Post on 05-Aug-2020
Resource Allocation in Full Duplex Heterogeneous
Networks with eICIC
Jun Zhang and Hui Tian State Key Laboratory of Networking and Switching Technology,
Beijing University of Posts and Telecommunication, Beijing 100876, China
Email: buptzhangjun@163.com; tianhui@bupt.edu.cn
Abstract— In this paper, we consider a two-tier heterogeneous
Orthogonal Frequency Division Multiple Access (OFDMA)
network which consists of one Full Duplex (FD) macrocell and
several FD femtocells. Enhanced Inter-Cell Interference
Coordination (eICIC) by means of Almost Blank Subframe
(ABS) is considered to reduce the cross-tier interference. We
formulate a resource allocation problem to maximize the sum-
rate of the network by jointly allocating subcarrier and power
among user equipments (UEs). The problem is solved by a sub-
optimal algorithm, which greedily allocates subcarriers based on
necessary conditions of the optimization problem and then
distributes power of Base Stations (BSs) and UEs by Iterative
Water Filling (IWF) procedure. Simulation results reveal that
the proposed subcarrier and power allocation algorithm improve
the throughput over the Time Division Duplexing (TDD) Half
Duplex (HD) system and the FD system with a round-robin
scheduler. The FD gain of the macrocell can achieve 45% when
the ABS scheme is applied, while the FD technique almost
doubles the throughput of the femtocells.
Index Terms—Full duplex, OFDMA, resource allocation,
subcarrier and power allocation
I. INTRODUCTION
Recent studies have shown that next generation
wireless communication systems are expecting to
accommodate greatly growing mobile traffic [1], [2].
However, traditional bi-directional communication is
achieved by either frequency division duplexing (FDD)
or Time Division Duplexing (TDD) [3], which suffers
from low spectrum efficiency compared with Full Duplex
(FD) technology. FD technology enables a node to
transmit and receive simultaneously on the same
frequency band and thus potentially doubles the spectral
efficiency. Recent progress in signal processing can
efficiently cancel the Self-Interference (SI) in FD and
brings in-band wireless FD into reality [4], [5]. FD
technology, as a candidate of future 5th Generation (5G)
radio access technology, has received lots of attention
from the academic and industrial community [6]-[10].
Manuscript received January 11, 2016; revised June 21, 2016.
This work was supported by the National Science and Technology Major Project under Grant No. 2015ZX03001025-002 and Huawei
Company. Corresponding author email: buptzhangjun@163.com.
doi:10.12720/jcm.11.6.550-557
The main difficulty of FD technology to realize the
almost doubled spectral efficiency lies in the strong SI
imposed by the signal leakage from the transmitter
antennas of a FD node to its own receiver antennas. The
techniques to handle SI cancellation can be generally
categorized into three different types, namely the antenna
cancellation, the analog cancellation and the digital
cancellation. Antenna cancellation technique relies on
directional isolation, absorptive shielding and cross-
polarization to isolate the transmit and receive antennas
[11], [12]. Analog cancellation methods generate a
delayed and attenuated copy of the transmit signal, and
then subtract it from receive signal [13], [14]. After
digitalize the analog receive signal, digital cancellation
can clean out residual SI caused by non-linear RF chains
[14], [15]. The three cancellation techniques are usually
combined to obtain high SI cancellation gain. Recently SI
cancellation value up to 110dB has been reported [13].
The FD Medium Access Control (MAC) layer design
has been investigated in cellular networks. A single small
cell was considered in [16]-[19]. A hybrid scheduler that
adapts to FD or Half Duplex (HD) operation according to
the link channel and interference conditions was proposed
in [16], [18]. Sub-optimal joint subcarrier and power
allocation algorithms were considered to maximize the
system throughput in [17], [19]. On the other hand, the
multi-cell scenarios were also investigated in [7], [20]
and [21]. The performance of FD communication was
evaluated in a dense small cell scenario and 30-40%
mean throughput gain over HD transmission was
observed [7]. A heuristic greedy scheduling algorithm
and power allocation based on geometric programming
were proposed in [20] and 60% FD gain was show in
outdoor pico cell deployment. A utility function defined
by the sum of the logarithm of users’ data rates was
maximized to maintain fairness among FD users [21].
However, few literatures addressed the resource
allocation in FD heterogeneous networks. The network
throughput of a heterogeneous network was derived by
accounting the node density, and SI cancellation value
using the stochastic geometry [22]. The closed-form
approximations for sum ergodic capacity have been
derived in interference coordinated heterogeneous
networks but Equal Power Allocation (EPA) was
assumed [23]. The FD gain achieved by joint subcarrier
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 550
and power allocation in heterogeneous networks has not
been exploited, which is the aim of this paper.
The 3GPP has proposed the enhanced Inter-Cell
Interference Coordination (eICIC) to alleviate the cross-
tier interference in co-channel deployed heterogeneous
networks in Release 10 [24]. One of the features is the
almost blank subframe (ABS). The strong interfering
nodes are muted in both data and control channels and
only reference signal is transmitted during the ABS
subframe. This allows the victim nodes to schedule their
users without cross-tier interference. The ABS scheme is
also considered in this paper.
In this paper, we consider a joint subcarrier and power
allocation problem in a FD heterogeneous network. A
sub-optimal algorithm is proposed by firstly assigning
subcarrier to uplink and downlink users while considering
the users’ minimum data rate constraints. Secondly the
power is distributed by Iterative Water Filling (IWF)
algorithm. Furthermore, we investigate the effect of
interference coordination scheme on the FD gain. We
compare the performance of the proposed algorithm with
that of a TDD HD system as well as a FD system with
Round-Robin (FD-RR) scheduler. The simulation results
show that the proposed resource allocation scheme can
improve the throughput over TDD HD and FD-RR
systems. The FD gain is enhanced when the interference
coordination scheme is applied in the heterogeneous
network.
The rest of this paper is organized as follows. Section 2
provides the system model. In Section 3, we formulate
the joint subcarrier and power allocation problem. A sub-
optimal algorithm is proposed in Section 4 followed by
simulation results in Section 5. Finally, Section 6
concludes this paper.
II. SYSTEM MODEL
We consider a heterogeneous network consists of one
macrocell and a set of femtocells. The heterogeneous
network operates in a single-input and single-output
(SISO) orthogonal frequency division multiple access
(OFDMA) system with N subcarriers. A FD base station
(BS) is located at the center of the macrocell and serves
randomly distributed HD uplink and downlink macrocell
users (MUE). The number of the uplink and downlink
MUE are U and D , respectively. The radius of the
macrocell is MR . The femtocells are co-channel deployed
with the macrocell. Each femtocell communicates with its
femtocell user (FUE) in its Close Subscriber Group (CSG)
within a radius FR . For simplicity, we assume there is
only one FUE in every femtocell. Both the femtocell and
the FUE operate in FD mode. Let us denote the set of
uplink MUEs, downlink MUEs and subcarriers as UL , DL and respectively. Since we assume there is a
single FUE per femtocell, we use to denote the set of
femtocells or the set of FUEs without causing much
ambiguity. In the considered heterogeneous network, the
femtocells are strong disrupters and should apply ABS to
avoid causing severe cross-tier interference to the MUEs
in their proximity. The ABS ratio is denoted as .
III. PROBLEM FORMULATION
In this section, we present the data rates in normal
subframes and ABS subframes. Then, we formulate a
joint subcarrier and power allocation problem to
maximize the sum rate of the heterogeneous network. We
also derive a criterion to assign subcarriers based on
Karush-Kuhn-Tucker (KKT) conditions.
A. Date Rates in Normal Subframes
The macrocell BS simultaneously receives signal from
one of its uplink user and transmits signal to one of its
downlink user. At the same time, the co-channel
femtocells generate interference to the macrocell in
normal subframes. The macrocell schedules the thi uplink
MUE, and thk downlink MUE on the thn subcarrier. The
data rate of the thi uplink MUE on the thn subcarrier is
given as
, ,,UL
, 2
,
0,
1
MU UL
i n i nM
ik n MB
k n
FB FU nMB
SI
PR log
P
H
I IC
(1)
where 1
, ,
FB
FB l n l n
l
I P H and 2
, ,
FU
FU l n l n
l
I P H are the
interference generated by FD femtocell BSs and FUEs,
respectively. ,
MU
i nP , ,
MB
k nP , ,
FB
l nP and ,
FU
l nP denote the
transmit power of the thi uplink MUE, the macrocell BS,
the thl femtocell BS and the thl FUE on the thn subcarrier,
respectively. ,
UL
i nH , 1
,l nH and 2
,l nH are the channel from
the thi uplink MUE to the macro BS, the thl femtocell BS
to the macro BS and the thl FUE to the macro BS,
respectively. MB
SIC denotes the SI cancelation value at the
macro BS [16]. 0, n is the noise power on the thn
subcarrier.
The data rate of the thk downlink MUE on the thn
subcarrier is given as
, ,,DL
, 2 ' '
, , 0,
(1 )
MB DL
k n k nM
ki n MU
i n ki n FB FU n
PR log
P H I I
H (2)
where ' 3
, ,
FB
FB l n l n
l
I P H and ' 4
, ,
FU
FU l n l n
l
I P H are the
interference generated by the FD femtocell and FUEs,
respectively. ,
DL
k nH , ,ki nH , 3
,l nH and 4
,l nH denote the
channel from the macrocell BS to the thk downlink MUE,
the thi uplink MUE to the thk downlink MUE, the thl
femtocell BS to the thk downlink MUE and the thl FUE
to the thk downlink MUE, respectively.
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 551
The data rate at the thl femtocell BS in the uplink
transmission on the thn subcarrier is given as
, ,,UL
, 2
, 7
, , 0,
1
FU UL
l n l nF
l n FB
l n MB
FB FU k n k n nFB
SI
P HR log
PJ HJ P
C
(3)
where 5
, ,
,
FB
FB k n k n
k k l
J P H and 6
, ,
,
FU
FU k n k n
k k l
J P H are
the interference generated by co-channel femtocell BSs
and FUEs, respectively. ,
UL
l nH , 5
,k nH , 6
,k nH and 7
,k nH are
the channel from the thl FUE to the thl femtocell BS, the
thk femtocell BS to the thl femtocell BS, the thk FUE to
the thl femtocell BS and the macrocell BS to the thl
femtocell BS, respectively. FB
SIC is the SI cancelation
value at the femtocell BS.
The data rate at the thl femtocell BS in the downlink
transmission on the thn subcarrier is given as
, ,,DL
, 2
, ' ' 10
, , 0,
1
FB DL
l n l nF
l n FU
l n MB
FB FU k n k n nFU
SI
P HR log
PJ J P
CH
(4)
where ' 8
, ,
,
FB
FB k n k n
k k l
J P H and ' 9
, ,
,
FU
FU k n k n
k k l
J P H are
the interference from co-channel femtocell BSs and FUEs
to the thl FUE, respectively. ,
DL
l nH , 8
,k nH , 9
,k nH and 10
,k nH
the channel from the thl femtocell BS to the thl FUE, the
thk femtocell BS to the thl FUE, the thk FUE to the thl
FUE and the macrocell BS to the thl FUE, respectively.
FU
SIC is SI cancelation value at the FUE.
B. Data Rates in ABS Subframes
In ABS subframe, the femtocells are muted with data
transmission. As a consequence, there is no cross-tier
interference from the femtocell tier to the macrocell. The
data rates at the uplink and downlink channels by
scheduling the thi uplink MUE and thk downlink MUE
on the thn subcarrier are given as
, ,,DL
, 2
, , 0,
1
MB DL
k n k nM
ki n MU
i n ki n n
PD log
P
H
H (5)
, ,,UL
, 2
, 0,
1
MU UL
i n i nM
ik n MB MB
k n SI n
HPD log
P C
(6)
C. Joint Subcarrier and Power Allocation Problem
Formulation
Given the ABS ratio , the sum rate of the thi uplink
MUE and thk downlink MUE on the thn subcarrier is
given as
,UL ,DL ,UL ,DL
, , , , ,1 M M M M M
ik n ik n ki n ik n ki nC R R D D (7)
The sum rate of the heterogeneous network on the thn
subcarrier is given by
,DL ,UL ,DL ,UL
, , , , ,1
M M F F
ik n ki n ik n l n l n
l
C R R R R
,DL ,UL
, , M M
ki n ik nD D (8)
In this paper, our goal is to maximize the sum rate of
the heterogeneous network by optimizing scheduling and
power allocation scheme. We formulate the problem as
,max
UL DL
nik ik n
i kn
a CMB MU
A,P ,P
(9)
s.t.
C1: ,
1
DL
NMB
k n MB
n k
P P
C2: ,
1
N
MU
i n U
n
P P
C3: ,
,
1
,
UL
NM UL UL UL
ik n t
n k
C R i
C4: ,
,
1
,?
DL
NM DL DL DL
ik n t
n i
C R k
C5: 1
UL DL
nik
i k
a
C6: 0,1 , , nika n
, UL DLi k
C7: , 0, , MB UL
k nP k n
C8: , 0, , MU DL
i nP i n
where constraints C1 and C2 limit the transmit power of
each femtocell BS and UE to be below MBP and UP ,
respectively. C3 and C4 set the QoS requirement UL
tR
and DL
tR to ensure the minimum date rates of the uplink
and downlink MUEs, respectively. We assume the
minimum date rate constraints are feasible for every user.
C5 and C6 are imposed to guarantee that each subcarrier
can only allocated to at most one uplink and downlink
MUE pair. C7 and C8 represent the non-negative power
constraint of the transmit power on each subcarrier. The
variable matrices A , MBP and MU
P are obtained by
stacking all nika , ,
MB
k nP and ,
MU
i nP , respectively. Note that
we assume the femtocell BSs and FUEs employ EPA
among the subcarriers for simplicity (i.e., , FU Ul n
PP and
, FB FBl n
PP , l , where is the cardinality of the
sets and FBP is the maximum transmit power of FBSs).
Due to the exclusive nature of subcarrier assignment,
the optimization problem (9) is an integer optimization
problem. Although the problem can be optimally solved
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 552
via exhaustive search, the complexity involved increases
exponentially as the number of users and subcarriers
increase. Therefore, we relax the constraints C6 as
0,1nika . The relaxed problem is still not a convex
problem, because it is not jointly concave in their solution
space. However, any optimal solution still satisfies the
KKT conditions, so we can obtain a criterion to schedule
UE pairs based on the KKT condition. In particular, we
have the following proposition.
Proposition 1: To maximize the objective function (9),
the subcarrier n is allocated to the UE pair with thi uplink
and thk downlink users selected by
* *
,,
, arg max
UL DL
ik ni k
i k C (10)
Proof: The Lagrangian function of the optimization
problem (9) is defined by
MB MUA,P ,P ,μ,λ,β (11)
, 0 ,
1 1
UL DL DL
N NMB
nik ik n k n MB
n ni k k
a C P P
,
, ,
1 1
UL UL UL
N NMU M UL UL
i i n U i ik n t
n ni i k
P P C R
,
,
1
1
DL DL UL DL
NM DL DL
k ik n t n nik
n nk i i k
C R a
where μ , λ and β are the stacked Lagrange multipliers.
Deriving partial derivative of with respect to nika , we
get
,
0, 0
0, 0
nik
ik n n
niknik
if
aif
aC
a (12)
From (12), if subcarrier n is allocated to the UE pair
with thi and thk users (i.e. 0nika ), the UE pair has the
largest ,ik nC among other UE pairs, which implies (10).
IV. PROPOSED RESOURCE ALLOCATION ALGORITHM
In this section, we design a heuristic algorithm based
on Proposition 1. The proposed algorithm greedily
allocates the subcarriers to the UE pairs maximizing the
increase of the sum rate of the UE pair in step 1. The UE
pairs are selected from the uplink and downlink UEs pairs,
whose minimum data rates are not satisfied. In order to
eliminate the dependence on transmit power of the MUEs
and MBS, the uplink and downlink dates are calculated
by assuming the EPA of the transmit powers. In step 2,
the residual subcarrier is allocated to UE pairs, whose
minimum data rate requirement is not satisfied only in the
uplink or the downlink, to further increase the sum
throughput. Then given the subcarrier allocation, the
power of the MBS and MUEs are distributed using IWF
algorithm in step 3[25].
Let us denote UL
iS and DL
kS as the subcarrier assigned
to the thi uplink MUE and thk downlink MUE,
respectively. Let denote the unassigned subcarriers.
We use UL and DL to denote the uplink and downlink user whose minimum data rates are not satisfied, respectively. The data rates of the communication links
are obtained under EPA, i.e. , MB MB
k n
PP ,
, MU Ui n
PP ,
, FB FBl n
PP and
, FU Ul n
PP . Data rate of the
thi uplink
MUE and thk downlink MUE is given as
, ,UL ,UL
, ,1
ULi
M UL UL M M
i i ik n ik n
n S
R S R D (13)
, ,DL ,DL
, ,1
DLk
M DL DL M M
k k ki n ki n
n S
R S R D (14)
The sum data rate of the femtocell tier on the
subcarrier set UL DL
i kS S is given as
,DL ,UL
, ,1
UL DLi k
F UL DL F F
i k l n l n
ln S S
R S S R R (15)
The detailed algorithm is given in Table I. Note that
there are two IWF algorithms, namely IWF1 and IWF2,
in step 3, since the interference from femtocells is
different in normal subframes and ABS subframes. In
Table I, denotes the empty set and ,0x max x .
TABLE I: PROPOSED RESOURCE ALLOCATION ALGORITHM
Data: Channel gains: 1 2 3
, , , , , , , ,, , , , , , ,UL DL UL DLi n k n ki n l n l n l n l n l nH H H H H H HH .
4 5 6 7 8 9 10
, , , , , , , , , , , , ,l n k n k n k n k n k n k nH H H H H H H
Maximum power constraints: MBP , FBP and UP .
Initialization: UL
iS , DL
kS , ,
UL UL , DL DL , 0n
Step 1: Subcarrier allocation for UL
tR and DL
tR
While UL and DL and
1) 1 n n , UL UL
i iS S n , DL DL
k kS S n .
2) ik
, , M UL UL M DL DL F UL DLi i k k i kR S R S R S S
, , M UL UL M DL DL F UL DLi i k k i kR S R S R S S
3) * *
,
, arg max
UL DL
iki k
i k
4) * * * * * *1,?,? UL UL DL DL
ni k i i k ka S S n S S n
5) Update * *
,M UL UL
i iR S and * *
,M DL DL
k kR S
6) If * *
, M UL UL ULti i
R S R , */UL UL i end if
7) If * *
, M DL DL DLtk k
R S R , */DL DL k end if
8) / n
End loop
Step 2: Residual subcarrier allocation
Residual subcarrier allocation is almost the same as step 1. The only difference is that line 3 must be replaced by:
If UL and DL
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 553
9) * *
,
, arg max
UL DL
iki k
i k
Else if UL and DL
10) * *
,
, arg max
UL DL
iki k
i k
Else
11) * *
,
, arg max
UL DL
iki k
i k
End if
Step 3: Iterative water-filling algorithm
In normal subframes, perform IWF1 algorithm.
In ABS subframes, perform IWF2 algorithm.
IWF1: For iteration 1: max _iterationm
For UL DLi
If ULi
12) , 0,
,
,
MB MB
k n SI FB FU nMU
i n i ULi n
P C I IP
H,
s.t. ,
ULi
NMU
i n U
n S
P P
Else
13)
' '
, , 0,
,
, ,
MU
k n ik n FB FU nMB
i n i MB DLk n k n
P H I
H
IP
P,
s.t. ,
DLi
DLN
i MBMBi n
n S
S PP
N
End if End loop End loop
IWF2: For iteration 1: max _iterationm
For UL DLi
If ULi
14)
,
0,
,
,
MBk n
nMBMU SI
i n i ULi n
P
CP
H,
s.t. ,
ULi
NMU
i n U
n S
P P
Else
15) , , 0,
,
, ,
MU
k n ik n nMB
i n i MB DLk n k n
P HP
P H
s.t. ,
DLi
DLN
i MBMBi n
n S
S PP
N
End if End loop End loop
Result: subcarrier indicator A , power allocation MBP and MU
P .
V. SIMULATION RESULTS
In this section, we present a simulation analysis
comparing the throughput of the proposed joint FD
scheduling and power allocation algorithm (denoted as
FD-P) with that of a FD-RR system and a TDD HD
system in the same simulation setting. We evaluate the
throughput of TDD HD by firstly assuming the all
subcarriers are dedicated to uplink or downlink
communication respectively. Then the throughput of
TDD HD system is obtained by halving the sum of the
uplink and the downlink throughput. The throughput of
TDD HD system is equivalent to that of a TDD HD
system with symmetric uplink and downlink time slot
configuration. The SI cancellation values are the same for
the MBS, the FBSs and FUEs.
We adopt 3GPP LTE specifications for heterogeneous
network simulation [26]-[28]. The channel in femtocells
is assumed to experience the path loss model for indoor
hotzone. In particular, path losses for line-of-sight (LOS)
and non-line-of-sight (NLOS) are given with the
probability as
LOS
1, 0.018
0.018 / 0.027 ,0.018 0.037
0.5, 0.037
d
P exp d d
d
where d is in kilometers. The penetration loss between
MBS and FBSs and between FBSs is set to 10dB . All the
channels are subject to independent identical distributed
(i.i.d.) Rayleigh fading coefficients with unit mean.
Without loss of generality, we assume identical power
constraints for the MUEs and FUEs. All the results are
obtained over 500 drops. Detailed simulation parameters
are shown in the Table II.
TABLE II: S
Parameter Setting
MR 500m
FR 50m
4
,U D 10,10
System bandwidth 10MHz
Number of subcarriers 50
MBP 43dBm
FBP 20dBm
UP 23dBm
Thermal Noise Density -174dB/Hz
Path loss for Macrocell 128.1+37.6log10(d) Path loss for femtocell LOS: 89.5 +16.9log10(d)
NLOS: 147.4 + 43.4log10(d)
,UL DL
t tR R 1 Mbps, 2 Mbps
TABLE III: THE TOTAL FD GAIN OF THE HETEROGENEOUS NETWORK
SI cancellati
on value
110dB 100dB 90dB 80dB 70dB
FD-P 69% 66% 65% 62% 56%
FD-RR 58% 57% 55% 53% 45%
The total FD gain is defined by the average increase of
sum rate of the heterogeneous network over TDD HD in
all subframes. Table III depicts the total FD gain of the
heterogeneous networks at various SI cancellation values
when 0.1 . It shows that both the FD-P and FD-RR
scheme cannot double the sum rate and can achieve 69%
and 58% FD gain at SI cancellation value 110dB and
100dB, respectively. The FD-P scheme obtained about
11% more total FD gain over the FD-RR scheme.
However, since the interference conditions are different
between the macroll and the femtocells and between ABS
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 554
IMULATION ARAMETERSP
subframes and normal subframes, it’s beneficial to
inspect the FD gain of the macrocell and the femtocells,
respectively.
The sum rate of the macrocell of the FD-RR and FD-P
scheme in normal subframes are shown in Fig. 1, where
SI x dB means the SI cancellation value is x dB .
When FD-RR scheme is employed, there is no gain of the
sum rate of macrocell compared with the TDD HD
systems. This is due to there are strong Inter-User
Interference (IUI) since RR scheduler is used. What’s
more, the co-channel femtocells cause severe interference
to the downlink reception of the MUEs in their vicinity.
The FD-P scheme schedules UE pairs with small IUI and
can obtain FD gain. However, under the best practical SI
cancelation value of 110dB, FD-P achieves about 21%
gain. In order to further exploit FD gain, eICIC schemes
such as ABS should be employed to reduce the cross-tier
interference.
0 100 200 300 400 500 6000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Sum rate of the macrocll (Mbps)
CD
F
SI=110dB
SI=100dB
SI=70dB
TDD HD
FD-P
FD-RR
Fig. 1. The sum rate of the macrocell in FD-P and FD-RR in normal
subframes
100 200 300 400 500 600 7000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Sum rate of the Macrocell (Mbps)
CD
F
SI=110dB
SI=100dB
SI=70dB
TDD HD
FD-P
FD-RR
Fig. 2. The sum rate of the macrocell in FD-P and FD-RR in ABS
subframes
In ABS subframes, the sum rate of the macrocell is
shown in Fig. 2. The figure confirms that the FD-RR
cannot provide any FD gain due to the strong IUI. The
FD-RR only provides comparable performance with the
TDD HD system on 110dB SI cancelation value.
However, notable FD gain is achieved by FD-P scheme.
The FD gains are 46% and 32% at SI cancelation value of
110dB and 100dB, respectively. The FD gains of the FD-
P scheme are much higher than that of Fig. 1. This shows
that FD systems are more prone to cross-tier interference
and eICIC is required to mitigate severe cross-tier
interference in heterogeneous deployment.
The total FD gain of the macrocell in the FD-P scheme
is defined by the average increase of the sum rate of the
macrocell over TDD HD mode in all the subframes. The
total FD gain under various ABS ratios is indicated in Fig.
3. It is illustrated that the total FD gain of the FD-P
scheme increases with ABS ratio. However, the total FD
gain increases faster with the ABS ratio at higher SI
cancellation value. The reason is that the FD gain of the
FD-P scheme at high SI cancellation values(e.g. 110dB)
in ABS subframes is much bigger than that at low SI
cancellation value(e.g. 70dB) as depicted by Fig. 2. The
total FD gain can achieve about 45% at SI cancellation
value of 110dB.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.95
10
15
20
25
30
35
40
45
50
ABS Ratio
Tota
l F
D G
ain
(%
)
SI=110dB
SI=100dB
SI=70dB
Fig. 3. Total FD gain of the marocell of the FD-P scheme
0 100 200 300 400 500 600 700 800 900 10000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Sum rate of femtocells (Mbps)
CD
F
FD-P,SI=110dB
FD-P,SI=70dB
TDD HD
FD-RR,SI=110dB
FD-RR,SI=70dB
Fig. 4. The sum rate of the femtocells in normal subframes
Fig. 4 depicts the sum rate of the femtocells in normal
subframes when the macrocell employs the FD-P and
FD-RR scheme. It is seen that both the schemes achieve
almost doubled gain of sum rate of femtocells over the
TDD HD scheme. There are two reasons: First, the link
distances in the femtocells are short, so the signal is much
stronger than the interfrerence and SI. Secondly, since
both the femtocell BSs and FUEs operated in FD mode,
there are no IUI. This observation also shows that
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 555
femtocell is suitable for deployment of FD technology.
What’s more, the FD-P degrades a small amount of the
sum rate of the femtocells, since the FD-P scheme
generates more cross-tier interference to the femtocells
compared with the FD-RR scheme.
VI. CONCLUSION
In this paper, we have formulated a joint subcarrier and
power allocation problem in the heterogeneous network.
To solve the problem, we proposed a three-step algorithm,
which greedily allocates the subcarriers to satisfy the
minimum data rate requirement of the MUEs under EPA
and then perform IWF procedure to distributed power
under individual BSs’ and UEs’ power constraints. We
considered the eICIC by means of ABS to alleviate cross-
tier interference. Performance evaluation demonstrates
that the proposed scheme can exploit FD gain in the
heterogeneous network. What’s more, a high FD gain up
to 45% can be obtained on ABS subframes, which proves
that FD technology is prone to cross-tier interference.
ACKNOWLEDGMENT
This work was supported by National Science and
Technology Major Project under grant No.
2015ZX03001025-002 and Huawei Company.
REFERENCES
[1] Cisco White Paper, Cisco Visual Networking Index:
Global Mobile Data Traffic Forecast, 2009-2014 Reports
[2] ITU Report, ITU-R M.2243(IMT.UPDATE), WP5D, 2011
[3] A. Goldsmith, Wireless Communications, Cambridge, U.K.:
Cambridge University Press, 2005.
[4] M. Heino, D. Korpi, T. Huusari, et al., “Recent advances in
antenna design and interference cancellation algorithms for
in-band full duplex relays,” IEEE Communications
Magazine, vol. 53, no. 5, pp. 91-101, May 2015.
[5] L. Laughlin, M. Beach, K. Morris, et al., “Electrical
balance duplexing for small form factor realization of in-
band full duplex,” IEEE Communications Magazine, vol.
53, no. 5, pp. 102-110, May 2015.
[6] X. Zhang, W. Cheng, and H. Zhang, “Full-duplex
transmission in phy and mac layers for 5G mobile wireless
networks,” IEEE Wireless Communications, vol.22, no. 5,
pp.112-121, Oct. 2015.
[7] N. H. Mahmood, G. Berardinelli, F. M. L. Tavares, et al.,
“On the Potential of Full Duplex Communication in 5G
Small Cell Networks,” in Proc. IEEE 81st Vehicular
Technology Conference (VTC spring), Glasow, 2015, pp.
1-5.
[8] S. Han, I. Chih-Lin, Z. Xu, et al., “Full duplex: Coming
into reality in 2020?” in Proc. IEEE Global
Communications Conference (GLOBECOM), Austin, 2014,
pp. 4776-4781.
[9] L. Wang, F. Tian, T. Svensson, et al., “Exploiting full
duplex for device-to-device communications in
heterogeneous networks,” IEEE Communications
Magazine, vol. 53, no. 5, pp. 146-152, May 2015.
[10] B. Debaillie, B. van Liempd, B. Hershberg, et al., “In-
Band full-duplex transceiver technology for 5G mobile
networks,” in Proc. 41st European Solid-State Circuits
Conference (ESSCIRC), San Francisco, 2015, pp. 84-87.
[11] E. Everett, A. Sahai, and A. Sabharwal, “Passive self-
interference suppression for full-duplex infrastructure
nodes,” IEEE Transactions on Wireless Communications,
vol. 3, no. 2, pp. 680-694, Feb. 2015.
[12] J. I. Choi, M. Jain, K. Srinivasan, et al., “Achieving single
channel, full duplex wireless communication,” in Proc.
16th Annual International Conference on Mobile
Computing and Networking, Chicago, 2010, pp. 1-12.
[13] D. Bharadia, E. McMilin, and S. Katti, “Full duplex
radios,” ACM SIGCOMM Computer Communication
Review, vol. 43, no. 4, pp. 375-386, 2013.
[14] M. Duarte and A. Sabharwal, “Full-duplex wireless
communications using off-the-shelf radios: Feasibility and
first results,” in Proc. 44th Asilomar Conference onSignals,
Systems and Computers, Pacific Grove, 2010, pp. 1558-
1562.
[15] M. Duarte, C. Dick, and A. Sabharwal, “Experiment-
driven characterization of full-duplex wireless systems,”
IEEE Transactions on Wireless Communications, vol. 11,
no. 12, pp. 4296-4307, Dec. 2012.
[16] S. Goyal, P. Liu, S. Panwar, et al., “Improving small cell
capacity with common-carrier full duplex radios,” in Proc.
IEEE International Conference on Communications,
Sydney, 2014, pp. 4987-4993.
[17] A. C. Cirik, K. Rikkinen, Y. Rong, et al., “A subcarrier and
power allocation algorithm for OFDMA full-duplex
systems,” in Proc. European Conference on Networks and
Communications (EuCNC), Palais des Congrès, 2015, pp.
11-15.
[18] A. C. Cirik, K. Rikkinen, and M. Latva-aho, “Joint
subcarrier and power allocation for sum-rate maximization
in OFDMA full-duplex systems,” in Proc. IEEE 81st
Vehicular Technology Conference (VTC Spring), Glasow,
2015, pp. 11-15.
[19] C. Nam, C. Joo, and S. Bahk, “Joint subcarrier assignment
and power allocation in full-duplex OFDMA networks,” in
Proc. International Conference on Information and
Communication Technology Convergence, Busan, 2014, pp.
924-927.
[20] S. Goyal, P. Liu, et al. Full Duplex Operation for Small
Cells. [Online]. Available:
http://arxiv.org/pdf/1412.8708.pdf
[21] A. Cirik “Fairness considerations for full duplex multi-user
MIMO systems,” IEEE Wireless Communications Letters,
vol. 4, no. 4, pp. 2162-2337, Aug. 2015.
[22] J. Lee and T. Q. S. Quek, “Heterogeneous network
throughput with hybrid-duplex systems,” in Proc. IEEE
Global Communications Conference (GLOBECOM),
Austin, 2014, pp. 3635-3640.
[23] M. O. Al-Kadri, A. Aijaz, and A. Nallanathan, “Ergodic
capacity of interference coordinated hetnet with full-duplex
small cells,” in Proc. 21th European Wireless (EW)
Conference, Budapest, 2015, pp. 1-6.
[24] R1-104968, Summary of the Description of Candidate
eICIC Solutions, 3GPP Standard, 2010.
[25] W. Yu, G. Ginis, and J. M. Cioffi, “Distributed multiuser
power control for digital subscriber line,” IEEE Journal on
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 556
Selected Areas in Communications, vol. 20, no. 5, pp.
1105-1115, July 2002.
[26] TR 36.814, Evolved Universal Terrestrial Radio Access
(E-UTRA); Further Advancements for E-UTRA Physical
Layer Aspects, 3GPP Standard, 2010.
[27] H. Zhang, C. Jiang, N. C. Beaulieu, et al., “Resource
allocation in spectrum-sharing OFDMA femtocells with
heterogeneous services,” IEEE Transactions on
Communications, vol. 62, no. 7, pp. 2366-2377, July 2014..
[28] H. Zhang, C. Jiang, and J. Cheng, “Cooperative
interference mitigation and handover management for
heterogeneous cloud small cell networks,” IEEE Wireless
Communications, vol. 22, no. 3, pp. 92-99, June 2015.
Jun Zhang received the B.S. degree from
Hubei University of Arts and Science,
Xiangfan, China, in 2007 and the M.S.
degree from Shanghai Normal University,
Shanghai, China, in 2010. Currently, he is
a Ph.D. candidate in Beijing University of
Posts and Telecommunications (BUPT),
China. His research interests mainly
include heterogeneous networks,
interference cancelation, radio resource management and
stochastic geometry.
Hui Tian received the M.S. and Ph.D.
degrees both from Beijing University of
Posts and Telecommunications (BUPT)
of China in 1992 and 2003, respectively.
Now she is a professor in BUPT and the
director of the Media Access Lab. (MAT)
of Wireless Technology Innovation
Institute (WTI). Her research interests
mainly include LTE and 4G system design, MAC protocols,
resource scheduling, ad hoc and sensor networks, radio resource
management.
Journal of Communications Vol. 11, No. 6, June 2016
©2016 Journal of Communications 557