clea
ran
c
e
int
e
rval. It also i
ndicates th
e t
i
me interval
b
e
twee
n the
st
arting
of green
for o
ne
app
ro
ach
until th
e
next time the
gree
n
start
s
.
Gree
n
split fo
r a
si
gnal
in a
given
dire
cti
on
is define
d
as
the fraction of
cycle len
g
th wh
e
n
the sig
nal is green i
n
that directio
n.
Pre-time
d traffic cont
roll
ers
ope
rate
in a pred
etermin
ed a
nd reg
u
la
rly repeate
d
seq
uen
ce
of
sign
al in
dications.
Gen
e
rally, this ty
pe
o
f
traffic
cont
roller is chea
p
e
r to
pu
rcha
se,
and ea
sie
r
to
install an
d m
a
intain than t
he othe
r
type
s such as
act
uated traffic
controlle
rs. Th
eir
repetitive nat
ure al
so facili
tates co
ordi
n
a
tion wi
th adj
ace
n
t signal
s, and they are useful whe
r
e
prog
re
ssion i
s
de
sire
d [11].
An additional
variable of
coo
r
din
a
tion is offset. The offset (
) can
be defined as the
interval from
an offset refe
ren
c
e p
o
int a
t
one si
g
nal t
o
the followi
n
g
nea
re
st on
e at the othe
r
sign
al.
The
of
fset refe
re
nce point i
s
a p
o
int wh
ere th
e controll
er
make
s
a d
e
ci
sion
to termi
nate
the coo
r
din
a
ted pha
se. Th
e cycle le
ngt
h can b
e
me
asu
r
ed from the su
cce
ssiv
e
offset reference
point define
d
by the operat
or [11].
2. Rese
arch
Metho
d
2.1 The Sy
st
em Hard
w
a
r
e
and Soft
w
a
re
The prototype of the sy
stem con
s
i
s
ts
of two traffic cont
rolle
rs, o
ne actin
g
as
a maste
r
controlle
r a
n
d
the
other
as
a lo
cal
controlle
r.
As use
d
by pre
v
ious re
sea
r
che
r
s
[6],[13],
the
system
u
s
es a di
strib
u
ted
syste
m
s ap
proa
ch
fo
r b
u
ilding
a tim
e
-ba
s
e
d
coo
r
dination. A t
r
affic
controlle
r can
be operated
for stand
-alo
ne or coor
din
a
ted mode.
Each traffic
controlle
r ha
s all
the
featu
r
e
s
d
e
sired
f
o
r sig
nal cont
rol at the
in
te
rsecti
on su
ch as
CPU,
a
c
curate
clo
c
k
ge
nerator,
and commu
n
i
cation mo
du
le as sho
w
n
in Figure 1
.
Each traffic controll
er h
a
s a data
b
a
s
e
contai
ning si
gnal-timi
ng pl
ans that will be allocat
ed to manag
e vehicle flows in
the lane for the
sign
alized intersectio
n
. Each ho
ur of the
wee
k
is cove
red by the sig
nal-timin
g pla
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Pre-Timed and Coordinated Tra
ffic
Cont
roller Sy
s
t
em
s
bas
ed on
AVR .... (Freddy
Kurniawan)
789
Figure 1. Th
e system blo
c
k diag
ram
At the heart
of the traffic controll
er
i
s
an ATm
e
g
a128A AVR
microcontroll
er. Th
e
microcontroll
er is
ru
nning
at
a clo
c
k freque
ncy of 1
1
.0592M
Hz. The mi
cro
c
o
n
trolle
r conta
i
n
s
main pe
rip
h
e
r
als:
an 8
-
bit
RISC mi
croproce
s
so
r,
128
kB flash me
mory spa
c
e,
4kB SRAM,
and
4kB EEPROM. The
algorithm for
cont
rolling traffic
is impl
ement
ed to be a Traffic
Cont
roll
er
Program written in C la
ngu
age u
s
ing
Co
deVisio
n
AVR 2.05.3 Sta
ndard Versio
n. The progra
m
is
store
d
in th
e
flash m
e
mo
ry and th
e d
a
taba
se
c
ont
aining
sig
nal-timing plan i
s
sto
r
e
d
in t
he
EEPROM of
the microcontroller.
In order to save the memo
ry
resource and reduce the
comp
utation
load, all vari
able
s
are formatted
a
s
8 or 16
-bit i
n
teger
and
all mathemat
ical
comp
utation
s
are pe
rform
e
d usin
g integ
e
r ope
ratio
n
s.
A KYL-200
U
wirel
e
ss
com
m
unication m
odule i
s
u
s
e
d
for
commu
ni
cation
at ea
ch traffic
controlle
r a
n
d
op
erato
r
te
rminal.
The
o
perato
r
te
rmi
nal
can
do
wn
load th
e
sign
al-timing
pla
n
to
each traffic controlle
r and
read the pla
n
from each tr
affic controlle
r. In order to
coordinate the
local
co
ntrolle
r, the ma
ster
controlle
r also se
nd
s
syn
c
hroni
zatio
n
d
a
ta to the lo
cal co
ntrolle
r. The
module tra
n
smits the data usin
g Freq
ue
ncy Shi
ft Keying modul
atio
n at baud rat
e
of 9600bp
s.
A
frame
of the
data
contai
ns
of 8
-
bit
d
a
ta, a
start
bit,
a stop bi
t,
and no pa
rity.
All
types
of
comm
uni
cati
on are d
one
wirel
e
ssly at a freque
ncy o
f
433MHz wit
h
different he
ader d
a
ta.
2.2 Signal-ti
ming Plan
The re
se
arch
is focu
sed o
n
a coupl
e of a four
-way intersectio
n
(cro
ssroa
d
)
with a traffic
sign
al at each intersectio
n
.
For each in
terse
c
tion
with a set of entry and exit r
oad
s, the traffic
sign
al cycl
es
throug
h a fixed seq
uen
ce
o
f
phase
s
. Wit
h
a sig
nal-tim
ing plan, a d
a
y
is segm
ent
ed
to ten time sl
ots. To a
c
co
mmodate
da
y-varian
ce
of traffic volum
e
in a
wee
k
,
there
are
th
ree
sign
al-timing
plan
s avail
abl
e that
ca
n b
e
allo
cate
d
to
several type
s of d
a
y: we
ekdays, Satu
rd
ay,
and Sun
day. All plans o
f
the master and the lo
cal controller are defin
ed
at the Traffic
Manag
eme
n
t Ce
nter
pro
g
ram
in th
e
ope
rato
r te
rminal
an
d
can
be
do
wnload
ed to
both
controlle
rs
wi
rele
ssly
as
shown in Fi
gu
re 1. A
ll inte
rse
c
tion
s in th
e area
can
b
e
co
ntroll
ed
by
different timin
g
-pla
ns
simul
t
aneou
sly tha
t
were me
rge
d
into an ove
r
all sa
map
horic cycl
e. Once
the plan
s
are
set, they re
main fixed u
n
t
il they
are
ch
ange
d ma
nu
ally from the
terminal. T
abl
e 1
sho
w
s the sig
nal-timin
g pla
n
for wee
k
d
a
y
s that is use
d
for experim
ent.
Synchronizatio
n
data
D
o
w
n
l
o
a
d
s
i
g
n
a
l
-
t
i
m
i
n
g
p
l
a
n
R
e
a
d
s
i
g
n
a
l
-
t
i
m
i
n
g
p
l
a
n
D
o
w
n
l
o
a
d
s
i
g
n
a
l
-
t
i
m
i
n
g
p
l
a
n
R
e
a
d
s
i
g
n
a
l
-
t
i
m
i
n
g
p
l
a
n
Th
e opera
t
or ter
m
inal
K
Y
L-
200U
Commu
nication
mo
dule
RT
C
DS
13
0
7
C
l
ock generat
or
AT
mega128A
micro
c
ontr
o
ller
In
terface
T
he master contro
ller
KYL-200U
Co
mmunication
module
KYL-
200U
Communication
modu
le
RTC
DS1307
Clo
c
k gener
ator
AT
mega128A
micr
ocon
troller
Interfa
c
e
The local contr
o
ller
T
r
affic light
T
r
affic light
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 787
– 794
790
Table 1. A sig
nal-timin
g pla
n
for the mast
er and the lo
cal controlle
r
Parameter
Ti
me
slot
Ti
me
Master control
l
er
Local co
ntrol
l
er
P
1
P
2
P
3
P
4
P
1
P
2
P
3
P
4
p
G
r
een interv
al
(
g
)
1
4:30
a.m.
10
10
10
10
10
10
10
10
0
0
2
6:00
a.m.
12
15
20
15
12
14
20
15
0
0
3
6:30
a.m.
15
20
30
20
15
19
30
20
20
40
4
7:10
a.m.
17
20
26
20
17
19
26
20
20
20
5
8:00
a.m.
20
22
26
20
20
20
27
20
5
20
6
10:00
a.m.
25
25
25
25
21
25
28
25
5
20
7
3:30
p.m.
30
25
22
25
29
25
22
25
10
20
8
6:00
p.m.
25
25
20
25
22
22
20
22
10
20
9
9:30
p.m.
15
15
15
15
0 0
0 0 0 0
10
12:00
p.m.
0
0
0
0
0 0
0 0 0 0
Yello
w int
e
rv
al (
Y
)
All
time
3
3 3 3
3
3
3
3
Red-c
l
ear
ance i
n
te
rval (
R
)
All
time
5
5 5 5
5
5
6
5
A time sl
ot o
f
the ma
ster
controlle
r
co
nsi
s
ts
of tim
e
to
start th
e
time
slot an
d green
interval (
g
)
of all phase
s
, while a time slot of the local controller h
a
s an ad
ditio
nal offset (
)
and
limit data (
p
)
.
The
r
e a
r
e
some timing
constraints for safety an
d f
a
irne
ss at
ph
ase
i
(P
i
)
: t
he
gree
n split (
g
i
) ru
ns fo
r
a variabl
e in
terval that can ra
nge
bet
wee
n
a mini
mum (
G
min
) and
maximum (
G
max
)
whil
e the
yellow
and
red-clea
ra
nce
splits ru
n fo
r
a fixed inte
rval (
Y
i
and
R
i
). If
the green int
e
rval of
all
phases at a ti
me sl
ot is zero, the
c
ontroll
er
will not control the traffi
c at
the time slot and the sig
nal displ
a
ys
flashin
g
-ye
llo
w. Offset ca
n vary from zero to the cycle
length. All values h
a
ve bee
n roun
ded int
o
numbe
rs of time step of o
ne-se
con
d
re
solutio
n
.
As mentione
d previou
s
ly, in order to
sync
h
r
oni
ze
the local controlle
r, the maste
r
controlle
r se
n
d
s the
synch
r
oni
zation
dat
a (syn
c).
Wh
en re
ceivin
g the data, the
local
controll
er
read
s the tim
e
from the
RTC then
save
s the time a
s
. Afterward, a
t
the end of l
o
cal
cycl
e (
),
the microcon
troller of the
local
control
l
er ex
e
c
ute
s
algorith
m
1 for syn
c
h
r
oni
zing the
sig
n
a
l-
timing of the local
controlle
r to the si
gnal
-timing of the maste
r
co
ntro
ller.
Algorithm 1: S
y
n
c
hronize the signal-timing of the local controller to the master
=
read_time(
R
T
C)
4
1
.
′
′
if
=
0 or
>
then return
if
>
0
then go to 13
if
ξ
, then
ξ
∑
4
1
ξ
ξ
ξ
∆
4
1
ξ
→
for i =1 to 4 do
return
if
ξ
–
, then
–
∑
4
1
ξ
ξ
ξ
∆
4
1
ξ
→
for i =1 to 4 do
return
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Pre-Timed and Coordinated Tra
ffic
Cont
roller Sy
s
t
em
s
bas
ed on
AVR .... (Freddy
Kurniawan)
791
The first ste
p
of the algorithm (line 1
)
is to
read the time from the
RTC a
nd sto
r
e it as
16-bit inte
ger
. After that, line 2 compute
s
a total of the gre
en inte
rval of all pha
se
s (
G
). Line
3
comp
utes th
e
that will be
used to limit
the maximu
m amount of
adjustm
ent that can
be
made in one
cycle. The
para
m
eter i
s
used to
pre
v
ent an exce
ssive g
r
ee
n interval or g
r
een
intervals
that are
to
o sho
r
t durin
g
tra
n
siti
oning.
If a lim
it is impo
se
d, the si
gnal
m
a
y not be
abl
e
to compl
e
te
offset tran
sition within
on
e cycl
e.
The
algorith
m
co
n
s
train
s
to a
d
d
or
subtract
the
gree
n inte
rval
no mo
re th
a
n
a
certai
n p
e
rcentag
e of
that range
s f
r
om 0
to 99.
The valu
e of
0
mean
s the lo
cal controll
er
is not sy
n
c
hronized to the maste
r
co
ntro
ller.
Line 4 comp
utes the
current offset tha
t
is
defined a
s
the time dif
f
eren
ce b
e
tween the
end of the m
a
ster
cycl
e (
) and the e
n
d
of the local
cycle
(
). The value is
store
d
in a 16
-bit
unsi
gne
d inte
ger
(
’
). If the cu
rrent offse
t
is e
qual
to
t
he offset defi
ned i
n
the
da
tabase (
), th
e
sign
al-timing
of the local controlle
r h
a
s
been
syn
c
hronized to the
sign
al-timin
g
of the ma
ster
controlle
r. A synchroni
zed l
o
cal timing
ca
n be sh
own a
s
ca
se 1 in
Figure
2
.
Let
den
otes the time diff
eren
ce
bet
we
en
’
and
in line
5. If
is le
ss than
zero,
t
L
is
leadin
g
, then
lines 8 –
12
shift next
t
L
progre
s
sively la
ter by timin
g
the next
cycl
e len
g
th
sligh
t
ly
longe
r than
th
e programme
d cycl
e len
g
th
. Synchro
n
izi
ng a le
adin
g
l
o
cal timi
ng
ca
n be
sho
w
n
a
s
ca
se 2 in
Figure
2
. On
the other h
a
n
d
, if
is grea
ter than zero,
t
L
is lagging,
then lines 1
4
– 18
s
h
ift next
t
L
progre
s
sively e
a
rlie
r by timing the next cy
cle
len
g
th sli
g
htly shorte
r. T
he ca
se
ca
n be
s
h
ow
n
as
ca
se
3
in
Figure
2
.
Figure 2. Shifting the end o
f
local cycl
e
Whe
n
the ne
xt
t
L
will be shifted progressively later, li
ne 9 di
stributes extra
green interval
prop
ortio
nally among
st all
pha
se
s. T
he highe
r the green interval o
f
a phase (
) thus the high
e
r
the addition in the green interval (
g
i
). The equatio
n
in line 9
is
done by multiplying
and
ξ
first, then the corre
s
po
nd
ing re
sult in an 16 bit integer i
s
divided by
∑
. Note that the
comp
utation i
s
perfo
rme
d
usin
g intege
r division. An
8-bit intege
r part of the division re
sult i
s
a
cha
nge in th
e value of the gree
n interval (
h
i
); me
anwhile the remaind
e
r i
s
distrib
u
ted to
the
addition
of a
n
y gre
en i
n
terval
startin
g
from th
e ph
ase
that h
a
s the g
r
eate
s
t
value of
gre
en
interval. Final
ly, line 12 co
mputes the g
r
een interv
al t
hat will be all
o
cate
d to the next cycle.
Line
s 14 – 1
8
are to shift
t
L
prog
re
ssive
ly earlier by subtra
cting tim
e
from one o
r
more
gree
n inte
rval
in the
seq
u
e
n
ce i
n
the
si
milar
m
anne
r. When
this
method i
s
u
s
ed, the ne
w
g
r
een
interval may
be
clo
s
e to
the minim
u
m pha
se
gre
en inte
rval (
G
min
), meani
ng only
a small
adju
s
tment in
cycl
e len
g
th
can
be
made
by sh
orte
ni
ng. In the
c
a
se, it may tak
e
many
c
y
c
l
es to
c
o
mplete an
offs
et trans
i
tion.
3. Results a
nd Discu
ssi
on
Ma
st
er
tim
i
n
g
A
sy
nc
hr
on
i
z
e
d
local
timing
(c
as
e 1
)
’=
’=
’=
Synchronize
a
le
ading
local
t
i
ming
(c
as
e 2
)
’<
(lead
ing)
’=
’=
is longer
Synchronize
a
la
gging
local
t
i
ming
(c
as
e 3
)
’>
(lagging)
’=
’=
is shorter
=
cycle lengt
h
=
l
ocal off
s
et
d
e
fi
ned in plan
’
=
current local
of
fse
t
=
t
he end of
master cycle
=
t
he end of
l
o
cal
cycle
Sinc data
S
inc
data
S
inc data
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 787
– 794
792
The syst
em is tested u
s
in
g a sign
al-tim
ing pl
an for th
e maste
r
and
the local
cont
rolle
r as
sho
w
n in T
a
b
l
e 1. The ma
ster co
ntrolle
r
controls th
e vehicl
e flow at
time slot 1 un
til 9 (4.30 a.m
.
~ 11.5
9
p.m.) and the l
o
cal
cont
rolle
r co
ntrols at ti
me
slot 1 u
n
til 8 (4.30 a.m. ~
9
.
29 p.m.). Ti
me
slot 1 and 2 o
f
the local co
ntrolle
r are n
o
t c
oordinate
d
to the master co
ntroll
er b
e
ca
use
= 0
an
d
p
= 0.
3.1 Transitio
n
Time
In orde
r to m
easure th
e p
e
rform
a
n
c
e, the sy
stem is
tested by o
p
e
r
ating it on
di
fference
value of offset error
(
). It operates
usi
ng time sl
ot 5 that
the co
rre
sp
ondi
ng time is from 8
.
00
a.m. to 9.59 a.m. Both controlle
rs
call
a 120-se
co
nd
cycle len
g
th while the lo
cal controlle
r also
call
s for a 40
-se
c
o
nd offse
t. The master controll
er re
start
s
at 08:00:00 and the
local controll
er
resta
r
ts at
08
:00:40. In th
e
ca
se,
the
cu
rrent l
o
cal offset is
equ
al to
the offset
defi
ned
at the
time
slot. A dashe
d-g
r
ay line in
Figure
3
sho
w
s th
e diag
ra
m of the cycl
e length of
th
e ca
se. Th
ere is no tran
si
tion time
becau
se there is no offset
error value o
c
curs in the tra
n
sition (
= 0s
).
Figure 3. The
fluctuation of
cycle len
g
th versu
s
differe
nt of
In the
se
con
d
ca
se, th
e lo
cal controll
er
rest
art
s
at
08:
00:55 th
us it
has
a 55-se
cond
l
o
cal
offset. The
of
fset e
rro
r val
ue
(
) i
s
15
seco
nd
s. In
order to
sy
nchronize the
timi
ng of
the l
o
cal
controlle
r, the following
cycle is su
btra
cted by 15 seco
nd
s to be 105 se
con
d
s
. The tran
sit
i
on
time is 105 seco
nd
s.
In the n
e
xt ca
se
s, the l
o
cal
controll
er
re
s
t
art
s
latter thus
the current loc
a
l
offs
et
respe
c
tively,
are 30, 45, 7
5
, 90, and 10
5 se
con
d
s. It can b
e
sh
own in
Figure
3
that the diagra
m
of the cy
cle l
ength at the tran
sition time
vary with the offset
error
. When the offs
et error
is le
ss or eq
ual to a
half of
a cycl
e length, the
next cycle le
ngth
is d
e
crea
se
d
in order to
sh
ift the next en
d-of-a-cyc
le
e
a
rlie
r. On th
e
other ha
nd,
whe
n
the
offset
error
is g
r
ea
te
r
th
an
a ha
lf o
f
th
e
c
y
c
l
e
le
ng
th
,
the
next cycl
e le
ngth i
s
in
cre
a
se
d in
orde
r to
shift the next end-of-a
-cy
c
l
e
later.
Figure
4
shows the
comparison
of th
e transition time
will be
taken versus t
he
current
offset in the d
i
fference of
p
.
The value
of
p
will limit the amount of t
he change in
the total green
interval. Whe
n
the
system
use
s
the
p
va
lue of 20%,
G
lim
it
is 16 se
con
d
s
(comp
u
ted on li
ne
3 of
algorith
m
1). The amou
nt of addition or
subtra
ction o
f
the cycle length (
) will be no more than
16 se
co
nd
s. It can be sho
w
n in the
90
100
110
120
130
140
150
8:0
0
:00
8:0
1
:00
8:0
2
:00
8:0
3
:00
8:0
4
:00
8:0
5
:00
8:0
6
:00
8:0
7
:00
Cycl
e
le
ng
t
h
(s
)
Ti
m
e
(a
.
m
.
)
= 0s
= 15s
= 30s
= 45s
= 75s
= 90s
= 1
05s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Pre-Timed and Coordinated Tra
ffic
Cont
roller Sy
s
t
em
s
bas
ed on
AVR .... (Freddy
Kurniawan)
793
Figure
4
(c
a
s
e
p
= 2
0
%), when the o
ffset erro
r (
) is greate
r
th
an 16 secon
d
s, the
transitio
n
will
take
mo
re t
han
one
cy
cl
e. In ge
ne
ral,
wh
en
is e
qual
or l
e
ss t
han
a h
a
lf of
the
cycle l
ength,
the larg
er th
e
value of
, the longer the transition
will
take. In the
case, the
cycle
will be
slightl
y
short
e
r duri
ng t
he transition. Otherwi
s
e, when
is
greate
r
tha
n
a half of the
cycle
length, the l
a
rge
r
the val
u
e of
, the shorter the transition
will ta
ke. In the
case, the next cy
cles
will be sli
ghtly longer d
u
ri
ng
the transition
.
Figure 4. The
transition tim
e
versu
s
offset
Figure
4
also
sho
w
s th
at the larger the
value of
p
, the shorter the tr
ansition will take.
Whe
n
the
system
co
nstrai
ned to
a
dd
or su
btra
ct
the
total green
in
tervals no
mo
re th
an
10%,
a
900-se
co
nd
s is
req
u
ire
d
to achieve ti
ming
synchronization
wh
en the
value
of
is from
60
se
con
d
s t
o
6
8
se
con
d
s.
H
o
wev
e
r,
a low
e
r v
a
lue of
p
will make the transition sm
oother.
3.2 Memor
y
Usage and
CPU Utiliz
ation
The Traffic
Controlle
r Progra
m
at the master
cont
rolle
r use
s
6
.
5% of flash memory
spa
c
e
an
d 6.
2% of SRAM
sp
ace o
n
ly
as
sh
own
in
Table
2. Th
e
memo
ry u
s
a
ge of th
e lo
cal
controlle
r is
greate
r
tha
n
the ma
ster b
e
ca
use
there
is a
coo
r
din
a
tion alg
o
rith
m at the lo
cal
controlle
r. M
ean
while, the
databa
se
co
ntains th
e
sig
nal-timin
g pla
n
s of
ma
ster
controlle
r takes
place 4.7% of EEPROM space only
and the database
at the local c
ontroller takes pl
ace a larger
spa
c
e d
ue to the addition d
a
ta for offset (
θ
) and limite
r
(
p
). The re
p
o
rt of memory usage
can
be
sho
w
n in Ta
b
l
e 2.
Table 2. Mem
o
ry usa
ge of the micro
c
ontroller
Memory types
Capac
ity
(Byte)
Memory use
d
Master
Local
(Byte) (%)
(Byte)
(%)
Flash
1310
72
8574
6.5 1091
4
8.3
SRAM
4351
271
6.2
337
7.7
EEPROM 4096
192
4.7
252
6.2
In general, when exe
c
utin
g the Traffic
Contro
lle
r Progra
m
, in the period of on
e-second
interval avail
able, the CP
U only spends l
e
ss
than 10 milli
seco
nds on processing the T
r
affi
c
Controlle
r Progra
m
. The CPU utilizatio
n is less than
1%; this means that in most of the tim
e
the
0
10
0
20
0
30
0
40
0
50
0
60
0
70
0
80
0
90
0
0
2
04
06
08
0
1
0
0
1
2
0
Tr
a
n
s
i
t
i
o
n
Tim
e
(s
)
C
u
rre
n
t
Of
f
s
e
t
(s
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 787
– 794
794
CPU in idl
e
state. From t
he fa
ct that
memory
usage and
CPU
utilization are low, it can be
con
c
lu
ded th
at Traffic Con
t
roller Prog
ra
m can b
e
exp
ande
d wid
e
ly by adding
a
n
y
features of t
h
e
traffic co
ntroll
er sy
stem wit
hout
ch
angi
n
g
the microco
n
trolle
r.
4. Conclusio
n
A pre-tim
ed a
nd coordinate
d
traffic cont
roller
system
can
be impl
e
m
ented b
a
se
d on an
ATmega1
28A
micro
c
ont
rol
l
er. Th
e
syst
em u
s
e
s
th
e
sig
nal
-timing
plan
s f
o
r
da
ily and
we
ekl
y
basi
s
in a databa
se that will be allocated to
manage vehicl
e flow. The timing of the loca
l
controlle
r ca
n be synchro
n
ize
d
to the timing of
the master
cont
rolle
r. The transitio
n time for
synchro
n
ization only take
s one to several cycle
s
. Th
e memory u
s
age of the micro
c
o
n
troll
e
r i
s
lowe
r than 1
0
%, while the
CPU utilization is no
m
o
re than 1%, thus the
syste
m
can b
e
wi
dely
develop
ed.
Ackn
o
w
l
e
dg
ement
The re
sea
r
ch
is fully supp
orted by The
Mi
nistry of Educatio
n
an
d Culture Indone
sia
unde
r g
r
ant
“
Hiba
h Bersai
ng
Dikti
”. T
h
e
autho
rs woul
d like
to tha
n
k
to
“
Kopertis Wilayah V
” a
nd
“
STT Adisutjipto
” for thei
r sup
port.
Referen
ces
[1]
F
a
zli S., Moha
mmadia S., R
ahma
n
i M. Ne
ural N
e
t
w
ork b
a
sed V
ehic
l
e
Classific
a
tio
n
for Intelli
ge
nt
T
r
affic Control.
Internation
a
l J
ourn
a
l of Softw
are Eng
i
n
eeri
n
g & Applic
atio
n
s
. 2012; 3(3): 1
7
-22.
[2]
Khan BA, La
i
NS. An Advance
d
F
u
zz
y
Log
ic bas
ed
T
r
affic Control
l
er.
Internatio
n
a
l Jour
nal
of
Innovati
on i
n
the Digit
al Econ
o
m
y
.
20
14; 5(1)
: 31-40.
[3]
Askerzad
e IN,
Mahmo
od M.
Contro
l the E
x
tensio
n T
i
me o
f
T
r
affic Light i
n
Sin
g
le
Juncti
on b
y
Usin
g
Fuzz
y
Logic.
In
ternatio
nal J
o
u
r
nal of El
ectrical
& Co
mp
uter
Scienc
es IJECS – IJENS
. 2010; 10(2): 4
8
-
55
[4]
Z
hu H, F
a
n H,
Guo S. M
o
vin
g
Ve
hicl
e D
e
te
ction
and
T
r
acking
in
T
r
affic Images
bas
ed
on
Horiz
onta
l
Edges.
T
E
LKOMNIKA Indone
sian Jo
urna
l of Electrical E
ngi
neer
ing
. 2
013;
11(1
1
): 647
7-6
483.
[5]
Xi
e
XF
, Smith
S, Barl
o
w
G.
Sche
dul
e-Driv
en
Coor
din
a
tio
n
for
Rea
l
-T
ime T
r
af
fi
c Net
w
ork Co
ntrol.
Procee
din
g
s 2
2
nd
Internation
a
l Co
nferenc
e
on Auto
mat
ed
Plan
nin
g
an
d Sched
uli
ng
(ICAPS). Atibaia,
Sao Pau
l
o, Bra
z
il. 201
2: 323-
331.
[6]
Shamsh
irba
nd
SS. A Distributed Ap
proac
h
for
Coordi
nati
on bet
w
e
e
n
T
r
affic Lights b
a
s
ed on Gam
e
T
heor
y
.
T
he Internati
ona
l Arab
Journa
l of Informati
on T
e
ch
n
o
lo
gy
. 201
2; 9(2): 148-1
53.
[7]
W
u
L, Yang L
,
Liu H, Z
han
g Y. Bee Insp
ired
Z
o
n
a
l V
e
hicle
Ro
uting
Algorit
hm in U
r
ban T
r
affic.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2014; 1
2
(9): 6
699-
671
0.
[8]
Dotoli M, F
anti
MP, Meloni C
.
Coordi
natio
n
and
Re
al T
i
me Optimizatio
n
of Signal-timi
ng pl
ans for
Urba
n T
r
affic
Contro
l.
IEEE International Conf
erence on Networking,
S
ensing and
Control
. 20
04; 2
:
106
9-10
74.
[9]
Dobr
e C. Usi
ng Intel
l
i
gent
T
r
affic signa
ls
to Red
u
ce V
ehicl
e Emiss
i
o
n
s.
Internati
o
n
a
l Jo
urna
l o
f
Innovativ
e Co
mp
utin
g, In
formati
on a
nd C
o
ntrol
. 201
2; 8(9
)
: 6283-6
3
0
2
.
[10]
W
ang K, Hua
n
g
L, Z
hai Y, W
ang
Y. T
r
affic
Sign
al S
y
nc
hro
n
izati
on i
n
Do
u
b
le R
i
ngs
Roa
d
Structure.
Journ
a
l of Infor
m
ati
on
& Co
mputatio
na
l Scie
nce
. 201
2; 9(2)
: 395-40
2.
[11]
F
edera
l
Hi
gh
w
a
y
A
d
min
i
strati
on.
T
r
affic Sig
nal-ti
m
i
ng Ma
n
ual
.
F
e
d
e
r
a
l
H
i
g
hw
ay
A
d
mi
ni
stration. U.S.
Dep
a
rtment of T
r
ansportation.
2008; 6
1-6
49.
[12] Primantari
LF
A.
Koordin
a
si
Pengat
uran
Lamp
u
La
lu L
i
ntas (Studi
k
a
sus: Ruas J
a
la
n Prof. Dr.
Sorhars
o
- A
d
i
Sucipto - A.Yani - A
d
i S
oemarmo).
Jurnal T
e
knik S
i
pil
dan Arsitektur
. 20
10; 8(
12): 1
3
-
30.
[13]
Jatmiko W
,
Azurat A, W
i
bo
w
o
AH, Marih
o
t H,
W
i
caksana
M,
T
a
kaga
w
a
I, Seki
y
a
m
a
K, F
u
kuda T
.
Self-Organizing Ur
ban T
r
affic
Control Arc
h
it
ecture
w
i
th S
w
a
rm-Self Organizing M
a
p in
J
a
karta: Signal
Contro
l S
y
ste
m
and Simul
a
tor,
Internation
a
l
Journ
a
l on S
m
art Sens
in
g a
nd Intell
ige
n
t Systems
. 20
10;
3(3): 443-
46
5.
Evaluation Warning : The document was created with Spire.PDF for Python.