TELKOM
NIKA
, Vol.14, No
.1, March 2
0
1
6
, pp. 86~9
0
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i1.2358
86
Re
cei
v
ed
Jul
y
25, 201
5; Revi
sed
No
ve
m
ber 23, 201
5; Acce
pted
De
cem
ber 1
0
,
2015
Resear
ch on Load Balancing in C-RAN with Femtocells
Zhanjun Liu*
1
, Xia Peng
2
,Yujing
He
2
, Qichao Ma
2
1
Colle
ge of Co
mmunicati
on E
ngi
neer
in
g,Cho
ngq
ing U
n
iv
ers
i
t
y
,Ch
o
n
gqi
ng
400
03
0,Chi
n
a
2
Chon
gqi
ng Ke
y L
abor
ator
y
of
Mobil
e
Comm
unic
a
tions T
e
chno
log
y
,
Cho
ngq
in
g Uni
v
ersit
y
of Posts
and T
e
lecom
m
unic
a
tion, Ch
ong
qin
g
40
00
6
5
, P. R. China
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: liuzj@c
qu
pt.edu.cn, 45
466
0
432
@qq.com
A
b
st
r
a
ct
T
he trad
itio
nal
lo
ad
ba
la
ncin
g o
p
ti
mi
z
a
ti
on
sche
m
e
ca
nno
t guar
ante
e
th
e us
er
’
s
Q
o
S i
n
the
C
-
RAN w
i
rel
e
ss
communic
a
tio
n
netw
o
rks w
h
e
n
al
l the
ne
i
g
h
bori
ng macroc
ells are un
der
the
situ
atio
n of
overl
oad. In or
der to resolv
e this prob
le
m, w
e
im
pl
e
m
e
n
t the femtoc
ell t
o
the macroce
ll netw
o
rk, w
h
ich
consi
ders t
he
a
d
vanta
ges
of f
e
mtoc
el
l,
i
.
e
., sm
al
l
co
ve
ra
ge
,
l
o
w co
st
a
n
d
be
tte
r si
gn
al
qua
l
i
t
y. Th
e
r
e
f
o
r
e, i
n
this pa
per, the
joi
n
t loa
d
b
a
la
ncin
g
mech
an
i
s
m is
pro
pose
d
to ach
i
ev
e the j
o
int o
p
ti
mi
z
a
ti
on
of femt
o
c
el
l
and
macr
oce
ll
netw
o
rk by all
o
w
i
ng the us
ers
of overlo
ad
ed
macr
ocel
l to h
and
over to the
femtoc
ell. F
i
na
l
l
y,
the si
mu
lati
on
results sh
ow
th
at the
prop
ose
d
mech
an
is
m greatly i
m
pr
ov
es
the user
sat
i
sfaction
as w
e
ll as
the resourc
e
ut
ili
z
a
ti
on rate of
the femtoc
ell n
e
tw
ork
w
h
ile decreas
es the bl
ockin
g
rate an
d drop
pin
g
rate
.
Ke
y
w
ords
: F
e
mtoce
ll, Lo
ad
Bala
ncin
g, Join
t
Optim
i
z
a
tion, Reso
urce Util
i
z
ation
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
To meet the
gro
w
ing
cap
a
city dema
n
d
s in
wi
rel
e
ss co
mmuni
ca
tion netwo
rks, Cloud
Radi
o
Acce
ss Network (C-RA
N
) ha
s
b
een pro
p
o
s
e
d
for th
e p
u
rpose of
satisf
ying the
cap
a
city
requi
rem
ents
and
scala
b
ility probl
em
s [1
-2]. C-RA
N i
s
co
mpo
s
ed
of
a Ba
seBa
nd
Unit (BB
U
) a
nd
a remote
radio unit (RRU), an
d it also has better flexibility because it breaks the fixed
con
n
e
c
tion
relation
ship
b
e
twee
n BBU and
RRU [
3
]. More
over, som
e
n
o
ta
ble featu
r
e
s
are
inclu
ded
in
C-RA
N, e.g.,
centrali
zed
p
r
oce
s
sing,
coll
aborative
rad
i
o,
re
al-time
clou
d com
put
ing
infrast
r
u
c
ture,
and cle
an sy
stem [4].
Since the
di
stribution a
n
d
se
rvice
req
u
i
reme
nt
of the users a
r
e
st
ocha
stic in
C-RAN
wirel
e
ss a
ccess net
work
[5-6], therefo
r
e, the
loa
d
balan
cing te
chonol
ogy is i
nevitable in t
he
ca
se of
solvi
ng the p
r
o
b
le
m of unb
alan
ced
cell
and i
m
provin
g the
resou
r
ce utili
zation.
Ho
we
ver,
the tra
d
itional
load
bal
an
ci
ng
schem
es
only allo
w
th
e
users
of ove
r
loade
d ma
cro
c
ell to
ha
ndo
ver
to the n
e
igh
borin
g ma
cro
c
ell
s
. Con
s
e
quently, there will
be
no
suita
b
le n
e
i
ghbo
ring
cell
s to
sup
port the u
s
ers’ ha
ndov
er wh
en the
neigh
bori
ng ce
lls are also overloa
d
, an
d it also can
not
alleviate the unbal
an
ced p
r
oble
m
[9-11]
.
With the in
creases of th
e data u
s
age
of
wirele
ss
commu
nic
a
t
i
on sy
st
e
m
o
c
c
u
r
s
in
indoo
r, the rese
arch of in
door
se
rvice
are be
co
min
g
more a
nd
more imp
o
rta
n
t [12]. As the
femtocell
ha
s the adva
n
ta
ges
of sm
all
coverage, l
o
w cost
and
b
e
tter si
gnal
q
uality, therefo
r
e, it
has
bee
n se
lected
as th
e prim
ary
ca
nditate techn
o
logy to resolve the p
r
o
b
lem of in
do
or
coverage
in
the
future 5G wirel
e
ss communi
ction.
The Femto
c
ell Acce
ss
Point (FAP) of
femtocell an
d
existing networks as a b
a
ckha
ul
con
n
e
ctivity can meet the upcoming dem
an
d of
high d
a
ta rate and
extend
the coverag
e
area[13
-15]
. More
over, t
he coexiste
n
c
e of
ma
cro
cell
and femtocell
can also all
o
w the users of over
loade
d macro
c
ell to handove
r
to the femtocell.
Therefore, a j
o
int optimizati
on schem
e of load bal
a
n
ci
ng in C-RA
N two tier networks is
pro
p
o
s
ed
The re
main
d
e
r of this p
a
per i
s
organi
zed a
s
follo
ws: In Se
ctio
n 2, we de
ccrib
e
the
system m
ode
l. Section 3 a
nalyze
s
the
deta
iled p
r
o
c
edure of the
prop
osed
Joi
n
t Optimizati
on
Load Bla
n
ci
n
g
Base
d on
The Coexiste
nce
of Femt
ocell
s
(JO
L
B
)
me
chni
sm.
The si
mulati
on
results a
r
e di
scusse
d in Section 4.
We con
c
lu
de the
pape
r in Secti
on 5.
2. Sy
stem Model
In the C-RA
N radio a
c
cess network, wh
en the ce
ll is
overloa
d
, the netwo
rk tri
e
s
to move
the mobile u
s
ers of the
overloa
ded
cell to t
he ap
prop
riate fe
mtocell to a
c
hieve the l
oad
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Load Bala
nci
ng in C-RA
N with Fem
t
oce
lls (Zh
anju
n
L
i
u)
87
balan
cing
whi
l
e ma
ke
re
asonabl
e u
s
e
of
femtocell
resource
s. Th
e
system mo
del
is a
s
shown i
n
Figure 1.
Figure 1. System model
As we
can
se
e form Figu
re
1, there are t
h
ree m
a
cro
c
e
lls
RR
U1
,
RR
U2
an
d
RR
U3
in the
C-
RAN wi
rel
e
s
s
a
c
c
e
s
s
netwo
rk.
RRU
1
i
s
ove
r
l
oade
d,
RR
U2
and
RRU3
are the t
w
o
neigh
bori
ng
macro
c
ell
s
with different load level
s
.
APi-j
denote
s
the
i-th
femtocell in the
j-th
macroc
ell. If the
RR
U2
an
d
RR
U3
are
all und
er the
situation
of heavy overlo
ad, the u
s
e
r
s of
RR
U1
are n
o
t
allowed to
h
andove
r
to th
e
RRU
2
and
RR
U3
,
and
th
en the
u
s
ers
of
RR
U1
will be
hand
overe
d
to the femto
c
ell. For
example,
the u
s
ers of RRU1
will be
han
dovere
d
to t
h
e
femtocell
s
su
ch a
s
AP1-1
,
AP1-2
and
AP
1
-
3
for the p
u
rpo
s
e of rea
c
hin
g
the loa
d
balan
cing.
3. Joint Opti
mization Loa
d Blancing Based
on Th
e Coexis
t
en
ce
of Femtoc
ells(JOL
B)
The detaie
d
p
r
ocedu
re of JOLB algo
rith
m is as follo
ws:
Step1
: Co
mp
are the cell lo
ad with the predefine
d
overload thre
sh
ol
d. If
the load value is
greate
r
th
an
the thre
sh
old
value, it will
be tre
a
ted
as the ove
r
loa
d
ed
cell. Th
en
, to identify the
neigh
bori
ng
cell
s wh
ose load is
small
e
r than the
t
h
re
shol
d, an
d gene
rate a
target cell li
st
cell_li
st.
Step2
: Analy
z
e the
loa
d
state of neig
h
bor
cell
s in
the cell_li
st, a
nd then
arra
nge the
cell_li
st in ascen
d
ing o
r
de
r accordi
ng to the load state
.
Step3
: Get th
e ue_li
s
t in descen
d
ing o
r
der ba
se
d on
the sign
al st
rength of users .
Step4
: If the
cell_li
st is not
empty, hand
over the first
use
r
of the u
e_list to the first cell of
the cell_list,
and u
pdate
the u
e_list
an
d cell_list
ba
s
ed o
n
the
sig
nal
stren
g
th
of users a
nd
load
state of cell
s, sepa
rately. If the
cell_list
is empty, handover the first use
r
of the ue_li
s
t to the
adja
c
ent femtocell, an
d up
date the ue_li
s
t
based on t
he sig
nal stre
ngth of use
r
s
.
Step5
: Ch
eck if the cell is still overl
oad.
If the
cell is ove
r
l
oad, rep
eat
Step2-4
,
otherwise go
to
Step6
;
Step6
: Te
rmi
nate the load
balan
cing a
n
d
then wa
it fo
r the next load balan
cing
p
e
riod.
3. Simulation Resul
t
s an
d Analy
s
is
In this se
ctio
n, we describ
e the simulati
on enviro
n
me
nt and re
sult
s of the perfo
rman
ce
of the p
r
o
p
o
s
ed
scheme.
We
assu
me t
hat 37
RRUs
are
located
in
the n
e
two
r
k,
and th
e n
u
m
ber
of cells is 3
7
.
Morever, e
a
ch
RRU co
vers on
e cell
, and the users a
r
e rand
omly distribut
ed in
each cell. Th
e system
mo
del is a
s
sho
w
n in Fi
gure
1, whi
c
h ind
i
cate
s the di
stributio
n of the
use
r
s a
nd ba
se statio
ns, a
nd the simul
a
tion para
m
ete
r
s a
r
e liste
d in Table 1.
The sim
u
lation results m
a
inly focus on t
he resources ut
ilization, the number of the
unsatisfied u
s
er, blo
ckin
g
rate
a
nd cal
l
drop
pi
ng
ra
te of the JO
LB algo
rithm. The
comp
ared
algorith
m
s in
clud
e No L
o
ad Balan
c
ing
(NLB), Lo
ad
Balancin
g o
n
ly with macrocell
s
(LB
)
, and
Joint Otimi
z
at
ion Loa
d Blan
cing b
a
sed o
n
the coexi
s
tence of femtocell (JOLB
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 86 – 90
88
Table 1. Simulation pa
ram
e
ters
Parameter
Value
RRU distance
500m
RRU tra
n
smit pow
e
r
46dBm
Mobility
mode
velocity
：
5km/h,3
0
km/h,60km/h
direction
：（
0,360
）
Load balancing p
e
riod
200ms
The pass-loss model
L=32.44+20lgd(k
m
)+ 20lgf(Mhz)
In Figure
2, we can
see t
hat with the i
n
cr
ea
se
s of the user a
rriv
a
l rate, the a
v
erage
resou
r
ces util
ization of JO
LB is large
r
than t
he NLB
and LB sche
mes. Thi
s
is becau
se wh
e
n
the
neigh
bori
ng
macro
c
ell
s
are in heavy lo
ads, the LB
can only tra
n
sf
er a
small a
m
ount of users to
the ma
crocell
s
u
n
til it is in
overloa
d
. Ho
wever,
the
JOLB
can
shift the u
s
e
r
of the ove
r
loa
d
e
d
cell
to femtocell
whe
n
the n
e
i
ghbo
ring
ma
crocells are
in
overloa
d
. Th
erefo
r
e, the
JOLB sch
e
me
can
improve the resource ut
ilization of both macrocell
s
and femtocell
s
Figure 2. Average
re
sou
r
ce utilization rate
In Figure 3,
we
can
se
e that with the
increa
se
s of t
he u
s
e
r
arriv
a
l rate, the
n
u
mbe
r
of
unsatisfied
u
s
er of
JO
LB i
s
smalle
r th
a
n
the
NLB
an
d LB
schem
e
s
. Thi
s
i
s
be
cause
whe
n
th
e
neigh
bori
ng
cell
s are in o
v
erload, the
LB and
NL
B
scheme
s
can
not tran
sfer t
he u
s
ers to t
h
e
neigh
bori
ng cells, whe
r
ea
s
the
JOLB scheme ca
n
tr
a
n
sfer
the use
r
s of
the ove
r
load
ed cell
t
o
femtocell, whi
c
h en
su
re
s the use
r
’s q
uali
t
y of service (QoS).
Figure 3. Nu
mber of un
sat
i
sfied u
s
ers
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
0.
7
0.
75
0.
8
0.
85
0.
9
0.
95
A
rriv
a
l R
a
te
(
UE
/
s
)
A
ver
a
g
e
R
e
s
ou
r
c
e
U
t
ili
za
ti
on
R
a
t
e
(%
)
JO
L
B
LB
NL
B
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
0
10
20
30
40
50
60
70
80
A
rri
v
a
l R
a
t
e
(
UE
/
s
)
Nu
mb
e
r
o
f
Un
s
a
t
i
s
f
i
e
d
Us
e
r
JO
LB
LB
NL
B
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Load Bala
nci
ng in C-RA
N with Fem
t
oce
lls (Zh
anju
n
L
i
u)
89
Figure 4 is the blocking
rat
e
of the network by ap
plying di
fferent
mech
ani
sm
s. From the
Figure
4,
we can se
e
that with
the
in
cre
a
se of
the u
s
er a
rrival
rate
, the JO
LB schem
e h
a
s t
he
minimum bl
o
cki
ng rate, which in
dicates that the JOL
B
can de
crea
se the bl
ocki
ng rate of u
s
ers
to meet the access requi
re
ment of use
r
s.
Figure 4. Blocki
ng rate
In Figure
5, we can
see t
hat with the i
n
crea
se
of th
e use
r
a
rrival
rate, the net
work
call
drop
rate
also incre
a
se
s.
Und
e
r th
e sa
me arrival rate, the call droppin
g
rate
o
f
the JO
LB is the
least
while
th
at of NLB is the bi
gge
st. F
u
rthe
rmo
r
e,
whe
n
the
use
r
a
rrival
rate i
s
1, th
e
call d
r
op
rate of
JOLB
is zero, whi
c
h indi
cate
s th
at the
han
do
ver re
quirem
ents of all
users are satisfi
ed.
Whe
n
the arri
val rate is 3, the call drop rate of
JOLB drop
s by 15
%, which me
ans that the JOLB
improve
s
the
hand
over succe
ss
ratio.
Figure 5. Call
droppi
ng rate
4. Conclusio
n
In the C-RA
N wirel
e
ss
net
work, a j
o
int
optimizatio
n l
oad b
a
lan
c
in
g algo
rithm b
a
se
d on
the co
existen
c
e of femto
c
e
lls is propo
se
d for th
e
pu
rp
ose
of providi
ng bette
r u
s
e
r
service
whe
n
all the neighb
oring m
a
cro
c
ells are overl
oade
d.
Our p
r
opo
se
d algo
rothm can m
e
et the incre
a
sing
servi
c
e
dem
a
nds an
d h
u
g
e
data
excha
nge
by intr
o
d
u
cin
g
the
C-RAN
wi
rele
ss acce
ss n
e
twork.
More
over, th
e loa
d
bal
an
cing
problem
is i
nevitable
in
C-RAN d
ue to th
e ra
ndomn
e
ss
of the
use
r
’s lo
catio
n
an
d
servi
c
e re
que
st. A
s
the
tra
d
itional lo
ad
bala
n
cin
g
me
ch
a
n
ism
s
of
C-RAN
can
not work
well
whe
n
all
the neig
hbo
ring m
a
cro
c
e
lls are ove
r
lo
aded. T
herefore, in
ord
e
r to
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
0.
45
0.
5
A
r
r
i
v
a
l
ra
t
e
(U
E
/
s
)
B
l
oc
k
i
ng R
a
t
e
(
%
)
JOLB
LB
NL
B
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
A
r
ri
val
R
a
t
e
(U
E
/
s
)
C
a
ll
D
r
oppi
ng
R
a
te
(%
)
JO
LB
LB
NL
B
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 86 – 90
90
resolve that
kind of
hand
over p
r
o
b
lem,
we take th
e
fe
mtocell
as th
e targ
et ha
nd
over
cell to
all
o
w
the use
r
s of overload
ed ma
cro
c
ell to h
a
n
dover to
the femtocell. Sim
u
lation re
sult
s sh
ow that t
h
e
prop
osed me
cha
n
ism
com
pare
d
with
t
he
LB and
NLB schem
e
s
can
redu
ce
the numb
e
r
of
unsatisfied
u
s
ers, lo
we
r the blo
c
king
rate and
drop
ped
call
rate
while i
m
pro
v
e the avera
g
e
network resource
utilization.
Ackn
o
w
l
e
dg
ements
This work wa
s su
ppo
rted
by 863 proj
ect (2014AA01
A
701) an
d Progra
m
for Ch
angjia
ng
Schola
r
s and
Innovative Re
sea
r
ch T
eam
in University (IRT129
9).
Referen
ces
[1]
Limin
g
Che
n
, Hu
Ji
n,
H
aomi
ng Li.
An
Energ
y
Effi
ci
en
t Im
pl
em
en
ta
ti
on o
f
C
-
R
A
N
i
n
H
e
tN
e
t
. IEEE
Vehic
u
lar T
e
chnol
og
y Co
nfere
n
ce (VT
C
F
a
ll). 2014: 1-5.
[2]
Henrik
Ho
lm, Aleksa
ndr
a
Check
o
, Rami
Al-ob
a
id
i.
Optimal Ass
i
gn
me
nt of C
e
ll
s in C-
R
A
N
Depl
oy
me
nts w
i
th Multiple BBU Pool
s
. IEEE Eur
opean Confer
ence on Net
w
orks a
n
d
Commun
i
cati
o
n
s (EuCNC). 2
015: 15-
19.
[3]
Z
hanj
un
L, Z
h
i
c
hao
Z
,
Shi
y
a
n
Z
.
Co
op
erati
v
e
R
RU S
e
lec
t
ion A
l
gor
ithm
for Multic
ast S
e
rvice
in
C-
RAN Wireless
Net
w
ork.
Jour
n
a
l of XiDi
an Un
iversity
. 201
3; 40(6): 16
8-1
7
3
.
[4]
Li G, Zhang S,
Yang
X
.
Arch
i
t
ecture of GPP
Base
d, Scal
a
b
le,
Larg
e
-scal
e
C-RA
N BBU
Poo
l
. IEEE
Globec
om W
o
rkshops. 20
12:
267-
272.
[5]
Xi
ao
hu Ge,
T
ao Han, Yan Z
han
g, Guoqi
an
g M
ao. Spectrum and En
erg
y
Efficienc
y E
v
alu
a
tion o
f
T
w
o-T
i
er F
e
mtocell N
e
t
w
ork
s
w
i
t
h
Partia
ll
y Ope
n
Ch
an
nels.
IEEE Transactions on Vehicular
T
e
chno
logy
. 2
014; 63(
3): 130
6-13
19.
[6]
Hua
n
D, Qi
a
nbi
n C,
Xian
w
e
i
C. A P
o
w
e
r
C
ontro
l
Selecti
on A
l
g
o
rithm of
Ad
aptive
Servic
e
Enviro
nment in
LT
E SON.
Jo
urna
l of Cho
n
g
q
in
g Univ
ersity
of Posts and T
e
leco
mmunic
a
tions
. 20
11
;
23(6): 66
1-6
6
4
.
[7]
Andre
a
s Lo
bi
n
ger, Sz
y
m
on
Stefanski.
Lo
a
d
Bal
anci
ng i
n
Dow
n
link
LT
E
Self-Optimi
z
i
n
g Netw
orks
.
IEEE Vehicul
a
r
T
e
chnol
og
y
C
onfere
n
ce. 20
1
0
: 1-5.
[8]
Shia
o-Li
T
s
ao, Ch
en-W
e
i
W
ang, Y
u
n
-
Ciou
Li
n.
A
Dyna
mic L
oad-
bal
anc
ing
Sche
me for
Hetero
gen
eo
u
s
Wireless Ne
tw
orks
. IEEE Wireless
Com
m
unic
a
tions
and Net
w
orking Conference.
201
4: 307
0-30
75.
[9]
Den
g
y
ua
n
Xu, Shi
x
u
n
W
u
, Benni
u Z
han
g, Xi
aoq
in
Qin. Po
w
e
r Bal
ance A
O
DV Algorithm
of W
S
N in
Agricult
ure Mo
nitori
ng.
TEL
K
OMNIKA
. 2013
; 11(4): 811-8
1
8
.
[10]
XIN K, RUI Z
.
Price-b
a
sed re
source a
lloc
a
ti
on for spectru
m-sharin
g femtocell
net
w
o
rks:
a stackelb
e
r
g
game a
ppr
oac
h.
IEEE Journal on Selected A
r
eas in Comm
unications
. 20
12
; 30(3): 538-5
4
9
.
[11]
Zhanjun Liu, Q
i
chao Ma, Cong
Ren, Qianbin Chen. Load B
a
lancin
g Bas
e
d on the Specific Offset of
Han
dover,
T
E
L
K
OMNIKA Indones
ian J
ourn
a
l of Electrica
l
Engi
neer
in
g
. 2014; 12(
8): 628
1-62
90.
[12]
Cha
ndras
ekh
a
r
V, Andre
w
s
J, Gat
herer A.
F
e
mtocell
Ne
t
w
orks:
a Surv
e
y
.
IEEE Comm
unic
ations
Maga
z
i
ne
. 2
0
0
8
; 46(9): 59-6
7
.
[13]
Xia
o
h
u
Ge, T
ao H
an, Y
a
n
Z
hang, Gu
oq
ia
ng M
ao.
Sp
ectrum an
d E
ner
g
y
Efficie
n
c
y
Evalu
a
tion
of
T
w
o-T
i
er F
e
mtocell N
e
t
w
ork
s
w
i
t
h
Partia
ll
y Ope
n
Ch
an
nels.
IEEE Transactions on Vehicular
T
e
chno
logy
. 2
014; 63(
3): 130
6-13
19.
[14]
Yida
n Z
hua
ng, Su Z
hao, Xia
o
r
ong Z
hu.
A ne
w
handov
er mecha
n
is
m for femtoc
el
l-to-fe
m
toce
ll
. 20
12
Internatio
na
l C
onfere
n
ce o
n
W
i
reless Com
m
unic
a
tions &
Sign
al Process
i
ng (W
CSP). 2012: 1-4.
[15]
KUANG X, YANG Q,
F
U
F
.
Resource al
loca
ti
on for fe
mtocell net
w
o
r
ks
w
i
th imperf
e
ct channe
l
kno
w
l
e
d
ge.
IEEE Signal Proc
essing Comm
unications and
Com
p
uting
. 2
0
11: 1-5.
Evaluation Warning : The document was created with Spire.PDF for Python.