TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.5, May 2014, pp
. 3943 ~ 39
5
4
DOI: http://dx.doi.org/10.11591/telkomni
ka.v12i5.4276
3943
Re
cei
v
ed O
c
t
ober 1
4
, 201
3; Revi
se
d Decem
b
e
r
17, 2013; Accept
ed Ja
nua
ry 7,
2014
An Ada
p
tive Vertical Handoff Algorithm Based on
UMTS and WLAN
Jin Ling, Zhang Hui
*
, Ya
ng Long-Xia
ng, Zhu Hon
g
-Bo
Jian
gsu Ke
y L
ab of W
i
reless
Commun
i
cati
o
n
s, Nanj
ing U
n
i
v
ersit
y
of Posts
and T
e
lecom
m
unic
a
tions,
Nanji
ng, 2
100
03, Chi
n
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: zhhjo
i
ce@
1
2
6
.com
A
b
st
r
a
ct
Accordi
ng to t
he ch
aracter
i
stics of het
erog
ene
ous netw
o
rks
and user d
e
man
d
for
se
a
m
l
e
ssl
y
conn
ectio
n
, the
nec
essity of
h
and
off to W
L
A
N
w
a
s rese
ar
c
hed. F
i
rst
of al
l, in
order
to re
duce
the
influ
e
n
ce
of shad
ow
fadi
ng o
n
rece
ive
d
sign
al stren
g
th (R
SS), a s
m
oothi
ng a
l
g
o
rit
h
m
bas
ed o
n
supp
ort functio
n
s
w
a
s propos
ed.
T
hen, co
nsid
e
r
ing th
e co
ntex
t of mo
bil
e
ter
m
i
nals
an
d n
e
tw
orks, and by
usin
g pro
b
a
b
il
i
s
tic
know
led
ge, w
e
pro
pos
e a
n
ada
ptive
vert
ical
ha
ndoff a
l
gorit
hm b
a
se
d on
co
mp
en
sating ti
me. T
h
e
simulati
on r
e
s
u
lts show
that
, compare
d
w
i
t
h tw
o al
g
o
r
i
t
hm
s, th
e p
r
op
ose
d
al
go
ri
thm
ca
n
m
a
ke
m
o
re
effective an
d a
ccurate h
a
n
dof
f decisi
ons, re
duce th
e
pi
ng-
pon
g effect an
d interr
uptio
n p
r
oba
bil
i
ty, thereb
y
greatly i
m
provi
ng ha
nd
off performanc
e.
Ke
y
w
ords
:
UMTS, WLAN, v
e
rtical h
andoff, com
p
ensating tim
e
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
With the
ra
pid devel
op
ment of wi
reless n
e
two
r
k an
d
com
m
unication t
e
ch
nolo
g
y,
different
wirel
e
ss n
e
two
r
k
system
s
su
ch
as WiM
a
x, WMAN,
UMT
S
, and
WLA
N
h
a
ve a
ppe
ared.
As the
develo
p
ment trend
of next
gen
eration net
wo
rk, hetero
gen
e
ous
net
work i
n
tegratio
n i
s
t
h
e
integratio
n of merits of diffe
rent networks, in
order to obtain the most s
a
tis
f
ac
tory QoS.
Although the
r
e have be
en
a lot of resea
r
ch
es
abo
ut vertical
han
dof
f algorithm [1
-5], but
they are mo
stly fixed thresh
old
algo
ri
thms. In
fa
ct
, netwo
rk
co
ndition
s p
o
se an
imp
o
rt
ant
influen
ce
on
vertical
han
d
o
ff deci
s
io
ns.
Literature
[6]
define
s
a ve
rtical
han
doff
criterio
n b
a
sed
on RSS, whi
c
h will be adaptively adjust
ed based
on
mobile terminal movement
speed, in order
to effec
t
ively improve the hit ra
te and to reduc
e
th
e unnecess
a
ry handoff. A
cross
layer handoff
manag
eme
n
t mech
ani
sm f
o
r the
scene
of mobile u
s
e
r
s m
o
ving fro
m
WLA
N
to 3
G
is put fo
rward
in literature [7], where an
adaptive thre
shol
d is a
dop
ted to effectively redu
ce th
e hand
off failure
and ha
ndoff
error p
r
ob
abil
i
ty, but it doesn’t evaluat
e
the necessity of mob
ile no
de (M
N) a
c
ce
ss
to WLAN. In literature [8], acco
rdi
ng to information
su
ch as mo
bi
le terminal m
o
vement sp
e
e
d
and ha
ndoff d
e
lay, for MN,
the neces
sity of acce
ss to
WLAN
and v
e
rtical
han
dof
f time of moving
out of WLAN are e
s
timate
d, thereby effect
ively redu
cing u
nne
ce
ssary ha
ndoff
and conn
ecti
on
interruption.
Ho
wever, the
netwo
rk
con
t
ext is
not co
nsid
ere
d
wh
e
n
analyzi
ng the ne
ce
ssity of
ac
ce
ss t
o
W
L
A
N
.
In order to
re
duce the
sha
dow fadin
g
ef
fect
on
RSS, and to
imp
r
ov
e the
a
c
cura
cy of the
hand
off deci
s
ions,
we p
r
op
ose
an al
gorit
hm ba
s
ed
on
sup
port fun
c
ti
on to smooth
RSS, whe
r
ea
s
traditional
alg
o
rithm
s
mo
stly utilize the a
v
erage
or
wei
ghted ave
r
ag
e method. M
ean
while, ba
sed
on UMTS a
n
d
WLA
N
inte
gration frame
w
ork, inco
rpo
r
ating the mo
bile termin
als and the network
context, and
by using p
r
o
bability theory, this paper
prop
oses a n
e
w ad
aptive vertical ha
nd
off
algorith
m
ba
sed on
comp
e
n
satin
g
time, to analyz
e t
he ne
ce
ssity
of acce
ss to
WLAN, the
r
e
b
y
makin
g
more effective and
accurate han
doff deci
s
ion
s
.
2 Res
earc
h
Metho
d
2.1. Relate
d Work a
nd Ne
t
w
o
r
k Model
The top
o
logy
model
of het
erog
ene
ou
s
netwo
rk in
teg
r
ation
i
s
sho
w
n
in Figu
re 1,
taking
UMTS a
nd
WLAN net
wo
rk fo
r
examp
l
e, wh
erei
n t
he
UMTS a
n
d WLAN
access p
o
ints a
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3943 – 39
54
3944
referred to
a
s
ba
se
statio
n (BS) an
d AP respe
c
tively, and the mobile statio
n
are m
o
ving from
UMTS to WL
AN at a con
s
t
ant spe
ed.
Figure 1. Het
e
rog
ene
ou
s Network Top
o
logy Model
In order to better realize se
amless handoff, the
vertical handoff algorithm put forward in
this pap
er is
operating on
a cro
s
s layer
vertical
switch control mod
e
l, as sh
own in Figure 2.
Figure 2. A Cross Laye
r
Vertical S
w
itch
Control M
o
d
e
l
RSS mea
s
urements: M
N
detect
s
cu
rre
n
t RSS of
acce
ss n
e
two
r
k with the help
of network, whe
n
link qu
ality is detecte
d to decrea
s
e to a certai
n thre
sh
old, hand
off is trigge
re
d.
)
,
(
log
10
f
d
d
P
P
RSS
ref
Lref
Tx
(1)
W
h
er
e
Tx
P
is tra
n
smitting
po
wer of
WLA
N
ba
se
station
,
ref
d
is the
dista
n
ce
bet
wee
n
referen
c
e p
o
i
n
t and th
e b
a
s
e
station,
Lref
P
is the refere
nce point
of the
path lo
ss,
is a path
loss ind
e
x, which
usually i
s
2
-
4d
b in th
e
urb
an
enviro
n
ment, an
d
d
is the
dista
n
ce bet
ween
M
N
and th
e reference poi
nt.
)
,
(
f
is the
sh
ado
w fadi
ng
whi
c
h i
s
Gau
ssi
an di
strib
u
tio
n
with
mean 0 an
d varian
ce
2
.
Speed
e
s
timation: By u
s
i
ng GPS
or t
he e
n
ve
lop
e
of
po
we
r spe
c
tral
de
nsity of
RSS,
movement sp
eed of MN is
estimated.
Han
doff time delay estimat
i
on: The ha
n
doff
signal d
e
l
ay requi
red
by the netwo
rk laye
r
and layers ab
ove the network laye
r is e
s
timated.
Han
doff de
ci
sion: By u
s
in
g information
su
ch
a
s
the
han
doff del
ay, terminal
velocity,
curre
n
t network
re
sou
r
ce
s and RSS,
make ha
nd
off decisio
n based on propo
sed ad
ap
tive
vertical ha
nd
off algorithm i
n
this pap
er.
Han
doff execution: By usi
ng the
re
sult
of
hand
off de
cisi
on, exe
c
u
t
e the corre
s
pondi
ng
regis
t
ration.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An Adaptive
Vertical
Han
d
o
ff Algorithm
Based o
n
UM
TS and WLA
N
(Jin Ling
)
3945
2.2. The RSS Smoothing Algorithm Based on Sup
port Fun
ctio
ns
Due
to the
shado
w
effect, mea
s
u
r
eme
n
t of
RSS wil
l
be flo
a
ting,
thereby
impa
cting th
e
perfo
rman
ce
of vertical ha
ndoff algorith
m
. In this paper, we a
d
o
p
t the smoot
hing algo
rith
m
based on
su
pport fun
c
tio
n
to minimize mea
s
urem
ent errors, thereby imp
r
ov
ing the de
cision
ac
cur
a
cy
.
In
the
terminal
mov
e
ment
proces
s, by usin
g the uniform
sa
mpling on th
e
received si
g
nal
s
t
r
e
ng
th
o
f
U
M
T
S
an
d
W
L
AN r
e
s
pec
tive
ly, w
h
ere
s
T
is th
e sa
mpling inte
rv
al and
N i
s
t
h
e
wind
ow
si
ze,
RSS se
que
nce
)
1
,
1
,
0
](
[
N
k
k
RSS
ca
n be
obtaine
d, an
d
]
1
[
N
RSS
is
the recent sa
mpling result. By Equation (1), RSS sam
p
led value at
th
k
samplin
g time can b
e
expre
s
sed a
s
follows:
(2)
After weighte
d
averag
e:
(3)
Whe
r
e
i
w
is the wei
ght for RSS at time
i
k
whi
c
h
satisfie
s the
equation:
1
1
0
N
i
i
w
.
Assu
ming tha
t
MN is a sl
o
w
mobil
e
use
r
, movement
within a
certai
n time distan
ce is limite
d
, and
correl
ation b
e
twee
n adja
c
ent data
e
x
ists. The
r
ef
ore,
i
w
is determin
ed by usin
g su
ppo
rt
function[9] in
this pap
er,
and the relia
bility of
]
[
i
RSS
is di
rectly p
r
opo
rt
ional to the d
egre
e
of
being
supp
orted by the rest
data. The differen
c
e be
tween
]
[
i
RSS
and
]
[
j
RSS
is
ij
d
, which
can b
e
expre
s
sed a
s
:
(4)
The su
ppo
rt functio
n
is defi
ned a
s
:
ij
ij
ij
d
d
r
max
1
(5)
Whe
r
e
ij
r
repre
s
ent
s su
ppo
rt relation
ship
betwe
en
]
[
i
RSS
and
]
[
j
RSS
, if
j
i
, then
1
ij
r
.
That is, the d
a
ta is fully su
pporte
d by itself. T
he supp
ort function in
matr
ix form is as follo
ws:
(6)
In sup
p
o
r
t d
egre
e
mat
r
ix R, the i
n
fluence facto
r
of the
i
-th m
easure
m
ent
data for oth
e
r
measurement
data i
s
)
,
,
0
(
N
i
w
i
, an
d
1
1
0
N
i
i
w
.
T
N
w
w
w
w
]
,
,
[
1
1
0
and it
can b
e
cal
c
ulate
d
by equatio
n:
Rw
w
max
(7)
)
,
(
]
[
log
10
]
[
f
d
k
d
P
P
k
RSS
ref
Lref
Tx
1
0
]
[
)
(
N
i
i
i
k
RSS
w
k
RSS
1
,
2
,
0
,
,
]
[
]
[
N
j
i
j
k
RSS
i
k
RSS
d
ij
1
1
11
10
1
1
11
10
1
0
01
00
N
N
N
N
N
N
r
r
r
r
r
r
r
r
r
R
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3943 – 39
54
3946
Whe
r
e
max
is th
e maximum
value of
eige
n
v
ectors, an
d
i
w
is th
e corre
s
pondi
ng ei
ge
nvector.
In combin
atio
n with Equati
on (3
), weig
hted avera
ge of
]
[
k
RSS
i.e.
]
[
k
RSS
can be
cal
c
ulate
d
.
2.3. The Vertical Ha
ndo
ff Decision Alg
o
rithm
Whe
n
WLA
N
is dete
c
ted to be availabl
e to a MN in
the WLA
N
an
d UMTS ove
r
lappin
g
coverage, th
e corre
s
po
nd
ing netwo
rk entity can ju
d
ge wheth
e
r it
can acce
ss to WLAN or
not.
Two f
a
cto
r
s
need
to b
e
consi
dered: o
ne i
s
WLAN
availability, i.e. RSS m
u
st
be
gre
a
ter than
min
RSS
and the othe
r is the ne
ce
ssity of handof
f to WLAN.
Becau
s
e
the
coverage
of
WLAN is limited, MN
move
ment spee
d i
s
a
n
imp
o
rta
n
t factor
to judge the necessity of hand
off to WLAN. Whe
n
MN moveme
nt spee
d is hi
gh, once switchin
g
to WLAN, it will so
on mo
ve throug
h the AP co
verin
g
radi
us a
n
d
handoff ba
ck to UMTS.
Th
e
‘ping-pon
g ef
fect’ lea
d
to
fluctuatio
ns in
RSS
and
he
avy intera
ctio
n si
gnali
ng i
n
co
re
net
wo
rk,
thereby influe
ncin
g busi
n
e
s
s contin
uity,
wa
sting
sig
n
a
ling and net
work
resou
r
ces. To avoid such
unne
ce
ssary
hand
offs, MN must have a
certai
n dwe
ll time in WLA
N
. If MN movement speed
is
very high, dwell time in WLAN is less than
the hand
o
ff signaling d
e
lay, handoff failure eme
r
g
e
s.
Therefore, if the terminal v
e
locity is sma
ll,
handoff to
WLAN i
s
feasible; otherwise, still acce
ss to
UMTS is a b
e
tter choi
ce. Condition
s re
q
u
ired fo
r acce
ss to WLAN a
r
e as follo
ws:
max
min
]
[
v
v
RSS
k
RSS
(8)
Whe
r
e
]
[
k
RSS
is the RSS after smoothi
ng at
th
k
sampling time,
min
RSS
is the minimu
m
value of
RSS
that MN can
comm
uni
cate
with
WLA
N
,
v
rep
r
e
s
ent
s m
o
vement
spe
ed of
MN,
an
d
max
v
is spee
d thresh
old. The correspon
ding
algorith
m
ana
lysis is
sho
w
n
in Figure 3.
1
l
B
1
2
M
A
P
O
(
)
R
d
2
l
Figure 3. Analysis Di
agram
of Handoff Process
In Figure 3, MN move
s in
and out of
WLAN
at a con
s
tant spee
d a
l
ong a
straig
h
t
line.
1
l
,
2
l
are p
o
sitio
n
s
of
MN mo
ving in
and
out of
WLA
N
re
sp
ectively.
1
l
,
2
l
can
be
s
e
lect
e
d
at
rand
om withi
n
the WLA
N
cove
rag
e
,
a
l
ong with co
rrespon
ding
a
ngle
1
an
d
2
, whi
c
h
are
evenly di
strib
u
ted in
]
2
,
0
[
. It’s
as
sumed that
MN’s
RSS is
min
RSS
when
at
po
sition
1
l
, and
that it is
out
th
RSS
_
whe
n
at position
B. In the proce
ss of
MN
moving away from AP, RSS is to
decrea
s
e, an
d it decre
ase
s
to
min
RSS
when m
o
ves to positi
on
2
l
, where is the covera
g
e
edge
of
AP.
1
,
2
are e
s
timat
ed han
doff delay for MN
moving in an
d out of WLA
N
re
spe
c
tivel
y
.
t
is
th
e r
e
s
i
de
nc
e
time o
f
MN
in WL
AN
. If
t
is
sh
orte
r t
han
1
, that is
,
MN moves out of WLA
N
before th
e ha
ndoff pro
c
e
ss is co
mpleted
,
handoff failu
re eme
r
g
e
s.
Otherwise, if
t
is long
er tha
n
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An Adaptive
Vertical
Han
d
o
ff Algorithm
Based o
n
UM
TS and WLA
N
(Jin Ling
)
3947
1
but i
s
sho
r
ter than th
e
sum
of
1
and
2
, ‘ping-po
ng
effect’
eme
r
ge
s; M
e
anwhile, the
hand
off
is unn
ecessa
ry.
2.3.1. The Compensa
tin
g
Time
Con
s
id
erin
g t
hat ha
ndoff p
r
ocess nee
d
extra
signali
n
g ove
r
hea
d, i
t
is n
e
cessa
r
y for M
N
to hand
off to WLAN,
whe
n
it dwell
s
in
WLAN fo
r a
prop
er time.
On the oth
e
r ha
nd, if the
perfo
rman
ce
of WLAN i
s
very good
a
nd han
doff d
e
lay is small,
a sho
r
ter
re
side
nce time
is
enough for M
N
to
be beneficial in
WLA
N
. In order to evaluate
necess
ity of vert
ical
handoff, it’s
necessa
ry to
determi
ne th
e shorte
st
re
side
nc
e t
i
me
f
o
r M
N
in
W
L
A
N
,
w
h
ich
co
mpri
se
s h
and
of
f
delay for MN moving i
n
a
nd out
of WL
AN i.e.
1
,
2
, and the
com
p
e
n
satin
g
time
T
. By this
way, the sh
o
r
test resi
den
ce time i
s
de
sign
ated by
2
1
T
.The comp
en
sating time
T
relates
to
1
,
2
, the netwo
rk
context [1
0] and net
work fe
e, as
sho
w
n in
Fig
u
re 2. Th
e
cal
c
ulatio
n of
T
is com
p
lied
with the follo
wing meth
od.
Con
s
id
erin
g data tran
smi
s
sion rates p
r
ovided by wireless net
works a
nd fee
s
for a
cce
ss
to them, the revenue fun
c
ti
on for u
s
er
i
in the network
j
can b
e
expre
s
sed a
s
follo
ws:
)
,
(
,
,
,
j
i
j
i
j
i
C
B
f
E
(9)
Whe
r
e
j
i
B
,
is the band
width provided by net
work
j
to user
i
,
j
i
C
,
is the fees requi
red to p
a
y
for net
wo
rk
j
.
The
reven
u
e
functio
n
j
i
E
,
is t
he result of n
o
rmali
z
e
d
an
d weighted
summatio
n
of ea
ch
pa
rameter. A
ccordin
g to
th
e u
s
e
r
's de
mand, th
e
correspon
ding
wei
ghts to
each
para
m
eter
which a
r
e refe
rre
d to as
B
i
w
,
and
C
i
w
,
are assign
ed re
sp
e
c
tively. Also, these
weig
hts satisf
y
equation:
1
,
,
C
i
B
i
w
w
and
can b
e
a
d
juste
d
app
ro
priately a
c
cording to u
s
e
r
s’
busi
n
e
ss n
e
e
d
s an
d prefe
r
ences. Fo
r co
nvenien
ce,
lo
garithm to the
param
eter of
the formula:
j
i
C
i
j
i
B
i
j
i
C
w
B
w
E
,
,
,
,
,
ln
ln
(10)
Her
e
in
1
,
i
E
,
2
,
i
E
is d
e
fined
as the
reve
nue
fun
c
tion
of u
s
e
r
i
ac
cess to
U
M
TS
and
WLAN
res
p
ec
tively. Then for us
er
i
, th
e revenu
e di
fference bet
wee
n
acce
ss to UMTS and
WLAN
can b
e
expre
s
sed
as:
3
,
1
,
2
,
,
1
,
2
,
,
3
,
1
,
2
,
ln
ln
ln
ln
i
i
i
C
i
i
i
B
i
i
i
i
i
E
C
C
w
B
B
w
E
E
E
E
(11)
Whe
r
e
3
,
i
E
represe
n
ts ha
nd
off signalin
g overhe
ad [11
]. By Equation (10
)
, we can se
e the
greate
r
i
E
is, i.e. the revenu
e for user
i
in WLAN i
s
rel
a
tive higher,
the grate
r
ne
ce
ssity o
f
hand
off to WLAN will b
e
. If MN sw
itch
es to WLAN
for enjoying
WLAN
re
sou
r
ce, the short
e
st
time comprises
compensating time
T
an
d the
han
doff delay fo
r
mo
ving in
and
o
u
t of
WLA
N
i.e.
1
,
2
. Where c
o
mpensating time
T
can b
e
determi
ned
by the followi
ng equ
ation
s
:
1
2
)
2
1
3
,
1
,
3
,
2
,
1
,
)
(
(
i
i
i
i
i
E
E
E
E
E
e
T
e
T
e
(12)
else
e
E
e
e
E
T
E
i
E
E
i
i
i
,
0
)
1
1
ln(
0
,
1
1
0
,
1
2
2
1
3
3
,
(13)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3943 – 39
54
3948
Whe
r
e the
si
gnalin
g overh
ead resulting
from
vertical
hand
off is ref
e
rred to a
s
3
,
i
E
, and it can b
e
set to
a fixe
d value
grea
ter than
0.
By Equation
(13
)
, the
greater
i
E
is,
th
e shorte
r th
e
comp
en
sat
i
n
g
t
i
me
T
will be. If
i
E
is a neg
ative numbe
r,
T
will be infinite
and the verti
c
al
hand
off is un
necessa
ry; If
i
E
is not too large,
T
c
h
a
nge
s
w
i
th
i
E
and
hand
off delay; If
i
E
is large
r
than
a function va
lue,
T
can b
e
ig
nore
d
.
2.3.2. Speed Thresh
old
The sp
eed th
reshold
max
v
in Equation (8)
can be calcul
ated by the followin
g
met
hod.
Acco
rdi
ng to
sectio
n 3.2.1, if MN’s
d
w
ell time in
WLAN i
s
larger than
2
1
T
, vertic
al
hand
off to WLAN i
s
ne
ce
ssary.
Herein
a
p
is defin
ed a
s
unne
ce
ssary
hand
off pro
b
ability and it
can b
e
expre
s
sed a
s
:
)
(
)
(
2
2
1
t
p
p
T
t
p
p
f
a
(14)
Whe
r
e
t
is e
s
timated resid
e
n
ce time i
n
WLAN
and th
e pro
bability den
sity functi
on (P
DF) fo
r
t
can b
e
determined by the followin
g
method.
The PDF a
n
d
the joint prob
ability density
at position
1
l
and
2
l
can b
e
expre
s
sed a
s
:
otherwise
f
f
l
l
,
0
]
2
,
0
[
,
,
2
1
)
(
)
(
2
1
2
1
2
1
(15)
otherwise
f
,
0
2
0
),
2
1
(
1
)
(
(16)
MN’s m
o
vem
ent distan
ce
2
1
l
l
can b
e
expre
s
sed a
s
:
)
cos
1
(
2
2
2
R
vt
(17)
)
cos
1
(
2
)
(
2
2
v
R
g
t
(18)
i
t
is the root for
)
(
g
t
can be expressed a
s
:
i
i
i
t
t
t
g
f
t
f
)
(
'
)
(
)
(
(19)
)
cos
1
(
2
sin
)
(
'
v
R
g
(20)
2
2
2
4
1
)
(
'
R
t
v
v
R
g
(21)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An Adaptive
Vertical
Han
d
o
ff Algorithm
Based o
n
UM
TS and WLA
N
(Jin Ling
)
3949
)
2
1
(
1
)
(
i
i
t
t
f
(22)
The PDF of
t
can b
e
cal
c
ul
ated by Equa
tion (19
)
ca
n be expre
s
sed
as:
otherwise
v
R
t
t
v
R
v
otherwise
v
R
t
g
f
g
f
t
f
t
t
t
t
,
0
2
,
4
2
,
0
2
,
)
(
)
(
)
(
)
(
)
(
2
2
2
2
2
1
1
(23)
The dist
ributi
on functio
n
of
t
can be expressed a
s
:
v
R
T
v
R
T
dt
t
v
R
v
dt
t
f
T
F
T
T
2
,
1
2
0
,
4
2
)
(
)
(
0
2
2
2
0
(24)
By the distri
bution fun
c
tion (2
4) a
nd
Equat
ion (14
)
, unne
ce
ssa
r
y hand
off probability ca
n
be
obtaine
d:
2
1
2
1
2
1
2
1
2
,
1
2
0
,
2
arcsin
2
)
(
R
v
R
v
R
v
T
t
p
p
a
(25)
The han
doff failure p
r
o
babi
lity is as follows:
1
1
1
1
2
0
),
2
arcsin(
2
2
,
1
)
(
R
v
R
v
R
v
t
p
p
f
(26)
By Equation (8), (25
)
and
(26), t
he spee
d threshold
can be obtai
ne
d:
)
2
sin(
2
2
1
1
a
p
T
R
v
(27)
)
2
sin(
2
1
2
f
p
R
v
(28)
)
,
min(
2
1
max
v
v
v
(29)
The propo
se
d hand
off algorithm flow
chart is
sho
w
e
d
in Figure 4.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3943 – 39
54
3950
Figure 4. Han
doff Algorith
m
Flow Chart
3. Results a
nd Discu
ssi
on
3.1. The RSS Smoothing Algorithm
Figure 5. Vari
ations in
RSS with Dista
n
ce betwe
en AP of WLAN a
nd MN
To validate t
he efficien
cy
of prop
osed
RSS smo
o
thing alg
o
rith
m based o
n
sup
port
function
s,
co
mpared with the
sm
ooth
coefficient filte
r
ing al
go
ri
th
m, the simul
a
ting exampl
e of
variation
s
in RSS with distan
ce betwe
en AP and
MN is given.
As shown in Figure 5, the
prop
osed alg
o
rithm can effectively redu
ce
the influen
ce of shad
ow
effect on RS
S.
3.2. The Vertical Ha
ndo
ff Decision Alg
o
rithm
In the literatu
r
e [8], the
HNE algorith
m
i.e.
adaptive ti
me thre
sh
old
adju
s
tment al
gorith
m
is ado
pted in
the moving i
n
WLA
N
sce
nario,
while t
he traditio
nal
algorithm
m
o
stly adopt fi
xed
RSS thre
sh
ol
d. By usin
g M
a
tlab
simulati
on, alo
ng
with expe
riment
al pa
ramete
rs sh
own in
Ta
ble
1, the handoff
performan
ce
of three algo
rithms i
s
co
m
pare
d
with ea
ch othe
r.
1
2
3
4
5
6
7
8
9
10
-6
5
-6
0
-5
5
-5
0
-4
5
-4
0
-3
5
-3
0
-2
5
-2
0
-1
5
d
/
dref
R
S
S
(
dbm
)
T
he al
gori
t
hm
T
he i
deal
av
erage of
R
S
S
S
m
oot
hi
ng c
oef
f
i
c
i
ent
f
i
l
t
eri
ng al
gor
i
t
hm
M
eas
ur
ed R
S
S
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
An Adaptive
Vertical
Han
d
o
ff Algorithm
Based o
n
UM
TS and WLA
N
(Jin Ling
)
3951
1000
0 rand
o
m
traje
c
tori
e
s
a
r
e
gen
era
t
ed within
WLAN
cove
rag
e
, and
MN
movement
spe
ed rang
e
s
bet
ween
h
km
h
km
/
100
~
/
1
with an interval
of
h
km
/
2
. For each trajec
tory, a
positio
n
1
l
and
movement d
i
rectio
n
1
are
sele
cted, wh
en MN m
o
ve
s in WLAN,
whi
c
h a
r
e all
with ran
dom
uniform di
stri
bution withi
n
]
2
,
0
[
.
Table 1. System Simulatio
n
Paramete
rs
Radius of WLAN
R
150m
Transmission pow
e
r
of AP
Tx
P
20dbm
Distance betwee
n
AP and the r
e
f
e
rence point
ref
d
1m
Pass loss at reference point
Lre
f
P
40dbm
Pass loss
index
3.5
Standard deviati
on for shado
w eff
e
ct
4.3db
Handoff dela
y
for
moving from UM
TS to WLAN
1
2s
Handoff dela
y
for
moving from WALN to UM
TS
2
2s
Tolerable hand
of
f failure probability
f
p
0.02
Tolerable unnec
essary
hando
ff p
r
obabilit
y
a
p
0.04
In orde
r to evaluate ha
ndof
f efficiency, herein a
n
effici
ency fun
c
tion
i
U
is defined:
)
10000
(
)
(
1
,
3
,
2
,
N
N
N
e
N
N
e
U
failure
unnecessay
E
y
unnecessar
E
E
i
i
i
i
(30
)
Whe
r
e
2
,
i
E
and
1
,
i
E
are
reven
ue f
unctio
n
in
WLAN a
nd
UM
TS re
spe
c
tively for u
s
e
r
i
,
3
,
i
E
is
s
i
gnaling overhead res
u
lti
ng from vertic
al handoff,
N
is the total h
andoff times,
unnecessay
N
is
the unne
ce
ssary hand
off times, and
failure
N
is hand
off failure times.
I
f
3
,
i
E
is
set to
be a
fixed value, fo
r
example
2
ln
2
1
, the c
o
mpens
a
ting time in Equation
(13
)
ca
n be e
x
presse
d:
else
E
e
E
T
i
E
i
i
,
0
2
3
ln
0
,
)
1
(
2
0
,
1
(31
)
Figure 6
~ 9
sho
w
h
and
o
ff perform
an
ce co
mpa
r
iso
n
for th
ree
a
l
gorithm
s in
different
netwo
rk cont
ext. Where
0
i
E
, As sho
w
n in
Figure 6,
co
mpared
with
traditional
alg
o
rithm,
our proposed
algorithm can
effe
ctivel
y reduce handoff failure
probability
a
nd unnecessary
hand
off prob
ability to 0. Where
0
i
E
, the han
doff performa
n
ce i
s
sh
own in Figure 7.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3943 – 39
54
3952
Figure 6. The
Hand
off Performa
nce wh
e
n
MN Move
s into WLA
N
,
0
i
E
Figure 7. The
Hand
off Performa
nce wh
e
n
MN Move
s into WLA
N
,
0
i
E
Whe
r
e
2
3
ln
0
i
E
, as shown in Figu
re 8, the pro
posed alg
o
rit
h
m ca
n effectively reduce
hand
off failure prob
ability, and imp
r
ove
hand
off perfo
rman
ce.
0
50
100
0
0.
02
0.
04
0.
06
0.
08
0.
1
0.
12
v(
km
/
h
)
a
)
h
a
n
d
o
ff fa
i
l
u
re
p
r
o
b
a
b
i
l
i
t
y
h
a
nd
of
f
f
a
i
l
u
r
e
pr
oba
bi
l
i
t
y
0
50
100
0
0.
05
0.
1
0.
15
0.
2
0.
25
v(
km
/
h
)
b
)
u
n
n
eces
s
a
r
y
h
a
n
d
o
f
f
p
r
o
b
a
b
ilit
y
un
ne
ce
s
s
a
r
y
h
a
nd
of
f
pr
oba
bi
l
i
t
y
0
50
100
0.
8
1
1.
2
1.
4
1.
6
1.
8
2
v(
km
/
h
)
c
)
h
a
n
d
o
f
f
ef
f
i
ci
en
cy
ha
nd
o
f
f
effi
cien
cy
HNE
a
l
gor
ithm
T
h
e
a
l
gor
it
hm
T
r
a
d
i
tiona
l a
l
gor
it
hm
HNE
a
l
gor
ith
m
T
h
e
a
l
gor
ithm
T
r
a
d
ition
a
l a
l
gor
i
t
hm
HNE
a
l
gor
ithm
T
h
e
a
l
gor
it
hm
T
r
a
d
i
tiona
l a
l
gor
ithm
0
50
10
0
0
0.
0
2
0.
0
4
0.
0
6
0.
0
8
0.
1
0.
1
2
v(
k
m
/
h
)
a
)
ha
nd
of
f
f
a
i
l
ur
e
p
r
ob
a
b
i
l
i
t
y
ha
nd
of
f fai
l
ure
probabi
li
ty
0
50
10
0
0
0.
0
5
0.
1
0.
1
5
0.
2
0.
2
5
v(
km/
h
)
b
)
u
n
n
e
c
e
s
s
a
r
y
ha
nd
of
f
pr
ob
a
b
i
l
i
t
y
unnecessa
ry handof
f
probabil
i
t
y
0
50
10
0
1.
7
5
1.
8
1.
8
5
1.
9
1.
9
5
2
v(
km/
h
)
c)
h
a
n
d
o
f
f
ef
f
i
ci
en
c
y
handoff eff
i
ci
ency
HNE
a
l
go
r
i
th
m
T
h
e al
g
o
r
i
t
h
m
T
r
a
d
it
io
na
l
a
lg
or
i
t
h
m
HN
E
a
l
g
o
r
i
t
h
m
T
h
e al
g
o
r
i
t
h
m
T
r
a
d
it
io
na
l
a
l
go
r
i
th
m
HN
E
a
l
g
o
r
i
t
h
m
T
h
e al
g
o
r
i
t
h
m
T
r
a
d
i
t
i
o
n
a
l a
l
gor
it
hm
Evaluation Warning : The document was created with Spire.PDF for Python.