Академический Документы
Профессиональный Документы
Культура Документы
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/271445369
CITATION
READS
27
1 AUTHOR:
Mohamed Rihan Elmeligy
Egypt-Japan University of Science and Techn
15 PUBLICATIONS 24 CITATIONS
SEE PROFILE
AbstractInterference Alignment (IA) emerged on the communication scene as a solution to the interference problem in all
interference-limited networks, including heterogeneous cellular
systems. However, the performance of IA is greatly related to the
accuracy of the channel state information at transmitters (CSIT),
namely the number of feedback bits. Accordingly, in order to
improve the performance of IA, it would be useful to analyze
the number of feedback bits with respect to the sum rate loss.
Motivated by that, this paper studies a limited feedback-based IA
scheme suitable for two tier macrocell-femtocell heterogeneous
networks. First, an approximate analytical expression for the
upper bound on the total sum rate loss due to limited feedback
in the studied IA system, is derived. Then, a simulation based
evaluation of the sum-rate loss due to the implementation of
limited feedback IA in heterogeneous networks is obtained.
Simulation results confirmed the severe effect of quantization
of CSI on the interference alignment performance.
KeywordsInterference Alignment, Femtocell (IA), Macrocell,
Limited Feedback, Loss rate, Heterogeneous Networks (HetNets).
I.
I NTRODUCTION
Rtot
4
dk
=
log2 1 +
k=1 i=1
dk
(i)
pjk k,f (k) wk
j=1
j=i
2
(i)
(i)
2
2
4 dm
(j)
(i)
(l)
pjm k,f (m) wk Hk,f (m) vm + 1
Hk,f (k) vk +
m=1 l=1
(3)
m=k
4
(2)
m=1
A. System Model
We build on the work presented in [6], which introduced
a hierarchical IA (HIA) suitable for macrocell-femtocell HetNets. We consider a macrocell-femtocell HetNet with one
macrocell served by a basestation (BS) BS2 and two shared
spectrum femtocells each served with one femtocell access
point/BS, namely BS1 and BS3 . The femtocell BS (FBS)
serves one user per cell and the macrocell BS (MBS) serves
two users simultaneously, as shown in Fig.1. Each served user
receives d independent data streams. As a typical antenna configuration for macrocell-femtocell scenarios [6], it is assumed
that each FBS and MBS are equipped with N and 2N transmit
antennas, respectively, and all users have N receive antennas.
The received signal at the k th receiver is expressed as:
yk
4
=
pm k,f (m) Hk,f (m) Vm sm + nk
N/2
(5)
i
H1
3,1 H3,3 v4
=
1
H3,1 H3,3 vi4
(6)
N/2
(1)
m=1
W1 = N ((H1,3 V4 ) ),
(7)
W4 = N ((H4,1 V1 ) ),
(8)
W2 = N ((H2,1 V1 ) ) = N ((H2,3 V4 ) ),
(9)
W3 = N ((H3,1 V1 ) ) = N ((H3,3 V4 ) )
(10)
N
2
(11)
(12)
III.
IV.
(2) Each user k will send the vector indices of all its cross
channels obtained from (1), to its corresponding BS f (m).
(3) Using the same codebooks Ck,f (m) , Each BS construct
k,f (m) .
the quantized version of the channels H
(4) Each of the FBSs forwards the quantized channels of the
FUs users to the MBS through the backhual links. The MBS
will add the quantized cross channels CSIs of its MUs and
forward all the quantized channels to the IA design unit.
k, V
m) ,
(5) Calculate the HIA algorithm transceivers, (W
according to Eqs. (5-12) by using the quantized channels.
hkj +
1 ekj
ekj skj
(13)
kj ,
where skj is isotropically distributed in the nullspace of h
kj is the quantized unit vector
and is independent of ekj . h
codeword selected from the codebook, selected to represent
kj . Also skj is the unit vector representing the direction of the
h
quantization error, and ekj is the corresponding quantization
error magnitude calculated as [14]:
ekj = min
ci C
2
1 H
kj ci
kj + ekj Skj
1 ekj H
(i)
R
k
(16)
2
(i)
i (i)
pk wk
Hk,f (k) vk
2
(i)
i (i)
k
vk
pk w
Hk,f (k)
E log2
dk
2
(j)
j (i)
+ E log2
w
v
1
+
p
H
k,f
(k)
k
k
k
j=1
j=i
2
dm
4
(i)
(l)
k
vm
+
pjm w
Hk,f (m)
m=1 l=1
RLtot = E
log2
(17)
m=k
= T1 + T2
(18)
T1 = E
log2
(19)
(14)
2
(i)
i
k,f (k) v(i)
H
(pk Mk,f (m) ) wk
k
2
(i)
(i)
i
k
(pk Mk,f (m) ) 1 ek,f (k) w
vk
Hk,f (k)
E log2
2
(i)
w
k,f (k) v(i)
H
= E log2
k
k
2
(i)
(i)
1 ek,f (k) w
k
vk
(20)
Hk,f (k)
E log2
(i)
Rk
(i)
(i) are the perfect CSIs and limited feedback
where Rk and R
k
based CSIs sum rate for the ith stream of the k th user
respectively. By substituting in Eq.(16) using expression in
Eq.(3) and taking into account that we are trying to get an
upper bound on the loss, we can loosly consider that in the
case of perfect CSI, the interference terms in the denominator
of Eq.(3) are negligible. The attenuation factor related to the
pathloss and shadowing effects can be included in the channel
matrices, so the term k,f (k) Hk,f (k) will be replaced by only
Hk,f (k) . Therefore,
k=1 i=1
(1) Each user k, uses its codebooks Ck,f (m) with Bk,f (m)
k,f (m) .
bits, to quantize each Hk,f (m) , to H
hkj =
dk
4
(15)
T1 = E
log2
(i)
w
k
(i)
w
E log2
k
T1 = E
log2
2
k,f (k) v(i)
(21)
H
k
2
(i)
k,f (k)
vk
E log2 (1 ek,f (k) ) ,
H
(i)
(i) 2
In Eq.(21), the two terms |wk H
and
k,f (k) vk |
(i)
2
Hk,f (k)
vk | , in which the channels Hk,f (k) and Hk,f (k)
are drawn randomly from a complex unit hypersphere and the
(i) (i)
(i)
(i)
corresponding transceiver vectors (wk , vk ) and (w
vk )
k ,
are formed using the channel information of the entire network,
so they should follow similar distributions with the same
parameters [14]. Therefore T1 term can be expressed as:
T1 = E log2 (1 ek,f (k) )
(22)
= log2 (e) E n(1 ek,f (k) )
2
2
(i)
(i)
(j)
(l)
k,f (m)
k
k
H
vm + ek,f (m) w
vk
Skj
1 ek,f (m) w
!
0
(i)
(i)
(j)
(l)
k,f (m)
k
vm W
vk ,
H
+ 2 1 ek,f (m) ek,f (m) w
Skj
k
!
A2 = E
(i)
|
wk
E
(28)
2
(i)
(j)
k
ek,f (m) w
vk
Skj
(29)
1
n
B
1+ 2
, 2Bk,f (k) .
= log2 (e) 2 k,f (k)
n
M
1
n=1
dk
2
(i)
(j)
k
1 +
vk
pjk w
T2 = E log2
Hk,f (k)
j=1
j=i
2
dm
4
(i)
(l)
k
vm
+
Hk,f (m)
pjm w
m=1 l=1
A1
E ek,f (k)
(32)
N2
j=1
j=i
(24)
dk
4
UB
=
E RL
(33)
k=1 i=1
Bk,f (k)
(log2 (e)) 2
+ log2 1 +
dk
1
n
B
1+ 2
, 2 k,f (k)
n
N 1
n=1
pk
Bk,f (k)
j=1
j=i
2
dk
(i)
(j)
k,f (k)
k
vk
H
T2 log2 1 +
pjk E(Mk,f (k) ) E w
j=1
!
j=i
A1
2
dm
4
(i)
(l)
k
(25)
vm
pjm E(Mk,f (m) ) E w
+
Hk,f (m)
m=1 l=1
!
A2
A2 = E
(i)
(l)
1 ek,f (m) w
k
vm
Hk,f (m)
E
2
(i)
(j)
ekj w
k
vk
+
Skj
(31)
m=k
m=k
(30)
(26)
(27)
dm
4
m=1 l=1
m=k
j
pm
Bk,f (m)
N2
B
, 2 k,f (k)
N2 1
(34)
N2
Bk,f (m)
,
2
N2 1
(35)
(36)
V. S IMULATION R ESULTS
In this section, the performance of the limited feedback
HIA scheme to the heterogeneous network, is studied under
limited feedback design. The simulation parameters used for
such scenario are set as { Total number of users (K) = 4,
number of antennas per user (N ) = 2, each FBS serves one
user and have 2 antennas, the MBS serves two users and have
4 anetnnas }. The pathloss and shadowing parameters for the
proposed scenario are the same as listed in [18][17].
In Fig. 2, we compare the total network sum rate of the
HIA algorithm with perfect CSI case and with RVQ based
limited feedback case with different bit resolution ( B = 3,
7, 15 bits), where each CSI is quantized with 2B bits. It is
apparent that the HIA algorithm is very sensitive to the CSI
accuracy, where at (SNR = 30 dB), for example, the sum rate
decreases to 25% of its value with using high resolution value
of (B = 15bits), and an even larger decrease with lower bit
resolution values. Employing high bit resolution values (B =
20
15
10
0
0
VII.
ACKNOWLEDGEMENT
R EFERENCES
[1] V. R. Cadambe, and S. A. Jafar, Interference Alignment and Degrees of
Freedom of the K-user interference channel, IEEE Trans. Inf. Theory,
vol. 54, no. 8, pp. 3425-3441, Aug. 2008.
[2] C. Suh and D. Tse, Interference Alignment for Cellular Networks, in
proc. Allerton Conf. on Comm., Cont., and Comp., Berkeley, CA, Sept.
2008, pp. 1037-1044.
[3] C. Suh, M. Ho, and D. Tse, Downlink Interference Alignment, IEEE
Trans. Commun., vol. 59, pp. 2616-2626, Sept. 2011.
[4] N. Saquib, E. Hossain, Bao Le Long, In Kim Dong, Interference
management in OFDMA femtocell networks: issues and approaches,
IEEE Trans. on Wireless Comm., vol.19, no.3, pp.86-95, June 2012.
[5] Mohamed Rihan, Maha Elsabrouty, Osamu Muta, and Hiroshi Furukawa,
Iterative Interference Alignment in Macrocell-Femtocell Networks: A
Cognitive Radio Approach, accepted at IEEE inter. Symposium on
Wireless Comm. Systems (ISWCS), Barcelona-Spain, August 2014.
[6] Wonjae Shin, Wonjong Noh, Kyunghun Jang, Hyun-Ho Choi, Hierarchical Interference Alignment for Downlink Heterogeneous Networks,IEEE
Trans. on Wireless Comm., vol. 11 no. 12 pp. 4549 - 4559, Oct. 2012 .
[7] S. Cho, K. Huang, D. Kim, H. Seo, Interference Alignment for Uplink
Cellular Systems with Limited Feedback,IEEE Commun. Lett., vol. 16,
pp. 960-963, July 2012.
[8] N. Lee, W. Shin, R. W. Heath Jr., and B. Clerckx, Interference Alignment with Limited Feedback for Two-Cell Interfering MIMO-MAC, in
Proc. IEEE ISWCS, Aug. 2012, pp. 566-570.
10
15
20
25
SNR (SignaltoNoise Ratio in dB)
30
35
40
Fig. 2.
Sum rate performance with different bit resolution levels in
comparison with perfect CSI case.
40
35
30
25
20
15
10
5
0
0
10
15
20
25
SNR (SignaltoNoise Ratio in dB)
30
35
40
Fig. 3. Behavior of both the simulation based sum rate loss and the sum rate
loss upper bound obtained from the closed form expression.
25
40
35
Sum rate Loss (Bits/Sec/Hz)
VI.
30
15bits) can lead to double the sum rate value achieved with
low resolution counterparts (B = 3,7 bits), but we are still far
apart from the perfect CSI case due to both the quantization
error resulted from the limited feedback design, and the nature
of the proposed system model where the network suffer from
different types of interference including co-tier and cross-tier
interference.
In Fig. 3, it is apparent that the average sum rate obtained
from the simulations coincides with that obtained from the
closed form expression as the number of feedback bits (B)
becomes larger, where we assume in Eqs. (19) and (21) that
high resolution quantization of the CSI values (large number
of feedback bits for each of the cross channels) is assumed.
Similarly, Fig.4 shows the effect of increasing the number of
feedback bits (B) on the system sum rate loss with different
SNR values. At all SNR values, as the number of feedback
bits increases, the sum rate loss decreases accordingly. Additionally, the simulation results (average summ rate loss) obeys
the results calculated using the derived approximate rate loss
upper bound.
30
25
20
15
10
5
0
4
8
10
12
B (Number of feedback bits)
14
Fig. 4. Effect of bit resolution value (B bits) on the total sum rate loss with
different SNR values (SNR = 35, 25, 15 dB.)
[9] X. Chen and C. Yuen, Performance Analysis and Optimization for
Interference Alignment over MIMO Interference Channels with Limited
Feedback, IEEE Trans. Signal Process., vol. 99, pp. 1-10, Feb 2014.
[10] H. Gao, J. Leithon, C. Yuen, and H. A. Suraweera, New Uplink
Opportunistic Interference Alignment: An Active Alignment Approach,
in Proc. IEEE WCNC, Shanghai, China, Apr. 2013, pp. 1123-1127.
[11] D. Love, R. Heath Jr., V. Lau, D. Gesbert, B. Rao, and M. Andrews,
An overview of limited feedback in wireless communication systems,
IEEE J. Sel. Areas Commun., vol. 26, no. 8, pp. 1341-1365, Oct. 2008.
[12] R. T. Krishnamachari, and M. K. Varanasi, Interference Alignment
under Limited Feedback for MIMO Interference Channels, IEEE Trans.
Signal Process. vol. 61, no. 15, pp. 3908-3917, Aug. 2013.
[13] N. Jindal, MIMO Broadcast Channels With Finite-Rate Feedback,
IEEE Trans. on Inf. Theory, vol. 52, no. 11, pp. 5045-5060, Nov. 2006.
[14] R. Bhagavatula and R. W. Heath, Adaptive Bit Partitioning for Multicell Interference Nulling with Delayed Limited Feedback, IEEE Trans.
Signal Proc., vol. 59, no. 8, pp. 3824-3836.
[15] C. K. A. Yeung and D. J. Love, On The Performance of Random Vector
Quantization Limited Feedback Beamforming in a MISO System, IEEE
Trans. Wireless Commu., vol. 6, no. 2, pp. 458-462, Feb. 2007.
[16] Press WH, Teukolsky SA, Vetterling WT, Flannery BP, Numerical
Recipes 3rd Edition: The Art of Scientific Computing, Cambridge University Press, 2007.
[17] A. Abdelnasser, E. Hossain, Dong In Kim , Clustering and Resource
Allocation for Dense Femtocells in a Two-Tier Cellular OFDMA Network, IEEE Trans. on Wireless Comm., vol. 13, no. 3, pp.1628-1641,
March 2014.
[18] Mohamed Rihan, Maha Elsabrouty, Osamu Muta, and Hiroshi Furukawa, Interference Mitigation Framework Based on Interference
Alignment for Femtocell-Macrocell Two Tier Cellular Systems, IEICE
Trans. on Communications, vol. E98-B, No.03, pp.-, March 2015.