Link Performance Improvement Using Reconfigurable Multiantenna Systems

Abstract

In this work, we consider the link performance improvement achievable due to the use of multifunctional reconfigurable antennas (MRAs). Each MRA element used in this work has a reconfigurable radiation beam that can be dynamically steered in one of N_beams possible directions. Depending on the channel conditions, the direction that maximizes capacity or minimizes symbol error rate (SER) is chosen for each MRA. Simulation results are presented for four different transmission schemes - beamforming (BF), maximum ratio combining (MRC), spatial multiplexing (SM), and space-time block coding (STBC). The results show improvements in both capacity and SER are achievable for typical indoor channels (IEEE-802.11n model F and B). A key result is that this reconfigurable antenna system requires fewer number of antennas (therefore, fewer number of RF chains) to obtain performance similar to those of legacy multiantenna systems with larger number of RF chains.

Full text

Page 1

IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 8, 2009873
Link Performance Improvement Using
Reconfigurable Multiantenna Systems
Chitaranjan P. Sukumar, Student Member, IEEE, Hamid Eslami, Student Member, IEEE,
Ahmed M. Eltawil, Member, IEEE, and Bedri A. Cetiner, Member, IEEE
Abstract—In this work, we consider the link performance
improvement achievable due to the use of multifunctional recon-
figurable antennas (MRAs). Each MRA element used in this work
has a reconfigurable radiation beam that can be dynamically
steered in one of ?????? possible directions. Depending on the
channel conditions, the direction that maximizes capacity or
minimizes symbol error rate (SER) is chosen for each MRA.
Simulation results are presented for four different transmission
schemes—beamforming (BF), maximum ratio combining (MRC),
spatial multiplexing (SM), and space-time block coding (STBC).
The results show improvements in both capacity and SER are
achievable for typical indoor channels (IEEE-802.11n model F
and B). A key result is that this reconfigurable antenna system
requires fewer number of antennas (therefore, fewer number
of RF chains) to obtain performance similar to those of legacy
multiantenna systems with larger number of RF chains.
Index Terms—Multifunctional reconfigurable antenna (MRA),
multiple-input–multiple-output (MIMO), spatial multiplexing.
I. INTRODUCTION
T
the reconfigurable properties of an MRA that can be used as
additional degrees of freedom in adaptive system parameters
as first proposed in [1] and [2], which was further studied in
[3]–[5].Inparticular,itwasshownthatanMRA-equippedadap-
tive multiple-input–multiple-output (MIMO) wireless commu-
nication system can provide gains up to 30 dB as compared to
conventional fixed antenna MIMO systems [2].
A significant amount of work has been done on character-
izing performance gains of MRAs with various reconfiguration
capabilities. However, most of the published work has not been
in the context of statistical channel models that are typically
used in evaluating wireless algorithms. With this motivation,
in this work, we study capacity and symbol error rate (SER)
of beam-tilting MRAs utilizing statistical channel models. This
letter presents a framework linking a generic channel model, re-
configurable antennas, and the performance gains realized.
Capacity and SER improvements using MRAs are presented
for the following multiple-antenna systems—maximum ratio
HE multifunctional reconfigurable antenna (MRA) con-
cept has gained significant interest. This is mainly due to
Manuscript received May 27, 2009; revised July 03, 2009. First published
July 24, 2009; current version published August 11, 2009.
C. P. Sukumar, H. Eslami, and A. M. Eltawil are with the Department of
Electrical Engineering and Computer Science, University of California, Irvine,
Irvine, CA 92707 USA.
B.A.CetineriswiththeDepartmentofElectricalandComputerEngineering,
Utah State University, Logan, UT 84322 USA (e-mail: bedri@engineering.usu.
edu).
Color versions of one or more of the figures in this letter are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LAWP.2009.2028300
combining (MRC), beamforming (BF), spatial multiplexing
(SM), and space-time block coding (STBC). The motivation
behind choosing these four schemes is to quantify the perfor-
mance gains spanning the diversity-multiplexing space. For
high data rate systems, the best choice is SM, which maximizes
capacity as shown in the letter, albeit at a price of reduced di-
versity. In the other extreme is STBC, where the diversity order
of the system is increased, resulting in more reliable transmis-
sion. Finally, the BF and MRC schemes are discussed for the
case where there is a limitation on the number of antennas at
either the transmit or the receive site. Optimal mode selection
algorithms based on channel conditions have been studied in
literature, for example in [6], and the references therein.
The results presented in this letter confirm the results gener-
ated in [3], which highlight the benefits of incorporating pattern
diversity at the antenna level. Furthermore, our results are in
agreement with those presented in [5], which used ray-tracing
simulations and measurements to indicate clearly that substan-
tial benefits are realized by coupling both diversity and direc-
tivity of the antennas. While both [3] and [5] focus on capacity,
[4] calculates symbol error rates for a generic channel in the
context of a MIMO system that chooses the optimal mode from
a large set of generic modes. However, it does not emphasize
the effect of channel parameters such as angle of arrival (AoA),
number of clusters, power angular spread (PAS), etc. This letter
extends prior work described in [3]–[5] by utilizing statistical
channel models that incorporate effects such as AoA, PAS, etc.,
to quantify the impact on channel capacity and SER, which are
highly dependent on the underlying channel and the reconfig-
urable antenna properties.
Our results show that by using a reconfigurable multiple-an-
tennasystem,itispossibletomaintainsystemperformancewith
a reduced number of RF chains. In particular, we observed that
forthelowSNRregion,theperformanceofsystemswithMRAs
with fewer number of RF chains is superior to that of legacy
multiple antenna system with more RF chains. At high SNR,
however, this advantage is overshadowed due to gains obtained
from higher diversity order. Nonetheless, for the same number
of antennas, systems equipped with MRAs outperform systems
with conventional antennas across all SNRs.
II. CHANNEL MODEL
In cluster-based models, scatterers are grouped into clusters
spread around both transmit and receive antennas [7]. As a re-
sult, rays arriving at the receive antenna are also clustered with
a double-decaying exponential power delay profile (PDP). Typ-
ically, a ray from the th cluster and the th multipath has a PAS
in space that has a Laplacian distribution [8] and is represented
as follows:
1536-1225/$26.00 © 2009 IEEE
Authorized licensed use limited to: Univ of Calif Irvine. Downloaded on October 1, 2009 at 18:30 from IEEE Xplore. Restrictions apply.

Page 2

874IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 8, 2009
(1)
where
th cluster,
tion factor that satisfies the integral equation
is the AoA in the azimuth plane corresponding to the
is the angular spread, and is the normaliza-
(2)
The power received in the th direction of the MRA is therefore
given by
(3)
where
rection,
sibledirectionstowhichthereconfigurablebeamcanbesteered,
and
is the power of the th multipath from the th cluster.
Channel coefficients are generated using the following:
is the MRA radiation intensity in the th di-
being the number of pos-
(4)
where
unchanging line-of-sight (LOS) coefficient,
gainoftheantennaintheAoAoftheLOScomponent,
number of clusters, and
with zero mean and unit variance. As a case study, we choose
theIEEE-802.11nchannelmodels[9]thataredefinedforindoor
environments. We use this model to accurately characterize and
quantify the performance of MRAs. It should be noted that this
choice for channel model does not constitute a special case and
is chosen only for simulation purposes.
is the Ricean factor for the th multipath, is the
is the
isthe
is a Gaussian random variable
III. SYSTEM MODEL
A. MRA System
We consider a multiple antenna system with
tional transmit antennas and
MRA is assumed to steer its single beam into one of the pos-
sible beam directions from which it can receive energy
from the clusters. The radiation intensities of MRAs are chosen
as follows:
omnidirec-
MRA receive antennas. Each
for(5)
The 3-dB beamwidth
in azimuth plane is given as,
. We consider 12 different radiation scenarios cor-
responding to different values of
teger values from 1 to 12. For each scenario, at first, the pa-
rameter
is found based on 3-dB beamwidth. Parameter
then determined by the assumptions that the radiation intensity
of the isotropic antenna is unity and the radiated power is the
same for each scenario. Note that
to the isotropic antenna case, i.e.,
As an example, for the case of
the radiation pattern will be
figurable radiation beam can be steered into eight different di-
, which can take in-
is
corresponds
with
and
. This recon-
.
,
rections over the azimuth angles corresponding to
. Recent advancements in MRA
design and micro/nanofabrication technologies [10] enable a
single antenna element to tilt its beam into different directions.
In particular, an active antenna element combined with recon-
figurable passive parasitic elements can be used to accomplish
thisfunctionality[11].Onemustnote,however,thattheanalysis
performed in this work presents the best-case scenario for an-
tenna reconfigurability performance since the beam directions
and shapes are considered to be ideally perfect, i.e., no side-
lobes with exact beam directions. On the other hand, the ca-
pacity and SER analysis results obtained in this work provide
significant insight and guidance into optimum MRA design, its
microfabrication, and adaptation algorithm used to perform op-
timum reconfiguration in response to the time variations in the
communication system.
B. Capacity
We assume a frequency-selective channel with impulse re-
sponse of length
. The channel consists of
with different AoAs. If thefirst MRA is pointing in direction
the second is pointing in direction
define the set
It is understood that the th antenna can point in one of
directions. The total number of modes is given by
Thus, in the th mode, the th channel tap is givenby an
matrix . The frequency response of the channel for a
system in this mode is given by
clusters
,
, and so on, then one can
to represent the th mode.
.
(6)
and the capacity is given by
(7)
where
is the rank, i.e., number of eigenmodes, of
. For the th eigenmode of theis
the waterfilling power [12] and
The noise variance is
the frequency spectrum into a large number
frequency bins (similar to OFDM). If we assume that the
frequency bin corresponds to frequency
approximation
optimum set of directions for the MRAs that maximizes the
capacity is the one that maximizes the sum of all capacities of
all subcarriers. This can be represented as
is its eigenvalue.
. For ease of computation, we divide
of small
th
, we can make the
and . The
(8)
Thus,
lated using (8)].
represents the mode that maximizes capacity [calcu-
C. Symbol Error Rate (SER)
To quantify the reliability of a wireless link with MRA, we
measure the SER. We assume that the modulation scheme used
Authorized licensed use limited to: Univ of Calif Irvine. Downloaded on October 1, 2009 at 18:30 from IEEE Xplore. Restrictions apply.

Page 3

SUKUMAR et al.: LINK PERFORMANCE IMPROVEMENT USING RECONFIGURABLE MULTIANTENNA SYSTEMS875
Fig. 1. Capacity versus transmit SNR. (a) Model F. (b) Model B.
is OFDM with
cyclic prefix is
number of subcarriers and the length of the
. The received signal is given by
(9)
where
sponse of the channel, and the AWGN noise at the th receive
antenna from the th transmit antenna in the th direction. In
the frequency domain, this can be represented as
, andare the received signal, time re-
(10)
with the signals in the capital letters representing the frequency
domain transformation of the corresponding time-domain sig-
nals and
represents the subcarrier number under considera-
tion.
To choose the optimum direction, we perform the following
operation:
(11)
Thus,
of the Frobenius norms of the channel over all subcarriers. This
minimizes the SER [4].
represents the set of directions maximizing the sum
IV. RESULTS AND DISCUSSIONS
Inthissection,wequantifytheperformancegainsofMRAsin
termsofcapacityandSER.Thechannelmodelsusedthroughout
this section are the IEEE 802.11n models F (highly scattered)
and B (low scattering with LOS) [9]. In model F, the number
of clusters is six and the maximum delay spread is 1.05
s. In
Fig. 2. SER versus transmit SNR. (a) Model F. (b) Model F.
model B, the number of clusters is two and the maximum delay
spread is 80 ns. The antenna spacing is assumed to be half a
wavelength in the simulations. All simulations were performed
using the Monte-Carlo technique.
A. Capacity
The capacity calculation is performed as given by (8). We
compare the following systems: 2
MRC. For each case, the capacity of the system with MRAs
is compared to that of a system with omnidirectional antennas.
Fig. 1(a) and (b) depict the capacity of the systems described
above as a function of transmit SNR (the ratio of the transmit
power to the average noise power) for
models F and B, respectively. Transmit SNR was chosen over
receiveSNRbecause: 1)number of receiveantennasfor various
systems are different; and 2) systems with directional antennas
have channel coefficients of greater magnitude than that of sys-
tems with omnidirectional antennas. However, the channel en-
vironment is constant for all systems under consideration. As
previously discussed in [2]–[5], the use of MRA provides sig-
nificant capacity improvements. In particular, for model F, SM
with directional antennas has 2.5 b/s/Hz more capacity than that
of SM with omnidirectional antennas as shown in Fig. 1(a). In
the case of model B [shown in Fig. 1(b)], the absolute value for
capacity for the omnidirectional antennas is lower since there
are fewer scatterers in the channel, while that changes slightly
for MRA-based system. This can be explained as follows. In
both model F and B, the statistical mean of the eigenvalues av-
eraged across all subcarriers are similar leading to similar ca-
pacity values. Finally, it should be noted that BF and MRC have
2 SM, 2 1 BF, and 12
for channel
Authorized licensed use limited to: Univ of Calif Irvine. Downloaded on October 1, 2009 at 18:30 from IEEE Xplore. Restrictions apply.

Page 4

876 IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 8, 2009
Fig. 3. SER versus transmit SNR. (a) Model B. (b) Model B.
the same capacity characteristics, and their capacities therefore
coincide for all transmit SNR.
B. SER
OFDM system is used for SER analysis with the following
parameters. The number of subcarriers is 1024, the length of
cyclicprefixistakentobe128,thesystembandwidthis40MHz,
and transmit signal power is taken to be unity. For each cluster,
the AoA is uniformly distributed with its mean defined in the
standard [9].
In Fig. 2(a) and (b), SER versus transmit SNR for the fol-
lowing transmission schemes for systems in a model F environ-
ment are shown:
1) STBC using a) 2
1 (Alamouti code [13]); b) 2
(stacked Alamouti); c) 4
2) 4
1 BF;
3) 1 4 MRC.
To have a fair SER comparison, each system must transmit
the same number of bits per subcarrier. To this end, all schemes
use 8-PSK modulation except for STBC 4
QAM because the space-time code used is of rate 3/4. This en-
sures that each subcarrier carries 3 bits. All simulations are car-
ried out for
. Clearly, the performance of MRA
systems is better than that of their counterparts with omnidirec-
tional antennas. It can also be seen in Fig. 2(a) that STBC 2
with directional antennas outperforms STBC 2
directional antennas at SNRs less than 4 dB. At higher SNR, the
MRA advantage is lost due to higher diversityorder of the 2
system.
It is preferable to have fewer number of antennas at the mo-
bile station due to space and power constraints. Toward that
end, we compare different schemes with the goal of achieving
2
1 ([14]);
1, which uses 16
1
2 with omni-
2
better or similar performance with fewer number of antennas
at the mobile. In Fig. 2(b), we demonstrate that better perfor-
manceisachievableusingSTBC2
thanMRC1
4withomnidirectionalantennasacrossallSNRs.
In other words, using fewer number of MRA, it is possible to
achieve similar or better performance as compared to systems
with more omnidirectional antennas. Note also that the perfor-
mance of BF and MRC are the same as expected. Results for
channel model B are depicted in Fig. 3(a) and (b). It can be
seen that the MRAs perform better than the omnidirectional an-
tennas across all SNR and all transmission schemes. The ab-
solute values of the SERs are higher due to presence of fewer
number of scatterers for the omnidirectional antennas. The per-
formance of the MRAs in model B is better than that of the
MRAs in model F due to the existence of the LOS component
that guarantees a higher receive SNR.
2withdirectionalantennas
V. CONCLUSION
The results presented in this letter indicate that by choosing
the right beam direction of an MRA, significant capacity in-
creases and SER reductions can be obtained. In particular, al-
most 25% increase in capacity was reported for 2
tems at dB for IEEE 802.11n channel model F.
This gain increases to 66% for IEEE 802.11n channel model B.
Furthermore, it was shown that better SER performances are
achievable using fewer number of antennas (RF chains) at the
receiverendfora2
2STBCwithMRAsthan1
legacyomnidirectionalantennasinamodelF-typeenvironment.
2 SM sys-
4MRCwith
REFERENCES
[1] B. A. Cetiner et al., “Multifunctional reconfigurable MEMS integrated
antennas for adaptive MIMO systems,” IEEE Commun. Mag., vol. 42,
no. 12, pp. 62–70, Dec. 2004.
[2] B. A. Cetiner, E. Sengul, E. Akay, and E. Ayanoglu, “A MIMO system
with multifunctional reconfigurable antennas,” IEEE Antennas Wire-
less Propag. Lett., vol. 5, pp. 463–466, 2006.
[3] D. Piazza et al., “Design and evaluation of a reconfigurable antenna
array for MIMO systems,” IEEE Trans. Antennas Propag., vol. 56, no.
3, pp. 869–881, Mar. 2008.
[4] A. Grau, H. Jafarkhani, and F. De Flaviis, “Reconfigurable MIMO
communication system,” IEEE Trans. Wireless Commun., vol. 7, no.
5, pp. 1719–1733, May 2008.
[5] J.BoermanandJ.T.Bernhard,“Performancestudyofpatternreconfig-
urable antennas in MIMO communication systems,” IEEE Trans. An-
tennas Propag., vol. 56, no. 1, pp. 231–236, Jan. 2008.
[6] R. W. Heath and A. J. Paulraj, “Switching between diversity and mul-
tiplexing in MIMO systems,”IEEE Trans. Commun., vol. 53, no. 6, pp.
962–968, Jun. 2005.
[7] A. A. Saleh and R. A. Valenzuel, “A statistical model for indoor mul-
tipath propagation,” IEEE J. Sel. Areas Commun., vol. SAC-5, no. 2,
pp. 128–137, Feb. 1987.
[8] K. Pedersen et al., “A stochastic model of the temporal and azimuthal
dispersion seen at the base station in outdoor propagation environ-
ments,” IEEE Trans. Veh. Technol., vol. 49, no. 2, pp. 437–447, Mar.
2000.
[9] V. Erceg et al., “TGn channel models,” IEEE 802.11-03/940r4, May
2004.
[10] N. Biyikli, Y. Damgaci, and B. A. Cetiner, “A low-voltage small-size
double-arm MEMS actuator,” Electron. Lett., vol. 45, no. 7, pp.
354–356, Mar. 2009.
[11] S. Zhang, G. H. Huff, J. Fend, and J. T. Bernhard, “A pattern reconfig-
urable microstrip parasitic array,” IEEE Trans. Antennas Propag., vol.
52, no. 5, pp. 1845–1848, May 2005.
[12] A. Goldsmith, “Wireless communications,” Cambridge, 2005.
[13] S. M. Alamouti, “A simple transmit diversity technique for wireless
communications,” IEEE J. Sel. Areas Commun., vol. 16, no. 8, pp.
1451–1458, Oct. 1998.
[14] V. Tarokh et al., “Space-time block codes from orthogonal designs,”
IEEE Trans. Inf. Theory, vol. 45, no. 5, pp. 1456–1467, Jul. 1999.
Authorized licensed use limited to: Univ of Calif Irvine. Downloaded on October 1, 2009 at 18:30 from IEEE Xplore. Restrictions apply.