TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 11, Novembe
r
2014, pp. 78
3
2
~ 784
4
DOI: 10.115
9
1
/telkomni
ka.
v
12i11.65
80
7832
Re
cei
v
ed
Jul
y
5, 2014; Re
vised Augu
st
19, 2014; Accepted Septe
m
ber 5, 2014
Handover Analysis of Data and VoIP Services in
802.11b/g/n Wireless L
A
N
Imam Prasety
o
*
1
, Muham
m
ad Anif
2
, Ari Sri
y
anto
3
, Subuh
Pramono
4
, Sarono Widodo
5
,
Sidiq S. Hida
y
a
t
6
Dep
a
rtment of Electrical E
ngi
neer
ing, Po
lite
k
nik Ne
geri Se
maran
g
Jl. Prof. H. Soedarto, S.H.
T
e
mbal
ang,
Sem
a
ran
g
, Central
Java, Indo
nesi
a
T
e
lp: +
62-24-7
473
41
7, F
a
x:+
62 24 7
4
7
2
3
9
6
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: imam.praset
@
gmai
l.com
1
,
muhamm
ad.an
if@pol
in
es.ac.i
d
2
,
ari.sri
y
ant
o@p
o
lin
es.ac.id
3
, subu
hpram
on
o
@
pol
in
es.ac.id
4
, sarono.
w
i
d
o
do@
pol
ines.
a
c
.
id
5
,
sidiqs
h@
pol
ine
s
.ac.id
6
A
b
st
r
a
ct
IEEE 802.1
1b/
g/n WLAN w
i
t
h
co
nventi
o
n
a
l
Extend
ed S
e
r
v
ice Set (ESS
) infrastructure
do
esn
’
t
supp
ort ha
ndo
ver feature.
Ha
ndov
er f
eatur
e
all
o
w
s
user to
mi
grate s
e
rv
ic
es betw
e
e
n
A
P
s w
i
thout los
i
ng
conn
ectio
n
. Ha
ndoff latency i
s
one pro
b
le
m in W
L
AN to
perform re
al-ti
m
e app
licati
on s
u
ch as VoIP. T
o
brin
g se
a
m
les
s
ha
ndov
er
an
d QoS
guar
an
tee i
n
ESS
ne
tw
ork, w
e
propose
five
net
w
o
rk opti
m
i
z
at
i
o
n
meth
ods i.e.co
nfigur
ing ov
erl
app
ing
area of
APs, pl
acing
APs in one r
o
amin
g do
ma
in,
using the s
a
me
SSID and s
e
cu
rity mo
de, ch
o
o
sin
g
APs cha
nne
l by
mar
g
in
of 5, and c
onfi
guri
ng APs as
DHCP forw
ard
e
r
.
Han
dover test
is don
e by sn
iffing on th
e cli
e
nt that
experi
e
nced h
a
n
dov
er
. In the result, there are thr
e
e
steps in Lay
er 2 handoff, i.
e. probing,
authenticati
on, and r
e
-assoc
i
ation with m
a
xim
u
m
handoff latenc
y is
325.0
2
ms i
n
data servic
es and 6
7
.41
2
ms in VoIP
. Overall thr
oug
hp
ut is 1.955 –
3.268M
bps i
n
da
t
a
services a
nd 2
00.70
4 – 23
0.
4 Kbps in V
o
IP. In Vo
IP services, one w
a
y
delay is 3
9
.98
5
- 49.18
ms, o
n
e
way jitter is 9.45 - 19.
57
m
s
, and the pack
e
t loss is 0 -
0.548%. Overall, the built
network system
is
able
to
guar
ante
e
QoS in han
dov
er case, both i
n
da
ta and VoIP se
rvices.
Ke
y
w
ords
: W
L
AN, han
dov
er
, QoS,
Sniffing, VoIP
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
IEEE
802.11 Wirel
e
ss LAN
a.k.a. Wi-Fi
have seen
im
mense in last fe
w years.
Wirel
e
ss
LAN offer se
veral advanta
ges ove
r
fixed (or "w
i
r
ed"
) networks i.e. mobility, ease and sp
eed
of
deployme
nt, flexibility, and
co
st [1-2]. T
h
e pop
ula
r
an
d
most
used 8
02.11
stand
ards fo
r
Wirel
e
ss
LAN or
Wi-Fi
now a
r
e 80
2.11b/g/n [3-4].
There a
r
e t
w
o ope
ratio
n
mode
s in
WL
AN i.e.inde
pe
ndent
(ad
-
ho
c)
and
infra
s
t
r
uctu
re.
Extended Se
rvice
Set (E
SS) Netwo
r
k is
cla
s
sified
into inf
r
a
s
tructure m
ode,
built by
mul
t
iple
Bas
i
c Service Sets
(BSS) that interc
onnec
ted by
Dis
t
ribution System (D
S
)
. For the mos
t
part,
DS are
wire
d Ethernet [5].
The p
r
o
c
e
s
s of migratin
g co
nne
ctivity from one
Acce
ss Point
(AP) to a
n
o
ther i
s
comm
only referred to handoff,
handover, or
roami
ng
[6]. Genera
lly, ESS network
conventional
doesn’t support handov
er
feature. Handover f
eature i
n
crease the
mobility factor in the
Wirel
e
ss
LAN, but han
doff latency i
s
one of p
r
o
b
lem in Wi
rel
e
ss LAN to p
e
rform
real
-ti
m
e appli
c
atio
ns
like Voice over Intern
et Protocol (V
oIP) [7].
Seamle
ss ha
ndover i
s
n
e
e
ded to
solve
hand
o
ff laten
c
y case
at re
al-time
co
mm
unication
in Wirel
e
ss
LAN and g
u
a
rante
e
QoS
during h
and
over [8]. To bring
sea
m
le
ss h
and
over in
Wirel
e
ss LA
N, we pro
p
o
s
e
five network
optimizat
io
n
method
s i.e.configuri
ng ov
erlap
p
ing a
r
e
a
o
f
APs, pla
c
ing
APs in
one
roaming
do
m
a
in, u
s
ing
th
e same
SSID and
se
curity
mod
e
, choo
sing
APs cha
nnel
by margin of
5, and co
nfig
uring APs a
s
DHCP forwarder.
1.1.
Supporting
Handov
er to
ESS Net
w
ork
An Extended Services Set (
ESS) is defined as two or more Basic Servi
c
e Sets
c
o
nnec
ted by a c
o
mmon Dis
t
ribution
Sys
t
em (DS)
[9], as
s
h
own in Figure 1. IEEE 802.11
doe
sn’t
spe
c
i
f
y the DS te
chnolo
g
y.In ne
arly all
co
mm
erci
ally succe
ssfull
p
r
odu
ct
s, Ethernet i
s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Han
d
o
v
er An
alysis of Data
and VoIP Servi
c
e
s
in 802.
11b/g/n Wi
rel
e
ss LAN (Im
a
m
Prasetyo
)
7833
used as the
DS technol
ogy [1, 5]. The
BSS cells
coverage may
overlap to
provide roaming or
handoff capabilities [10].
With ov
erlapping area of BSS cells, the station (ST
A
) is possibl
e to
migrate
con
n
e
ctivity before lose
the conne
ction fro
m
old AP. Besid
e
s that, t
o
sp
eed u
p
the
hand
over pro
c
e
ss,
all
of th
e APs
mu
st
configured
in t
he
sam
e
Se
rvice Set Id
ent
ifier (SSID)
a
n
d
se
curity mod
e
[11]. The same SSID and
se
cu
rity modeco
n
figuratio
n can de
cre
a
se
the
synchro
n
ization laten
c
y be
tween AP an
d STA.
BSS
1
BS
S2
Co
v
e
r
a
g
e
ma
y
ov
e
r
la
p
to
pr
ov
i
d
e
r
o
am
i
n
g
/
ha
nd
of
f
c
a
pa
bi
l
i
t
i
e
s
Figure 1. The
Conventio
nal
Extended Service
s
Set
There are two hand
over
pro
c
e
s
ses
ba
sed
o
n
ro
am
ing dom
ain, i.e. handove
r
in one
roami
ng dom
ain and h
and
over between
roamin
g do
mains [5]. Thi
s
domai
n is a
l
so refe
rred a
s
a
Layer 2 n
e
twork. APs that
are in the same br
oad
ca
st domain
an
d config
ure
d
with the sa
m
e
SSID are
sai
d
to be i
n
on
e ro
aming
do
main. Hand
o
v
er in o
ne d
o
m
ain roami
n
g
sho
w
n
in Fig
u
re
2. Han
dover
betwe
en roa
m
ing do
main
s mea
n
ha
nd
over bet
wee
n
APs that located in different
sub
nets or di
fferent n
e
two
r
ks.
Hand
ove
r
bet
wee
n
ro
aming dom
ai
ns ca
n
imp
a
ct
the
ap
plication
se
ssi
on at cli
ent and ad
d the han
doff du
ration.
To preve
n
t IP segm
entatio
n in the STA whe
n
switchi
ng the APs
service
s
, the A
P
s mu
st
be
configu
r
e
d
a
s
Dyn
a
mi
c
Ho
st Co
nfiguratio
n
Protocol
(DHCP) forwarder. B
y
default, AP o
r
Wirel
e
ss Ro
uter co
nfigu
r
ed as DHCP
serve
r
[17]. DHCP forwarder means APs
forward th
e
DHCP proto
c
ol from one
dedi
cated server. This
techniqu
e can p
r
eventIP seg
m
entation in
the
STA during h
andove
r
or th
e IP addre
s
s of the
STA is the sam
e
, both before and
after han
dov
er
p
r
oc
es
s
.
Ne
t
w
o
r
k
S
u
bne
t
A
Ro
u
t
er
La
ye
r
2
Sw
it
c
h
Lay
e
r
2
Sw
i
t
c
h
R
a
om
i
n
g
/
Ha
ndof
f
Dom
a
i
n
AP
1
A
P
2
BS
S
1
B
S
S
2
ST
P
Blo
c
k
e
d
R
o
a
m
i
n
g
/
H
a
ndo
f
f
us
e
r
Figure 2. Han
dover in On
e Roami
ng Do
main
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 78
32 – 784
4
7834
1.2.
DSSS non-Ov
er
lapping Channel
Direc
t
Sequence Spread-s
pec
t
rum
(DSSS) is
spread
s
p
ec
trum bas
e
d on direc
t
s
e
quenc
e
that used in WLAN [10]. In
DSSS c
h
annels
,
bas
e
d on US
and E
u
rope
s
t
andards,
cha
nnel 1 fo
r instan
ce, op
erate
s
from
2
.
401 G
H
z to
2.423 G
H
z (2
.412 G
H
z
± 1
1
MHz),
cha
n
n
el
2 op
erate
fro
m
2.406
to 2
.
428 G
H
z (2.417
GHz
±
1
1
MHz), an
d
so fo
rth. DS
SS system
with
overlap
p
ing
chann
els shou
ld not
be
co
-l
ocate
d
b
e
ca
u
s
e th
ere
will
almost
always b
e
a
d
r
a
s
tic or
compl
e
te
red
u
ction
in t
h
ro
ughp
ut [10].
Becau
s
e
the
cente
r
frequ
e
n
cy a
r
e
5
M
H
z ap
art
and
the
cha
nnel
s a
r
e
22 M
H
z
wid
e
, ch
ann
els should
be
co
-l
oc
ate
d
o
n
ly
if
the cha
n
nel numbe
r
are
at
least five apa
rt, such a
s
1 and 6, 2 and 7, etc.
Theref
ore, there a
r
e three ch
an
nels can be co-
locate
d, i.e. 1, 6 and 11, as sho
w
n in Fig
u
re 3.
Figure 3. DSSS Non-overlapping Channel
1.3.
Qualit
y
of Se
r
v
ices Guara
n
tee
The
relia
ble
netwo
rk mu
st
provides Q
u
ality of
Servi
c
e
(Q
oS)
gu
arante
e
to
cli
ents. In
data servi
c
e
s
, the system
calle
d very re
liable if
theth
r
ough
put is
9
9
% from b
a
n
d
width. In Vo
IP
servi
c
e
s
, a su
mmary of the key QoS re
qu
ireme
n
ts an
d recomme
ndat
ions for VoIP
are [12]:
a)
One-way d
e
l
a
y (mouth
-
to-ear) sh
ould
b
e
no
mo
re th
an 150
ms, it is the same
as
ITU-T G.1
14
recomme
ndat
ion [13];
b)
One-way jitter sho
u
ld be ta
rgeted u
nde
r
30 ms;
c)
Packet loss should b
e
no
more tha
n
1
%
;
d)
Thro
ugh
put should b
e
gua
ranteed at 21
–
320 kbp
s
per
call.
1.4. Related
Wor
k
Figure 4. IEEE 802.11 Ha
n
doff
Proce
dure and Hand
off Latency
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Han
d
o
v
er An
alysis of Data
and VoIP Servi
c
e
s
in 802.
11b/g/n Wi
rel
e
ss LAN (Im
a
m
Prasetyo
)
7835
Many re
se
arches
have b
e
e
n
dedi
cate
d to
analy
z
e the
hand
over p
r
oce
dure an
d i
m
prove
the handoff
performanc
e
s in IEEE 802.11
WLA
N
[
14-16]. In Mis
h
ra
res
e
arch [14]
about an
empiri
cal a
n
a
l
ysis of han
d
o
ffs usin
g ca
rds an
d
AP from seve
ral vendo
rs, the
r
e
are two lo
gi
cal
step
s in hand
over pro
c
e
s
s: discove
ry an
d re-a
ut
henti
c
ation. Di
sco
v
ery is a sca
npro
c
e
s
s orti
me
to find the n
e
w AP to ha
ndover, a.
k.a
.
probin
g
pro
c
e
ss. Re-aut
henticati
on p
r
oce
s
s consi
s
t
o
f
authenti
c
atio
n and
re
-a
ssociatio
n
. The
hand
over
p
r
oce
dure an
d
hand
off latency illu
strate
d in
Figure 4.
In the expe
ri
ment results
[14], prob
e d
e
lay
take 90
% from total
hand
off laten
c
y. Total
hand
off latency best value
is 53.3 ms a
nd wo
rst
valu
e is 420.8 m
s
. The be
st value occu
rs at
cisco AP and
Lucent STA and the wo
rst
val
ue occu
rs
at cisco AP and ci
sco STA.
H.S. Kim et
al. [15] prop
ose
d
sele
ctiv
e chann
el
scannin
g
me
ch
anism
u
s
ing
neigh
bor
grap
h. Neigh
bor
graph
is sho
r
t scan al
gorithm
whi
c
h
ca
n
give the
nea
re
st
APs
and
th
eir
cha
nnel
s informatio
n. In the re
sults, total handoff
latency be
st value is 12 ms a
nd wo
rst valu
e is
332 m
s
.The
best valu
e o
c
curs at Sel
e
ct
ive Scanni
ng
with Unicast
method in
1 n
e
ighb
or a
nd t
h
e
worst value o
c
cursat Sele
ctive Scanning
method in 2 neigh
bors.
Chu
ng-S
hen
g Li et al. [16] prop
osed t
he
neig
hbo
r
grap
h cache
(NGC) me
ch
anism to
redu
ce
scan
ning laten
c
y while a mo
bile station t
r
ies to ma
ke
a link-laye
r handove
r
. The
simulatio
n
re
sults
sho
w
that the handov
er delay
by NGC is 2.61
4 to 50 ms and
able to meet the
criteria of VoIP application.
2. Rese
arch
Metho
d
Gene
rally, the built netwo
rk system fo
r th
is experi
m
e
n
t is 802.11
WLAN
with Extended
Service Set infrast
r
u
c
ture
that suppo
rt seam
le
ss h
andove
r
. As mentioned
earlie
r, to bri
ng
seaml
e
ss
ha
ndover in thi
s
net
wo
rk
sy
stem, we
p
r
opo
se five n
e
twork
optimi
z
ation
metho
d
s
i.e.configu
r
in
g overlap
p
ing
area of APs, placi
ng APs in one roam
ing domai
n, usin
g the sa
me
SSID and se
curity mod
e
, cho
o
si
ng APs ch
ann
el by
margi
n
of 5, and configu
r
i
ng APs as
DHCP
forwarder. In
this ESS network th
ere i
s
a local VoIP server that
dedicated to V
o
IP testbed.
The
overall sy
ste
m
desig
n sh
o
w
n in Figu
re
5.
AP
2
IP:
1
9
2
.
1
6
8
.
2.
4/24
;
SS
I
D
:
PN
_
P
a
t
i
;
Se
c
u
r
i
ty
:
No
;
Ch
an
n
e
l
:
6;
DH
C
P
:
di
s
a
bl
e
.
No
m
a
d
i
c
C
lie
n
t
No
m
a
d
i
c
Cl
i
e
n
t
’
BS
S
1
BS
S
3
BS
S
2
ES
S
DS
L
Mo
d
e
m
Mi
k
r
o
t
ik
Ro
u
t
e
r
(+
DH
C
P
Se
r
v
e
r
)
In
t
e
r
n
e
t
(S
p
e
ed
y
3M
b
p
s
)
Fi
x
e
d
C
lie
n
t
AP
1
IP:
19
2
.
16
8
.
2
.
3
/
2
4
;
SS
I
D
:
PN
_
P
a
t
i
;
Se
c
u
r
i
ty
:
No
;
Ch
an
n
e
l
:
1;
DH
C
P
:
di
s
a
bl
e
.
Vo
I
P
Se
r
v
e
r
AP
3
IP:
19
2
.
16
8
.
2
.
5
/
2
4
;
SS
I
D
:
PN
_
P
a
t
i
;
Se
c
u
r
i
ty
:
No
;
C
h
a
nne
l
:
11;
DH
C
P
:
di
sa
bl
e
.
Figure 5 Overall Netw
or
k S
y
st
em De
sign
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 78
32 – 784
4
7836
2.1.
The Ne
t
w
o
r
k
Sy
stem Explanation
The exp
e
rim
entwe
re
don
e in the
“Pe
n
gadila
n Nege
ri Pati”
office
at Pati, Cent
ral Java,
Indone
sia. T
h
e net
work
sy
stem, a
s
sh
o
w
n i
n
Fig
u
re
5, built a
nd t
e
sted
in thi
s
place. The
m
a
in
part of this
network
s
y
s
t
em are Extended Servic
e Set (ESS), VoI
P
s
e
rver, Mikrotik
router, and
Fixed & Nom
adic
Client th
at be explain
ed belo
w
.
2.1.1. Extended Serv
ice S
e
t (ESS)
The ESS c
o
ns
is
t of three
APs
with overlap
BSS c
e
lls
.
The APs
inte
rc
onnec
t
ed by
Layer
2 Di
stributio
n
System (DS),
it m
ean
s the
DS sy
stem b
u
ilt
by Layer
2 device, su
ch as
switch. T
he
APs pla
c
e i
n
one
roa
m
in
g dom
ain a
n
d
their IP ad
dre
s
s are
cl
assified in
o
ne
subn
et or ID
Network (1
9
2
.168.2.0/24
). The SSID and secu
rit
y
mode of all APs are
in the sa
me
config
uratio
n, the SSID i
s
PN_Pati a
nd
the securi
ty
mode i
s
ope
n
authe
nticatio
n o
r
di
sabl
e.
The
APs cha
nnel
have margin
of 5, i.e. AP1
in cha
nnel 1,
AP2 in cha
n
n
e
l 6, and AP3
in cha
nnel 1
1
.
The DHCP fe
ature in all A
P
s are di
sa
bl
e to perform DHCP forwarder.
The APs u
s
ed in thi
s
e
x
perime
n
t are TP
-Lin
k
WR841
ND v8 type. This prod
uct
spe
c
ification
s
are de
scrib
e
d
bello
w [17]:
a)
Hardware: Athero
s
AR934
1 Chip
set, 53
5 Mhz
CPU, 32 MiB RAM;
b)
Firmware: TP-Lin
k 3.13.33
Build 1305
06
Rel.48
660n;
c)
Standard: IEEE 802.11b/g/n, up to 300Mbps
(NB: In additi
on, to spe
ed
up the pa
ssiv
e
scanni
ng p
r
oce
s
s, the be
aco
n
interval
of APs is set
a
t
minimum val
ue= 4
0
ms. T
he default val
ue in TP-Lin
k is 100 m
s
)
2.1.2. VoIP
Ser
v
er
VoIP se
rver i
s
a
pe
rson
al
comp
uter de
skt
op th
at in
stalled Ela
s
tix
v.2.4.0 64bit
operating
system.
T
h
is person
a
l com
puter
(o
r thi
s
serve
r
) po
we
red by AM
D P
henom
II X2
pro
c
e
s
sor,
4
G
B
RAM, and 500GB hardi
s
k.
This
se
rverl
o
cated in l
o
cal network and
serve the V
o
IP facility on t
he
netwo
rk syste
m
coverage,
both wired an
d wirel
e
ss.
2.1.3. Mikrotik
Rou
t
er
This devi
c
e have three
main functio
n
s, i.
e. Routing, Netwo
r
k Address Transl
a
tion
(NAT
), and
DHCP serve
r
. This device
powe
r
e
d
by
Intel Dual Core 2.7
G
Hz
pro
c
e
s
sor, 1
G
B
RAM, 250
G
B
hardi
sk,
Broad
com
m
BCM57
0
4
C
Gigabit Ethe
rnet, and v5
.20 lv6 mikrotik
route
r
OS.
2.1.4.
Fixed and Nomadic Clie
nt
In this netwo
rk sy
stem th
ere a
r
e two
client
s with d
i
fferent cha
r
a
c
teri
stic, nam
ely fixe
d
and nom
adi
c. The charact
e
risti
c
differen
c
e
s
aret
h
e
mobility and the conn
ectio
n
type. Thefixed
client do
esn’t move (conn
ected to n
e
twork
usi
ng cab
l
e) an
d the n
o
madi
c cli
ent
moves bet
ween
APs cove
ra
ge an
d exp
e
rien
ce
d ha
ndover (con
necte
d to n
e
twork
usi
n
g
wirele
ss).
The
spe
c
ification
s
of clients a
r
e
describ
ed be
low:
a)
Fixed cli
ent:
Asus A43SV
powere
d
by
Wind
ows
8 P
r
o 6
4bit a
nd
Realte
k P
C
Ie
GBE
Family, 10/10
0/100Mb
p
s et
hern
e
t NIC;
b)
Noma
dic
clie
nt: Sony Vaio VPCEB25
F
G po
wered
by Windo
ws 8 Pro 64bit
and
Athero
s
AR9
285 Wi
rele
ss Adapter.
2.2. Testbed
Me
thodolog
y
In this sectio
n, we d
e
scri
bed h
o
w to t
e
st
the b
u
ilt netwo
rk
syst
em in ha
ndo
ver ca
se
,
both in
data
and VoIP
se
rvices.
Thi
s
te
stbed
aim
s
to
test the
a
b
ility of the b
u
ilt
netwo
rk
syst
em
in deliveri
ng
seaml
e
ss h
a
ndover fe
ature and g
uar
a
n
tee the Qo
S. To obtain
valid data on
the
results of testbed, we pe
rf
orme
d netwo
rk
sniffing
at the client. Th
e sniffer
software are Xirrus
Wi-fi Insp
ecto
r [18], Micro
s
oft Network
Monitor,
Cola
soft Cap
s
a [1
9], and Wire
shark [20-21]. The
testbed p
a
ra
meters are
sh
own in Ta
ble
1 belo
w
.
Table 1. Te
stbed Paramet
e
r
No.
Parameter
Value
1.
VoIP Codec
G.711 PC
M µ
-l
a
w
2.
Mobility
speed
Normal
w
a
lk ±1.
2192m/s [22]
3.
Mobility
path
Show
n in F
i
gur
e
6
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Han
d
o
v
er An
alysis of Data
and VoIP Servi
c
e
s
in 802.
11b/g/n Wi
rel
e
ss LAN (Im
a
m
Prasetyo
)
7837
2.2.1.
Hand
ov
er at Data Serv
ice
s
Han
dover
at data se
rvice
s
testbe
d
i
s
d
one by
sniffing o
n
the
cli
ent that
expe
rien
ced
hand
over (wit
h
paramete
r
and
p
a
th
spe
c
ified) within ±60 se
con
d
s sniffing
time whe
n
the clie
nt
wa
s doin
g
the download
a
c
tivity from googlevide
o
.com.
2.2.2.
Hand
ov
er at VoIP Ser
v
ice
s
Han
dover at
VoIP servi
c
e
s
te
stbed
is
done
by sniffing o
n
the
cli
ent that exp
e
rien
ced
hand
over (wit
h
paramete
r
and
p
a
th
spe
c
ified) within ±60 se
con
d
s sniffing
time whe
n
the clie
nt
wa
s doin
g
the VoIP telephone a
c
tivity
(
betwe
en nom
adic a
nd fixed client).
(NB: sa
mplin
g perfo
rmed
3 times on ea
ch path, both
for data and
VoIP service
s
)
2.2.3.
VoIP QoS per Receiv
e
Si
gnal Streng
th Indicatior (RSSI)
This te
stbe
d
mean
s Qu
ality of Service
(QoS)
te
sting i
n
VoIP based
on RSSI de
creme
n
t.
We
dete
r
min
edthe
RSSI d
e
creme
n
t val
ue i
s
1
0
dBm
with
the l
o
wer limit
is ra
n
ge
(-3
0) to
(-40)
dBm and up
per limit is range (-7
0
) t
o
hand
over
RSSI thresh
old. As an a
nalogy, firstly
we
measured Vo
IP QoS at RSSI range
(-3
0
)
to (-40)
dB
m, secondly
we me
asure
d
again at ran
ge (-
40) to
(-50)
d
B
m and
so o
n
. VoIP call d
u
ration i
s
±60
per
RSSI de
cre
m
ent. Thi
s
testbed
used
the
same p
a
rame
ter and path
as the p
r
evio
us testb
ed.
Figure 6. Path of Mobility
3. Results a
nd Analy
s
is
In this
se
ction
,
we de
scri
be
d and
analy
z
ed the
re
sults of the testb
e
d
. The te
stbe
dre
s
ults
and a
nalysi
s
are
abo
ut
hand
over
proce
s
s (i
n L
a
yer 1, L
a
yer 2,
and
u
pper Laye
r
OSI
perspe
c
tive),
the Quality of Service (Qo
S
) durin
g han
dover, and Q
o
S per RSSI decrem
ent.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 78
32 – 784
4
7838
3.1.
Handov
er at Data Serv
ice
s
in La
y
e
r 1
OSI Perspectiv
e
As sho
w
n in
Figure 7 bel
ow, the x-axis sh
ows time
in second
s
andthe y-axi
s
sho
w
s
RSSI in dBm. The first gra
ph sho
w
s RS
SI when c
lien
t
is served by
AP1 (SSID:
PN_Pati & MAC:
10-FE
-ED-9
C
-04
-
DA
) and
the se
con
d
g
r
aph
sh
ows
R
SSI when
client is se
rve
d
by AP2 (SSID
:
PN_Pati & MAC: C0
-4A
-
00
-EC-93
-10
)
. Can
be
seen that no
madic
clie
nt never lo
se
s the
con
n
e
c
tion d
u
ring
han
dov
er. On
ce th
e
nomadi
c
cli
ent moves a
w
ay from AP
1 cove
rag
e
and
RSSI getting
down u
n
til RSSI handove
r
thre
sh
old
(-7
2
dBm
at thi
s
sa
mple
), in
the n
e
xt seco
nd,
the noma
d
ic
client di
re
ctly serv
e
d
by AP2 at -52 dB
m RSSI with
out losi
ng th
e co
nne
ction.
It
mean
s the ha
ndover h
a
s b
een succe
s
sfully perform
e
d
in Layer 1
OSI perspe
c
tive.
Figure 7. RSSI during Han
dover at Data
Service
s
(pat
h 1, sample 1
)
All of other sample re
sult
s aren’t much differ
ent with
the first sam
p
le that explained in
the previous,
they sho
w
e
d
the same
pro
c
ed
ure. T
he ha
ndove
r
pro
c
e
dure i
n
Laye
r
1 O
S
I
perspe
c
tive is on
ce the cli
ent moves a
w
ay
from old
AP and the
RSSI getting down until RSSI
hand
over th
resh
old, in
ov
erlap
p
ing
APs a
r
e
a
, the
cl
ient will
switch di
re
ctly to
new AP servi
c
e
s
that have hig
her RSSI. Overall in the re
sults,
the RS
SI threshol
d value is mo
re t
han -7
0 dBm.
3.2.
Handov
er at Data Serv
ice
s
in La
y
e
r 2
OSI Perspectiv
e
Figure 8. Layer 2 Ha
ndove
r
Pro
c
edu
re
(data se
rvice
s
, path 1, sam
p
le 1)
Figure 8
sh
o
w
n the
sniffing re
sult u
s
in
g Micro
s
oft Network M
onit
o
r a
nd a
naly
z
ed
with
Wire
sh
ark [2
0-21] at
sam
p
le 1 p
a
th 1
testbed.
Can
be seen in t
he Figu
re
8
above, there
are
three lo
gical
step
s in a
hand
over: Probing, Au
the
n
tication, a
n
d
Re
-Asso
c
i
a
tion. At packet
number 13416, AP1 (M
AC: Tp-Li
n
kT_9c:04:da
) still sent
RTS frame to client (M
AC:
Hon
H
aiP
r
_d5
:7f:09). At the next pa
cke
t, AP2 (MAC:
Tp-Lin
kT_e
c:93:10)
se
nt
prob
e
re
spon
se
as
re
spo
n
se f
r
om
clie
nt’s
p
r
obe
requ
est
in a
c
tive sca
n
n
ing. T
hen
at
pa
cket nu
mb
er
1341
8, AP2
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Han
d
o
v
er An
alysis of Data
and VoIP Servi
c
e
s
in 802.
11b/g/n Wi
rel
e
ss LAN (Im
a
m
Prasetyo
)
7839
sent a
u
thent
ication
message to valid
ate the
clie
nt before jo
in the net
work. After t
he
authenti
c
atio
n su
cce
ssfull
y
, at packet n
u
mbe
r
13
419
, AP2 sent re
-asso
c
iatio
n
messag
e to client
that mean
s
a
llows cli
ent to
re
-join to
the
netwo
rk. After that, the
n
e
xt 802.11
p
a
cket i
s
cont
ain
action info
rm
ation and the
data traffic (TCP)
workin
g again. It mean
s the han
dover ha
s be
en
su
ccessfully perfo
rmed in
Layer 2 OSI perspe
c
tive.
All of other sample results aren’tmu
c
h
differ
ent from
the first sam
p
le that explained in
the previou
s
,
they have the same p
r
oce
dur
e. Th
e hand
over
pro
c
ed
ure in
Layer 2 OSI
perspe
c
tive
con
s
i
s
t of th
ree l
ogi
cal
steps: Pr
obin
g
, Authentica
t
ion, and
Re
-Asso
c
iation.
By
analyzi
ng the
timestamp of
handove
r
ste
p
at the sn
iffing re
sult
s, we can
get the
hand
off latency,
inclu
de Prob
e Delay, Authenticatio
n Del
a
y, and Re-a
ssoci
a
tion De
lay [14]. Table 2 sho
w
n tot
a
l
hand
off latency that generated in the testbed.
Table 2. The
Han
doff Late
n
cy at Han
d
o
v
er of Data Service
Test
bed
Probe Del
a
y
(m
s)
A
u
th
. Dela
y
(m
s
)
Re-ass. Dela
y
(ms)
Total
(m
s)
path sample
1
1 29.022
4.383
0.489
33.894
2 282.93
4.282
6.167
293.379
3 314.27
4.26
6.49
325.02
2
1 41.896
4.519
20.983
67.398
2 80.986
4.282
6.712
91.98
3 94.709
4.298
10.3
109.307
From T
able 2
above, kn
own that
most la
tency is
cau
s
ed by pr
obing
phase. Thi
s
result i
s
simila
r toMishra et al. [14] resea
r
ch, that probin
g
de
lay up to 90% from total
hand
off latency.
The “bi
g
” lat
ency in p
r
obi
ng pha
se i
s
cau
s
e
d
by
scannin
g
ch
an
nel time (fre
q
uen
cy cha
n
g
e
s)
before
client
swit
ch the AP services.
Based o
n
Ta
ble 2, total handoff laten
cy at data service han
dove
r
best value i
s
67.39
8
ms
and
worst value i
s
32
5.02 m
s
. By
comp
ari
ng
th
is total
ha
nd
off latency
wi
th the
previo
us
resea
r
che
s
[14-1
5
], the to
tal hand
o
ff latency worst value in thi
s
t
e
stbe
d is bet
ter, ho
wever
the
best value i
s
lowe
r.
3.3.
Hand
ov
er at Data Serv
ice
s
in Upper L
a
y
e
r OSI Perspec
t
iv
e
Figure 9. Through
put (dat
a se
rvi
c
e
s
, path 1, sample
1)
From Fig
u
re
9 above can be se
en that
the
handove
r
pro
c
e
s
s in built netwo
rk
system
doe
sn‘t have
significant effect on throu
ghput.
The th
roug
hput at h
andove
r
time
stillat 378.24
5
KBps. Howe
ver, wh
en vi
ewe
d
fro
m
t
he flu
c
tuat
io
n pattern of
throug
hput, t
he throug
hpu
t at
hand
over tim
e
is lo
wer th
an afte
r a
nd
before.
The
overall th
ro
u
ghput val
ue i
n
this testb
e
d
is
3.268 Mb
ps.
This valu
e is good b
e
cau
s
e the ISP Speedy (i
ntern
e
t con
nectio
n
sou
r
ce in thi
s
system
)have
band
widthu
p
to 3 Mb
ps. T
he throug
hpu
t in all of
sa
mples at h
a
n
dover
of dat
a
servi
c
e
s
sh
o
w
n in Tabl
e 3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 78
32 – 784
4
7840
Table 3. The
Thro
ugh
put a
t
Handove
r
of Data Service
s
Test
bed
Thro
u
g
hpu
t at
hand
o
v
er (KB
p
s)
O
v
erall thro
ug
h
put
(Mbps
)
path sample
1
1 378.245
3.268
2 234
3.245
3 216.24
1.955
2
1 12.5
1.866
2 43.75
3.238
3 32.083
3.2043
From
Table
3 above, th
e
wo
rst th
rou
ghput
at
han
dover
occu
rs in sample
1
path 1
testbed, b
u
t its value still
at 12.5 KBps or 10
0
Kbp
s
. It is not a
b
ig problem
becau
se ove
r
all
throug
hput
sti
ll rea
c
h
1.86
6
Mbp
s
. Ove
r
a
ll, the bu
ilt
ne
twork
system
give Q
o
S gu
arante
e
in
da
ta
servi
c
e
s
du
ri
ng han
dover i
n
uppe
r Laye
r
OSI perspe
c
tive.
3.4.
Handov
er at VoIP Ser
v
ice
s
in La
y
e
r 1
OSI Perspectiv
e
Figure 10. RSSI during Handove
r
at Vo
IP Services
(path 1, samp
le 1)
Gene
rally, the hand
over p
r
ocedu
re in L
a
yer
1 OSI perspe
c
tive at VoIP and data servi
c
e
have same p
a
ttern, a
s
sh
own i
n
Figu
re
10. On
ce th
e clie
nt move
s a
w
ay from
old AP and
RSSI
getting d
o
wn
until
RSSI hand
over th
resh
old, in
ov
erlap
p
ing
APs a
r
ea, th
e
client
will
switch
dire
ctly to ne
w AP service
s
that h
a
ve h
i
gher R
SSI without losi
ng t
he conn
ectio
n
. Overall, th
e
RSSI thresho
l
d value in VoIP service
s
h
andove
r
is m
o
re than
-70
dBm.
3.5.
Handov
er at VoIP Ser
v
ice
s
in La
y
e
r 2
OSI Perspectiv
e
Figure 11. La
yer 2 Han
dov
er Pro
c
ed
ure
(VoIP servi
c
e
s
, path 1, sa
mple 1)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Han
d
o
v
er An
alysis of Data
and VoIP Servi
c
e
s
in 802.
11b/g/n Wi
rel
e
ss LAN (Im
a
m
Prasetyo
)
7841
As sh
own in
Figure 11 a
bove, the ha
ndover
pro
c
e
dure i
n
Laye
r
2 OSI pe
rspective
betwe
en dat
a and VoIP
servi
c
e
s
a
r
e
identical. Thi
s
re
sult i
s
sh
own th
e sam
e
patterni
n
al
l o
f
sampl
e
results. The
r
e a
r
e
three l
ogi
cal
step
s
in h
a
ndover
of Vo
IP se
rvicesproce
s
s: Probi
ng,
Authenticatio
n, and
Re
-A
ssoci
a
tion. T
able 4
sho
w
n the h
and
off latency val
ue at h
and
over of
VoIP s
e
rvic
e.
Table 4. The
Han
doff Late
n
cy at Han
d
o
v
er of VoIP Service
Test
bed
Probe Del
a
y
(m
s)
A
u
th
. Dela
y
(m
s
)
Re-ass. Dela
y
(ms)
Total
(m
s)
path sample
1
1
32.088
4.402
2.255
38.745
2
31.68
4.45 3.753
39.883
3 19.269
4.26
2.763
26.292
2
1 40.443
4.273
0.6
45.316
2
46.755
15.98
4.677
67.412
3
26.158
4.362
2.573
33.093
Based
on dat
a above, it can be
said th
at this
re
sults are ide
n
tic with the data service
s
testbed
re
sult
s, thatmo
st h
andoff del
ay is cau
s
ed
by prob
e del
ay. The p
r
ob
e de
lay is up to
9
0
%
from total handoff latency.
The hand
off latency
in this tesb
ed isra
nge of 26.29
2 to 67.412 ms
and the
average i
s
41.7
9
ms.By com
p
a
r
ing thi
s
h
and
off latency va
lue withth
e h
andoff laten
cy in
previou
s
re
se
arche
s
[14
-
1
6
], this ha
ndo
ff latency va
l
ue is the b
e
st
. This
re
sult i
ndicates th
atour
system su
ce
ssfully
delive
r
se
amle
ss
hand
over for VoIP app
lication better
than previou
s
resea
r
che
s
[14-16].
3.6.
Hand
ov
er at VoIP Ser
v
ice
s
in Upper L
a
y
e
r OSI Perspec
t
iv
e
Figure 12. Th
roug
hput (VoI
P services, p
a
th 1, sample
1)
As sh
own in
Figure 12
ab
ove,
the han
dover at VoI
P
servi
c
e
s
in
built network system
doe
sn‘t h
a
ve
signifi
cant
effect o
n
th
rou
g
hput. At ha
nd
over time,
th
e throug
hput
still rea
c
h
22.
2
KBps. Howe
ver, wh
en vi
ewe
d
fro
m
t
he flu
c
tuat
io
n pattern of
throug
hput, t
he throug
hpu
t at
hand
over tim
e
is lo
we
r tha
n
after a
nd b
e
fore. Ove
r
all
throu
ghput i
n
this te
stbed i
s
0.21
5 Mb
ps or
equivalent
wi
th 220.1
6
Kbp
s
. Th
e throu
ghput i
n
al
l
of sa
mple
s a
t
hand
over
o
f
VoIP se
rvices
sho
w
n in Ta
b
l
e 5 belo
w
.
Evaluation Warning : The document was created with Spire.PDF for Python.