SlideShare uma empresa Scribd logo
1 de 5
Baixar para ler offline
1
Euclidean Distance Antenna Selection for Spatial
Modulation with Multiple Active Transmit Antennas
Lloyd Blackbeard, Hongjun Xu and Fengfan Yang
Abstract—
I. INTRODUCTION
Spatial Modulation (SM)[1] could be argued to be
known fairly ubiquitously in communications academia.
Much work has been done in improving its throughput,
bit error rate (BER) performance and detection complexity
[2][3][4][5][6][7][8][9][10][11].
In one such improvement, called Euclidean Distance An-
tenna Selection (EDAS)[9], 2n
n ∈ N transmit antennas are
selected from a larger set in a fashion that maximises the Eu-
clidean distance between any two symbols. In Multiple Active
Spatial Modulation (MASM)[11], more than one antenna is
selected to be active in each symbol period, usually allowing
for more information to be mapped to the spatial domain
and allowing the simultaneous transmission of unique symbols
from each active antenna, thereby improving throughput. To
the best of the authors’ knowledge, no system has been
implemented to improve the performance of MASM in a
fashion similar to EDAS. Thus, in this paper, EDAS will be
implemented for SM systems with multiple active antennas.
The remainder of the paper follows: in section II, a model
for MASM is presented that will be used in subsequent
sections; in section III, we implement the idea of EDAS
for MASM and summarise antecedent work with which the
proposed method will be compared; in section IV, we write
the computational complexity for all schemes implemented in
the paper and produce the analytical results; in section V, we
show simulation results in terms of bit error rate (BER) and
in section VI, concluding remarks are given.
Unless otherwise specified the notation convention follows:
(·)−1
, (·)T
, (·)H
, (·)†
, E[·], | · | and | · |F denote the matrix
inverse, transpose, Hermitian, Moore-Penrose pseudoinverse,
expectation, Euclidean norm and Frobenius norm operators
respectively; regular, bold face lower case, bold face upper
case and capital script/cursive text refer to scalars, vectors,
matrices and sets respectively; subscripts (·)ij denote the
ith
row, jth
column entry in the corresponding matrix and
subscript (·)i denotes the ith
entry in the corresponding vector
or the ith
column in the corresponding matrix.
II. SYSTEM MODEL
The following serves to summarise convential MASM. We
consider a MIMO environment with Nt transmit antennas and
Nr receive antennas. We observe that a maximum of NΓ =
2
log2(Nt
Np
) antenna groups can be defined for Np ≤ Nt active
transmit antennas1
. For each active antenna, we map log2M
bits from the input bitstream to an MQAM symbol selected
from an MQAM constellation of size M. A further log2NΓ
bits select which of the Nt antennas will be active, or rather,
which antenna group will be used in transmission.
Once mapping is complete for the symbol period, we have
the transmit symbol column vector x with MQAM symbols
in positions corresponding to the active antennas and zeros
elsewhere. This vector is transmitted over a channel matrix H
with entries CN(0, 1), which are independently and identically
distributed (i.i.d.). At the receive side, additive white Gaussian
noise is added in the form of a column vector n with i.i.d.
entries CN(0, σ2
), where σ2
is determined by the signal to
noise ratio (SNR). We can thus write the received signal vector
(RSV) y as:
y = Hx + n (1)
Detection on the receive side, in our case, is performed
using Maximum Likelihood (ML) detection, the explanation
of which follows.
Let us call the set of MASM symbols from which the
symbol x is chosen X. Detection is performed using an
exhaustive search over all X. We write the ML decision metric
as the following, where ˜x is the estimated symbol and xi is
an element of X:
˜x = arg min
xi
y − Hxi
2
F xi ∈ X (2)
III. PROPOSED ANTENNA GROUP SELECTION SCHEME
AND SIMPLIFICATIONS
We move to apply the work of [10] and [9], which are
proposed for improving the performance of conventional SM,
to MASM. In conventional SM, the number of active transmit
antennas is one, and therefore the spectral efficiency is given
by log2 MNt. By increasing the number of transmit antennas
to NT > Nt and choosing Nt antennas from the total NT
in an intelligent fashion, the bit error rate (BER) performance
can be improved [10][9]. We can intuit that this method is
indeed valid as a column in the channel matrix undergoing
fading can be replaced by another column that has a larger
norm.
Now, in MASM, if the number of transmit antennas is in-
creased from Nt to NT , the number of possible antenna groups
is increased from Nt
Np
to NT
Np
. Unlike the conventional SM
case, the question that must be asked in the MASM case is
1Fractional bit encoding can be used to overcome this limit [3].
2
how to select those antenna groups that maximise the BER
performance.
A. Proposed Scheme - Euclidean Distance Antenna Group
Selection (EDAGS)
We take note in the MASM system described in II, that the
total number of possible antenna groups is NΨ = Nt
Np
and
that an integer power of two number of them NΓ, must be
selected for the construction of the symbol set X. However,
in contrast to conventional MASM, for the proposed scheme
and similar to conventional EDAS, we have a number of
transmit antennas greater than the minimum required for
NΓ = 2
log2(Nt
Np
) , written as NT ≥ Nt and thus, we have
NΨ = NT
Np
.
Now, in conventional MASM, the selection of NΓ from
NΨ is made arbitrarily, however, in the work presented in
this paper, we select those antenna groups which maximise
Euclidean distance between any two MASM symbols. We
call this method Euclidean Distance Antenna Group Selection
(EDAGS) and elucidate further with a mathematical descrip-
tion.
Let us note the ML metric (2) and rewrite it after substitut-
ing (1).
˜x = arg min
xi
Hx + n − Hxi
2
F xi ∈ X (3)
From (3), we can see that the performance of the system can
be improved by selecting a subset of MASM antenna groups
that maximises the minimum Euclidean distance between
symbol vectors.
Enumerating the NΨ
NΓ
ways of selecting valid antenna
groups as Ψ, we can write the selection as:
ΓED = arg max
Γ∈Ψ
{ min
γ1,γ2∈Γ
x1,x2∈S
x1=x2 iff γ1=γ2
Hγ1
x1 − Hγ2
x2
2
F } (4)
Where: Hγi has Np columns taken from those in the
channel matrix H which correspond to the active antennas
in the γi antenna group of Γ; xi is an Np × 1 vector from
the MNp
sized set S of all possible Np dimensional MQAM
symbols and ΓED is the resultant set of antenna groups.
B. Euclidean Distance Antenna Selection (EDAS)
Unfortunately, the complexity of the proposed EDAGS
scheme is very large. In order to reduce complexity, we apply
the results of [9] and [10] and apply them to the MASM
system. In order to do this, we find those Nt antennas of the
full NT antenna set that maximise the minimum Euclidean
distance between MASM symbols and subsequently create
antenna groups from the Nt antennas selected. To reiterate, this
is in contrast to selecting those antenna groups that maximise
the minimum Euclidean distance between symbols.
We obtain the equation for the EDAS method as described
when applied to MASM as:
IED = arg max
I∈I
{ min
x1=x2∈X
HI(x1 − x2) 2
F } (5)
Where I is the enumeration of each combination of NT
Nt
and HI has Nt columns described by I. Once IED is
found, we create the NΓ antenna groups required for MASM
transmission from the antennas described in IED.
C. Capacity Optimised Antenna Selection
In [9], a second scheme aside from EDAS is proposed that
is based upon the capacity of the channel matrix instead of the
Euclidean distance between symbols. The approach results in
drastically reduced complexity for the price of performance.
Being based upon capacity, the scheme is dubbed Capacity
Optimised Antenna Selection (COAS).
In this paper, we propose capacity optimisation by consid-
ering the equation for the capacity of a MIMO channel. This
capacity is given by:
C = EH log2 det INr +
ρ
η
HHH
(6)
Let us call the sum in the determinant function A, i.e.:
A = INr
+
ρ
η
HHH
(7)
Using QR decomposition, we can write A as the product of
a unitary, orthogonal matrix Q and an upper triangular matrix
R. The determinant function in (6) can thus be expressed as:
det INr
+
ρ
η
HHH
= det (QR) (8)
Using established properties of determinants, specifically,
the determinant of a product and the determinant of a unitary
matrix, we can further write:
det (QR) = det (Q) · det (R) = 1 · det (R) = det (R) (9)
Since R is upper triangular, its determinant is equal to the
product of its diagonal elements. This is written as:
det INr +
ρ
η
HHH
=
Nr
i=1
rii r ∈ R (10)
Now, since the objective of capacity optimised antenna
selection is to improve performance by selecting Nt antennas
from the larger set of NT antennas, the ensuing step, logically,
is to compute the upper triangular matrix in such a way
that the diagonal entries are monotonically decreasing. This
may be done intuitively by permuting the columns of the
channel matrix H and calculating the determinant for each
permutation. Once the diagonal is as desired, we select those
Nt antennas corresponding to the first Nt columns in the R
matrix.
Applying this method as is, we are presented with the
obstacle of very large computational complexity due to the
number of possible permutations, given as NT !. For example,
if NT = 6, there are 720 possible permutations. We are
able to drastically decrease the computational complexity by
altering the way in which QR decomposition is performed.
This method is presented in what follows.
α ≤ CSM ≤ α + log2(NSM ) (11)
3
Where α = 1
NSM
NSM
i=1 log2(1 + ρ hi
2
). It is clear
from (11) that capacity is maximised if the Nt antennas
corresponding to the columns in the H with the largest norms
are chosen out of the NT columns.
Although COAS is intended for conventional SM, with a
single symbol being transmitted in each symbol period, we use
the scheme as a comparison to proposed schemes by creating
the NΓ antenna groups from those Nt that are selected.
D. Improved Capacity Optimised Antenna Selection
We note that COAS is intended to be implemented as part of
a conventional SM scheme. This can be readily seen from the
equation for capacity given in (11), which exhibits the norm
of a single column at a time. It is therefore advantageous to
improve upon COAS by considering the equation for capacity
that is applicable to simultaneous transmission from multiple
transmit antennas, since, discarding antenna group selection in
MASM, that which remains is spatial multiplexing (SMX).
We write the equation for the capacity of the MASM scheme
as:
α ≤ CMASM ≤ α log2(NΓ) (12)
Where, critically, α is given as:
We continue by analysing the term HHH
, that is, the chan-
nel matrix multiplied by its Hermitian or conjugate transpose.
If we apply QR decomposition to the channel matrix, the term
becomes:
HHH
= QRRH
QH
(13)
Where QQH
= I is a unitary matrix with orthogonal
columns and R is upper triangular. Trivially, the conjugate
transpose of an upper triangular matrix is a lower triangular
matrix, and thus we find RRH
as a diagonal matrix consisting
of the squares of the diagonal entries in the original upper
triangular matrix with unchanged positions.
Further, considering the intrinsic properties of the matrix Q
mentioned as unitary and orthogonal, we discard it in (6).
With such insight elucidated regarding the capacity of
the MASM system, it is readily observed that performing
QR decomposition in such a way that the columns of H
are permuted to produce an upper triangular matrix R with
monotonically decreasing entries on the diagonal is a simple
method to improve the capacity of the system.
Let us consider the ML metric (2). If we substitute the
received signal vector (1) into the ML metric and assume
x = xi, we can rewrite (2) as:
˜x = arg min
xi
H∆x + n 2
F xi ∈ X (14)
Where ∆x = x − xi. In the proposed method of this paper
named EDAGS, we maximise the minimum distance of (14)
(when x = xi) by considering simultaneously each antenna
and symbol combination. However, this exhaustive method is
computationally intensive. In order to reduce complexity, we
assume that the symbols resulting from ∆x are unknown. With
this assumption in mind, it is apparent that selecting those
transmit antennas which maximise the Euclidean distance
represented by H∆x + n is a method for minimizing the
BER of the system. Since, in this method, n is also unknown,
the objective is pursued in the following:
I = arg max
I∈I
HI∆x 2
F (15)
In COAS [9], as mentioned in Section III-C, we select those
columns in the channel matrix with the highest absolute norms
in order to maximise the capacity and thus the BER of the
system. If this method is used to solve for (15), we obtain a
fairly optimal result. However, the use of such a method would
assume that all columns in the channel matrix are orthogonal,
an assumption which is fairly accurate if the number of receive
antennas is large. However, in simulations, it is shown that
better performance can be achieved by pursuing (15) whilst
not assuming perfect orthogonality between every column in
H.
We thus propose a solution of (15) which assumes does
not assume perfect orthogonality between the columns in the
channel matrix, a method dubbed Sorted QR Decomposition
Based Antenna Selection (QRAS), which, to the best of the
authors’ knowledge, has not been offered elsewhere.
To support and give motivation for QRAS, we trace what
follows. To begin, we consider a MIMO system with Nt = 2.
The square of the triangle inequality gives:
h1 + h2
2
≤ h1
2
+ 2 h1 · h2 + h2
2
(16)
Since in uncorrelated flat-fading MIMO systems, column
vectors are pseudo-orthogonal, we can disregard the inner
product term 2 h1 · h2 and write a new triangle inequality
for the considered MIMO system:
h1 + h2
2
≤ h1
2
+ h2
2
(17)
Now, for the two column channel matrix in the considered
system, we decompose H∆x into the following:
H∆x = QR∆x = [q1 q2]
r11 r12
0 r22
∆x1
∆x2
(18)
Where, as is customary with QR decomposition, Q is
a unitary orthogonal matrix and R is an upper triangular
matrix. If we find the square of the Frobenius norm for the
decomposition in (18), we may begin by discarding Q as it is
unitary and write the remainder as:
H∆x 2
F = R∆x 2
F =
r11∆x1 + r12∆x2
r22∆x2
2
F (19)
Furthering (19), we are given:
H∆x 2
F = (r11∆x1 + r12∆x2)2
+ (r22∆x2)2
(20)
Nearing completion of the motivation, we consider two ex-
treme cases of (20): that in which both columns of the channel
matrix are parallel and that in which the columns of the chan-
nel matrix are completely orthogonal - over the comparison,
the amplitude of both vectors in the channel matrix remain
constant. In the first case, (20) is given by (r11∆x1+r12∆x2)2
and in the second, we have (r11∆x1)2
+ (r22∆x2)2
4
Let us note that in a system with Nr receive antennas,
we can find Nr orthogonal vectors, provided no two column
vectors in the channel matrix are exactly parallel.
We decompose the channel matrix into a unitary orthogonal
matrix Q and an upper triangular matrix R. This decompo-
sition is performed for each permutation of the columns of
the channel matrix H until the diagonal of the R matrix is
monotonically decreasing. In this way, we find the optimal
antenna selection by choosing those columns with the greatest
orthogonal distance from other columns.
Note in QRAS that the number of times QR decomposition
must performed is upper bounded by NT ! and thus the
computational complexity rises with the factorial of NT .
Once again, for comparison to the proposed scheme, we
implement QRAS for MASM. After the implementation, we
are left with Nt chosen antennas, from which we construct
the NΓ antenna groups.
IV. COMPUTATIONAL COMPLEXITY FOR PROPOSED AND
ANTECEDENT SCHEMES
Drawing on the work in [], we quantify the computa-
tional complexity of the antecedent and proposed schemes
in floating point operations (FLOPs), where each multipli-
cation, division, addition and subtraction count as a single
FLOP. Each scheme was analysed by the author in the
same manner in order to ensure continuity in method and
for brevity, such is excluded in this paper as the deriva-
tions are trivial. Thus, formulae for each are simply stated:
For EDAS, we have NT
Nt
N2
t M2
[2NtNr + Nt + Nr − 1]; for
QRAS, 2NT !NrN2
T ; for COAS, 2NT Nr −NT and for EDAGS
(proposed),
(NT
Np
)
NΓ
N2
ΓM2Np
[4NpNr + Nr − 1].
Fig. 1: Computational Complexity for 4 Antenna Selection
Schemes in 2 MASM Configurations
Figure 1 shows the complexity of the proposed and an-
tecedent schemes in the configurations described in Table I.
The proposed scheme is clearly orders of magnitude more
complex than that of the schemes to which it is compared.
However, in the subsequent section, it can be seen that the
NT Nt Np Nr NΓ M
6bits/s/Hz 6 4 2 5 4 4
10bits/s/Hz 6 4 2 5 4 16
TABLE I: MASM Configurations
proposed outperforms those to which it is compared in terms
of bit error rate (BER) performance.
V. RESULTS
The proposed scheme was simulated in a Rayleigh flat
fading environment with AWGN for the configurations in
Table I.
For comparison, we include those methods explained in III.
Note that in all cases other than the proposed scheme, we
select the minimum number of transmit antennas Nt, needed
to attain NΓ (i.e. 4 out of NT = 6), whereas in the proposed
scheme, the antenna groups may include any of the NT = 6
transmit antennas - this is where the advantage of the proposed
scheme lies. The results are shown in Figure 3.
Fig. 2: Simulation Results for MASM 6bits/s/Hz
We can see from the simulation results that the proposed
scheme offers a significant performance gain over the other
schemes considered. However, intrinsically, the computational
complexity of the proposed method is far larger than any of
the schemes used for comparison. Thus, there is room for
further work in finding a simplified method of implementing
the proposed method.
VI. CONCLUSION
Building upon the work of [9] and [11], the authors propose
a method to select optimal antenna groups in MASM. The
proposed method is found to perform significantly better than
conventional EDAS and two other benchmarks considered.
REFERENCES
[1] R. Mesleh, H. Haas, C. W. Ahn, and S. Yun, “Spatial modulation
- a new low complexity spectral efficiency enhancing technique,” in
Communications and Networking in China, 2006. ChinaCom ’06. First
International Conference on, oct. 2006, pp. 1 –5.
5
Fig. 3: Simulation Results for MASM 10bits/s/Hz
[2] H.-W. Liang, R. Chang, W.-H. Chung, H. Zhang, and S.-Y. Kuo, “Bi-
space shift keying modulation for mimo systems,” Communications
Letters, IEEE, vol. 16, no. 8, pp. 1161–1164, 2012.
[3] N. Serafimovski, M. Di Renzo, S. Sinanovic, R. Mesleh, and H. Haas,
“Fractional bit encoded spatial modulation (fbe-sm),” Communications
Letters, IEEE, vol. 14, no. 5, pp. 429 –431, may 2010.
[4] R. Chang, S.-J. Lin, and W.-H. Chung, “New space shift keying
modulation with hamming code-aided constellation design,” Wireless
Communications Letters, IEEE, vol. 1, no. 1, pp. 2–5, 2012.
[5] S. Sugiura, C. Xu, and L. Hanzo, “Reduced-complexity qam-aided
space-time shift keying,” in Global Telecommunications Conference
(GLOBECOM 2011), 2011 IEEE, 2011, pp. 1–6.
[6] M. Di Renzo and H. Haas, “Transmit-diversity for spatial modulation
(sm): Towards the design of high-rate spatially-modulated space-time
block codes,” in Communications (ICC), 2011 IEEE International Con-
ference on, june 2011, pp. 1 –6.
[7] M. DiRenzo and H. Haas, “Performance comparison of different spatial
modulation schemes in correlated fading channels,” in Communications
(ICC), 2010 IEEE International Conference on, May 2010, pp. 1–6.
[8] R. Mesleh, M. D. Renzo, H. Haas, and P. M. Grant, “Trellis coded
spatial modulation,” Wireless Communications, IEEE Transactions on,
vol. 9, no. 7, pp. 2349 –2361, july 2010.
[9] R. Rajashekar, K. Hari, and L. Hanzo, “Antenna selection in spatial
modulation systems,” Communications Letters, IEEE, vol. 17, no. 3, pp.
521–524, March 2013.
[10] N. Wang, W. Liu, H. Men, M. Jin, and H. Xu, “Further complexity
reduction using rotational symmetry for edas in spatial modulation,”
Communications Letters, IEEE, vol. PP, no. 99, pp. 1–1, 2014.
[11] J. Wang, S. Jia, and J. Song, “Generalised spatial modulation system
with multiple active transmit antennas and low complexity detection
scheme,” Wireless Communications, IEEE Transactions on, vol. 11,
no. 4, pp. 1605 –1615, april 2012.

Mais conteúdo relacionado

Mais procurados

Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...
Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...
Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...graphhoc
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...IOSR Journals
 
QRC-ESPRIT Method for Wideband Signals
QRC-ESPRIT Method for Wideband SignalsQRC-ESPRIT Method for Wideband Signals
QRC-ESPRIT Method for Wideband SignalsIDES Editor
 
2014.03.31.bach glc-pham-finalizing[conflict]
2014.03.31.bach glc-pham-finalizing[conflict]2014.03.31.bach glc-pham-finalizing[conflict]
2014.03.31.bach glc-pham-finalizing[conflict]Bách Vũ Trọng
 
Topo intro wsn
Topo intro wsnTopo intro wsn
Topo intro wsnRishu Seth
 
Performance of cognitive radio networks with maximal ratio combining over cor...
Performance of cognitive radio networks with maximal ratio combining over cor...Performance of cognitive radio networks with maximal ratio combining over cor...
Performance of cognitive radio networks with maximal ratio combining over cor...Polytechnique Montreal
 
Simultaneous_VTC_Qian
Simultaneous_VTC_QianSimultaneous_VTC_Qian
Simultaneous_VTC_QianQian Han
 
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...Multi carrier equalization by restoration of redundanc y (merry) for adaptive...
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...IJNSA Journal
 
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...ijsrd.com
 
Resource allocation in OFDM based cognitive radio system
Resource allocation in OFDM based cognitive radio systemResource allocation in OFDM based cognitive radio system
Resource allocation in OFDM based cognitive radio systemGautham Reddy
 
A Subspace Method for Blind Channel Estimation in CP-free OFDM Systems
A Subspace Method for Blind Channel Estimation in CP-free OFDM SystemsA Subspace Method for Blind Channel Estimation in CP-free OFDM Systems
A Subspace Method for Blind Channel Estimation in CP-free OFDM SystemsCSCJournals
 
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...Polytechnique Montreal
 
TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013Clay Stanek
 

Mais procurados (17)

Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...
Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...
Performance Analysis of Ultra Wideband Receivers for High Data Rate Wireless ...
 
9517cnc05
9517cnc059517cnc05
9517cnc05
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
 
QRC-ESPRIT Method for Wideband Signals
QRC-ESPRIT Method for Wideband SignalsQRC-ESPRIT Method for Wideband Signals
QRC-ESPRIT Method for Wideband Signals
 
2014.03.31.bach glc-pham-finalizing[conflict]
2014.03.31.bach glc-pham-finalizing[conflict]2014.03.31.bach glc-pham-finalizing[conflict]
2014.03.31.bach glc-pham-finalizing[conflict]
 
Topo intro wsn
Topo intro wsnTopo intro wsn
Topo intro wsn
 
Performance of cognitive radio networks with maximal ratio combining over cor...
Performance of cognitive radio networks with maximal ratio combining over cor...Performance of cognitive radio networks with maximal ratio combining over cor...
Performance of cognitive radio networks with maximal ratio combining over cor...
 
Simultaneous_VTC_Qian
Simultaneous_VTC_QianSimultaneous_VTC_Qian
Simultaneous_VTC_Qian
 
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...Multi carrier equalization by restoration of redundanc y (merry) for adaptive...
Multi carrier equalization by restoration of redundanc y (merry) for adaptive...
 
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...
Bit Error Rate Performance of MIMO Spatial Multiplexing with MPSK Modulation ...
 
FK_icassp_2014
FK_icassp_2014FK_icassp_2014
FK_icassp_2014
 
10.1.1.666.9435
10.1.1.666.943510.1.1.666.9435
10.1.1.666.9435
 
E04923142
E04923142E04923142
E04923142
 
Resource allocation in OFDM based cognitive radio system
Resource allocation in OFDM based cognitive radio systemResource allocation in OFDM based cognitive radio system
Resource allocation in OFDM based cognitive radio system
 
A Subspace Method for Blind Channel Estimation in CP-free OFDM Systems
A Subspace Method for Blind Channel Estimation in CP-free OFDM SystemsA Subspace Method for Blind Channel Estimation in CP-free OFDM Systems
A Subspace Method for Blind Channel Estimation in CP-free OFDM Systems
 
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...
Using Subspace Pursuit Algorithm to Improve Performance of the Distributed Co...
 
TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013
 

Destaque

Electromyography Analysis for Person Identification
Electromyography Analysis for Person IdentificationElectromyography Analysis for Person Identification
Electromyography Analysis for Person IdentificationCSCJournals
 
Optical Spatial Modulation with Transmitter-Receiver Alignments
Optical Spatial Modulation with Transmitter-Receiver AlignmentsOptical Spatial Modulation with Transmitter-Receiver Alignments
Optical Spatial Modulation with Transmitter-Receiver AlignmentsMarwan Hammouda
 
Concept of Flip OFDM and its applications
Concept of Flip OFDM and its applicationsConcept of Flip OFDM and its applications
Concept of Flip OFDM and its applicationsDarshan Bhatt
 
Optical Spatial Modulation OFDM using Micro LEDs
Optical Spatial Modulation OFDM using Micro LEDsOptical Spatial Modulation OFDM using Micro LEDs
Optical Spatial Modulation OFDM using Micro LEDsBasil Jacob
 
Analog and digital modulation formats of optical fiber communication within a...
Analog and digital modulation formats of optical fiber communication within a...Analog and digital modulation formats of optical fiber communication within a...
Analog and digital modulation formats of optical fiber communication within a...IAEME Publication
 
Multi Carrier Modulation and Single Carrier Modulation
Multi Carrier Modulation and Single Carrier ModulationMulti Carrier Modulation and Single Carrier Modulation
Multi Carrier Modulation and Single Carrier Modulationfernandomireles
 
Spatial Modulation
Spatial ModulationSpatial Modulation
Spatial ModulationArvin Moeini
 
Digital & analog transmission
Digital & analog transmissionDigital & analog transmission
Digital & analog transmissionJeffery Vava
 
Different types of Modulation Techniques
Different types of Modulation TechniquesDifferent types of Modulation Techniques
Different types of Modulation TechniquesHimel Himo
 
Optical fiber communication Part 2 Sources and Detectors
Optical fiber communication Part 2 Sources and DetectorsOptical fiber communication Part 2 Sources and Detectors
Optical fiber communication Part 2 Sources and DetectorsMadhumita Tamhane
 

Destaque (11)

Electromyography Analysis for Person Identification
Electromyography Analysis for Person IdentificationElectromyography Analysis for Person Identification
Electromyography Analysis for Person Identification
 
Optical Spatial Modulation with Transmitter-Receiver Alignments
Optical Spatial Modulation with Transmitter-Receiver AlignmentsOptical Spatial Modulation with Transmitter-Receiver Alignments
Optical Spatial Modulation with Transmitter-Receiver Alignments
 
Concept of Flip OFDM and its applications
Concept of Flip OFDM and its applicationsConcept of Flip OFDM and its applications
Concept of Flip OFDM and its applications
 
Optical Spatial Modulation OFDM using Micro LEDs
Optical Spatial Modulation OFDM using Micro LEDsOptical Spatial Modulation OFDM using Micro LEDs
Optical Spatial Modulation OFDM using Micro LEDs
 
Analog and digital modulation formats of optical fiber communication within a...
Analog and digital modulation formats of optical fiber communication within a...Analog and digital modulation formats of optical fiber communication within a...
Analog and digital modulation formats of optical fiber communication within a...
 
Multi Carrier Modulation and Single Carrier Modulation
Multi Carrier Modulation and Single Carrier ModulationMulti Carrier Modulation and Single Carrier Modulation
Multi Carrier Modulation and Single Carrier Modulation
 
Spatial Modulation
Spatial ModulationSpatial Modulation
Spatial Modulation
 
Digital & analog transmission
Digital & analog transmissionDigital & analog transmission
Digital & analog transmission
 
(Ofdm)
(Ofdm)(Ofdm)
(Ofdm)
 
Different types of Modulation Techniques
Different types of Modulation TechniquesDifferent types of Modulation Techniques
Different types of Modulation Techniques
 
Optical fiber communication Part 2 Sources and Detectors
Optical fiber communication Part 2 Sources and DetectorsOptical fiber communication Part 2 Sources and Detectors
Optical fiber communication Part 2 Sources and Detectors
 

Semelhante a Masters Report 3

18 15993 31427-1-sm(edit)nn
18 15993 31427-1-sm(edit)nn18 15993 31427-1-sm(edit)nn
18 15993 31427-1-sm(edit)nnIAESIJEECS
 
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...ijcnac
 
MIMO Channel Estimation Using the LS and MMSE Algorithm
MIMO Channel Estimation Using the LS and MMSE AlgorithmMIMO Channel Estimation Using the LS and MMSE Algorithm
MIMO Channel Estimation Using the LS and MMSE AlgorithmIOSRJECE
 
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...IJECEIAES
 
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelCapacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelIOSR Journals
 
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelCapacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelIOSR Journals
 
Optimal pilot symbol power allocation in lte
Optimal pilot symbol power allocation in lteOptimal pilot symbol power allocation in lte
Optimal pilot symbol power allocation in lteDat Manh
 
Iterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderIterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderCSCJournals
 
Iterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderIterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderCSCJournals
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...IOSR Journals
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...IOSR Journals
 
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...Editor IJCATR
 
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...Editor IJCATR
 
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...Editor IJCATR
 
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding Algorithm
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding AlgorithmFixed Point Realization of Iterative LR-Aided Soft MIMO Decoding Algorithm
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding AlgorithmCSCJournals
 
Performance evaluation with a
Performance evaluation with aPerformance evaluation with a
Performance evaluation with aijmnct
 
Paper id 26201482
Paper id 26201482Paper id 26201482
Paper id 26201482IJRAT
 
Hybrid Low Complex near Optimal Detector for Spatial Modulation
Hybrid Low Complex near Optimal Detector for Spatial Modulation Hybrid Low Complex near Optimal Detector for Spatial Modulation
Hybrid Low Complex near Optimal Detector for Spatial Modulation IJECEIAES
 
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...ijwmn
 

Semelhante a Masters Report 3 (20)

18 15993 31427-1-sm(edit)nn
18 15993 31427-1-sm(edit)nn18 15993 31427-1-sm(edit)nn
18 15993 31427-1-sm(edit)nn
 
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...
Performance Comparison of Modified Variable Step Size Leaky LMS Algorithm for...
 
MIMO Channel Estimation Using the LS and MMSE Algorithm
MIMO Channel Estimation Using the LS and MMSE AlgorithmMIMO Channel Estimation Using the LS and MMSE Algorithm
MIMO Channel Estimation Using the LS and MMSE Algorithm
 
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...
Computationally Efficient Multi-Antenna Techniques for Multi-User Two-Way Wire...
 
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelCapacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
 
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading ChannelCapacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
Capacity Enhancement of MIMO-OFDM System in Rayleigh Fading Channel
 
Optimal pilot symbol power allocation in lte
Optimal pilot symbol power allocation in lteOptimal pilot symbol power allocation in lte
Optimal pilot symbol power allocation in lte
 
Technical details
Technical detailsTechnical details
Technical details
 
Iterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderIterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO Decoder
 
Iterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO DecoderIterative Soft Decision Based Complex K-best MIMO Decoder
Iterative Soft Decision Based Complex K-best MIMO Decoder
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
 
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
Performance Analysis of Various Symbol Detection Techniques in Wireless MIMO ...
 
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...
Dynamic Chunks Distribution Scheme for Multiservice Load Balancing Using Fibo...
 
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...
Mathematical Approach to Complexity - Reduced Antenna Selection Technique for...
 
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...
Mathematical Approach to Complexity-Reduced Antenna Selection Technique for A...
 
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding Algorithm
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding AlgorithmFixed Point Realization of Iterative LR-Aided Soft MIMO Decoding Algorithm
Fixed Point Realization of Iterative LR-Aided Soft MIMO Decoding Algorithm
 
Performance evaluation with a
Performance evaluation with aPerformance evaluation with a
Performance evaluation with a
 
Paper id 26201482
Paper id 26201482Paper id 26201482
Paper id 26201482
 
Hybrid Low Complex near Optimal Detector for Spatial Modulation
Hybrid Low Complex near Optimal Detector for Spatial Modulation Hybrid Low Complex near Optimal Detector for Spatial Modulation
Hybrid Low Complex near Optimal Detector for Spatial Modulation
 
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...
3D METALLIC PLATE LENS ANTENNA BASED BEAMSPACE CHANNEL ESTIMATION TECHNIQUE F...
 

Masters Report 3

  • 1. 1 Euclidean Distance Antenna Selection for Spatial Modulation with Multiple Active Transmit Antennas Lloyd Blackbeard, Hongjun Xu and Fengfan Yang Abstract— I. INTRODUCTION Spatial Modulation (SM)[1] could be argued to be known fairly ubiquitously in communications academia. Much work has been done in improving its throughput, bit error rate (BER) performance and detection complexity [2][3][4][5][6][7][8][9][10][11]. In one such improvement, called Euclidean Distance An- tenna Selection (EDAS)[9], 2n n ∈ N transmit antennas are selected from a larger set in a fashion that maximises the Eu- clidean distance between any two symbols. In Multiple Active Spatial Modulation (MASM)[11], more than one antenna is selected to be active in each symbol period, usually allowing for more information to be mapped to the spatial domain and allowing the simultaneous transmission of unique symbols from each active antenna, thereby improving throughput. To the best of the authors’ knowledge, no system has been implemented to improve the performance of MASM in a fashion similar to EDAS. Thus, in this paper, EDAS will be implemented for SM systems with multiple active antennas. The remainder of the paper follows: in section II, a model for MASM is presented that will be used in subsequent sections; in section III, we implement the idea of EDAS for MASM and summarise antecedent work with which the proposed method will be compared; in section IV, we write the computational complexity for all schemes implemented in the paper and produce the analytical results; in section V, we show simulation results in terms of bit error rate (BER) and in section VI, concluding remarks are given. Unless otherwise specified the notation convention follows: (·)−1 , (·)T , (·)H , (·)† , E[·], | · | and | · |F denote the matrix inverse, transpose, Hermitian, Moore-Penrose pseudoinverse, expectation, Euclidean norm and Frobenius norm operators respectively; regular, bold face lower case, bold face upper case and capital script/cursive text refer to scalars, vectors, matrices and sets respectively; subscripts (·)ij denote the ith row, jth column entry in the corresponding matrix and subscript (·)i denotes the ith entry in the corresponding vector or the ith column in the corresponding matrix. II. SYSTEM MODEL The following serves to summarise convential MASM. We consider a MIMO environment with Nt transmit antennas and Nr receive antennas. We observe that a maximum of NΓ = 2 log2(Nt Np ) antenna groups can be defined for Np ≤ Nt active transmit antennas1 . For each active antenna, we map log2M bits from the input bitstream to an MQAM symbol selected from an MQAM constellation of size M. A further log2NΓ bits select which of the Nt antennas will be active, or rather, which antenna group will be used in transmission. Once mapping is complete for the symbol period, we have the transmit symbol column vector x with MQAM symbols in positions corresponding to the active antennas and zeros elsewhere. This vector is transmitted over a channel matrix H with entries CN(0, 1), which are independently and identically distributed (i.i.d.). At the receive side, additive white Gaussian noise is added in the form of a column vector n with i.i.d. entries CN(0, σ2 ), where σ2 is determined by the signal to noise ratio (SNR). We can thus write the received signal vector (RSV) y as: y = Hx + n (1) Detection on the receive side, in our case, is performed using Maximum Likelihood (ML) detection, the explanation of which follows. Let us call the set of MASM symbols from which the symbol x is chosen X. Detection is performed using an exhaustive search over all X. We write the ML decision metric as the following, where ˜x is the estimated symbol and xi is an element of X: ˜x = arg min xi y − Hxi 2 F xi ∈ X (2) III. PROPOSED ANTENNA GROUP SELECTION SCHEME AND SIMPLIFICATIONS We move to apply the work of [10] and [9], which are proposed for improving the performance of conventional SM, to MASM. In conventional SM, the number of active transmit antennas is one, and therefore the spectral efficiency is given by log2 MNt. By increasing the number of transmit antennas to NT > Nt and choosing Nt antennas from the total NT in an intelligent fashion, the bit error rate (BER) performance can be improved [10][9]. We can intuit that this method is indeed valid as a column in the channel matrix undergoing fading can be replaced by another column that has a larger norm. Now, in MASM, if the number of transmit antennas is in- creased from Nt to NT , the number of possible antenna groups is increased from Nt Np to NT Np . Unlike the conventional SM case, the question that must be asked in the MASM case is 1Fractional bit encoding can be used to overcome this limit [3].
  • 2. 2 how to select those antenna groups that maximise the BER performance. A. Proposed Scheme - Euclidean Distance Antenna Group Selection (EDAGS) We take note in the MASM system described in II, that the total number of possible antenna groups is NΨ = Nt Np and that an integer power of two number of them NΓ, must be selected for the construction of the symbol set X. However, in contrast to conventional MASM, for the proposed scheme and similar to conventional EDAS, we have a number of transmit antennas greater than the minimum required for NΓ = 2 log2(Nt Np ) , written as NT ≥ Nt and thus, we have NΨ = NT Np . Now, in conventional MASM, the selection of NΓ from NΨ is made arbitrarily, however, in the work presented in this paper, we select those antenna groups which maximise Euclidean distance between any two MASM symbols. We call this method Euclidean Distance Antenna Group Selection (EDAGS) and elucidate further with a mathematical descrip- tion. Let us note the ML metric (2) and rewrite it after substitut- ing (1). ˜x = arg min xi Hx + n − Hxi 2 F xi ∈ X (3) From (3), we can see that the performance of the system can be improved by selecting a subset of MASM antenna groups that maximises the minimum Euclidean distance between symbol vectors. Enumerating the NΨ NΓ ways of selecting valid antenna groups as Ψ, we can write the selection as: ΓED = arg max Γ∈Ψ { min γ1,γ2∈Γ x1,x2∈S x1=x2 iff γ1=γ2 Hγ1 x1 − Hγ2 x2 2 F } (4) Where: Hγi has Np columns taken from those in the channel matrix H which correspond to the active antennas in the γi antenna group of Γ; xi is an Np × 1 vector from the MNp sized set S of all possible Np dimensional MQAM symbols and ΓED is the resultant set of antenna groups. B. Euclidean Distance Antenna Selection (EDAS) Unfortunately, the complexity of the proposed EDAGS scheme is very large. In order to reduce complexity, we apply the results of [9] and [10] and apply them to the MASM system. In order to do this, we find those Nt antennas of the full NT antenna set that maximise the minimum Euclidean distance between MASM symbols and subsequently create antenna groups from the Nt antennas selected. To reiterate, this is in contrast to selecting those antenna groups that maximise the minimum Euclidean distance between symbols. We obtain the equation for the EDAS method as described when applied to MASM as: IED = arg max I∈I { min x1=x2∈X HI(x1 − x2) 2 F } (5) Where I is the enumeration of each combination of NT Nt and HI has Nt columns described by I. Once IED is found, we create the NΓ antenna groups required for MASM transmission from the antennas described in IED. C. Capacity Optimised Antenna Selection In [9], a second scheme aside from EDAS is proposed that is based upon the capacity of the channel matrix instead of the Euclidean distance between symbols. The approach results in drastically reduced complexity for the price of performance. Being based upon capacity, the scheme is dubbed Capacity Optimised Antenna Selection (COAS). In this paper, we propose capacity optimisation by consid- ering the equation for the capacity of a MIMO channel. This capacity is given by: C = EH log2 det INr + ρ η HHH (6) Let us call the sum in the determinant function A, i.e.: A = INr + ρ η HHH (7) Using QR decomposition, we can write A as the product of a unitary, orthogonal matrix Q and an upper triangular matrix R. The determinant function in (6) can thus be expressed as: det INr + ρ η HHH = det (QR) (8) Using established properties of determinants, specifically, the determinant of a product and the determinant of a unitary matrix, we can further write: det (QR) = det (Q) · det (R) = 1 · det (R) = det (R) (9) Since R is upper triangular, its determinant is equal to the product of its diagonal elements. This is written as: det INr + ρ η HHH = Nr i=1 rii r ∈ R (10) Now, since the objective of capacity optimised antenna selection is to improve performance by selecting Nt antennas from the larger set of NT antennas, the ensuing step, logically, is to compute the upper triangular matrix in such a way that the diagonal entries are monotonically decreasing. This may be done intuitively by permuting the columns of the channel matrix H and calculating the determinant for each permutation. Once the diagonal is as desired, we select those Nt antennas corresponding to the first Nt columns in the R matrix. Applying this method as is, we are presented with the obstacle of very large computational complexity due to the number of possible permutations, given as NT !. For example, if NT = 6, there are 720 possible permutations. We are able to drastically decrease the computational complexity by altering the way in which QR decomposition is performed. This method is presented in what follows. α ≤ CSM ≤ α + log2(NSM ) (11)
  • 3. 3 Where α = 1 NSM NSM i=1 log2(1 + ρ hi 2 ). It is clear from (11) that capacity is maximised if the Nt antennas corresponding to the columns in the H with the largest norms are chosen out of the NT columns. Although COAS is intended for conventional SM, with a single symbol being transmitted in each symbol period, we use the scheme as a comparison to proposed schemes by creating the NΓ antenna groups from those Nt that are selected. D. Improved Capacity Optimised Antenna Selection We note that COAS is intended to be implemented as part of a conventional SM scheme. This can be readily seen from the equation for capacity given in (11), which exhibits the norm of a single column at a time. It is therefore advantageous to improve upon COAS by considering the equation for capacity that is applicable to simultaneous transmission from multiple transmit antennas, since, discarding antenna group selection in MASM, that which remains is spatial multiplexing (SMX). We write the equation for the capacity of the MASM scheme as: α ≤ CMASM ≤ α log2(NΓ) (12) Where, critically, α is given as: We continue by analysing the term HHH , that is, the chan- nel matrix multiplied by its Hermitian or conjugate transpose. If we apply QR decomposition to the channel matrix, the term becomes: HHH = QRRH QH (13) Where QQH = I is a unitary matrix with orthogonal columns and R is upper triangular. Trivially, the conjugate transpose of an upper triangular matrix is a lower triangular matrix, and thus we find RRH as a diagonal matrix consisting of the squares of the diagonal entries in the original upper triangular matrix with unchanged positions. Further, considering the intrinsic properties of the matrix Q mentioned as unitary and orthogonal, we discard it in (6). With such insight elucidated regarding the capacity of the MASM system, it is readily observed that performing QR decomposition in such a way that the columns of H are permuted to produce an upper triangular matrix R with monotonically decreasing entries on the diagonal is a simple method to improve the capacity of the system. Let us consider the ML metric (2). If we substitute the received signal vector (1) into the ML metric and assume x = xi, we can rewrite (2) as: ˜x = arg min xi H∆x + n 2 F xi ∈ X (14) Where ∆x = x − xi. In the proposed method of this paper named EDAGS, we maximise the minimum distance of (14) (when x = xi) by considering simultaneously each antenna and symbol combination. However, this exhaustive method is computationally intensive. In order to reduce complexity, we assume that the symbols resulting from ∆x are unknown. With this assumption in mind, it is apparent that selecting those transmit antennas which maximise the Euclidean distance represented by H∆x + n is a method for minimizing the BER of the system. Since, in this method, n is also unknown, the objective is pursued in the following: I = arg max I∈I HI∆x 2 F (15) In COAS [9], as mentioned in Section III-C, we select those columns in the channel matrix with the highest absolute norms in order to maximise the capacity and thus the BER of the system. If this method is used to solve for (15), we obtain a fairly optimal result. However, the use of such a method would assume that all columns in the channel matrix are orthogonal, an assumption which is fairly accurate if the number of receive antennas is large. However, in simulations, it is shown that better performance can be achieved by pursuing (15) whilst not assuming perfect orthogonality between every column in H. We thus propose a solution of (15) which assumes does not assume perfect orthogonality between the columns in the channel matrix, a method dubbed Sorted QR Decomposition Based Antenna Selection (QRAS), which, to the best of the authors’ knowledge, has not been offered elsewhere. To support and give motivation for QRAS, we trace what follows. To begin, we consider a MIMO system with Nt = 2. The square of the triangle inequality gives: h1 + h2 2 ≤ h1 2 + 2 h1 · h2 + h2 2 (16) Since in uncorrelated flat-fading MIMO systems, column vectors are pseudo-orthogonal, we can disregard the inner product term 2 h1 · h2 and write a new triangle inequality for the considered MIMO system: h1 + h2 2 ≤ h1 2 + h2 2 (17) Now, for the two column channel matrix in the considered system, we decompose H∆x into the following: H∆x = QR∆x = [q1 q2] r11 r12 0 r22 ∆x1 ∆x2 (18) Where, as is customary with QR decomposition, Q is a unitary orthogonal matrix and R is an upper triangular matrix. If we find the square of the Frobenius norm for the decomposition in (18), we may begin by discarding Q as it is unitary and write the remainder as: H∆x 2 F = R∆x 2 F = r11∆x1 + r12∆x2 r22∆x2 2 F (19) Furthering (19), we are given: H∆x 2 F = (r11∆x1 + r12∆x2)2 + (r22∆x2)2 (20) Nearing completion of the motivation, we consider two ex- treme cases of (20): that in which both columns of the channel matrix are parallel and that in which the columns of the chan- nel matrix are completely orthogonal - over the comparison, the amplitude of both vectors in the channel matrix remain constant. In the first case, (20) is given by (r11∆x1+r12∆x2)2 and in the second, we have (r11∆x1)2 + (r22∆x2)2
  • 4. 4 Let us note that in a system with Nr receive antennas, we can find Nr orthogonal vectors, provided no two column vectors in the channel matrix are exactly parallel. We decompose the channel matrix into a unitary orthogonal matrix Q and an upper triangular matrix R. This decompo- sition is performed for each permutation of the columns of the channel matrix H until the diagonal of the R matrix is monotonically decreasing. In this way, we find the optimal antenna selection by choosing those columns with the greatest orthogonal distance from other columns. Note in QRAS that the number of times QR decomposition must performed is upper bounded by NT ! and thus the computational complexity rises with the factorial of NT . Once again, for comparison to the proposed scheme, we implement QRAS for MASM. After the implementation, we are left with Nt chosen antennas, from which we construct the NΓ antenna groups. IV. COMPUTATIONAL COMPLEXITY FOR PROPOSED AND ANTECEDENT SCHEMES Drawing on the work in [], we quantify the computa- tional complexity of the antecedent and proposed schemes in floating point operations (FLOPs), where each multipli- cation, division, addition and subtraction count as a single FLOP. Each scheme was analysed by the author in the same manner in order to ensure continuity in method and for brevity, such is excluded in this paper as the deriva- tions are trivial. Thus, formulae for each are simply stated: For EDAS, we have NT Nt N2 t M2 [2NtNr + Nt + Nr − 1]; for QRAS, 2NT !NrN2 T ; for COAS, 2NT Nr −NT and for EDAGS (proposed), (NT Np ) NΓ N2 ΓM2Np [4NpNr + Nr − 1]. Fig. 1: Computational Complexity for 4 Antenna Selection Schemes in 2 MASM Configurations Figure 1 shows the complexity of the proposed and an- tecedent schemes in the configurations described in Table I. The proposed scheme is clearly orders of magnitude more complex than that of the schemes to which it is compared. However, in the subsequent section, it can be seen that the NT Nt Np Nr NΓ M 6bits/s/Hz 6 4 2 5 4 4 10bits/s/Hz 6 4 2 5 4 16 TABLE I: MASM Configurations proposed outperforms those to which it is compared in terms of bit error rate (BER) performance. V. RESULTS The proposed scheme was simulated in a Rayleigh flat fading environment with AWGN for the configurations in Table I. For comparison, we include those methods explained in III. Note that in all cases other than the proposed scheme, we select the minimum number of transmit antennas Nt, needed to attain NΓ (i.e. 4 out of NT = 6), whereas in the proposed scheme, the antenna groups may include any of the NT = 6 transmit antennas - this is where the advantage of the proposed scheme lies. The results are shown in Figure 3. Fig. 2: Simulation Results for MASM 6bits/s/Hz We can see from the simulation results that the proposed scheme offers a significant performance gain over the other schemes considered. However, intrinsically, the computational complexity of the proposed method is far larger than any of the schemes used for comparison. Thus, there is room for further work in finding a simplified method of implementing the proposed method. VI. CONCLUSION Building upon the work of [9] and [11], the authors propose a method to select optimal antenna groups in MASM. The proposed method is found to perform significantly better than conventional EDAS and two other benchmarks considered. REFERENCES [1] R. Mesleh, H. Haas, C. W. Ahn, and S. Yun, “Spatial modulation - a new low complexity spectral efficiency enhancing technique,” in Communications and Networking in China, 2006. ChinaCom ’06. First International Conference on, oct. 2006, pp. 1 –5.
  • 5. 5 Fig. 3: Simulation Results for MASM 10bits/s/Hz [2] H.-W. Liang, R. Chang, W.-H. Chung, H. Zhang, and S.-Y. Kuo, “Bi- space shift keying modulation for mimo systems,” Communications Letters, IEEE, vol. 16, no. 8, pp. 1161–1164, 2012. [3] N. Serafimovski, M. Di Renzo, S. Sinanovic, R. Mesleh, and H. Haas, “Fractional bit encoded spatial modulation (fbe-sm),” Communications Letters, IEEE, vol. 14, no. 5, pp. 429 –431, may 2010. [4] R. Chang, S.-J. Lin, and W.-H. Chung, “New space shift keying modulation with hamming code-aided constellation design,” Wireless Communications Letters, IEEE, vol. 1, no. 1, pp. 2–5, 2012. [5] S. Sugiura, C. Xu, and L. Hanzo, “Reduced-complexity qam-aided space-time shift keying,” in Global Telecommunications Conference (GLOBECOM 2011), 2011 IEEE, 2011, pp. 1–6. [6] M. Di Renzo and H. Haas, “Transmit-diversity for spatial modulation (sm): Towards the design of high-rate spatially-modulated space-time block codes,” in Communications (ICC), 2011 IEEE International Con- ference on, june 2011, pp. 1 –6. [7] M. DiRenzo and H. Haas, “Performance comparison of different spatial modulation schemes in correlated fading channels,” in Communications (ICC), 2010 IEEE International Conference on, May 2010, pp. 1–6. [8] R. Mesleh, M. D. Renzo, H. Haas, and P. M. Grant, “Trellis coded spatial modulation,” Wireless Communications, IEEE Transactions on, vol. 9, no. 7, pp. 2349 –2361, july 2010. [9] R. Rajashekar, K. Hari, and L. Hanzo, “Antenna selection in spatial modulation systems,” Communications Letters, IEEE, vol. 17, no. 3, pp. 521–524, March 2013. [10] N. Wang, W. Liu, H. Men, M. Jin, and H. Xu, “Further complexity reduction using rotational symmetry for edas in spatial modulation,” Communications Letters, IEEE, vol. PP, no. 99, pp. 1–1, 2014. [11] J. Wang, S. Jia, and J. Song, “Generalised spatial modulation system with multiple active transmit antennas and low complexity detection scheme,” Wireless Communications, IEEE Transactions on, vol. 11, no. 4, pp. 1605 –1615, april 2012.