InternetworkingIndonesiaJournal

Volum

e 3N

umber 1

Spring 2011

ISSN: 1942-9703IIJ ©

2011w

ww

.InternetworkingIndonesia.org

Editors’ Introduction 1

Wavelet Approach on Frequency Energy Distribution of 3Electrooculograph Potential Towards Direction W. M. Bukhari W. Daud and R. Sudirman

Test Case Generation using GOM Algorithm 11S. Subramanian and R. Natarajan

A Review of Existing Traffic Jam Reduction and 19Avoidance TechnologiesB. Hardjono

Analisa Pengaturan Channel untuk Perbaikan Performansi 25Pengiriman SMSRoessobiyatno, S. Prasetio, A. Qiantori and W. Agung

The Indonesian Journal of ICT and Internet Development

Internetworking Indonesia JournalThe Indonesian Journal of ICT and Internet Development

ISSN: 1942-9703

Manuscript Language: The Internetworking Indonesia Journal accepts and publishes papers in Bahasa Indonesia and English.

Manuscript Submission: Manuscripts should be submitted according to the IEEE Guide for authors, and will be refereed in the standard way. •Manuscript pages should not exceed 7 pages of the IEEE 2-column format. It should be submitted as a Microsoft-Word file, •using the IIJ Template document which can be found at the www.InternetworkingIndonesia.org website.Manuscripts submitted to the IIJ must not have been previously published or committed to another publisher under a copy-•right transfer agreement, and must not be under consideration by another journal.Papers previously published at conferences can be submitted to the IIJ, but must be revised so that it has significant differences •from the conference version.Authors of accepted papers are responsible for the Camera Ready Copy formatted using the same IIJ Template format.•Authors are advised that no revisions of the manuscript can be made after acceptance by the Editor for publication. The ben-•efits of this procedure are many, with speed and accuracy being the most obvious. Please email your papers (or questions) to: • editor@InternetworkingIndonesia.org

Submission Guidelines: Please review the descriptions below and identify the submission type best suited to your paper.Research Papers• : Research papers report on results emanating from research projects, both theoretical and practical in nature.Short papers• : Short research papers provide an introduction to new developments or advances regarding on-going work.Policy Viewpoints• : Policy Viewpoints explore competing perspectives in the Indonesian policy debate that are informed by aca-demic research.Teaching Innovation• : Teaching Innovation papers explore creative uses of information technology tools and the Internet to im-prove learning and education in Indonesia.Book Reviews• : A review of a book, or other book-length document, such as a government report or foundation report.

Prof. Edy Tri Baskoro, PhD (ITB, Indonesia)Mark Baugher, MA (Cisco Systems, USA) Lakshminath Dondeti, PhD (Qualcomm, USA)Paul England, PhD (Microsoft Research, USA)Prof. Svein Knapskog, PhD (NTNU, Norway) Prof. Merlyna Lim, PhD (Arizona State University, USA)

Prof. Bambang Parmanto, PhD (University of Pittsburgh, USA) Prof. Wishnu Prasetya, PhD (Utrecht University, The Netherlands)Graeme Proudler, PhD (HP Laboratories, UK) Prof. Jennifer Seberry, PhD (University of Wollongong, Australia) Prof. Willy Susilo, PhD (University of Wollongong, Australia)Prof. David Taniar, PhD (Monash University, Australia)

Thomas Hardjono, PhD (MIT Kerberos Consortium, MIT, USA)

Budi Rahardjo, PhD (ITB, Indonesia)

Kuncoro Wastuwibowo, MSc(PT. Telkom, Indonesia)

Internetworking Indonesia Journal is a semi-annual peer-reviewed electronic journal devoted to the timely study of Information & Com-munication Technology (ICT) and Internet development in Indonesia. The journal follows the open-access philosophy and seeks high-quality manuscripts on the challenges and opportunities presented by information technology and the Internet in Indonesia.

Editorial Advisory Board

Co-Editors

Journal mailing address: Internetworking Indonesia Journal, PO Box 397110 MIT Station, Cambridge, MA 02139, USA.

Technical Editorial Board

Sulfikar Amir, PhD (NTU, Singapore)Moch Arif Bijaksana, MSc (IT Telkom, Indonesia)Teddy Surya Gunawan, PhD (IIUM, Malaysia)Dwi Handoko, PhD (BPPT, Indonesia)Mira Kartiwi, PhD (IIUM, Malaysia)Bobby Nazief, PhD (UI, Indonesia)Anto Satriyo Nugroho, PhD (BPPT, Indonesia)

Yanuar Nugroho, PhD (Manchester University, UK)Bernardi Pranggono, PhD (University of Leeds, UK)Bambang Prastowo, PhD (UGM, Indonesia)Bambang Riyanto, PhD (ITB, Indonesia)Andriyan Bayu Suksmono, PhD (ITB, Indonesia)Henri Uranus, PhD (UPH, Indonesia)Setiadi Yazid, PhD (UI, Indonesia)

Vol. 3/No. 1 (2011) INTERNETWORKING INDONESIA JOURNAL 1

ISSN: 1942-9703 / © 2011 IIJ

ELCOME to the Spring 2011 issue of the Internetworking Indonesia Journal. This regular issue of the journal

brings four (4) papers from a diverse background, covering a range from biomedical signals recognition to CDMA networks.

To our delight one of the papers is written in Bahasa Indonesia, something that is in-line with our aims of promoting the culture of writing good scientific papers in Indonesia. We feel that developing one’s ability to express ideas, concepts and research methods/results in Bahasa Indonesia is an important step towards developing the habit of good research reporting – a skill that is transferable later when writing papers in English. As such, we see the IIJ also as “bridging” journal that fills a gap. Many Indonesian researchers are already able to produce good scientific papers in English, and thus are able to submit their papers to international conferences (e.g. IEEE sponsored conferences) and have access to these international conferences and journals. However, there remains the need for a national-level journal in Indonesia, one to which researchers who are comfortable writing papers in Bahasa Indonesia are able to submit their papers.

Another aim of the journal is to provide a publishing venue for graduate students who are completing their Masters (S2) or Doctoral (S3) studies. To that end we are delighted that one of the papers in the current issue of the journal was written by a Master’s degree student. We feel it is important for the coming generation of students to begin writing papers early in their careers and to obtain experience in submitting papers to journals.

The first paper in the current issue of the IIJ focuses on biomedical signals recognition research. The paper describes the identification of electrooculography (EOG) signals related to eye movements, and proposes the use of wavelet transforms instead of the usual Fourier transforms. The paper describes the data acquisition environment used to conducts the research work, and suggests that researchers in this field look into the details of the energy and frequency bands distribution (from the eye movement signals) in order to obtain better interpretation of the EOG signals.

The editors can be reached individually at the following email addresses.

Thomas Hardjono is at hardjono@mit.edu, Budi Rahardjo is at rahard@paume.itb.ac.id, while Kuncoro Wastuwibowo is at kwast@telkom.co.id.

The second paper addresses a relevant issue in the area of

software engineering. It focuses on the topic of test case generation, in which an automated approach to generating test-cases for a system is performed. The paper proposes the use of a three tier architecture containing several components, including a source code analyzer, an XML parser, a constraint analyzer and a test data generator. The approach employs the branch coverage criteria using the Generalized Optimization Meta heuristic (GOM) algorithm and code constraint graph (CCG) to efficiently maximize the coverage of all the branches in the test case. The work finds that test case generation is faster than with the simple genetic algorithm. This is due to the fact that the number of iterations for reaching the optimal solution is quick.

Traffic jam reduction and avoidance is the focus of the third paper, in which a literature survey is conducted on technologies that address the problem of traffic jams in highly populated areas. It surveys technical solutions that have proposed and adopted in some countries, including China and in Europe. The overall conclusion of the paper is that a combination of technical systems and regulatory approaches are needed in order to address the needs that are specific to the city or location in question.

Finally, the fourth paper – written in Bahasa Indonesia – looks into techniques to improve channel management in CDMA networks in order to provide a better success rate for SMS message transmissions and therefore better utilization of a given CDMA network. SMS transmissions experience high failures rates in the presence of a high volume of voice traffic. The paper, which is written by engineers and researchers at the TELKOM Research and Development Center (in Bandung, Indonesia), provides some research data that points to the need for CDMA operators to address channel management in anticipation of times of high voice traffic that may impact SMS message transmissions. Seeing that SMS messaging is today an important part of personal communications many developing nations, this topic is very relevant for the telecommunications industry in Indonesia.

Thomas Hardjono Budi Rahardjo

Kuncoro Wastuwibowo

Editors’ Introduction

W

2 INTERNETWORKING INDONESIA JOURNAL Vol. 3/No. 1 (2011)

ISSN: 1942-9703 / © 2011 IIJ

Vo

Abstuidemolot tratimchaphycapeyeto dtecaccsigrecdiranaTrchascaeneFroexhsca4H(4-

IPot

ocuReby prode-eleeye

T

ol.3/No.1 (2011)

bstract— Biomudied by manyentification of ovement potent

of informatiansform is to dme in differenaracteristic of ysical variableptured using ees to record thedetermine the c

chnique is adopcurate. New te

gnals to be cacorded data is rections, i.e. upwalysis for eachansform (WT) ange of time alogram is ploergy for each wom the result, hibit differenceale such as left

Hz), downward -8Hz).

Index Terms—tentials, Scalogr

The human 12mm. Themovement o

ulography (EOetinal Potential

the metabooduction of C-polarizations ectrical recordie has a restin

WD

T

medical signals y researchers. I

Electro-oculogtials by using wion than FFTistribute the sint frequency

the signals se in signal anlectrodes place

e eye movementcharacteristic o

pted because it ichnology enha

aptured using composed of a

ward, downwarh eyes signal by comparing tand frequenc

otted to displaywavelet coefficie

it is shown tes in signals ent with scale 6 (with scale 9 (1

—- Electro-oculam, Wavelet Tra

I. INTRO

eye is a sphere signals thatf the human e

OG). The EOGl (CRP) that isolically activRP comes froof the nervousing correspondg electrical po

Wavelistribu

F

IN

recognition aIn this study, graphy (EOG)

wavelet transforT. The capabignal energy wibands. This

since energy inalysis. The Eed on the forehts. The wavelet f eye movemenis non-invasive,ncement has athe Neurofax

an eye movemerd, left and righis analyzed bythe energy dist

cy of each sigy the different

ent towards diffhat the differe

nergy with thei(8-16Hz), right -2Hz) and upw

ogram, Eye Mansform.

DUCTION

rical structure wt can be sen

eyes can be knG is derived fs generated wive retinal epom the hyper-p cells in the re

ding to the eye otential, with

et Appution o

to

WFaculty of Ele

NTERNETWORK

ISSN: 194

ctivity has beewe describe th

) signals of eyrm which givesility of wavelith the change

will show this an importa

EOG signals ahead around thfeatures are uset waveform. Th, inexpensive an

allowed the EOEEG-9200. Th

ent towards fouht. The proposey using Wavelribution with thgnal. A wavelt percentages ferent movemenent EOG signair correspondinwith scale 8 (

ward with scale

Movement, Sign

with a radius nsed from th

nown as Electrfrom the Corneithin the eyebapithelium. Thpolarization an

etina. EOG is thmovement. Ththe front of th

proachof Elecoward

W. M. Bukhariectrical Engin

81310 Skud

KING INDONE

42-9703 / © 201

en he ye

s a let of he

ant are he ed his nd

OG he ur ed let he let of

nt. als ng (2-e 7

nal

of he

ro-ea all he nd he he he

globe phenom1848 a EOthe eyeare genthe forresult movemrespectdependwhen tpresenabnormsupran[2]. EOthe reoculogdiagnopotentigeneraenvisarelativ

This EacquisVertica

h on Frctroocuds Dire

i W. Daud andneering, Univdai, Johore, M

ESIA JOURNAL

11 IIJ

positive and menon was firsand has been thOG are taken ue. Exact electrnerally placedrehead. When will occur. Th

ment can betively. The dent on the elethe eyes are m

nt, EOG is usmalities suchnuclear occulom

OG is a techniqetina. The regram. The maosis and in recoial produced b

ation of the aging dipoles loely positive po

EOG signal ition system al (V) channe

requenulograpection

d R. Sudirmaversiti TeknoloMalaysia

L

the back wst observed byhe foundation ousing bipolar erode placemen

d on the templen the eyes movhe magnitudese seen betwpolarity of

ectrode setup smoving towardsed for the eh as nystagmotor dysfunct

que for measuresulting signaain applicationording eye movby movement

EOG signal ocated in the eyotential with re

Fig. 1. Position of

is picked upconsisting of

els. The placem

ncy Enph Pot

an ogi Malaysia

with globe negy Emil du Bois-of electrooculoelectrodes on thnts vary, but thes or on the dve, a differents of the right ween -75µV

movement psince the signa

d positive electevaluation of gmus, strabition is briefly e

ring the restingal is called s are in ophthvements [3]. Tof the eye or

can be undyes with the co

espect to the ret

f electrodes

by a bi-chaf the Horizonment of the el

nergy tential

3

gative. This -Reymond in

ography [1]. he outside of he electrodes distal ends of tial potential and left eye to 150µV

potentials is al is positive trode [1]. At occulomotor ismus, and explained by

g potential of the electro-halmological

The EOG is a eye lid. The derstood by ornea having tina [4].

annel signal ntal (H) and lectrodes are

4 shotwH2

andrefcon

NsigandDineye[6]Nemoanascaspefouthe

Wabecinfextinswoovan frerep 14

own in Figure o electrodes to

2) to detect hod below the eyference electrodnnection is illu

Fig. 2. (1) (3) Disposable

Numerous othgnal processingd extract the fenesh [5] used te gaze for con] used the EOGeural Networkovement for talysis. Study dalogram decomecific frequencund that 90% oe lower or high

avelet transforcause of its abformation. Thetension of thstead of workiorks on a mercome the lim

analyzing fuequency. It hpresented on a In the previo

, 15, 16, 17],

1. The EOG so the right andorizontal moveye (V1, V2) tode is placed on

ustrated in Figu

Data Acquisitione Ag/AgCl bioelect

her techniques g have been ueatures of interthe EOG signa

ntrolling a comG signals for thk (ANN), and the classificatidone by the Bmposition to dcy bands of vof the signal enher scales and s

II. WAVELET

rm is a powerbility to extrace wavelet tranhe classic Foung on a single

multi-scale basmitations of Ffunction that as a finite etransient signaous studies as wavelet analy

INTER

signals are obtad left of the oement and anoo detect vertican the forehead ure 2.

n system. (2) Etrode

from the theorused to obtainrest for classifials for determin

mputer while Ayhe classificatio

Sudirman [7ion by using Bhander [8] udetermine the vertical eye mnergy (90%) issignal denoisin

T TRANSFORM

rful tool in anct time and fresform could burier transforme scale (time osis [10]. Waourier methodare localized

energy functioals.

demonstrated ysis has develo

RNETWORKING

ISSN: 194

ained by placinouter canthi (Hother pair aboval movement. (G). The overa

Electrode connecto

ry of biomedicn representationication purposening the angle ysegul and Kan with Artifici

7] used the eytime frequencsed the wavelmost energy

movement. Thes concentrated ng.

nalyzing signaequency domabe defined as am, except thaor frequency), avelet functionds by employind in time anon and can b

by [11, 12, 1oped exciteme

G INDONESIA

42-9703 / © 201

ng H1, ve A

all

or.

cal ns es. of

ara ial ye cy let in ey in

als ain an at, it

ns ng nd be

3, ent

for thsuccestransiesignal wavelein termexpandtrigonoin the called Wamethodtime anrepreseanalyssignalsdecomfrequendata fdecomThis is Inprocesbe inteset of a vaniAssumbrough In discret

version

wavele

a

where respect

wavele

space o

(s

tCab

where inner obtainecommo

paramewith j,wavele

JOURNAL

11 IIJ

he researchersssfully implement signal anal

processing apet analysis is s

ms of a set of bds the functioometric polyno

form of transthe mother waavelet functionds by employind frequency. ented on transiis is used imps. The wavel

mposition of thncy channels. from a cont

mpositions reprs the use of then signal processsing non-statioerpreted as a dindependent fr

ishing momentmption from tht us to this sol

both formste), the signal i

ns tab of

et:

a

tab

1

a and b aretively, with a

et transform (

of the square in

,)(,)(

)(

, tt

tst

ba

the symbol * product. The ed by discretizon form, the

eters a and b b,kZ. By subsets:

s in signal mented in varilysis, commun

pplications. Inset up within tbasic functionson in terms ofomials in Fouslation and di

avelet. ns overcome ing analyzing It is a finite e

ient signals. In ortantly in pro

let transform e original signIt is essentiallytinuous signaresented at di

e mother wavelssing, wavelet onary signals. decomposition frequency compt localized bothe study donlution. of wavelet

is decomposed

a single functi

a

bt

e the scale aa, b and

(CWT) of a s

ntegrable funct

1

a

bt

a

mean complediscrete wav

zing the paramDWT employ

based on powerstituting in Eq

DAUD & S

processing. Iious applicationication signalnstead of Fourthe expansion s. Different frof wavelets rath

urier. It generailation of a fix

the limitationsfunctions that

energy functionn signal processocessing the no

can be internal into set of y the method o

al into seriesifferent frequelet. analysis is moThe wavelet trof the origina

positions. The th in frequenc

ne by Magos

analysis (cond into scaled an

ion t called

and translationd a ≠ 0. The

signal s(t) Ltions) is define

dtb

x conjugation velet transformmeters a and bys a dyadic sa

rs of two: a =2q. (1), we obtai

SUDIRMAN

t has been ons such as ls and other rier analysis, of functions m Fourier, it her than the

ates wavelets xed function

s of Fourier t are local in n and can be sing, wavelet on-stationary rpreted as a

independent of processing s of signal ency ranges.

ostly used in ransform can al signal into

wavelet has cy and time. so [18] has

tinuous and nd translated

d the mother

(1)

n parameters e continuous

2L (the

ed as:

(2)

and < > the m (DWT) is b. In its most ampling with

2 j ; b = k2 j , in the dyadic

Vo

Th

d j

col

ort

theeff A.

conwader

wit

and

Hebe

locdepcomit ieacmesig B.

Wrepdisis pmarecran

ol.3/No.1 (2011)

2)( 2/, t jkj

he DWT can be

(,)(

)(

,

,

tts

ts

kj

kj

By appropria

llection of fu

thonormal basi

e original sigficiently withou

Wavelet Energ

The orthonorm

ncept of eneravelet decomporived from the

th coefficient s

, jkj dE

d the overall en

2

,k

kjEM

ence, the total eobtained as

totE Energy coeffi

calization andpending on thmpare the diffeis necessary to ch scale due ethods have bgnals.

Wavelet ScalogWavelet scalopresent the timscrete wavelet plotted as a filagnitude of thctangle are relnge for this c

)2(2 ktj

e written as

,)

2(2 2/j

ately selecting

functions { is for 2Lgnal to be rut any redunda

gy Distribution

mality of the se

rgy within theosition to be Fourier theory

series d kj , is g

2

,kj

nergy at resolu

2

,

1

0kjd

j

energy associat

12

01j

k

M

j

djM

ficients as compd density ovehe scale. Thererent movemencompensate foto the down

been applied

gram ograms (refer me frequency transform. In tled rectangle we coefficient. ated to the timcoefficient. Co

IN

).

)dtktj

the mother w

j,k(t) j,k 2

. The correlate

reconstructed ancy.

n

et { kj , (t)j,ke framework linked with thy. The energy s

given by

ution j is

ted with the en

.2

,k

puted by Eq. (7er different frefore, in ordent of energy at or the halved tin sampling opto the analys

to Figure 6 localization p

this plot each dwhose color coThe location

me interval anoefficients at

NTERNETWORK

ISSN: 194

(3)

(4)

wavelet )(t th

Z} forms a

ed DWT allow

accurately an

Z} allows th

of the discrehe usual notionseries associate

(5)

(6)

ntire signal can

(7)

7) have differefrequency baner to study andifferent scale

ime resolution peration. Theis of the EO

and Figure property of thdetail coefficieorresponds to th

and size of thnd the frequenc

low levels a

KING INDONE

42-9703 / © 201

he

an

ws

nd

he

ete ns ed

ent nd nd es, at se

OG

7) he

ent he he cy

are

plottedlocalizfrequen In plottedrangesthe recThe baenergythe sqscalogrdata is

The syNeurofand thsystemIndepebut as to meathe hochosento othbelow This pand he

ESIA JOURNAL

11 IIJ

d as wide andze a wide tincies in the dacontrast, recta

d thin and tall s but large freqctangles grow ar shown on thy for each scalquares of the rams reveal tha

s captured in th

III. PROPO

ystem setting infax EEG-9100

he sampling intm is used to rendent measureboth eyes mov

asure the verticorizontal motion as their half her types such

shows the ovrocess was don

ence get better

Fig. 3. Flo

L

d short rectanime interval ata. angles for coefto indicate tha

quency ranges as a power o

he scalogram ple. This energydetail coeffic

at most of the he details coeff

OSED WORK AND

ncludes the EE0 software [19terval is 1ms. record EOG sements can beve in the verticcal motion of oon of both eyecell potential

h as silicon ruverall workingne in a quiet rorecorded signa

ow works of EOG

ngles to indicabut a narrow

fficients at higat they localizin the data. Th

of 2 as the levplot indicates y is defined ascient for eachenergy of the s

ficient.

D METHODOLO

EG data acquisi] with EOG eThe EEG data

signals from te obtained fromcal direction, itonly one eye tes. Ag/AgCl elis closer to zer

ubber electrodegs for the propoom to minimals as shown in

G signal processing

5

ate that they w range of

gh levels are e small time he heights of

vel increases. the range of s the sum of h scale. The signals in the

GY

ition system; lectrodes set a acquisition the subjects. m both eyes, t is sufficient together with lectrodes are ro compared es. Figure 3 posed study.

mize the noise n Figure 4.

g

6

insThupwfousubforsinrecfro

or subor unwby toomobas

Subjects werestructor who ghe instruction ward, downwaur successive bject repeated r 20 to 30 seconce the subjeccording duratioom artefacts ob

Figure 4: Vedi

The eyeball mstatic eye b

bjects were alsany disturban

wanted artefacusing wavel

olbox applicatiovement of EOsic characterist

e seated on a ave instruction

composed oard, left and rigeye movemenfor three times

onds for each ects were showon was reducedbserved in long

rtical (pink) and Hsplayed from data

moved to the deecomes the rso asked to avonces during thcts. EOG signaet analysis frion. Figure 5

OG signals by tic of EOG pot

INTER

chair and suns on how to mof four moveght. The recordnts for 10 subs. Initially, EO

eyes movementwing signs ofd to 10 secondser traces by vis

Horizontal (green) acquisition system

esired directionreference poinoid blinking, bhe recording tal captured warom MATLABis the real sigusing MATLAtentials.

RNETWORKING

ISSN: 194

upervised by amove their eyements that a

ding was done bjects and eac

OG was recordet. Unfortunatelf tiredness, ths which was fresual inspection

EOG signal m

n and the centnt. Furthermorbody movemento minimize thas then analyseB software angnal of the rigAB. It shows th

G INDONESIA

42-9703 / © 201

an es. are in ch ed ly, he ee

n.

tre re, nts he ed nd

ght he

It wasscalogrcoefficdecomorder 4becauscompodepend

Whetransfosignal of timeorder waveleused bpropercoefficcorrespthe scalevel. Tdetailsthe scasubject

Am

plitu

d (u

V)

JOURNAL

11 IIJ

Figure 5: Righ

s then uploaderam in order cient energy in

mposed down t4 (db4) as a mse it has two vaonent. The nud on the sampl

IV. R

en it comes toorm, we are inenergy of wav

e. Hence, we pto identify th

et energy coefbecause it reprty of waveletcient is plottepond to the malogram plot This energy is

s coefficient foalogram of sigt; left, right, do

0 100 200-400

-300

-200

-100

0

100

200

p(

)

ht EOG signal plo

ed into a progto present th

n scale and spato 10 level of

mother waveletanishing momeumbers of leve rate of origin

RESULTS AND D

o the time freqnterested in knvelet details c

plotted a scaloghe dominant sfficient for thepresents the timt transform. Ied as a filled

magnitude of thindicates the defined as the

or each scale. Fgnals of four diownward and u

300 400 500Time (ms)

DAUD & S

otted using MATLA

gram that runhe signal inace or time. Tf details using . The db4 has ent, i.e. constavel decomposinal signal recor

DISCUSSION

quency analysinowing the dioefficient with

gram for each mscales over the signal. The sme frequency In this plot, d rectangle whe coefficient.range of ener

e sum of the sqFigure 6 and Fiifferent movemupward.

600 700 800 90

Vertical EOG

Horizontal EO

SUDIRMAN

AB

ns a wavelet the wavelet

The signals is Daubechies been chosen

ant and linear ition strictly rded.

is in wavelet stribution of h the change movement in

he maximum scalogram is

localization each details

whose colors The bar on rgy for each quares of the igure 7 show ments from a

00 1000

OG

Vo

ol.3/No.1 (2011)

i) Lef

i) Upwa

IN

ft eye Movemen

ard eye Movem

NTERNETWORK

ISSN: 194

nt

Fig. 6. Scalogram

ment

Fig. 7. Scalogram

KING INDONE

42-9703 / © 201

m of horizontal eye

m of vertical eye m

ESIA JOURNAL

11 IIJ

ii) Rig

e movements

ii) Down

movements

L

ght eye Movem

nward eye Mo

ment

vement

7

8

capdifdetof consam

Aby domnumcalis nis diffre

modomupwsum

Dcoesigdif

The scalogramptured in thefferent EOG stails move fromwavelet coeffintent being hampling rate whAbout 15 data

using wavelminant energymber of extralculated and plnoticed that thslightly differe

fferent eye mequency bands.The percentagovements is minant energyward; scale 8 mmarized in TDominant enefficient energ

gnal. We use thfferent moveme

ms reveal that he different scasignals. Frequm high frequenficient increasealved at each ihich is 1000 Hza for each eye et scalogram y details coeffacted detail colotted as showne average percent from their movements a. ge of each dillustrated in

y is: scale 6 forfor right and sable II.

nergy level mgy that can be his parameter aent of EOG sig

Fig. 8: Ave

INTER

highest energyale of details

uency componncies to low frees from 1 to 10increment in az (see Table I).movement havin order to e

ficient and it oefficients fron in Figure 8.

centage for eacenergy level.

are associated

detail coefficieFigure 8.

r left eye movescale 9 for dow

means the mderived by sc

as the benchmagnals.

rage percentage

RNETWORKIN

ISSN: 194

y of the signalss coefficient ent extracted

equencies as sc0, with frequenaccordance to ve been analyzextract the mfrequencies. T

om each levelFrom Figure 8

ch movement dThis means t

d with differ

ent of four eStatistically,

ement; scale 7 wnward. They

maximum detaalogram for eaark to classify

s of energy level

NG INDONESIA

42-9703 / © 201

s is for by

cale ncy the

zed most The l is 8, it data that rent

eye the for are

ails ach the

EOSign

Lef

Righ

Up

Dow

l for 15 data of v

A JOURNAL

11 IIJ

SCALE

1

2

3

4

5

6

7

8

9

10 Table 1: Freque

G als

DominaEnergy S

ft Scale

ht Scale

p Scale 7

wn Scale 9Table 2: Dom

vertical and hori

FRRA

3

1

ency Content 10

ant Scale

AvePercen

6 8

8 9

7 9

9 9minant energy lev

izontal eye move

DAUD &

REQUENCY ANGE (Hz)

250-500

125-250

62.5-125

31.25-62.5

15.6-31.3

7.8-15.6

3.9-7.8

1.9-3.9

0.9-1.9

0.5-1.0 Level Decompo

erage ntage (%)

EFr

80.0 8

98.2 2

99.2

95.4 vel for 15 EOG D

ements

& SUDIRMAN

osition

Estimated requency

8-16 Hz

2-4 Hz

4-8 Hz

1-2 Hz Data

Vo

basdecthenotfreresbanintscahasdeswh

ThthaEleTethi

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10

[11

[12

ol.3/No.1 (2011)

V. CONCLU

This study hased on its estcomposition. Tey only focuset specifically i

equency bandssearcher to loonds distributionterpretation ofalogram. Resuls different fresign a supportheelchair by us

he authors are ankfulness toectronics Res

eknologi Malayis paper.

J. Malmivuo, Application of BUniversity PresK. Shinomiya, characteristics ooculogram. TheF. Jagla, M. JeRelated PotentCenter of ExceSciences, BratisR.B Knapp, H.Reality ApplicDisabilities. D. Kumar, E. PInterface, ProcHouston, TX, UG. Aysegul, K. using artificial n205. R. Sudirman, WPotentials UsingInternational C2009. 131-135. A. Bhandari, VNovel TechniquAnnual Indian CR. K. Armen,Electrooculografeature classificNew York City

0] N. Hazarika, J.EEG Signals Processing. 51,

] D. Lee, A. Applications, HCopyRight, 44-

2] I. Daubechies,Wavelets, Comm

USION AND FUT

as classified timated frequeThis is differened on the overinto the details. Hence, this pok into detailsn within four ef EOG signallt obtained indequency bandt machine for ping eye movem

ACKNOWL

so indebted ao the Biomearch Group

ysia for their id

REFER

R. Plonsey, BBioelectric and Bs, New York, 1995N. Itsuki, M. Ku

of potential and ce Journal of Medicergelova, I. Riecaials. Institute of llence for Cardiovslava, Slovakia. . Lusted, 1993. Bcations, Proc.

Poole, 2002. Classceedings of the SeUSA, October 23-2

Sadik, 2006. Clasneural network. Ex

W.M. Bukhari, 2g Time Frequency

Conference on App

V. Khare, M. Trikue for Signal CondConference INDIC, P. G. Joseph, am based system cation model. EM, USA, Aug 30-Se Z. Chen, C. TsoUsing the Wave Issue 1. 61-72. Yamamoto, 199Hewlett Packard 52. 1988. Orthonommunication on Pur

INT

TURE RECOMME

each EOG siency using want to previous rall wavelet des of each EOGpaper proposes of the energeye movement ls analysis bydicates that eacs and could paralyzed peopments.

LEDGMENT

and would likmedical Instru

(BMIE) at deas and comm

RENCES

BioelectromagnetisBiomagnetic Fields5,ch. 28. ubo, H. Shiota, 20cross-talk at each cal Investigation. Vansky, 2007. SaccNormal and Path

vascular Research

Biological Signal pVirtual Reality

ification of EOG econd Joint EMB26. ssification of electxpert System with

009. Classificatioy Analysis, Proceeplied Simulation a

kha, A. Anand. 2ditioning of ElectroCON.

Y. Minerva, E.for computer con

MBS Annual Interept 3. i, A. Sergejew, 19elet Transform. I

4. Wavelet AnaJournal. Hewle

mal Bases of Cre and Applied Ma

TERNETWORK

ISSN: 194

ENDATIONS

ignals movemavelet scalogrstudies in wh

ecomposition aG signals of eaes and targets gy and frequen

signals for bety using wavech eye movembe integrated ple to move th

ke to express oumentation athe Univer

ments to impro

sm: Principles s. New York: Oxf

008. Analyses of electrode in Elec

Vol. 55. 120-125.cadic Eye Movemhological Physiolo, Slovak Academy

processing in Virand Persons w

for Human CompBS/BMES Conferen

tro-oculogram signApplications 31, 1

on of Eye Movemedings of the IASTand Modelling, A

2006. Wavelet Bao-Oculogram Sign

K. Kristina, 20ntrol using a multrnational Conferen

977. ClassificationIEEE Trans. Sig

alysis: Theory At Packard Comp

Compactly Suppoath. 41, 909-996.

KING INDONES

42-9703 / © 201

ment ram hich and ach the ncy tter elet

ment to

heir

our and rsiti ove

and ford

f the ctro-

ment ogy, y of

rtual with

uter nce,

nals 199-

ment TED

ASM

ased nals.

006. tiple nce,

n of gnal

And pany

rted

[13] S. Gthe 674

[14] AugFreq

[15] I. Dloca100

[16] J. JTechColu

[17] AugFreq

[18] E. MEneand

[19] R. S“ EENetwColl

SIA JOURNAL

11 IIJ

G. Mallat, 1989. Awavelet represent-693.

gust W. Rihaczekquency. IEEE TranDaubechies, 199alization and signa5. . Carr, J. M. Brohnology, fourth edumbus Ohio.

gust W. Rihaczekquency. IEEE TranMagosso, M. Ursergetic Approach f

EOG”, Applied MSudirman, A.C. KoEG Different Freqwork and Fuzzyloquium on Signal

Wanhis MUnivinterTranComwmb

RubiM.ScUniveElectrmedicproceat rub

A theory for multitation, IEEE Trans

k, 1968. Signal Ensactions on Infor0. The Waveletal analysis, IEEE

own, 2001. Introdd. Prentice Hall. U

k, 1968. Signal Ensactions on Inforino, A. Zaniboni,

for the BiomedicalMathematics and Coh, N.M. Safri, W

quency Sound Resy Techniques” l Processing and i

n Mohd Bukhari Master of Electriversiti Teknologi rests are Signal pnsforms, Biomedicmputer Interface.bukhari41@gmail.

ita Sudirman recc. (1996) from Unersiti Teknologi trical Engineeringcal electronics essing and speech bita@fke.utm.my.

iresolution signal s. Pattern Anal. M

Energy Distributiormation Theory. 36t Transform. TimTrans. Inform. T

duction to BioMedUpper Saddle Riv

Energy Distributiormation Theory. 36, E. Gardella “A l Signals: Applica

Computation 207 (2W.B. Daud, N.H. Msponse IdentificatioProc. IEEE 6th

its Application (CS

Wan Daud is curical Engineering p

Malaysia Skudaiprocessing especiacal Instrumentatio He can becom.

ceived both the Bniversity of Tulsa

Malaysia, all dg. Her current inengineering, biomrecognition. She

9

decomposition: Mach. Intell. 11,

on in Time and 69-374. me –frequency

Theory. 41, 961-

dical Equipment ver, New Jersey

on in Time and 69-374. Wavelet Based

ation to the EEG 2009) 42-62.

Mahmood, 2010. on using Neural . International

SPA 2010).

rrently pursuing program at the i. His research ally in Wavelet on and Human e reached at

B.S. (1994) and and Ph.D. from degrees are in nterest includes medical signal can be reached

10 INTERNETWORKING INDONESIA JOURNAL Vol. 3/No. 1 (2011)

ISSN: 1942-9703 / © 2011 IIJ

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 11

ISSN: 1942-9703 / © 2011 IIJ

Abstract—Software testing involves appropriate validation and verification of a software component developed during the software lifecycle. Usually testing costs often account to high budget in the software development process. In order to minimize the testing costs, researchers and practitioners automate the testing process rather than carry out manual testing. Test Case Generation is the process of automatically generating a collection of test cases which are applied to a system under test. This paper utilizes branch coverage criteria using the Generalized Optimization Meta heuristic (GOM) algorithm and code constraint graph (CCG) to efficiently maximize the coverage of all the branches. The experimental results show that the proposed test generation technique is effective in generating tests for an application at large. Index Terms—Test case generation, branch coverage, evolutionary algorithm, Code Constraint Graph

I. INTRODUCTION

Testing is the process of exercising a software component using a selected set of test cases, with the intent of revealing defects. Testers need to detect these defects before the software becomes operational. Automating the testing process is a relevant issue since it will help reduce analysis costs by enabling a more systematic approach to testing [1]. A good test case is one that has a high probability of revealing a yet undetected defect. It requires the tester to consider the goal for each test case, that is, which specific type of defect is to be detected by the test case. Test Case Generation (TCG) is the process of automatically generating a collection of test cases which are applied to a system under test [28]. White-box TCG is usually performed by means of symbolic execution, i.e., instead of executing the program on normal values (e.g., numbers), the program is executed on symbolic values representing arbitrary values [9]. Test cases should be developed for both valid and invalid input conditions. That is, a tester must not assume that the software under test will always be provided with valid inputs. Inputs may be incorrect for several reasons. For example, software

Selvakumar. S. is with the Department of Information Technology, Thiagarajar College of Engineering, Madurai, India (Mobile: +91-9789916648; e-mail: sselvakumar@yahoo.com). Ramaraj N is with the Department of Computer Science &Engineering, G.K.M College of Engineering & Technology, Chennai, India (e-mail: prof.ramaraj@yahoo.in).

users may have misunderstandings, or lack of information about the nature of the inputs. A test case must contain the expected output or result and the results of the tests should be inspected meticulously. Branch coverage testing criterion encounters all the branches in a program i.e., the predicate of an ‘if’ statement should be evaluated to both true and false. The stronger criteria of condition, multiple-condition and path coverage are often infeasible to achieve for programs of more than moderate complexity, and thus branch coverage has been recognized as the basic measure for testing. A small number of test-data techniques have already been automated: random, static and dynamic, analysis-oriented, goal-oriented and structural or path-oriented test-data generators. Random generator is the simplest method of generation techniques, creates large amounts of test data; it could actually be used to generate input values for any type of program. Ultimately, a data type such as integer, string, or heap is just a stream of bits. However, because no information exists about the testing objectives, the generators often fail to find data that satisfy the stated objectives of the testing process. Since it merely relies on probability it has quite low chances in finding semantically small faults, and thus accomplishing high coverage. A semantically small fault is such a fault that is only revealed by a small percentage of the program input. Static and dynamic generators execute a program symbolically by means of variable substitution techniques instead of actual values. This technique requires plenty of computer resources. It also puts a lot of restrictions on the program. Symbolic execution also implies that a symbolic evaluator for the particular language is built which indeed requires a great amount of work. XML is now being used to replace large relational databases. Therefore, performance testing of XQuery implementations on very large documents is important [4]. Analysis-oriented generators have the ability to generate high quality test-data, but rely upon their designer with a great insight into the domain of operation, and hence are not readily extrapolate to arbitrary software systems. Goal-oriented generators provide guidance towards certain set of paths. Instead of letting the generator generate input that traverses from the entry to the exit of a program, it generates input that traverses a given unspecific path [11]. Because of this, it is sufficient for the generator to find input for any path. Two methods using this technique have been found: the chaining approach and the assertion-oriented approach. The latter is an interesting extension of the chaining approach. Typical for the chaining approach is the use of data dependence to find solutions to branch predicates. The characteristic of chaining is to identify a chain of nodes that are vital to the execution of

Test Case Generation using GOM Algorithm Selvakumar Subramanian and Ramaraj Natarajan

12 INTERNETWORKING INDONESIA JOURNAL SELVAKUMAR & RAMARAJ

ISSN: 1942-9703 / © 2011 IIJ

the goal node. This chain is built up iteratively during execution. Since this method uses the find-any-path concept it is hard to predict the coverage given as a set of goals. Assertion-oriented testing truly utilizes the power of goal-oriented generation. Certain conditions, called assertions are automatically inserted in the code. When an assertion is executed it is supposed to hold, otherwise there is an error either in the program or in the assertion. But they have serious problems associated with failing to find the global minima. The search space tends to ‘lack features’ and consists of large ‘flat’ areas which provide no information on the location or the direction of the true local minima. Although a number of different goal-oriented approaches and algorithms exist, it is difficult to judge exactly which approach represents the current state of the art. Structural or Path-oriented generation identifies the path for which the test data is to be generated. Unfortunately, if the path is infeasible that would cause the generator to fail to find an input that will traverse the path. Even though it has the merit of very thoroughly testing a specific path, it has two severe disadvantages. The first is that the number of paths is exponential to the number of branches. The second is that many paths are impossible to exercise due to relationships between the data. Branch coverage criterion measures which decision outcomes of an ‘if’ statement have been tested. Determining the number of branches in a method is also easy. The total number of decision outcomes in a method is hence equal to the entry branch in the method plus the number of branches that need to be covered. The rest of this paper is organized as follows. Section 2 discusses research related to the related work. Section 3 presents the details of the proposed approach. Section 4 describes an experimental study of the proposed criterion and observations. Section 5 presents conclusions and future work.

II. RELATED WORK

The work of M.F Bashir, and S.H.K. Banuri, [1]

extends the paradigm of the test data generation system to incorporate both specifications and model based testing which helps to perform the reclassification of the code, specification or model based techniques. Several attempts have been made to develop a system to generate test data automatically. The existing such system does not guarantee to generate test data in only feasible paths. Praveen Ranjan Srivastava et al. [2] proposed a method to generate feasible test data, using Genetic Algorithm. It is often desired that test data in the form of test sequences within a test suite can be automatically generated to achieve required test coverage. The work of Sushil K. Prasad et al. [3] proposes Genetic algorithm to test data generation for optimizing software testing. Ana Barbosa et al. [5] proposed a test case generation approach to model-based testing of graphical user interfaces from task models. [5] shows how task mutations can be generated automatically, enabling a broader range of user behaviors to be considered. More recently, Grammar-based test generation has been applied to many other testing problems, including the generation of eXtensible Markup Language (XML)

documents and the generation of packets for testing communications protocols [6].

Recent research has shown how to integrate covering-array techniques such as pairwise testing into Grammar-based test generation tools [6]. Their work proposed two case studies showing how to use grammars and covering arrays for automated software testing. Valentin Dallmeier et al. [7] combined systematic test case generation and typestate mining, static typestate verifier fed with enriched models report significantly more true positives, and significantly fewer false positives than the earlier proposed models. [8] proposed an automatic test generation solution using dynamic symbolic execution, uses distance in control-dependency graph to guide path exploration towards the change. [8] is effective in generating change-exposing inputs for real-world programs. [9] proposed a symbolic execution mechanism, by developing a fully Constraint Logic Programming based framework for test case generation of an OO imperative language. Rafael Caballero et al. [10] presented a general framework for generating SQL query test cases using Constraint Logic Programming. [11] presented an automated approach to generate unit tests that detect these mutations for object-oriented classes, the resulting test suite is optimized towards finding defects rather than covering code and the state change caused by mutations induces oracles that precisely detect the mutants. [12] proposed an approach for automated test case generation based on techniques from constraint programming. [13] proposed a scalable toolset using Alloy to automatically generate test cases satisfying T-wise from SPL models. [13] defines strategies to split T-wise combinations into solvable subsets.

III. OVERVIEW OF THE APPROACH

A. Framework

The block diagram of this proposed approach is depicted as in Fig.1. It consists of a three-tier architecture containing the following four blocks: Source code analyzer, XML parser, Constraint analyzer and Test data generator. Initially, a sample code consisting of only ‘if-else’ constraints is taken as input. As the name ‘constraint’ specifies, the input code that is to be tested should never contain or be allowed any loops such as ‘while’ or ‘for’ loops. The source code analyzer block analyses the input code and generates an XML document which separates the constraints and their outcomes and also neglecting the statements if any, present. The XML parser block initiates the XML document and creates a notepad file containing the constraints similar to those present in the XML document, given as the input file to the algorithm employed i.e., the evolutionary meta-heuristic algorithm. The primary portion of the third block which is the constraint analyzer takes the notepad file generated earlier as input and generates a graph called the Code Constraint Graph (CCG). This graph indicates the program flow of the source code as to which branches are present and which are to be tested. The CCG is no longer used because it just shows the control flow of the input source code. The secondary portion of the third

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 13

ISSN: 1942-9703 / © 2011 IIJ

block called the test data generator block implements the evolutionary algorithm namely the Generalized Optimization Meta-Heuristic (GOM) algorithm, which takes the notepad file to be checked against the constraints separated. The test data generator block first generates a random set of integers which may range from positive to negative. The set may contain an approximate number of 40-50 random integers. For each of the random set of integers, a fitness formula is evaluated in order to advance to the next population. After calculating the fitness for each of the random set of integers, fitness values are assigned to them. The chromosome with the optimal fitness is chosen as the base chromosome for the next generation of members. Each chromosome contains 24 bits. So the 24 members of the next population are generated from the base chromosome by flipping each bit of it. Those newly generated members are then calculated by the given fitness functions. Then the chromosomes are ranked according to their fitness values, from worst to the best. The optimal chromosomes of the next and the previous population are compared. The population that has a comparatively lesser optimal chromosome is deleted and the other one is kept as the base chromosome for the generation of the members of the next population. Cross over and mutation operations are applied to them if necessary.

Fig. 1. Proposed Framework for test case generation

B. Motivational Example

Consider the triangle classification program in Fig. 2, which accepts three variables say A, B and C each for the three sides of a triangle. The triangle classification accepts three integers as the three sides of a triangle, and decides which type of triangle it based upon the length of these three slides. The four possible results are: scalene, isosceles, equilateral and not a triangle. If all the three sides are equal, the program returns an equilateral triangle. If any two sides are equal, the program returns an isosceles triangle. If none

are equal, then the program returns a scalene triangle. So in this algorithm the above procedure of ranking and cross-over mutation operations are repeated three times, i.e., each for three sides of a triangle A, B, C. The generated members of the first population contain the chromosomes each with three genes that indicate the three variables used in that program. Every chromosome is then passed to the notepad file and checked against the constraints. When the chromosomes passed satisfy the particular constraints, a counter variable is incremented so as to count the number of branches satisfied by them. This procedure is repeated for each chromosome of the first population. After completing all the members of the first population, the same operations are performed for each pair of genes and then by applying cross over and mutation operations. The above procedure is repeated till the maximum number of branches is satisfied by the branch coverage criterion. The GOM algorithm generates three genes in each chromosome, as the number of variables involved in the source code is three. Before the generation of the appropriate genes in a chromosome, the program module is first parsed and all the ‘if-else’ branches involved are separated in a notepad file in order to easily give input to the GOM algorithm.

Fig. 2. Module of triangle classification.

C. Formulation of branch coverage

The proposed approach in this paper utilizes branch coverage criteria using a Generalized Optimization Meta heuristic (GOM) algorithm and code constraint graph (CCG). The process of generating test cases using GOM algorithm and CCG graphs is as follows; assume the input that is to be checked against the constraints in the source code to range I as: {I1, I2,…,In,}. Define and identify the test constraint set C as: {C1, C2,…,Cm}. For each and every input I, the test cases (Tc) are generated so that those inputs must satisfy the appropriate constraints encountered in the constraint set and this process is repeated until the maximum branch coverage is attained as output for the given input set. The amount of branch coverage (Tbc) criterion can be expressed mathematically as:

bc FITNESS = 100 – Tbc

Where BC - number of branches covered SC - number of statements covered MC - number of methods covered

14 INTERNETWORKING INDONESIA JOURNAL SELVAKUMAR & RAMARAJ

ISSN: 1942-9703 / © 2011 IIJ

B – Total number of branches S – total number of statements M – total number of methods The steps to be employed are as follows: A sample input code containing ‘if-else’ constraints is

taken as input. The constraints present in the source code are

separated by reading line by line and the unnecessary statements, header files, braces and new lines are removed.

The GOM algorithm is employed after generating a random set of integers say, up to 40-50 iterations.

The test cases are generated by the GOM algorithm indicating which branches are covered by appropriate chromosomes, each containing three genes.

The percentage of branch coverage by the chromosomes is obtained by applying the relevant formulas regarding the branch coverage.

The highest percentage of coverage by the chromosomes is returned as the best solution.

D. Proposed pseudo code of the algorithm

Fig. 3 depicts the pseudo code of a Generalized Optimization Meta-Heuristic (GOM)

Procedure TDGen Input: Program: code/program under test Output: CCG: A code constraint graph, which shows the control flow of the source code. Test cases: A set of test cases that is generated using GOM algorithm. begin

1. A sample code consisting of “if-else” conditions is taken as input.

2. The source code analyzer generates an XML document by analyzing the constraints in the source code and separates them in that document.

3. The XML parser goes through the XML document created and creates a notepad file containing only constraints so as to easily give as input against the GOM algorithm.

4. The constraint analyzer accepts the notepad as input and generates a tree called Code Constraint Graph (CCG) so that it shows a program flow of the source code with constraints as which branches are present and that are to be tested.

5. For each entry of the requirements. Initialize randomly the population of S species (bits) that are used to encode D program variables. loop

For each of the S bits of the species, find the fitness value.

Find the base chromosome that has the best fitness among the generated members.

Generate first population using bit changes to the base chromosome, as much times as its bits.

Evaluate the first population against the fitness function to find the fitness value, and assign to each of the members.

Perform ranking operation to rank the members according to their fitness value.

Apply either “one-point” or” two point” cross over operation by taking either best and its successor or worst and its predecessor.

Modify a single bit of the gene with probability P α k-µ (where k is the rank of the bit and µ, a free control parameter)

Compare the first population and next population fitness values. Discard the population which has worst base chromosome.

Repeat the above procedure for next population.

Update branch coverage criteria, test cases and iteration counter.

until stopping criterion is met 6. Return the branch coverage result and test cases.

end Fig. 3 Generalized Optimization Meta-Heuristic (GOM) pseudo code

Separation of constraints from the source code: The source code itself cannot be tested since it has irrelevant codes such as printing statements, other logical statements, etc. So it is necessary to separate the constraints alone from the source code, in this case the branches with their appropriate outcomes i.e., children. The methodology used to separate the branches alone from the source code consists of reading each and every line of the source code until a branch or constraint is encountered. A better way to implement this methodology is to read the source code and place them in an XML document. As seen in Fig. 1, the source code analyzer block performs this operation. As the branches are separated in that XML document, the unnecessary new lines, braces and other header files are removed so that only the constraints are filtered. The generated XML document is parsed and the appropriate operations are performed so that the unnecessary codes are removed. As seen in Fig.1, the XML parser block performs this operation. The filtered branches are placed in a notepad file, given as input to the GOM algorithm. In the notepad, an @ symbol is referenced before each branch so that a constraint is encountered. If any children’s are present in a branch, the @ symbol is incremented and if the line comes outside the ‘if-else’ branch, the @ symbol is appropriately decremented. The generated filtered branches and/or constraints from the source code are visualized as in Fig. 4 as:

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 15

ISSN: 1942-9703 / © 2011 IIJ

Fig. 4 Sample constraint generation

Generation of Code Constraint Graph (CCG): The Code Constraint Graph can be generated from the notepad file created as in Fig.3 to show the control flow of the source code consisting of constraints. As in fig. 1, the constraint analyzer block performs this operation of generating the CCG. The sample CCG generated from the triangle classification module can be visualized as in fig. 5 as shown below:

Fig. 5. CCG generation

Generating test cases using proposed evolutionary algorithm: The proposed Generalized Optimization Meta Heuristic (GOM) algorithm is a type of an evolutionary algorithm (EA) that makes use of evolutionary [16, 17] strategies (ES) and evolutionary programming (EP). The main idea behind the EAs is to evolve a population of individuals (candidate solutions for the problem) through competition, mating and mutation [18], so that the average quality of the population is systematically increased in the direction of the solution of the problem at hand. The evolutionary process of the candidate solutions is stochastic and “guided” by the setting of adjustable parameters. In an analogy with a natural ecosystem, in a EA different organisms (solutions) coexist and compete. The more adapted to the design space will be more prone to reproduce and generate descendants. On the other hand, the worst individuals will have fewer or no offspring. In an optimization problem, the fitness [19, 20] of each

individual is proportional to the value of the objective (cost) function, also called fitness function.

In a GOM algorithm, each bit is considered a species and a string of S bits is taken as the initial population of the species. The string consisting of S bits then encodes the D program variables to be represented in a binary format of 0’s and 1’s. In a variation of the canonical GOM described above, the bits are ranked separately for each substring that encodes each program variable, and N bits, one for each variable are flipped at each iteration of the algorithm. First, the GOM algorithm generates the random integers of up to 40-50 numbers. From the generated random numbers, each and every two pair of integers is taken into account. The second number of the pair is taken and the bits of that number are ranked according to their priority. The highest and the lowest bits are taken as an average to obtain a result. In the first number of the pair the shifting operation is performed, as many times as the result of averaging. The same procedure is repeated for the second number and the modified first and second numbers are kept aside. Then these numbers undergo appropriate cross over and mutation operations. The cross over method used here is the “two point” cross over.

The mutation operation is then performed with respect to the probability P α k-µ, where k is the rank of the bit and µ, a free control parameter. These optimized integers are then checked against the notepad file generated earlier. When the chromosomes satisfy a particular branch, a counter variable is incremented which indicates the number of branches that are satisfied by a single chromosome. In this case, a single chromosome consists of three genes since the number of variables encountered in the triangle classification is three namely A, B, C. The above procedure is repeated till the chromosomes in all populations by means of branch coverage criteria cover a maximum number of branches. The general representation of two-point cross over can be represented as: The two point crossover operator takes two vectors (a1,… , an) and (b1,… , bn) and yields two vectors (a1,… , ai, bi+1,… , bj, aj+1, ……, an) and (b1,… , bi, ai+1,… , aj, bj+1, ……, bn), where 1≤ i < j ≤ n-1 and i and j are randomly chosen. This means that both vectors are split at the same two positions and assembled with swapped middle parts. An example of a mutation operation performed is, Before: 1 1 0 1 1 0 1 0 0 1 1 0 1 1 1 0, After: 1 1 0 1 1 0 0 0 0 1 1 0 1 1 1 0, a change of bit in the gene takes place at bit position 6.

IV. THE EXPERIMENT

A. Subject Programs, Faulty Versions, and Test Case Pools

To examine the efficacy of our approach, the proposed approach was evaluated using real-world programs. In this section, we report the empirical evaluation results. We compared the GOM Coverage with the GA Coverage. In the experiments, the Siemens suite programs (Table 1), similar to those used by Dawei Qi et al. [8] and Rothermel et al. [15] were used to validate the performance of the proposed approach. Each program was hand-instrumented to record all the coverage information. Each program has a variety of

16 INTERNETWORKING INDONESIA JOURNAL SELVAKUMAR & RAMARAJ

ISSN: 1942-9703 / © 2011 IIJ

versions, each containing one fault. Each program also has a large universe of inputs. We obtained the subject programs from the Software-artifact Infrastructure Repository at UNL [14].

TABLE I SIEMENS SUITE SUBJECT PROGRAMS

Name Lines of code

Faulty version count

Test pool size

Program

Description

tcas 162 41 1608 Altitude separation

totinfo 346 23 1052 Information measure

schedule 299 9 2650 Priority scheduler

schedule2 287 10 2710 Priority scheduler

printtokens 378 7 4130 Lexical analyzer

printtokens2 366 10 4115 Lexical analyzer

replace 514 32 5542 Pattern replacement

Space 9127 38 13,585 Array definition language interpreter

B. Experimental Results and Observations

To examine the efficacy of our approach, we evaluated our approach using real-world programs. In this section, we report our empirical evaluation results. The obtained results of branch coverage criteria can be depicted by a graph as in Fig. 6 showing the convergence of coverage. The number of iterations is scaled along the X-axis and the percentage of branch coverage is scaled along the Y-axis. As compared to the simple genetic algorithm, GOM algorithm converges faster in less number of iterations. The maximum branch coverage obtained by applying the proposed algorithm is nearly 71%. As seen in Fig. 6, the applied GOM algorithm converges at a faster rate than the simple genetic algorithm i.e., at iteration 70 (number 7), the GOM reaches the maximum branch coverage of 71% and it is consistent in the further numbers of iterations, whereas the genetic algorithm reaches the maximum branch coverage of 64% only at iteration 100 (number 10).

Branch coverage

0

10

20

30

40

50

60

70

80

1 2 3 4 5 6 7 8 9 10 11 12

Number of iterations

Per

cen

tag

e o

f b

ran

ch

cove

rag

e

GOM cov

GA cov

Fig. 6. Comparison of branch coverage of GA and GOM

Fig. 5. Branch coverage of SIR objects

Fig. 7 shows the Branch coverage of the SIR objects. In terms of numbers, the vast majority of all test cases have at least one assertion. Although the achieved score is quite high, the search based approach offers potential for optimization. While the coverage based impact measurement guides the search towards assertions, in the experimental of SIR programs, for certain cases GOM is fair in coverage.

V. CONCLUSIONS AND FUTURE WORK

The GOM algorithm implemented gives a suitable way for automatic test case generation. The ease of test case generation is faster than with the simple genetic algorithm since the number of iterations for reaching the optimal solution is quick. The separation of constraints from the source code and then exporting them to a separate notepad file makes the implementation of this algorithm further easier. The code constraint graph (CCG) generated allows understanding the control flow of the source code to depict the amount of statements, branches and methods present and also which are covered. The cross-over and mutation operations are optional, since the algorithm has the capability to converge well without performing those operations. In terms of future work, we can extend our method by improving the fitness function to

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 17

ISSN: 1942-9703 / © 2011 IIJ

deduce a better result above the maximum amount of coverage obtained.

ACKNOWLEDGMENT

We thank Dr. Gregg Rothermel, Dept. of Computer Science, University of Nebraska, for providing the Siemens Suite of programs.

REFERENCES [1] Bashir, M.F, Banuri, S.H.K., "Automated model based software Test Data

Generation System", Proc. 4th International Conference on Emerging Technologies, pages 275-279, Oct.2008.

[2] Praveen Ranjan Srivastava, Priyanka Gupta, Yogita Arrawatia, Suman Yadav, Use of genetic algorithm in generation of feasible test data ACM SIGSOFT Software Engineering Notes, Vol. 34,Issue 2, March 2009

[3] Sushil K. Prasad, Susmi Routray, Reema Khurana and Sartaj Sahni, “Optimization of Software Testing Using Genetic Algorithm” Third International Conference Information Systems, Technology and Management, pages 350-351, March 2009

[4] Daniel Homan, David Ly-Gagnon, Paul Strooper, and Hong-Yi Wang, Grammar-Based Test Generation with YouGen, Software Practice And Experience, 1:1-20, 2009.

[5] A. Barbosa, A. Paiva and J.C. Campos, Test case generation from mutated task models, ACM Symposium on Engineering Interactive Computing Systems, 2011.

[6] Daniel Hoffman, Hong-Yi Wang, Mitch Chang, David Ly-Gagnon, Lewis Sobotkiewicz, Paul Strooper, Two case studies in grammar-based test generation, Journal of Systems and Software, Volume 83 Issue 12, December, 2010.

[7] Valentin Dallmeier, Nikolai Knopp, Christoph Mallon, Sebastian Hack, Andreas Zeller, "Generating test cases for specification mining", Proceedings of the 19th international symposium on Software testing and analysis ACM, NY, USA 2010.

[8] Dawei Qi, Abhik Roychoudhury, Zhenkai Liang, Test Generation to Expose Changes in Evolving Programs, 25th IEEE/ACM International Conference on Automated Software Engineering (ASE), September 2010.

[9] Miguel G´Omez-Zamalloa1, Elvira Albert1, Germ´ An Puebla, Test Case Generation for Object-Oriented", Imperative Languages in CLP, Journal of Theory and Practice of Logic Programming, Volume 10 Issue 4-6, July 2010.

[10] Rafael Caballero, Yolanda García-Ruiz, Fernando Sáenz-Pérez, " Applying Constraint Logic Programming to SQL Test Case Generation", Proceedings of 10th International Symposium, FLOPS 2010, Sendai, Japan, April 19-21, 2010.

[11] Gordon Fraser, Andreas Zeller, "Mutation-driven Generation of Unit Tests and Oracles" Proceedings of the 19th international ACM symposium on Software testing and analysis, NY, USA, 2010.

[12] François Degrave, Tom Schrijvers and Wim Vanhoof, "Towards a Framework for Constraint-Based Test Case Generation", Logic-Based Program Synthesis And Transformation, Lecture Notes in Computer Science, 2010, Volume 6037, 2010.

[13] Perrouin, G., Sen, S., Klein, J., Baudry, B., le Traon, Y., "Automated and Scalable T-wise Test Case Generation Strategies for Software Product Lines" , Proceedings of the 2010 Third International Conference on Software Testing, Verification and Validation, Paris, pages 459 - 468, 6-10, April 2010.

[14] Software infrastructure repository (SIR) http://www.cse.unl.edu/~galileo/sir.

[15] Gregg Rothermel,M. J. Harrold, J. Ostrin, and C. Hong. An empirical study of the effects of minimization on the fault detection capabilities of test suites. International Conference on Software Maintenance, pages 34–43, November 1998.

[16] S. Horwitz. Tool support for improving test coverage. In Proceedings of ESOP 2002: European Symposium on Programming, 2002.

[17] P. McMinn, M. Holcombe. The State Problem for Evolutionary Testing, GECCO, 2003, pp. 2488–2498.

[18] C.C. Michael, G. McGraw, M. Schatz. Generating Software Test Data by Evolution, IEEE transactions on software engineering 27 (12) (2001) 1085–1110.

[19] P. Thevenhod-Fosse, H. Waeselynck, STATEMATE: applied to statistical software testing, Proceedings of the 1998 International Symposium on Software Testing and Analysis, 1998.

[20] J.M. Voas, L. Morell, K.W. Miller, Predicting where faults can hide from testing, IEEE Software 8 (2) (1991) 41–48.

Selvakumar. S is completing his Ph.D research work in Computer Science & Engineering from the Anna University. He received the Masters Degree in Computer Science & Engineering from Madurai Kamaraj University in 1996. He received the Master of Business Administration from the same University. He has over 15 years experience in various institutions and Organizations. Currently he is an Assistant Professor in the Department of Information Technology,

Thiagarajar College of Engineering. He has long been interested in Software Engineering. His research interests include Software Testing, Software Quality Engineering, Software Project Management, Software Metrics and Data mining, Data Base Systems. He also had a carrier as a developer for real-time, business-critical systems. Thus he has experience both with the practical problems of software development and the theoretical underpinnings of Software Engineering and Computer Science. He has presented a number of papers in international journals and in various other journals. He has carried out various sponsored Short Term Training Programs and worked on various Projects. He is a Senior Member in the Computer Society of India, Member in the Institute of Engineers and the Indian society of Technical Education.

Ramaraj. N received bachelor degree in Electrical and Electronics Engineering from Madurai Kamaraj University, Madurai, Tamil Nadu, in the year 1976. Received Masters Degree in Power System Engineering from the Madurai Kamaraj University, Madurai, Tamil Nadu, in the year 1980. Received PhD degree from Madurai Kamaraj University, Madurai, Tamil Nadu in the year 1992 in Computer Applications. He is the Principal of GKM Engineering College, Chennai, Tamil Nadu,

India. He has more than 25 years of teaching experience and has presented a number of papers in international journals (20) and in various other journals (35). His area of interest is Artificial Intelligence, Software Engineering, Software Testing and Distributed Computing. He is a Member in the Institute of Engineers and the Indian society of Technical Education.

18 INTERNETWORKING INDONESIA JOURNAL Vol. 3/No. 1 (2011)

ISSN: 1942-9703 / © 2011 IIJ

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 19

ISSN: 1942-9703 / © 2011 IIJ

Abstract—Traffic congestion has become a growing problem in

many countries since the innovation of engines, and consequently, the mass production of commuter vehicles in the 19th century. A number of solutions have been sought to reduce its impact. Through a literature survey, this paper attempts to categorize a number of current approaches and identify the strengths and weaknesses of each solution or a combination of such solutions, in order to build a good background study. From this survey it has become clear that for success any solution most likely will have to integrate technologies from the different categories. For example, technical solutions must be combined with good traffic rules and regulations. Public education and the announcement of new regulations for commuters must be performed in stages and repetitively to increase public awareness.

Index Terms—traffic jam avoidance, congestion, integrated and intelligent system.

I. INTRODUCTION

URRENTLY there are numerous problems associated with traffic jams. Aside from the delays that a traffic

congestion may cause, it can also affect people psychologically (in terms of stress) as well as physically (causing tiredness, accidents, noise pollution and breathing problems due to air pollution). Traffic jams can be caused by obvious factors, such as the overflow of vehicles during busy hours or at special events, accidents, slow vehicles obstructing fast lanes, etc. and also by not so obvious reasons like (i.e. ripple effect which causes phantom traffic jams [1]).

Many researchers have offered various approaches to tackle the traffic jam problem. The cost to reduce its impact has also been growing, which makes many developing countries lag behind in their effort to overcome it. For example, in 2003 to 2004 alone Great Britain [2] has spent 242 million pounds to start its Making Better Use (MBU) program, which is part of its Highway Agency program. This program includes (but not limited to) the installation of automated incident detection and warning system, CCTV cameras, advanced road side message signs, spot improvements on needed areas (such as improving layouts of lanes and junctions and signaling at various junctions), finding novel approaches to traffic management

Benny Hardjono is with the Universitas Pelita Harapan, at the Tanggerang campus, Indonesia (e-mail: benny.hardjono@ uph.edu).

and the management information systems necessary to support the agency's work, as well as research and development of a national traffic control center.

Each approach can use one particular technology or a combination of them. A couple of common technologies used in many relatively advanced countries are ETC (Electronic Toll Collection) which was started in 1986 in Norway, and VICS (Vehicle Information and Communication System) which was started in the 1990s in Japan [3]. ETC allows drivers to pass a toll gate without stopping for payment and is installed in many cars. The VICS system [16] is a service using FM broadcast and optical beacons on the roadside to deliver traffic jam information to drivers so that their car navigation systems can display congested areas/roads on the map and navigate them avoiding the congested areas. Both systems can reduce the possibility of traffic jams, but requires supporting devices on the roads and on each vehicle. The ETC method not only requires a device on each vehicle, but it also assumes that each vehicle owner abides by the regulation; otherwise the automatic bank-account deduction mechanism will not work. Additionally, there is the extra cost incurred when the authorities want to persecute negligent vehicle owners.

Similarly in the VICS system, although VICS is useful there could be a time lag between the disseminated information and the real situation faced by the driver. This is because VICS collects all traffic jam information to one location (e.g., a central server), and disseminates it after processing. Also, if all cars in a certain road receive the same information and change their route in the same way according to the information, then the selected route will be congested quickly. Additionally, VICS requires many devices to be installed on the roadside for monitoring traffic conditions, and thus it is costly to deploy the VICS system on small roads (which can be used as alternative routes during a traffic jam). The cost is even higher when considering coverage for a large metropolitan city.

II. RESEARCH & REVIEW METHOD

This paper attempts to categorize a number of current approaches, as well as identify the strengths and weaknesses of each solution or a combination of solutions. The aim is to build a good background study, before proposing new ideas. The method used is through a literature survey. The survey

A Review of Existing Traffic Jam Reduction and Avoidance Technologies

Benny Hardjono Universitas Pelita Harapan, Indonesia

C

20

incIEEbo

app[5]tratra

A

sucdesmasteconof

R

jamtha

cludes browsinEE’s digital lioks.

III. A NUMBE

Three approacproach which ]-[6], [9] and [affic flow simuavel time on Va

A. Meticulous

The meticulouch as regionasign, pedestriaanagement meaeps have shownditions and eaWanjiali road,

Figure 1: In

Research showms have occurrat efficiency

ng about this ibrary, as well

ER OF APPROAC

REDU

ches are surveyis: a) based on

[6]; b) using inulators [3], [8]ariable Messag

Traffic Design

us traffic desial traffic plannan crossing optasurement (impwn to be abase the traffic j, Changsha, Ch

ntersection before

ws that 60% ored in intersect

can be inc

INTER

topic in the l as looking in

CHES IN TRAFFI

CTION

yed in this papn meticulous t

nter-vehicle com], [10], [12]; age Signs (VMS

n Approaches

ign approachening, intersecttimization desiplementing regble to improvjams, such as shina [6].

e and after meticu

of accidents anions [6]. The w

creased by 6

RNETWORKIN

ISSN: 194

internet and n to a number

C CONGESTION

per. Namely, traffic design [mmunication a

and c) using ro) [7], [14]-[15]

es involves stetion optimizatign, and operagulations). Thve urban trafshown in the c

ulous design

nd most of trafwork of [6] sho66.67%, throu

NG INDONESIA

42-9703 / © 201

in r of

N

the [2], and oad ].

eps ion

able ese ffic ase

ffic ows ugh

meticulincludesturning right of1 and T

Anotoutput lovercomcenter loriginalmade tonoticed approac

Traffby restapplyinfor examalso encthe toll other tatax, an ethis, palimited

Advatechnoloapproactraffic-idiscussetraffic fpublic t

Otherpublic e

(i)

(ii)

(iii)

(iv)

(v)

A JOURNAL

11 IIJ

ous transforms the reductiowait area, rea

f way between Table 1 [6]). ther reason for lanes does notme the over fline of the roadl lanes. The mo be 2.75m. Athat traffic ac

ch was considefic managementricting peak

ng single-doublmple during Bcouraged to trroad. Tax is a

axes are addedexhaust emissiarking in the parking time a

anced traffic ogy are usualches. This incinformation aved further beloflow control, atransport. r references [education, inclu

Implement the parking impact of redAbolish all aand pay for user fees. Sifees" and "dfees" that suse these finstead of encourages to own moreInstitute "Falanes, but chlanes of roaand trunk roEliminate p(and ideallyparking permin some citwork if gooindustrial areImplement tneighborhooThis will als

mation desigon of the inteasonable signavehicles and p

r traffic jams ist match that in flow of vehicldway is shiftedminimum widtAfter a long pecidents were n

ered to be a feant can also imp

hour traffic le license plateeijing Olympicravel by bus; applied for pri, such a mileagion tax, parking

heavy trafficand limited par

control melly adopted in cludes making vailable (suchow), applying mas well as provi

[2, 5, 9] discuuding the folloemployee parsubsidy). Thisducing car comautomobile sub

this from theimilarly, elimideveloper feessubsidize the fees for constr

automobile-rgreater auto ue vehicles). are Lanes." Thharges them a ads or making oads [2]. Do noparking requirey, everywhere)mits for neighbties). Note thaod public transeas. traffic calmingod and decreasso decrease aut

gn. The tranersection areaal timing, andpedestrian clea

s that the numb the intersectioes in the inte

d to reduce the th in single in

eriod of observnot increasing aasible solution prove the load

through regue, and set allowc Games). Cootherwise theyivately owned ge tax, a fuel cg fees and so o areas is con

rking spaces).ethods and

meticulous trdriving route

h as in the Vmanaged parkiiding automati

uss more reguowing: rking cash-outs would have ammuting by 25bsidies (direct gas tax and/oinate all "traffi" and "parkingautomobile. Aructing guide related constr

use (and make

hese let anyonefare per car. better use of

ot add more lanements in ind). If necessaryborhoods (alreaat this approacsport is availa

g to create a mse automobile to usage.

HARDJONO

nsformations a, wider left d making the ar (see figure

ber of input-on node. To rsection, the width of the nput lane is

vation, it was and thus this [6]. of the roads, ulation (e.g. wable times,

ommuters are y should use vehicles and consumption

on. Added to ntrolled (e.g.

information raffic design e guides and VMS system ing, applying ic control for

ulations and

t (equalizing an immediate %. and indirect)

or other auto fic mitigation g assessment Alternatively,

way transit ruction that people want

e use carpool Use existing motor ways

nes. dustrial areas y, implement ady common ch will only able in those

more livable dependency.

Vo

(v

(v

(vi

intall numopelimwhtrashoredcon

effthe

Intoptdes

Optre

Pu

Parma

ol.3/No.1 (2011)

vi) Use cremainicongestan amoAs russtages ithe sambe usedalso benalternathighwafor mor

vii) The goadopt arange otrials atwhilst m

iii) Before governmeducatebenefitsgaining

Qualitative antersection traffithe intersectiomber of outpuen-ends settingminate the int

hich in turn heaffic to flow ows that the duced by 20%nflict point is rIt is clear that fective, as it ce traffic conges

Step tersection timization sign

pen-ends eatment

ublic traffic

rking anagement

TABLE

congestion pring will be tion pricing. Th

ount that variesh hour approin order to kee

me optimal flowd to build guidenefit by payintives. Congesay capacity, whre highway conovernment (or a more expansof measures at t more sites to managing the r

implementingment or appoe the commutes they would

g public accepta

nalysis [2] shofic capacity andon input lanes aut lanes is kept g, adding signaterference betwelps pedestriansmoothly. Mtraffic capacit

%, travel speereduced by 80%the meticulousombines a numstion problem.

Increase icenter and

Set pedestrby markinrailings at c

Set left tursignal timin

Close a few oways, and adopen. Adjust the sfacilities. Prohibit parkin

I. Wanjiali road m

INT

ricing. Any eliminated b

his means chars so that trafficoaches, the prep total cars usw level. The moe way transit bg for potentialstion pricing hile reducing pnstruction.

its appointedive approach tits disposal, cincrease its ch

risks involved. g any new roointed agency ers or convincebring to themance.

ows that good d reduces delayare changed froinvariant. Als

al control, goodween vehicles ns cross safely

Moreover, quanty increases bed can increas% [2]. s traffic designmber of techno

Improvemennput lanes b reducing lanerian and bicycng / channecorners. rning waiting ang. open lanes, opdd signal con

station lay-ou

ng near road-si

meticulous traffic d

TERNETWORK

ISSN: 194

congestion sby the use rging for road uc is kept movirice increases ing the facilityoney raised couecause road usl motorists to u

also increapolitical press

d agency) shouto testing out

carrying out mhances of succe

oad policies should seek

e users about m, as a means

design improvy, as shown whom 4 to 6, and so, by optimizd design is able

and pedestriay and enables ntitative analy

by 15%, delayse by 10%, a

n approach is vologies to redu

nt by shifting lae width. cles waiting areling island a

area and optim

timize pedestrntrol at reserv

ut and impro

ide and across.

design [6]

KING INDONES

42-9703 / © 201

still of

use ng. in

y at uld

sers use

ases ure

uld the ore ess,

the to

the of

ves hen the ing e to ans, the

ysis y is and

ery uce

ane

eas and

mize

rian ved

ove

.

B. InSimul

Trafftraffic dwireless[3]. Codestinatmovemare equinformaIEEE 8additiononboarddestinatto pass pass eacamong destinat

In [3]

graph wand a roto a desestimateto the d

Conssquare sfrom onlinks thincominincominthe timmeasureindicatearea onboundarcar passthere arcrosses The areincomin

This simulatoToyota

SIA JOURNAL

11 IIJ

nter-vehicle Colators

fic informationdata using inters communicationsequently thtion) can be an

ment indicates thuipped to autoation using int02.11 wirelessnal devices od device to esttions. The stepeach route, (b

ch route by exccars, and (c)

tion.

Figure 2: Div

] it is assumed where each nodoad between instination from ed theoretically

destination. sequently, a tarshaped areas wne square to a

hrough which ang link and ong and outgoin

me needed to ed and is callede boundaries bn the center ory, indicated ases this area, thre 5 x 4 = 20 boundary of a

ea passage timng and outgoin

method wasor called NETCentral R & D

ommunications

n can be obtainr-vehicle commion, GPS and she data (the nalyzed to detehere is a traffi

onomously colter-vehicle coms communication the roads. timate the timeps involved areb) calculation changing the m estimation o

vision of traffic in

d that a given rode and link corntersections, re

the current loy by summing

rget geographiwith distances another, as sha car enters anoutgoing link,ng links is calpass each ared the area pass

between areas (of the figure, s α, β, γ, δ, andhe car passes tcombinations

an area, the came is the diffeng links of the a then evaluaSTREAM, whD Labs [12]. N

s & Traffic Flow

ned by gatherinmunication (i.esmall computertime to get tct movement –c jam. In this llect and sharemmunication bion protocol, w

This methode required to ge: (a) measuremof the statistic

measured time af time require

nto areas (after [3

oad map can berresponds to anespectively. Thocation of a veg up time to pa

cal region canof several hun

hown in Figurend exits an are, respectively.lled a link paiea for every sage time. The(see Figure 3)

there are 5 d ε (also Figurtwo of these lin

of link pairs. ar records the cerence of recorarea. ted using a

hich has been dNETSTREAM

21

w

ng statistical e. short range r in each car) to a certain – where slow method cars

e traffic jam based on the without using d allows the get to certain ment of time cs of time to and statistics ed to get to

3])

e treated as a n intersection he time to get ehicle can be ass each link

divided into ndred meters e 2 [3]. The ea are called . A pair of ir. Secondly, link pair is

e dotted lines . As for the links across e 3). When a nks, and thus

When a car current time. rded time at

traffic flow developed by

was used to

22

estwareaNErunto (linlanusia gtheinftheis eaclinlocexptimenttrahavarebousamtheRedifacc

AforsimFigsolactpreIt cpeo

timate traffic jaas successful alistic traffic fETSTREAM hnning on a giv

construct an nks), along wines, intersectioing specified tigraphical interfe given map formation on lie time intervalsconfigurable, ach link or rout

nk within the lecations of allperiment up to

me to pass antering/exiting p

affic informatiove passed the sea passage timundary, the timme link pair cae waiting time esults show tfference of lesscurate enough

Figure 3

Another simulr Aimsun [8]. mulation solutigure 4). Legilution and its mtual measuremeferences and claims to haveople will mo

am at Nagano in generating

flows on an achas the ability ten map simultarbitrary roa

ith legal speedns (nodes) witime intervals toface. NETSTR

as follows. inks and nodess of traffic lighalong with thete, and making

egal speed limil cars every o 300 cars aren area (called pair of roads (con statistics asame pair of rome for each pme difference ban be considerat traffic lightshat about 55s than 10%, wfor practical us

3: Links across ar

lation softwareIt is a fully inion which useion has patenmovement algoments of walthe elastic tole

e an accuracy oove in spaces

INTER

Olympic Gama good estim

ctual road systto support mortaneously. It ald system cond limits and cth traffic lightso change color

REAM simulateIt reads the

s, and other infhts. Its initial ine number of cag each car runt. It also recorsecond. In

e used. Each carea passage

called link pairare generated oads. By measpair of roads between multipred by taking s and/or turning5% of link-pawhich means thse [12].

rea boundaries (a

e currently avntegrated pedeses mainly vidnted its pedesorithms have blking speeds, erances of thesof 95% for a p

and to have

RNETWORKIN

ISSN: 194

mes in 1998 andmation (i.e. wtem). Moreovre than 1000 clso has a functnsisting of roaertain number

s. It is capablers and so on uses traffic flow

map data,formation suchnformation of cars which follon on the specifrds logs includ

this simulatcar measures e time) for ear) of the area, afrom cars whuring the averacrossing an a

ple routes with into account ag at intersectioairs time hashat this method

after [3])

vailable is Legstrian and vehideo cameras (strian simulatbeen calibrated

personal spase measuremenprediction of he been used

NG INDONESIA

42-9703 / © 201

d it with ver, cars ion ads

r of e of ing on the

h as cars ows fied ing ion the ach and

hich age

area the

also ons. s a d is

gion icle see ion

d to ace nts.

how by

indepenMetro aevaluatejourneymetrics.in [9,10

F

C. Ro

Unligenerateapproacof the rdrivers destinatimproveprocessreal timmay cho

Sensoin ordeincludestraffic dsuggestinformaminimizbypass more inalgorithinformamotorwconsidejams calong it colony Delhi, I

A JOURNAL

11 IIJ

ndent agencies and the Londone and predic

y times between. Other predic

0] and [12].

Figure 4: Inside aare used i

oad Travel Tim

ike inter-vehiced by onboach relies on senroad [14]-[15].

can know hotion. Integratee the flow of tr: the driving t

me, so that drivoose another leors embedded er to calculates data on trafdensity. In thted to be incluation on whaze or avoid dcertain conge

ntelligently, ushm [4] or fation to drive

ways. These er other routes,an be overcomis expected tialgorithm, whIndia, consisti

including Lonn Fire Brigade.ct evacuation n locations, dection software

a traffic commandin most major cit

mes on VMS

cle communicaard device onsors which ar These devicew much time ed with softwaraffic accordintime between vers can evaluess encumberein carriage-wae several varffic flow, speehe VMS systemuded in the d

at motorists sdelay, alternatiestions (thesesing various mfuzzy algorith

ers as they aptypes of info

, the length of me (perhaps duime to clear jhich is applieding of 25 plac

ndon Undergro. With Legion

times, densielay, route choie simulations a

d centre - video caties (after [6, 8])

ation which un each vehic

re embedded ones can provideis required to

are, this approng through a setwo points is

uate the state od route if need

ays are able to iables in real

ed and a reprem [7] the othedisplay (or FMshould consideive routes to e routes can bmethods – e.ghm [10]), anpproach junctiormation allowf traffic jams aue to accidentjams. In the d in North-Weces, it is fou

HARDJONO

ound, Madrid n the user can ity analysis, ice and other are discussed

ameras

uses beacons cle, another n the surface

e data so that o reach their oach helps to elf-regulating displayed in

of traffic and d be.

provide data l time. This esentation of er data items M radio) are er doing to help drivers

be suggested g. ant-colony nd relevant ions to join

w drivers to and when the ts), and how case of ant-

est region of und that, this

Vo

algimexi

appFigresdriVMcha[15thadesout23desroucomthesig

FappThA.

B.

C.

ol.3/No.1 (2011)

gorithm can mprovement of

isting jammed

Figure 5

In the VMS sproximately 50gure 5) and msearch by the tivers are for thMS service, mange their itin5]. One studyat the assumstinations as qt of 30 drivers% of participstination as quute because thmfort of drivine travel times gn board were n

From this liteproaches surve

hese categories Improving thones [6, 9]. Installing nei. On the

toll gateii. On ea

communwith sopredict various

Implementin[6].

provide newf distance 3%routes [4].

: VMS device (sigaround Paris (Ph

system [7] the00 meters apar

measurements atransportation he most part (9

most of the timnerary when thy [7], [11] conmption that dquickly as posss (all of whompants emphasuickly as posshey were famng; the remainto reach their not too long.

IV. DIS

erature revieweyed can be grare as follows

he usage of the

w hardware: roads: video,

es (e.g. Electronach vehicle: nication deviceoftware which

possible trafoptimization m

ng regulations

INT

w routes, r% to 47.9%

gnpost) on the exphoto. INRETS) [7

e sensors are rt, along with tare taken everyauthorities sho98%), glad to

me, they are nhe expressway

nducted in Paridrivers wish sible is not val

m know alternasized wantingsible, and cho

miliar it; 60% ning 17% opted

destination di

SCUSSION

w we believe rouped into thr.

e existing road

sensors [4], [7nic Toll Collec

devices ses (both are u

h can includeffic jams [8]methods (e.g. a

which manag

TERNETWORK

ISSN: 194

resulting in comparing w

pressway ]

typically, spacthe signposts (y minute. Furtows that althoubenefit from t

no more likelyy is crowded [is, France, sho

to reach thid for all drive

ative routes), o to reach th

ose an alternatemphasized

d to stay becausplayed on VM

that the varioree (3) categori

s or building n

7], [13]-[14], action [3]) such as GPusually integrae simulations -[10], [12] ant colony [4]))e traffic [2], [

KING INDONES

42-9703 / © 201

an with

ced see her ugh this

y to [7], ows heir ers: nly

heir tive the use MS

ous ies.

new

and

PS, ated

to and ). [5]-

The apracticaavailabisingle ceffectivfree roathe mosFor exaembeddimplemalternat

Theresurveyeassume a certainthe metnew lanconvertmajoritymay noSimilarlin factregulatireduce existencregulaticountriein certafor infrnations have sugovernm

Overvarious conductstone fomethodsucceedproven conditio

Whilcertain still a ntechnolovehiclescommutof familneed foalso cledevelopthe catcategoryannouncperformawarene

SIA JOURNAL

11 IIJ

applicability oality of implemility of funds. category whic

vely. These caads from trafficst appropriate tample, the VMded on the

mented on congtives (capacity e are numbeed in the curre

a certain qualin level of eduticulous approanes [9], becausted. In the Vy of commuter

ot be a valid aly, while certat create addiions prohibitin

traffic but ice of good puions imposinges may cause pain countries srastructure dev

because privaufficient fundment’s role in irall there are nu

approaches. ted in this topor other novel ds to mature. Md since it has

to give gooons.

le there are a approaches to

need to look foogies (e.g. uss. This enable

uters. Using exliarity on the p

or buyer educaear that whatep, most likely ttegories of apy of regulcement of new

med in stageess.

f these categormenting each Furthermore,

ch can addressategories shoulc jams. It is altechnology on

MS system whtarget roads

gested roads wand distance w

er of weakneent literature relity of existing

ucation among ach of [1] doese the existing VMS approacrs are keen to rassumption in ain regulationsitional problem

ng factory workits effectiveneublic transportg too much political unrestsubsidies are rvelopment. Thiately owned c

ds to perform infrastructure bumber of prom

Many inpic, and these solutions in thMeticulous descombined sev

od results in

V. CONCLUS

number of teo reduce and afor a solution tsing existing

es better integrxisting technopart of the useration regardingever novel solthe solution wpproaches. Thlations, publiw regulations fs and repeti

ries mainly deptechnology us

, it is clear thas the traffic jld be combinedlso important t

n the most conghich requires s

cannot be which do not hwise). esses in the eview. Certain

g infrastructurethe drivers. F

es not recommlanes are big e

ch it is assumread road-messall deploymen

s are practical,ms. Thus, fokers to use caress is dependtation. Along

transport taxt. In the case required to enais is notable inompanies typi

what is seenbuilding.

mising strong anvestigations can be used ae future, or forsign has a bettveral technolog

achieving b

SION

echnologies whavoid traffic jathat makes usehardware) in

ration and lesslogies also har (drivers), and

g the related prlution might s

will integrate ashus, for examic educationfor the commuitively to inc

23

pends on the sed, and the at there is no am problem d in order to to implement gested roads. ensors to be successfully

have suitable

approaches n approaches e and assume For example, end building enough to be

med that the sages, which nt situations. , others may or example, s/motors can dent on the these lines,

x in certain of subsidies, able funding n developing ically do not n to be the

aspects of the have been

as a stepping r the existing ter chance to gies and has better traffic

hich can aid ams, there is e of existing

n commuter-s cost to the s the benefit d reduces the roducts. It is subsequently spects across

mple, in the n and the uters must be crease their

24

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10

[11

[12

[13

[14

[15

[16

Ripple effect exhttp://www.thedphantom-trafficNational Audit of England’s mAnd Auditor GeN. Shibata, T. T(2006). A MethVehicle Commuon Mobile and P. Bedi, N. MedAvoiding trafficapproach, IEEEand MultimediaThe modern Trahttp://www.mod2010. X.Y. Gan, H.Y.Based on Metice-Business EngiB.C. Lavalette, Time InformatioThird InternatioCommunicationAisum : The intehttp://www.aimNovember 2010Cambridge SystTraffic CongestCongestion Mit

0] H. Yin, S.C. Wofuzzy-neural app

] S. Poitrenaud (1to action gram42, pp.31-69.

2] E. Teramoto, Met.al. (1998). PrWinter Games UWorld Congress

] P. Davies (1991urban traffic con

4] HuaYan Shang,variable messagTechnological S

] S. Cohen, H. Habouchons et temRecherche, Tran

6] Vehicle Informahttp://www.oki.November 2010

REFER

xplains Phantom trdailydust.co.uk/20c-jams/, last accessOffice (2004). Tacotorways and truneneral , HC 15 SesTerauchi, T. Kitanihod for Sharing Traunication (Invited Ubiquitous Systemdiratta, S. Dhand, Rc jam using Ant Co

E International Cona Applications (ICCansit Society : Soluderntransit.org/sol

. Yang (2009). Resculous Traffic Desiineering (ICEBE),C. Tijus, S. Poitreon on VMS and Tonal Conference ons (ICWM). egrated transport m

msum.com/site/cont0 tematics, Inc. withtion and Reliabilitytigation, prepared fong, etc. (2002). Uproach. Transport1995). The Procop

mmars, Internationa

M. Baba, H. Mori, Hrediction of TrafficUsing Traffic Simus on Intelligent Tr1). Assessment of angestion, Transpor, HaiJun Huang ange signs on traffic Sciences, Volume adj-Salem (1996).

mps de parcours. Lnsports, Sécurité, 5ation and Commun.com/jp/SSC/ITS/e0.

INTER

RENCES raffic jams 010/04/02/ripple-efsed 9 March 2011.ckling congestion

nk roads, Report Byssion 2004-2005, pi, K.Yasumoto, Maffic Jam InformatPaper), IEEE Thir

ms, July 2006, IEEER. Sharma, A. Sinolony Optimizationference on CompCIMA), pp.61-67.utions for traffic clution.html, last ac

search on Urban Tign, IEEE Internat, pp.465-466. enaud, C. Leproux raffic Congestion,n Wireless and Mo

modeling softwaretent/category/1/43

hTexas Transportaty Trends and Advafor Federal Highw

Urban traffic flow ptation Research Pape Semantic Netwal Journal of Hum

H. Kitaoka, I. Tanac Conditions for thulator: NETSTREansport Systems, Vadvanced technolortation Research B

nd ZiYou Gao (200congestion, Scien52, Number 2, SpEvaluation compa

L’expérience de la 51, pp.29-39. nication System, eng/vics.html, last

RNETWORKIN

ISSN: 194

ffect-explains- by making better uy The Comptrollerpp.1-44. . Ito, T. Higashinotion using Inter-rd Annual ConfereE.

nghal (2007). on - A novel putational Intellige

congestion, cessed 17 Novemb

Traffic Improvemetional Conference

(2007). Road Trav, Proceedings of thobile

e, 3/74/, last accessed

tion Institute (200anced Strategies fo

way Administrationprediction using a

art C, (10), pp.85-8work: an alternatiman-Computer Stud

ahashi, Y. Nishimuhe Nagano OlympiEAM, Proc. of 5th Vol.4, pp.1801-18ogies for relieving Board. 09). Impacts of

nce in China Seriesringer, pp.477-483arative des affichaville de Paris,

t accessed 21

NG INDONESIA

42-9703 / © 201

use r

o

ence

ence

ber

nt on

vel he

d 17

5). or n. a 89. ive dies,

ura, ic

06.

s E: 3. ages

stay in BSiemens Siemens between ranges telecommand Huma

A JOURNAL

11 IIJ

Bennydegree (MEngin MelbHe curDepartmin TangworkinAustralChristia

Bandung, he also TelecommunicatTelecommunicatianalog-radio andfrom software

munication technolan Computer Inter

Hardjono compl(with Honours)

g) degree by researbourne, Australia,rrently lectures ment at the Univergerang, Indonesia

ng as a tutor at the lia and seven yean University, in took the role of

tion Project Offion surveyors an

d optical communto hardware

logies to databaseraction).

leted the Bacheloand a Masters

rch, both at the RM, in 1989 and 199in the Informatirsitas Pelita Harap. Prior to this he sRMIT University

ears as a lecturerBandung, Indonesengineering cons

fice, developing nd building a DTnication networks

(e.g. compue, library manage

HARDJONO

r in Electronics in Engineering

MIT University, 92 respectively. cs Engineering

pan (since 2001) spent five years

y, in Melbourne, r at Maranatha sia. During his sultant with the

softwares for TMF converter s. His interest uter networks, ement software,

Vol.3/No.1 (2011) INTERNETWORKING INDONESIA JOURNAL 25

ISSN: 1942-9703 / © 2011 IIJ

Abstract—Tujuan studi ini adalah mengidentifikasi teknik

pengaturan channel dalam pengiriman SMS yang memungkinkan untuk mengatasi kegagalan pengiriman akibat padatnya traffic channel yang dipenuhi layanan voice pada network CDMA. Diharapkan dengan pengaturan channel pengiriman SMS dapat memperbaiki tingkat keberhasilan pengiriman SMS dan utilitas jaringan CDMA.

The purpose of this study is to identify channel management techniques for SMS message transmission that can overcome transmission failures caused by high traffic volumes within a voice service in a CDMA network. With the improved management of the SMS channel the desired goal is for a better success rate for SMS message transmissions and for better utilization of the CDMA network.

Index Terms— SMS, CDMA, channel, voice, performansi.

I. PENDAHULUAN

OTENSI pelayanan Potensi layanan SMS (Short Message Service) dalam meningkatkan pendapatan

operator telekomunikasi dapat terlihat dari usaha yang telah dan selalu dilakukan untuk meningkatkan performansi layanan ini. Segala upaya peningkatan layanan diusahakan agar dapat meminimalisasikan kegagalan proses pengiriman setiap SMS dengan meneliti setiap element network yang bertanggung jawab atas keberhasilan layanan ini.

Data statistik operasional menunjukkan bahwa terdapat peningkatan berarti jumlah kegagalan pengiriman SMS pada jaringan CDMA ketika promosi voice service diberlakukan. Walaupun kegagalan pengiriman SMS ini dapat terjadi karena banyak faktor, namun indikasi padatnya traffic channel yang selalu digunakan oleh voice service menjadi perhatian

Roessobiyatno is with the R&D Center of PT Telekomunikasi Indonesia

(PT TELKOM) in Bandung, Indonesia. He can be reached at: roesso@telkom.co.id.

Samudra Prasetio is with the R&D Center of PT Telekomunikasi Indonesia (PT TELKOM) in Bandung, Indonesia. He can be reached at: samudra@telkom.co.id.

Andri Qiantori, Ph.D. is with the R&D Center of PT Telekomunikasi Indonesia (PT TELKOM) in Bandung, Indonesia. He can be reached at: qiantori@telkom.co.id.

Dr. Wiseto Agung is with the R&D Center of PT Telekomunikasi Indonesia (PT TELKOM) in Bandung, Indonesia. He can be reached at: wiseto@telkom.co.id.

tersendiri. Standar pengiriman SMS yang diadopsi oleh operator

CDMA di Indonesia merujuk pada dokumen 3GPP2 [1]. Pada dokumen tersebut tidak dinyatakan secara khusus tentang keharusan menggunakan control channel dan traffic channel untuk pengiriman SMS. Keleluasaan ini diberikan pada operator untuk menentukan tipe channel yang tepat untuk melayani karakteristik SMS yang dikirimkan.

Setting length of SMS pada BSC menentukan suatu SMS dikirimkan melalui salah satu dari dua channel tersebut. Sementara setting default setiap BSC didasarkan pada best practice setiap provider, bukan disesuaikan dengan traffic aktual. Sehingga kemungkinan besar terdapat perbedaan besar dalam karakteristik traffic SMSnya.

Paper ini akan mengidentifikasi fungsi pengaturan panjang karakter untuk menentukan pengiriman paket SMS melalui control channel dan traffic channel serta akibatnya pada kegagalan proses pengiriman. Implikasi studi ini adalah untuk mendalami efek pengaturan channel untuk memperbaiki tingkat keberhasilan pengiriman SMS dan utilisasi jaringan CDMA.

II. ALUR PENGIRIMAN SMS PADA SISTEM CDMA

Fungsi utama layanan SMS adalah mentransfer pesan pendek antaran suatu aplikasi yang terletak pada suatu handset atau MS (Mobile Station) and suatu aplikasi dalam network seperti MSC (Message Service Center). MSC dan BSC menyediakan suatu pipa untuk pesan-pesan yang dikirimkan tersebut antara aplikasi yang ada di network, seperti di MC, dan aplikasi dalam handset.

Ada tiga tipe dasar pesan singkat yang didukung pada CDMA system network, yaitu mobile originated point-to-point, mobile terminated point-to-point, and broadcast. Tipe mobile originated point-to-point dan mobile terminated point-to-point membutuhkan mekanisme pertukaran pesan untuk dua arah pada air interface. Studi ini hanya akan membahas layanan SMS Regular yang meliputi SMS Mobile Originated (SMS MO) dan SMS Mobile Terminated (SMS MT) serta tipe-tipe channel yang digunakan untuk melewatkan layanan SMS tersebut.

Analisa Pengaturan Channel untuk Perbaikan Performansi Pengiriman SMS

Roessobiyatno, Samudra Prasetio, Andri Qiantori and Wiseto Agung Research and Development Center, TELKOM

Bandung, Indonesia

P

26

1) P

Pdapmemehandik

HchamesetBSyan'AD

JmachayanchaMepad'CMassterberMSMSmecha

sebtraJikakachake yan

sebtraJik

Pengiriman SM

Pengiriman SMpat dilakukan belalui traffic enggunakan acndset tidak dkirimkan melalHandset/MS mannel. Jika hanengirimkan atelah SMS ditSC akan menging diterimanyDDS User PartJika suatu haaka pengirimaannel yang seng sedang idleannel, maka hessage ke BSCda field serviceM Service Reqsignment requrmasuk traffic crhasil, BSC aSC. SelanjutnySC untuk penenunjukkan penannel. Setelah traffbelumnya, BSCaffic channel deka acknowledgan mengirimkannel. SelanjutMSC, dimana

ng diterima dar

Gambar 1:

Setelah traffbelumnya, BSCaffic channel deka acknowledg

MS MO

MS MO point-baik melalui co

channel. Jccess channel, dapat menerimlui 'ADDS tranmengirimkan Sndset memintacknowledgemeterima melaluirimkan pesan ya dari handt '. ndset sedang

an SMS-nya adang digunakae melakukan phandset akan mC dengan mene option. Selanquest' ke MSCuest. BSC akchannel yang t

akan mengirimya akan terbengiriman SMSngiriman SMS

fic channel C menerima suengan tipe bur

gement layer 2kan acknowletnya BSC mena elemen 'ADDri handset.

Pengiriman SMS

fic channel C menerima suengan tipe bur

gement layer 2

INTE

-to-point yaitu ontrol (access)

Jika pengirimmaka selama

ma/melakukan nsfer' dari BSCSMS ke jaringaa acknowledgement melalui

ui access chan'ADDS Transfset disertakan

menggunakanakan dilakukanannya. Sedangpengiriman SMmengirimkan snyertakan SMSnjutnya BSC akC, dan MSC akan mencari atersedia. Apab

mkan Assignmeentuk koneksi S. Gambar 1

S MO dengan a

terbentuk muatu pesan SMrst yang mengi2 diminta olehedgement ini ngirimkan pesaDS User Part' d

MO dengan Acce

terbentuk muatu pesan SMrst yang mengi2 diminta oleh

ERNETWORKIN

ISSN: 194

dari MS ke B) channel maup

man SMS Mproses transmpanggilan. SM ke MSC. an melalui accment layer 2, paging chan

nnel. Selanjutnfer', dimana SMn dalam elem

n traffic channn melalui traf

gkan jika handMS melalui trafsuatu OriginatS option numbkan mengirimkkan mengirimkalokasi resourila pengalokasent Complete dari handset dan Gambar

access dan traf

melalui prosedMS Delivery paindikasikan SM

h MS, maka Bmelalui traf

an ADDS Delidiisi dengan SM

ess Channel

melalui prosedMS Delivery paindikasikan SM

h MS, maka B

NG INDONESIA

42-9703 / © 201

SC pun MO misi MS

cess BS

nnel nya MS

men

nel, ffic dset ffic ion ber kan kan ces

sian ke ke

r 2 ffic

dur ada

MS. SC ffic ver MS

dur ada

MS. SC

akan mchannelke MSCyang dit

2) Peng

SedanBSC keataupun

Untu(Gamba

Ketikakan dikchannelmemuatBSS mchannelmengirikembalitidak boJika penharus mjuga hadengan

KemuterhadapKetika Tag di dengan di dalam

Sedanchanneldidahulupenentu

Pengpaging dapat di

ProsePaging BSC m

A JOURNAL

11 IIJ

mengirimkan l. Selanjutnya C, dimana elemterima dari han

Gambar 2: Pengi

giriman SMS M

ngkan pengirime MS dapat diln melalui traffiuk pengiriman ar 3) mekanismka suatu MSCkirimkan ke hal, maka MSC mt SMS pada e

mengirimkan Sl. Jika BSC tidimkan CDMAi data burst teoleh melebihi ngulangan tela

memberikan nearus memberikcause value yaudian handsp SMS yang dMSC memintadalam pesan Amengirimkan

mnya setelah Bngkan pengiril terdapat dualui penentuan tuan traffic changiriman SMS m

channel dengiuraikan sebagedurnya diawRequest ke BS

melakukan pagi

acknowledgemBSC mengirim

men 'ADDS Undset.

iriman SMS MO d

MT

man SMS MTlakukan dengac channel. SMS MT me

menya adalah sC mengetahui a

andset yang semengirimkan pelemen ADDSSMS tersebutdak menerima

A data burst, ersebut dimanbatas maksim

ah mencapai niegative acknowkan pesan ADang bersesuaiaset mengirimditerimanya dea response den

ADDS DeliverADDS Delive

BS menerima Aiman SMS Ma metode yantraffic channel nnel. melalui mekanigan didahului gai berikut: ali dengan M

S dan timer T3ing dengan m

ROESSOBIYA

ment ini memkan pesan ADser Part' diisi d

dengan Traffic Cha

T point-to-poinan control (pag

enggunakan traebagai berikutadanya sebuahedang menggunpesan ADDS D

S User Part ket melalui forw acknowledgemBSC harus m

na jumlah penmum yang telah

ilai maksimumwledgement LaDS Deliver A

an dengan kondmkan acknoengan suatu Angan memasukr, BSC harus mer Ack denganAck Layer 2 daMT menggunang digunakan dan dengan di

isme dengan mpenentuan tra

MSC mengirim113 mulai diakengirimkan Pa

ATNO ET AL.

elalui traffic DDS Deliver dengan SMS

annel

nt yaitu dari ging) channel

affic channel : h SMS yang nakan traffic Deliver yang e BSC. Lalu ward traffic ment setelah mengirimkan gulangannya h ditentukan.

m, maka BSC ayer 2. BSC

Ack ke MSC disi. owledgement Ack Layer 2. kkan elemen

membalasnya n elemen Tag ari handset. akan paging

yaitu tanpa idahului oleh

menggunakan affic channel

mkan pesan ktifkan. Lalu age Message

Vo

mepesCoMSmeack

E

RechaBSmeReper

ADinfakaAD

KdenmeLadit

Jelememedendia

denpenpenpad

Pselpadadamesel

ol.3/No.1 (2011)

elalui paging san Page Resp

omplete LayerSC. MSC kemengirimkan knowledgemen

Gambar 3:

Elemen Radio esponse yang annel oleh BS

SC. Dalam rengirimkan Reelease Order rmintaan releasSelanjutnya M

DDS Page memformation. Jikan memasukkDDS Page. MSKemudian BSngan melakukengetahui di sayer 2 Ack sebterimanya. Jika MSC mem

emen informaembalasnya denyertakan elengan yang daktifkan sebeluSedangkan prongan menggunentuan trafficnentuan trafficda saat BSC mPada mekanlanjutnya jika da pesan Paganya penentuengirimkan pelanjutnya mem

channel. Selanponse dan BSCr 3 yang beri

mudian mengheBase Statio

nt dari pesan Pa

Pengiriman SMS

Environment mengindikasik

C, MSC mengangka melepa

elease Order kke BS sebagse dari BS. MSC mengirimmbawa SMS d

ka MSC memkan elemen inSC juga akan mC mengirimka

kan prosedur el mana MS bbagai acknowl

minta acknowlasi Tag padadengan pesanemen informa

dikirimkan MSumnya, maka tiosedur pengiriunakan paginc channel hampc channel. Ked

mengirim Base nisme tanpa

elemen Radioge Response uan traffic esan ADDS

mbawa SMS d

INT

njutnya MS mC mengirimkansi pesan Pagi

entikan timer Ton Ack age Response d

S MT dengan Pagi

dan Resourceskan adanya pgirimkan pesanaskan konekske MS. Lalu Mgi acknowledg

mkan ADDS dalam elemen

minta acknowlnformasi Tag mengaktifkan tian SMS ke MS

seperti pagingberada. Lalu Mledgement terh

ledgement dena pesan ADDn ADDS Paasi Tag yangSC. Jika timimer akan dinoiman SMS meng channel pir sama dengaduanya prosesnStation Ack Or

penentuan o Environmen

yang menginchannel olePage ke BSCalam elemen A

TERNETWORK

ISSN: 194

membalas dengn pesan informing Response T3113. Lalu BOrder sebadari MS.

ing Channel

s pada pesan Papenentuan trafn ADDS Page si transport, MS mengirimkgement terhad

Page ke BSADDS User P

ledgement, Mke dalam pesmer T3113. S, yang didahug ke MS unMS mengirimkhadap SMS ya

ngan menyertakDS Page, Bage Ack yag sama nilain

mer T3113 teonaktifkan. elalui mekanistanpa didahu

an tanpa didahunya sama samrder. traffic chan

nt dan Resourndikasikan tidh BSC, MC. ADDS PaADDS User P

KING INDONES

42-9703 / © 201

gan masi

ke SC

agai

age ffic ke

BS kan dap

SC. Part

MSC san

ului tuk kan ang

kan SC ang nya elah

sme ului ului

mpai

nnel ces dak

MSC age Part

informaakan mADDS T3113.

Kemudengan mengetaLayer 2diterima

Jika Melemenmembalmenyertdengan diaktifkacknowsemua kpage res

Dalamselanjutmengiriacknow

Data beberapJakarta,karena yang be

Untupada sepengirimProses handsetSMS. Lmemeritersebutdilokasi

Pakettiba di hkaraktertersebutkarakterchannel

A. Pen

PenenMO sanpengirimbergantuTABELterpopumemilik

SIA JOURNAL

11 IIJ

ation. Jika Mmemasukkan e

Page. MSC

udian BSC memelakukan p

ahui di sel ma2 Ack sebagaianya. MSC meminta

informasi Tlasnya denga

rtakan elemenyang dikirim

kan sebelumnyawledgement di

koneksi transpsponse. m rangka mtnya mengirimimkan Rele

wledgement terh

III. D

transaksi SMSpa MSC antara, Surabaya, dterdapat indik

erlebihan ketikuk mengetahuietiap skenario man paket SMini diawali d

t yang bertugLalu paket SMSiksa log setiat, yaitu BSC, Vi pengirim maut SMS yang behandset penerir dan catatan lt selanjutnya dr SMS yang l yang berbeda

IV. HAS

ngiriman SMS M

ntuan channel ngat ditentukamnya. Sedantung tipe termL 1 yang menuler saat pengaki setting b

MSC meminta elemen inform

selanjutnya a

engirimkan SMprosedur sepeana MS beradi acknowledge

a acknowledgeTag pada pean pesan An informasi Tmkan MSC. a, maka timer iterima. Kemport untuk mem

melepaskan kmkan Release ease Order hadap perminta

DATA DAN ME

S yang diukur a bulan Mei sdan Makasar. kasi terjadiny

ka promosi voici tingkat kebe

channel yangMS dengan jum

dengan mempgas untuk meS yang dikirimap node jarinVLR, MSC, daupun lokasi penerhasil dikirim ima kemudianlog setiap SMSdianalisa deng

terkirim terha.

SIL PENELITIAN

MO

yang digunakan oleh settingngkan settingminal yang adnunjukkan bahwambilan data, hberbeda. Sett

acknowledgemasi Tag ke d

akan mengakt

MS ke MS, yanerti paging keda. Lalu MS mement terhadap

ement dengan mesan ADDS

ADDS Page Tag yang samJika timer Takan dinonakt

mudian MSC mbersihkan sta

koneksi transOrder ke MS

ke BSCaan release dar

ETODOLOGI

pada network sampai Juli tah

Lokasi-lokasiya kepadatan ce service dibeerhasilan pengig berbeda mak

mlah karakter yapersiapkan dungirimkan dan

mkan akan dipangan yang dilan SMSC baiknerima. oleh handset p

n dicacat. InforS yang berhasi

gan memperhathadap skenario

N DAN DISKUSI

kan pada pengg jumlah karakg jumlah kda. Hal ini twa dari 5 (lim

hanya tipe termting pada t

27

ement, MSC dalam pesan tifkan timer

ng didahului e MS untuk mengirimkan p SMS yang

menyertakan Page, BSC Ack yang

ma nilainya T3113 telah tifkan setelah

melepaskan atus pending

sport, BSC S. Lalu MS C sebagai ri BS.

diambil dari hun 2009 di i ini dipilih traffic SMS rlakukan. iriman SMS ka dilakukan ang berbeda.

ua perangkat n menerima antau dengan lewati paket

k yang terjadi

pengirim dan rmasi jumlah il dikirimkan tikan jumlah o pemilihan

giriman SMS kter terminal karakter ini terlihat pada ma) terminal minal C yang terminal C

28

memaHapen

Pmepenmepenter

Nmepenmeme

T

T

Ga

B.

Ppropenratackdenwa

enunjukkan bahaka paket SMal ini berbeda ngalihan channPengiriman Sempunyai mekndudukan chenggunakan bengiriman darrhadap penggunNamun demikenunjukkan prndek dari padempercepat proeningkatkan ke

TABEL I. JUMLA

PADA PENG

Tipe Terminal

A

B

C

D

E

ambar 4: Setting JuMekanis

Pengiriman S

Pengiriman SMoses pengirimangiriman lebihta-rata dibuknowledgemenngan menggunaktu 8-10 detik

hwa untuk jumS akan di kiri

dengan termnelnya pada jumSMS MO mkanisme yanghannel. Darieberapa jenis hri masing-manakan ADDS tkian dari TArosedur meka

da mekanisme oses pengirimaeberhasilan pen

AH KARAKTER SMGIRIMAN SMS MO

Jumlah K

5-161-4

5-161-4

11-11-1

5-161-4

5-161-4

umlah Karakter SMsme ADDS Pages u

SMS MT

MS MT melalan yang lebih h sederhana. Dutuhkan untnt SMS MT senakan traffic c

k.

INTE

mlah karakter simkan melalui

minal tipe lainnmlah karakter 4menggunakan kompleks kai hasil perhandset terlihaasing handsetransfer maupuABEL I, II, anisme ADDS

ADDS delivean SMS walaungiriman.

MS DAN CHANNEL

UNTUK BEBERAPA

Karakter

60 Tra4 Ac60 Tra4 Ac60 Tra0 Ac

60 Tra4 Ac60 Tra4 Ac

MS pada Pengirimuntuk Beberapa T

lui control chacepat dikaren

Dari hasil pentuk sampai kitar 4 detik. A

channel rata-ra

ERNETWORKIN

ISSN: 194

sampai dengan i access channnya yang sett4.

ADDS deliarena diperlukrcobaan dengat bahwa perilaet berbeda-beun ADDS deliv

III, IV, dan S transfer leber sehingga akupun tidak bera

YANG DIGUNAKAN

A TIPE TERMINAL

SMS MO

affic channel ccess channel affic channel

ccess channel affic channel

ccess channel affic channel

ccess channel affic channel

ccess channel

man SMS MT dengipe BSC.

annel mempunnakan mekanisngamatan, wa

mendapatkAdapun SMS Mata membutuhk

NG INDONESIA

42-9703 / © 201

10 nel. ing

ver kan gan aku eda

ver. V

bih kan arti

N

gan

nyai sme ktu kan MT kan

Gamba

Hasiltipe X, parametX akan selama BSC tippaging (GambaBSC tipketika la

Gamba

A JOURNAL

11 IIJ

ar 5: SMS dengan

l pengamatan p Y, dan Z mter yang berbemelewatkan pjumlah karaktpe Y dan Z channel ketik

ar 4). Hal ini mpe Z lebih muayanan voice d

ar 6: SMS dengaDikir

Jumlah Karakter PChannel.

pengiriman SMmenunjukkan beda untuk mengpengiriman pakter SMS masihmelewatkan pka jumlah kamenunjukkan budah penuh didigunakan seca

an Jumlah Karakterrim Melalui Traffi

ROESSOBIYA

Pendek Dikirim M

MS MT pada 3bahwa ketigangalihkan channket SMS ke pagh dibawah 79 kpengiriman pakarakter masih bahwa channelibandingkan tiara massal.

r Sangat Panjang Sfic Channel.

ATNO ET AL.

Melalui Paging

3 (tiga) BSC nya memiliki nel. BSC tipe ging channel karakter, tapi ket SMS ke 33 dan 10

l traffic pada pe X dan Y

Sering Gagal

Vo

lanjumcharentrayanmavodia

penpertrameke ketpadmeSM

BchaseptradapNapenMTSMterjtra

PperjarPabag

[1]

ol.3/No.1 (2011)

SMS dengan jngsung dialihamlah karakte annel (Gambarndah mekanismaffic voice sanng menjadi paka akan mudaice selalu paalihkan ke pagiOperator CDngaturan pengrubahan settin

affic channel dengubah settin

paging channtika operator ada lokasi yanenggunakan tMSnya.

Bagi operator annel pengirimpenuhnya diataffic antara MSpat diatur dengamun pengatungiriman antarT. Kemampuan

MS akan merjadinya kegagaffic channel yaPerbaikan mrformansi layringan dengan da akhirnya hagi peningkatan

3GPP2. 3rd GenInterfaces Inter

jumlah karakteakan ke traffi

pendek akanr 5). Ketika dame ini tidak akngat padat sedaparameter penah terjadi kegaadat sedangkaing channel (G

DMA pada giriman SMS

ng panjang kardan paging chg jumlah kara

nel maupun traakan mempromng dilayani BStraffic chann

V. KESI

CDMA tidak mman SMS Mtur oleh tipe S dan BSC ketgan perubahanuran channel ra BSC dan Mn system untuk

emudahkan opgalan pengirimang digunakan

mekanisme inyanan SMS s

menekan kegaasil riset ini akn layanan opera

REFER

neration Partnershroperability Specifi

INT

er sangat panjafic channel dan dikirimkan alam kondisi tkan bermasalaangkan setting

ngalihan channgalan pengirim

an SMS yangGambar 6).

dasarnya daMT ini den

rakter yang akhannel pada Bakter SMS yanaffic channel i

mosikan layananSC tipe Z yael untuk pe

IMPULAN

memungkinkanMO karena ke

terminal sehtika pengirimann setting jumla

dapat dilakuMS yaitu ketikak mengatur chperator untuk

man SMS akibalayanan voice

ni berpotensisehingga menagalan proses pkan memberikaator CDMA.

RENCES hip Project 2 (3GPPfication Release A.

TERNETWORK

ISSN: 194

ang secara umuan SMS deng

dengan pagtraffic voice yaah. Namun ketg jumlah karaknel tidak diubman karena trafg panjang tid

apat melakukngan melakukkan dialihkan SC. Kemampu

ng akan dialihkini akan bergun voice, teruta

ang lebih banyengiriman pa

n untuk mengaendali pemilih

hingga kepadan SMS MO tidah karakter SMukan pada sa mengirim SMannel pengirim

k mengantisipat padatnya ja. i meningkatk

ningkatan utilipengiriman SMan dampak pos

P2), Access Netwo June 2000.

KING INDONES

42-9703 / © 201

um gan ing ang tika kter bah ffic dak

kan kan ke

uan kan una ama yak aket

atur han atan dak MS. sesi MS

man pasi alur

kan itas

MS. sitif

ork

He is alsjoint reseclose join

Japanese interests aconsumerorganizatiprocedure

Telecommthe Conve

SIA JOURNAL

11 IIJ

Roessobifrom the Indonesiaengineer Indonesiaresearcheinterest ar

Samudradegree fBandungTechnolo(UGM) engineer IndonesiaProduct D

so actively involvearch work with rnt reasearch work w

Andri QTelekomPhilosopCommunIEEE anJapan Soreviewerinvolved

experts, sponsoredare in social interr behavior modelitions, and the deses and systems in o

Dr WTelecom(ITB), Iin TelCommuUK. He1988 inworkin

munity (APT) Wirergence Working G

iyatno received thTELKOM Scho

a in 2010. From in the R&D C

a (TELKOM). Hr in the TELKOre mostly related w

a Prasetio receifrom the TELKO

g, Indonesia in 19ogy degree from in 2006. From 1

in the R&D a (TELKOM). HDevelopment Lab

ved in several Asiresearchers from Jwith researchers f

Qiantori is an engmunikasi Indonesiaphy in Science fnications, in Toknd Japan Associaociety for Socio-Ir of several internd in many resead by the Asia Pacraction modeling, ing, adoption of sign and implemeorganizations.

Wiseto Agung rmmunications fromIndonesia in 1987lematics (in 199unication (in 2002e has been with PTn various engineer

ng in the TELKOMreless Forum (AWGroup Chairman.

he Management Bol of Managemen2002 to 2004 heCenter of PT T

He is currently OM R&D Center.with retail product

ived the ElectricOM School of 996 and a Master

the University o1997 to now he Center of PT T

He is currently as b in the TELKOMia Pacific TelecomJapan. Currently hfrom the ASEAN n

gineer at the R&Da since 1997. He hfrom the Universkyo, Japan. He isation for Social IInformation Studienational journals archer-exchange cific Telecommuniadvanced recommICT and related

entation of impro

received the Bm the Institut Tekn. He also received4) and a PhD 2) from the UniveT Telekomunikasi ring divisions, andM R&D. Within t

WF) he holds the r

29

Business degree nt in Bandung, e worked as an Telekomunikasi working as a His research analysis.

cal Engineering Technology in of Information

of Gajah Mada worked as an

Telekomunikasi a Manager of

M R&D Center. mmunity (APT) he is conducting nations.

D Center of PT has a Doctor of sity of Electro-s a member of Informatics and es. He is also a and is actively programs with ity. His research mender systems, technologies in ved forecasting

Sc degree in nologi Bandung

d an MSc degree in Multimedia

ersity of Surrey, Indonesia since

d he is currently the Asia Pacific responsibility of

30

TABE

TABE

TAB

TABE

INTER

EL II: SAMPLE P

EL III: SAMPLE

BEL IV: SAMPLE

EL V: SAMPLE P

RNETWORKING

ISSN: 194

ENGIRIMAN SM

PENGIRIMAN SM

E PENGIRIMAN

PENGIRIMAN SM

G INDONESIA

42-9703 / © 201

MS MO DENGAN

MS MO DENGAN

SMS MT DENGA

MS MT DENGAN

JOURNAL

11 IIJ

N MEKANISME A

N MEKANISME

AN MEKANISME

MEKANISME A

ADDS TRANSFER

ADDS DELIVER

E ADDS PAGES

ADDS TRANSFER

ROESSOBIYA

R

R

R

ATNO ET AL.

Internetworking Indonesia JournalThe Indonesian Journal of ICT and Internet Development

ISSN: 1942-9703

About the Internetworking Indonesia Journal

The Internetworking Indonesia Journal (IIJ) was established out of the need to address the lack of an Indonesia-wide independent academic and professional journal covering the broad area of Information and Communication Technology (ICT) and Internet development in Indonesia.

The broad aims of the Internetworking Indonesia Journal (IIJ) are therefore as follows:

Provide an Indonesia-wide independent journal on ICT• : The IIJ seeks to be an Indonesia-wide journal independent from any specific institutions in Indonesia (such as universities and government bodies). Currently in Indonesia there are numerous university-issued journals that publish papers only from the respective universities. Often these university journals experience difficulty in maintaining sustainability due to the limited number of internally authored papers. Additionally, most of these university-issued journals do not have an independent review and advisory board, and most do not have referees and reviewers from the international community.

Provide a publishing venue for graduate students• : The IIJ seeks also to be a publishing venue for graduate students (such as Masters/S2 and PhD/S3 students) as well as working academics in the broad field of ICT. This includes graduate students from Indonesian universities and those studying abroad. The IIJ provides an avenue for these students to publish their papers to a journal which is international in its reviewer scope and in its advisory board.

Improve the quality of research & publications on ICT in Indonesia• : One of the long term goals of the IIJ is to promote research on ICT and Internet development in Indonesia, and over time to improve the quality of academic and technical pub-lications from the Indonesian ICT community. Additionally, the IIJ seeks to be the main publication venue for various authors worldwide whose interest include Indonesia and its growing area of information and communication technology.

Provide access to academics and professionals overseas• : The Editorial Advisory Board (EAB) of the IIJ is intentionally com-posed of international academics and professionals, as well as those from Indonesia. The aim here is to provide Indonesian authors with access to international academics and professionals who are aware of and sensitive to the issues facing a develop-ing nation. Similarly, the IIJ seeks to provide readers worldwide with easy access to information regarding ICT and Internet development in Indonesia.

Promote the culture of writing and authorship• : The IIJ seeks to promote the culture of writing and of excellent authorship in Indonesia within the broad area of ICT. It is for this reason that the IIJ is bilingual in that it accepts and publishes papers in English and Bahasa Indonesia. Furthermore, the availability of an Indonesia-wide journal with an international advisory board may provide an incentive for young academics, students and professionals in Indonesia to develop writing skills appropriate for a journal. It is hoped that this in-turn may encourage them to subsequently publish in other international journals in the future.

Focus & Scope: The Internetworking Indonesia Journal (IIJ) aims to become the foremost publication for practitioners, teachers, researchers and policy makers to share their knowledge and experience in the design, development, implementation, and the man-agement of ICT and the Internet in Indonesia.

Topics of Interest: The journal welcomes and strongly encourages submissions based on interdisciplinary approaches focusing on ICT & Internet development and its related aspects in the Indonesian context. These include (but not limited to) information technology, communications technology, computer sciences, electrical engineering, and the broader social studies regarding ICT and Internet development in Indonesia. A list of topics can be found on the journal website at www.InternetworkingIndonesia.org.

Open Access Publication Policy: The Internetworking Indonesia Journal provides open access to all of its content on the principle that making research freely available to the public supports a greater global exchange of knowledge. This follows the philosophy of the Open Journal Systems (see the Public Knowledge Project at pkp.sfu.ca). The journal is published electronically (PDF format) and there are no subscription fees. Such access is associated with increased readership and increased citation of an author’s work.

Internetworking Indonesia JournalISSN: 1942-9703

www.InternetworkingIndonesia.org