Professional Documents
Culture Documents
Abstract— Quadrature spatial modulation (QSM) based transmissions for forthcoming fifth generation (5G) wireless
millimeter wave (mmWave) communication system design using communications need to be addressed with some novel tech-
frequency diverse array (FDA) is proposed in this article. nologies. With a tremendous bandwidth availability rang-
Since spatial modulation techniques (SMT) employ maximum
likelihood (ML) based detector at receiver, the computational ing from 30–300 GHz and enhanced data rates, millimeter
complexity increases vastly. Moreover, numerous existing SMT wave (mmWave) communication tends to be a vital tech-
methods utilize phased array (PA), where only angle dependent nology for 5G wireless communications designs [1]. Note
communication is possible. In this article, we propose to use a that, mmWave communications can support data rates up to
standard FDA that exploits a small linearly increasing frequency multi-gigabits per second (Gb/s) in device to device (D2D),
offset across the array for range-angle dependent QSM wireless
communications. That is, the in-phase and quadrature compo- indoor/outdoor and cellular communications [2]–[5]. Impor-
nents of transmission vector utilize slightly different frequencies tantly, wave propagation at high bands such as 70/80 GHz pro-
to deliver the information. Furthermore, we propose a suboptimal vides very high data rates and bears a negligible atmospheric
multi-stage (MS) detector in the receiver, which applies a matched attenuation. Moreover, it remains unaffected by several
filtering (i.e., bandpass filters matched to the corresponding channel deterioration originators like dust, snow and light
frequency offsets) approach to decode the index bits at first step,
whereas, merely, two most probable estimated indices are utilized rain [6], [7]. In short, a high-throughput wireless transmis-
further to decode associated data bits using a standard ML sion with a boosted overall achievable data rate is essential
method. Adhering an improved signal to noise ratio (SNR) due to meet the requirements of the future generation wireless
to an FDA based range-angle dependent transmission, simulation technologies.
and numerical results show the improved performance of the One of the vital technical advances include the multi-
proposed design over existing SM and QSM based schemes, while
MS approach reduces the receiver’s computational complexity. ple input single output (MISO)/multiple input multiple out-
Index Terms— Quadrature spatial modulation, frequency put (MIMO) based design, which have demonstrated the
diverse array, millimeter wave (mmWave) communications, wire- high-throughput transmission performances in various recent
less communications. standards such as WIMAX, LTE etc., [8], [9]. Therefore,
I. I NTRODUCTION hybridizing high data rate mmWave designs with MIMO
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4572 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
transmit antennas bits are transmitted, which is an added applications. Different from PA, a standard FDA utilizes a
benefit with all the inherent properties of SM. The QSM linearly increasing small frequency offset across its array
performance over Rayleigh channel has been presented in [16], elements to generate a time–range–angle dependent beam pat-
whereas a comprehensive framework for generalized fading tern [29], [30] for radar target localization and tracking appli-
scenarios has been proposed in [18]. Eventually, the overall cations [31], [32]. Since the FDA is relatively new technique,
improved data rate advantage makes QSM a potential candi- few works about its application in communication field can
date for 5G mmWave communications. be found in the literature [33]–[41]. In [33], the authors have
The conventional receivers of SMT employ maximum proposed a range-dependent retro-directive FDA focusing for
likelihood (ML) detector [13], which has to search jointly wireless information and power transfer (WIPT). Moreover,
overall deployed transmit antennas and constellation symbols physical layer security for communication has been achieved
to detect active transmit antenna for demodulation IM bits and using time modulated and the directional modulated FDAs
transmitted data symbol, respectively. Unfortunately, the com- in [34] and [35]–[37], respectively. In [38], the authors have
putational complexity of ML-based detector is very high. proposed a retro-directive FDA time modulation for detecting
Therefore, instead of jointly detecting, suboptimal detectors precise legitimate location of target for improved detection
are opted to detect the activated transmit antennas first and performance. Similarly, FDAs have been utilized in [39]for
demodulating the data symbol over this antenna later, which obtaining enhanced capacity and detection performance for
can significantly reduce the complexity for increasing number ambient backscatter communication. Moreover, FDA based
of deployed antennas and higher order constellations [19]. multi-beam index modulation has been exploited by cooper-
However, these suboptimal detectors don’t promise a low error ative legitimate users in [40]. Similarly, FDA based physical
rate for critical applications, which opens a new research area layer security has also been incorporated over Nakagami-m
in pursuit of near-optimal/suboptimal low complexity detec- fading channels for an overall improved performance [41].
tion methods [20]. Moreover, sparse transmission of SM based Moreover, mmWave communications using an FDAs have
structures pave a way for matched-filtering and multi-stage also been investigated and can be found in the literature
(MS) detection sub-optimal methods such as [20]–[22]. [42]–[48]. In [42], a secure mmWave UAV-Enabled SWIPT
Initially, a matched filtering approach was proposed in [20] network communication has been investigated, while secure
with a reduced complexity and near optimal detection. After- mmWave wireless communication using fixed region FDA
wards an improved matched filtering detector with further beamforming has been presented in [43]. Moreover, combina-
reduced complexity was proposed in [21], which makes the tion of spatial modulation with the FDA has been investigated
multiple most probable antenna indices decision at the first in [44], [45] for overall improved performance for fifth gener-
stage to particularly reduce the receiver complexity, while ation and beyond, wireless communication designs. Similarly,
improving the overall design performance in next stages. How- energy efficient design of mmWave communication networks
ever, it considers only a single receiver antenna. Furthermore, have been presented in [46], [47], where retro-directive FDA
another distance-based ordered detection (DBD) algorithm has been used to achieve the objective. Finally, FDA based
was proposed in [22], which introduces a parameter to make dynamic rotated angular beamforming for secure mmWave
a trade-off between computational complexity and detection wireless communication has been used in [48]. However,
performance. numerous potential benefits of FDA range-angle dependent
Although QSM warrants several promising potentials for functionality and beamforming are yet to be explored for an
future wireless communication technologies, yet most of the overall improved performance of these designs.
existing methods use phased array (PA) antennas at the Therefore, visualizing QSM as a powerful candidate for
transmitter [5], [23]–[28] that limit the QSM functionality mmWave communication applications and considering its
in angle dimensions, only, without taking into account any (standard QSM) limitations in angle dimension for energy
range dimension information. The main limitations include focusing and mitigating range-dependent interferences plus
high cost of electronic phase shifters, energy focusing on fixed noise have opened new possibilities to explore efficient meth-
angle directions for all the range values and no feasibility of ods for obtaining improved performance. Moreover, high com-
range-dependent interference suppression. Note that, the PA putational complexity of ML based receiver for QSM based
based design’s overall performance deteriorates drastically, designs warrants an improved, efficient and fast method with
if it has to discriminate between two or more receivers reduced computational complexity to achieve better data rate
that are highly correlated (e.g., legitimate and illegitimate and improved bit error rate (BER). These open problems have
communication receivers that reside on same direction but motivated us to hybridize QSM with an FDA for improved
different ranges). Therefore, a range-angle dependent QSM transmission, ICI avoidance and detection performances,
transmission may significantly increase the signal to noise while achieving a less computationally complex receiver
ratio (SNR) and signal to interference and noise ratio (SINR) as compared to the existing techniques with an improved
due to an improved range-angle dependent energy focusing performance.
and range-dependent interference cancellation, respectively. In this article, we utilize a frequency diverse array
In the recent years, a flexible array design with a range-angle (FDA) [29] in QSM for mmWave communications. Since two
dependent beamforming feature and more degrees of free- antennas are activated in a unique way at a single time instant
dom (DoF) known as frequency diverse array (FDA) [29] to transmit real and imaginary parts of the symbol, we propose
has been thoroughly investigated for radar and communication an FDA with linearly increasing frequency offset deployed at
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4573
the transmitter front end that will facilitate to transmit each given as fn (t) = f0 + Δfn (t). Note that, a time–
part of the symbol with a slightly different frequency. These range-angle dependent beampattern is generated due to a
frequency offsets not only originate a range-angle dependency linearly increasing small frequency offset across the array
but also a time variance is induced in FDA based transmission, i.e., Δfn (t) = (n − 1) Δf (t). The FDA transmitted signal,
which provides a potential multi-path utilization for mmWave that is received at a far-field point target, is given as [49]
wireless communications. More importantly, transmission on
different frequencies at one time opens the option of matched s (t, r, θ)
filtering at the receiver end to demodulate the IM and symbols N t −1 rn
bits in the proposed design. The main contributions can be = sn t − (1a)
n=0
c
summarized as follows:
t −1
rn
N
1) A time-range-angle dependent QSM transmission r − ndsin (θ)
= wn exp j2π fn t − t−
based on an FDA for mmWave wireless communication c c
n=0
MISO/MIMO channel is proposed for an improved signal to (1b)
noise ratio (SNR) performance. The linearly increasing fre-
quency offset scheme is utilized to transmit real and imaginary where sn (t) denotes the signal transmitted by the
parts of the communication symbol with slightly different nth element, rn = r−nd sin θ represents the distance of target
frequencies at one time for better ICI avoidance as well. from nth element and wn denotes the corresponding weight,
This approach also generates more flexibility for multi-path whereas c denotes the speed of light and θ represents the
utilization for mmWave wireless communications. angle.
2) We propose a suboptimal multi stage (MS) receiver The array factor of an FDA i.e., AFFDA that considers the
design with reduced computational complexity, which uses propagational delay is expressed as [49]
a bank of band pass filters matched to the corresponding
transmit frequency offsets for estimating the most probable AFF DA (t, r, θ)
t −1
rn
N
antenna indices to decoded index bits at first step, whereas, f0 ndsin (θ)
merely, two most probable estimated indices are utilized = exp j2π + Δfn t −
c c
to decode associated data bits using a standard maximum n=0
r − ndsin (θ)
likelihood (ML) method in the second step. × t− (2)
Adhering an improved SNR due to an FDA range-angle c
dependent transmission in a mmWave channel, simulation Note that, this time–range–angle dependent array factor
and numerical results show the improved performance of the facilitates to generate a beam pattern with multiple maxima
proposed design over existing SM and QSM based schemes, at different range and angle values, which is different from
while MS approach effectively reduces the receiver’s compu- PA where the maximum is constant for one angle and all
tational complexity. Moreover, time variance problem of FDA the ranges as shown in Fig. 1. Additionally, improved SNR
is avoided due to the fact that the in-phase and quadrature and signal to interference noise ratio (SINR) can be obtained
parts in the proposed design are transmitted only in one using an FDA transmission [31], [32], specifically when inter-
time slot. ferers have a same direction as of target but different ranges.
The rest of the article is organized as follows. Section II In contrast to PA, the FDA will illuminate, merely, the target
provides the details of standard FDA and QSM technologies. range-angle position at a specific time instant, avoiding all
Section III presents the proposed hybrid QSM–FDA design positions of the potential interferers.
and section IV is reserved for performance analysis. Finally,
simulation results are presented in section V and article is
concluded in section VI. B. Conventional QSM
In a standard
QSM transmission, a block of η-bits
II. F REQUENCY D IVERSE A RRAY (η = log2 M Nt2 ), where M denotes size of constellation
AND C ONVENTIONAL QSM and Nt represents the number of transmit antennas, is divided
into three blocks. The first block contains of log2 (M ) bits
A brief introduction of the standard FDA and the conven-
for symbol modulation, while the other two blocks contain
tional QSM based transmission has been presented for readers’
of log2 (Nt ) bits for activating two antennas for transmitting
ease in this section.
in-phase and quadrature parts of symbol [16].
For example, consider quadrature amplitude modulation
A. Frequency Diverse Array (4-QAM) MIMO scheme with 4 transmit antennas i.e.,
We consider a uniform linear array (ULA) with Nt antenna M = 4, Nt = 4 is used for QSM transmission. Using the
elements having uniform inter-element spacing d. Different above formula, total η = 6 data bits can be transmitted at a
2
from a phased array (PA), the frequency transmitted from particular time instant as η = log ⎡2 M N t = 6. We repre-⎤
nth element of an FDA is slightly different from its adjacent
element due to an increasing frequency offset across the array. sent these incoming bits as k = ⎣
11 01
1 1 ⎦,
Denoting Δf as frequency offset and f0 as a fundamental log2 (M) log2 (Nt ) log2 (Nt )
frequency, the frequency output fn of the nth element is where the first log2 (M ) bits i.e.,[1 1] modulate the 4-QAM
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4574 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
Fig. 1. Comparison of generated beam patterns towards target (0o and 200 m) by (a) an FDA (b) a PA.
constellation symbol as x = +1 − j that is decomposed into However, the complexity of ML-based receiver is very
its real and imaginary parts given as x = +1 and x = −j, high as it jointly searches all deployed transmit antennas and
respectively. The next log2 (Nt ) bits i.e.,[0 1] modulate the symbols for detecting active antenna indices and decoding data
transmit antenna = 2 to transmit real part of the symbol symbols.
x = +1. Finally, the last log2 (Nt ) bits i.e., [1 1] modulate
the transmit antenna = 4 to transmit imaginary part of III. P ROPOSED D ESIGN
the symbol x = −j. After coherently adding the vectors In this section, we present the proposed FDA based QSM
achieved from the mapping process, the resultant vector to be designs for mmWave communications. We consider both the
transmitted is given as MISO and the MIMO structures, where transmitters of both
T structures employ an array of antennas (i.e., FDA), whereas
x = [0 + 1 0 − j] (3)
receiver contains a single omnidirectional antenna (see Fig. 2)
Note that, if the sequence of incoming bits are all 1’s and multiple antennas (see Fig. 3) for MISO and MIMO,
i.e., k = 1 1 1 1 1 1 , which yields = = 4 with respectively. Note that, each transmit array element is trans-
resultant vector to be transmitted as x = [0 0 0 + 1 − j]T , mitting the signal with a slightly different frequency due to a
the ICI is still entirely avoided at the receiver input due to linearly increasing frequency offset Δf utilized in FDA.
quadrature and in-phase components of the carrier signal [5].
Eventually, the vector x is transmitted over a Nr × Nt A. Proposed Hybrid QSM-FDA Modulator
MIMO channel H, where Nr denotes the number of antennas
at the receiver. The signal experiences an additive Gaussian We consider a single bit sequence as the input of the
noise (AWGN) having zero mean and συ2 = N0 variance for proposed system, which is divided into three groups to gen-
both real and imaginary parts transmission (i.e., double side erate QSM based communication setup [16]. We obtain a
power spectral density equal to N0 /2. The received signal at transmit information vector x similar to (3) indicating the
output of the channel is given by active number of antenna elements from the array. Different
from the existing techniques, both real and imaginary parts
y = Pt (h x + jh x )+υ; , = 1, 2, . . . , Nt (4) are transmitted with slightly different frequencies due to an
T FDA based transmitter. The selected frequency offset vector
where h = h1, . . . hNr , represent the th column
Δfs for two activated antennas indices out of Nt elements is
of channel matrix H, whereas the channel entries can be
based on vector x and is given as
considered independent and identically distributed (i.i.d) com-
plex Gaussian random variables with zero mean and variance Δfs = [|x (1)| |x (2)| . . . |x (Nt )|][Δf 2Δf . . . Nt Δf ]
σh2 for line-of-sight (LOS) communications, while Rayleigh (6)
fading channel √ entries for Non-LOS (NLOS) communications.
Moreover, Pt and υ are transmitted energy and additive where |x (n)| represent the absolute value of n entry of th
noise vector with double sided variance N0 /2, respectively. vector x and denotes Hadamard product. For example,
With a perfect channel knowledge, the optimum ML decoder the activated antennas indices = 2 and = 4 in (3)
at the receiver of a QSM based MIMO system is given by [16] correspond to transmission of real and imaginary parts with
2Δf and 4Δf frequency offsets, respectively.
ˆ , ˆ , x̂ , x̂ Note that, slightly different frequency offsets transmitted
2 from in-phase and quadrature components do not destroy the
= arg min y − Pt (h x + jh x ) (5) orthogonality between them. Since FDA transmits identical
, ,x ,x
baseband waveform s(t) from each array element, the ij th
where . denotes the norm operator. Estimated antennas element of the correlation matrix denoted by βij is calculated
indices ˆ , ˆ and data symbols x̂ , x̂ are used to retrieve p
T
the transmitted sequence. as βij = wi s (t)wj∗ s∗ (t) ej2π(Δfi −Δfj )t dt, where w shows
0
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4575
the corresponding array element’s weight and Tp denotes the receiver (θd , rd ), a generalized channel impulse response
pulse duration. Therefore, the frequency offsets fulfilling the hn,m (t) for nth transmit and mth receive antennas can be
k
condition that Δfi − Δfj = Tijp ; kij ∈ Z will guarantee calculated similar to [50], [51] as
the signal orthogonality [31], [38].
We consider both cases, where this vector x is transmitted
L
towards a communication receiver located at (θd , rd ) over hn,m (t) = hln,m al exp {jϕl } δ
an Nt × 1 mmWave channel h and an Nr × Nt mmWave l=1
rdl
channel H for MISO and MIMO structures employing an × t − n δ θ − θdln δ φ − φdlm (8)
FDA, respectively. The received signals at the output of the c
MISO and MIMO channels are given by (7a) and (7b), shown where L denotes the total number of multi-path components,
at the bottom of the page, respectively. hln,m represents the lth subpath channel attenuation between
In (7a), xn represents the nth entry of resultant trans- rd l
nth transmit and mth receive antennas, whereas al , φl and cn
mit vector x that is modulated using a QSM scheme over
denote amplitude, phase and propagation delay of lth subpath.
a mmWave MISO channel h including h and h in
Moreover, θdln and φdlm denote the angle of departure (AOD)
accordance with the two active transmit antennas. With an
and angle of arrival (AOA) for nth transmit and mth receive
additive noise during propagation,
√ the output of the channel
antenna, respectively.
can be represented as x̃ = Pt (h x + jh x ) + υ, ,
For a FDA transmitter with a uniform inter-element dis-
= 1, 2, . . . , Nt . Moreover, in (7b), xn,m denotes the
tance d and a single receive antenna (i.e., MISO structure),
nth entry of vector x modulated over mth sub-channel of
the transfer function hn (f ) of this impulse response (8) can
an Nr × Nt MIMO channel H including h and h in
be written as
accordance with the two active transmit antennas. The output
of the channel
√ for a communication symbol can be represented
as x̃ =√ Pt (h x + jh x ) + υ, , = 1, 2, . . . , Nt , hn (f )
where Pt and υ are transmitted energy and additive noise = hn,1 (f )
with double sided variance N0 /2, respectively. Note that, for a L
MIMO channel, υ is a Nr ×1 vector. Moreover, Δfsn denotes = hln,1 al̃ exp jϕl̃ exp −j2πfn rdln
the nth value of vector Δfs presented in (6). l=1
f0
× exp −j2π nd sin θdln exp −j2π sin φdl1
c
B. 3D mmWave Channel Model for the Proposed Designs (9)
We consider a 3D mmWave channel for MISO and MIMO
designs, where antennas are operating at mmWave frequencies. Similarly, the transfer function hn,m (f ) for a receiver
Considering a known range-angle position of communication containing an FDA (i.e., MIMO structure) with uniform
Nt rd j2π
f0 (n−1)d(sin(θ)−sin(θd ))
+Δfsn (t− rcn ) t−
((r−rd )−(n−1)dsin(θ))
y(t,r,θ;rd,θd ) = t− e ∗ hn (t) + υn
c c
wn xn (7a)
c
n=1
Nt
Nr rd j2π
f0 (n−1)d(sin(θ)−sin(θd ))
+Δfsn (t− rcn ) t−
((r−rd )−(n−1)dsin(θ))
y(t,r,θ;rd,θd ) = t− e ∗ hn,m (t) + υn,m
c c
wn xn,m
n=1 m=1
c
(7b)
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4576 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
n d
× exp j2π Δfsn
l=1 c c
f0 (13)
× exp −j2π nd sin θdln
c we can see that (13) contains frequency offset information
Nt r q =p
d
≈ x̃ 3: Decode the information bits associated to these indices
c
n=1
i.e., { }log2 (Nt )−bits and { }log2 (Nt )−bits , respectively.
f0 (n − 1) d (sin (θ) − sin (θd )) 4: Assuming the correct indices estimation, the associated
×exp j2π
c data bits on these indices are estimated as i.e.,
r ((r−r )−(n − 1) dsin (θ))
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4577
using (12) and (13) by matching the specific output with each ≤
of transmitted frequency offset. Similarly, vector ỹ is obtained. n=1 k=1 l=1 n̂=1 k̂=1
Finally, the proposed Algorithm 1 is utilized to estimate the N (n, n̂) p (x,nl → x,n̂l) K k, k̂ p x,kl → x,k̂l
IM and data bits, respectively. ×
Nt2 M
(19)
IV. P ERFORMANCE A NALYSIS
In this section, we present the performance analysis of our where
x = (x + jx ) and p (x,nl → x,n̂l ) and
proposed design. p x,kl → x,k̂l are the pairwise error probabilities (PEP)
for selecting transmit vectors x,n̂l and x,k̂l when x,nl and
A. Probability of Error Analysis x,kl were transmitted, respectively. Moreover n, k and n̂, k̂
The proposed receiver estimates two quantities, 1) the active denote the true antenna and estimated indices, respectively,
antenna indices and 2) the transmitted symbols on these whereas N (n, n̂) and K k, k̂ represent number of error bits
indices. Importantly, the design performance depends upon between true and estimated indices.
r −1
rm
N
f0 md (sin (θ) − sin (θd )) (r−rd ) − mdsin (θ)
y(t,r,θ;rd,θd ) = ejϕ0 x̃g (t, rd , θd ) exp j2π +Δfm t − t−
m=0
c c c
(17)
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4578 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
Importantly, the PEP for QSM design with single receive searches data and spatial symbols assuming the perfect
antenna is given as given by [16] knowledge
2 of channel.
The complexity of QSM is given as
% & ' ∼ O 8N , whereas
1 t M N r the complexity of [21] is computed
p (xnl → xn̂l ) = 1−
ξ/2
(20) as ∼ O 8Nt2 + 8N 2 M for Nr = 1, where N ≤ Nt .
2 1 + ξ/2 However, in the proposed design, only two most prob-
able estimate of transmit antenna indices are obtained,
where ξ is exponential random variable. Since we assume
in the first stage, using the BPFs matching to the
the correct estimation of transmit symbols{x̂ , x̂ }, the ξ is
transmitted FDA frequency offset for each data symbol.
defined as
⎧ In the second stage, the two most probable estimates
⎪ Pt 2 2 2
⎪
⎪ σh |x̂ | + |x̂ | if h = hˆ , h = hˆ are forwarded for ML-based detection for data symbols.
⎪
⎨ 2N
⎪ 0
Pt 2 Hence the
complexity for the proposed
detector
is given
2
ξ= σ |x̂ | if h = hˆ , h = hˆ (21) as ∼O 8Nt2 Nr + 4Nt Nr + 8 22 M Nr , which can be
⎪
⎪ 2N0 h
⎪
⎪ Pt 2 reduced when Nr = 1. Note that, the first term in complexity
⎪
⎩ σ |x̂ |
2
if h = hˆ , h = hˆ
2N0 h expression is for bandpass filtering and active indices detec-
tion, whereas the second term is for ML detection, where only
For a receiver equipped with Nr receive antennas, 2 active indices are selected.
the instantaneous PEP is given by Since the first stage output of the BPFs significantly
⎛*
+ Nr ⎞
+ reduces the input set for the ML detection, it reduces
p (xnl → xn̂l ) = Q ⎝, ξm ⎠ (22) the second stage’s ML receiver complexity, significantly. Note
m=1 that, the overall complexity is less than the standard jointly
estimating ML receiver for a greater number of transmit
whereas the average PEP is given as antenna with M-QAM transmission as shown in the simu-
r −1
N lation. Providentially, a trade-off between performance and
Nr − 1 + m m
p (xnl → xn̂l ) = χNr (1 − χ) (23) computational complexity has to be made based on the
m
m=0 application.
where χ denotes the PEP with single receive antenna.
Importantly, assuming an accurate estimation of antenna
indices h , the proposed QSM-FDA receiver applies ML D. Capacity Analysis
criterion for symbol estimation as follows Since a QSM based technique transmit data bits through
√ spatial and signal constellation diagram, it adheres a specific
2
{h , xû } = argmax SN R |h xu | − 2Re {ỹ h xu } (24) nature of mmWave MIMO/MISO communication channel.
u
Therefore, capacity analysis considering both signal constel-
where xû denotes the ML based estimated symbol, while xu
lation and data over spatial symbols is presented for our
is uth symbol from M-QAM constellation i.e.,u ∈ [1 : M ].
proposed FDA-QSM design. At first, a MISO channel is
The average symbol error rate (SER) for the utilized M-QAM
considered, which can be extended to a MIMO channel easily.
constellation [54] can be used to get average symbol bit error
Moreover, the simulation results for both the SIMO and MIMO
probability pβ given as
designs used for the proposed method have been added in the
SERM−QAM simulations and results section.
pβ = (25)
log2 (M ) As QSM activates two transmit antennas to modulate bits in
Since a conventional ML detector jointly estimates the channels {h , h } along with the symbol bits s , the general
antenna indices and transmit symbol, which is computationally Ergodic capacity of the proposed MISO QSM-FDA design is
complex. Therefore, a multistage estimator is proposed using given as
the properties of an FDA. The complexity of the proposed
design is significantly reduced as the two most probable C = Ehd max I (x; ỹ|hd ) (26)
px
antenna indices are selected, at the first stage, using outputs
of the BPFs matched to a distinct frequency offset (see Fig 2). where E {.} is expectation operator and px denotes the
Hence a reduced set of input is then sent to the optimal ML probability density function (PDF) of transmitted vector x,
QSM detector to estimate the symbols on these antenna indices while mutual information is between x and received vector
channels in the second stage. Finally, both antenna indices and ỹ is maximized over px assuming a perfect knowledge of
data symbols {h , xû } are estimated. channel h. Since proposed QSM-FDA design use different
channels to convey information, no averaging over channel
C. Receiver Computational Complexity vector h is required. Therefore, maximization of mutual infor-
It has been proved [16] that receiver complexity of QSM mation between transmitted spatial and constellation symbol
is almost the same as SM, while the spectral efficiency with the received vector ỹ is the capacity of proposed design is
has increased by utilizing 2 antennas at single time instant. given as
Two extra complex multiplications for each receive antenna
C = max I (h , s ; ỹ) (27)
is needed to estimate data symbols, whereas ML jointly ph , ps ,
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4579
Analogous to [5], mutual information i.e, I (h , s ; ỹ) is Moreover, σx2 is the signal power, wr is the received array
derived as weights vector i.e.,wR = [w1 , w2 , . . . , wNr ] and u (t, θd , rd )
denotes the receive steering vector, whereas Ri+υ denotes the
I = H (ỹ) − H (ỹ|h , s ) (28a)
$ % $ /'/ interference plus noise covariance matrix.
ỹ−h s 2
− Since a mmWave based communication signal
= −log2 (e) − Eỹ log2 Eh ,s e 2
συ
is normally considered in a noise-dominant rather
(28b) than an interference-dominant environment, therefore,
Δ
SN RF DAmmW ave = SIN RF DAmmW ave and the SINR in
where H (.) denotes the entropy function. As the entropy (34) can be approximated as
doesn’t depend on h and s [5], the maximization of H (ỹ) is
enough. Since a zero mean Gaussian distribution CN (0, σ 2 ) is 0 H 02
best candidate to maximize entropy, where σ denotes variance, σx2 0wR u (t, θd , rd )0
SN RF DAmmW ave
(35a)
the receive vector must have CN (0, σỹ2 ), where σỹ2 denotes the συ2 wRHw
R
variance of ỹ. Hence entropy is given by σx2 |wR u (t, θd , rd )|
2
(35b)
H (ỹ) = −Eỹ {log2 (pỹ (ỹ))} (29a) συ2 wR Hw
R
2
= log2 πe 1 + συ2 (29b) |wR u (t, θd , rd )|
SN R0 (35c)
Note that, the capacity for QSM with one receive antenna wR Hw
R
is derived as [5] where συ2 denotes the noise power and SN R0 is a com-
C = log2 (1 + SN R) (30a) mon SNR. Additionally, it has been shown in [5] that a
3D mmWave-QSM channel can be approximated with a
1
= log2 1 + 2 (30b) log-normal channel. Provided a proper shaping of constel-
συ
lation symbols for 3D-mmWave-QSM channel, a theoretical
where SNR is signal to noise ratio, whereas the capacity for capacity can be reached with one input data stream. Note that,
channel dependent MISO SMT designs is given as [55]–[57] an FDA based transmission considers a range-angle position
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4580 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
Fig. 4. Comparison of average BER for Nt = 4 and 4 − QAM modulation (a) MISO structure Nr = 1 (b) MIMO structure Nr = 4.
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4581
Fig. 7. Comparison among the QSM capacity and the capacities derived in Fig. 9. Comparison of theoretical capacity and simulated mutual information
literature for Rayleigh and mmWave channels with Nt = 4 and Nr = 1. of the proposed FDA-QSM, QSM and SM over the mmWave channel model
with Nt = 4 and Nr = 1.
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4582 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
m=0
(r−rd )−mdsin(θ)
j2π Δfm (t− rm
c ) t−
×e
c
(37)
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
BASIT et al.: FDA BASED QSM FOR mmWAVE WIRELESS COMMUNICATIONS 4583
[10] P. Liu, Y. Li, W. Cheng, W. Zhang, and X. Gao, “Energy- [35] J. Xiong, S. Y. Nusenu, and W.-Q. Wang, “Directional modulation
efficient power allocation for millimeter wave beamspace MIMO- using frequency diverse array for secure communications,” Wireless Pers.
NOMA systems,” IEEE Access, vol. 7, pp. 114582–114592, 2019, doi: Commun., vol. 95, no. 3, pp. 1–11, Mar. 2017.
10.1109/ACCESS.2019.2935495. [36] S. Y. Nusenu, W.-Q. Wang, and S. Ji, “Secure directional modulation
[11] A. Agrawal, V. K. Mishra, A. Gupta, and B. Bansal, “Performance using frequency diverse array antenna,” in Proc. IEEE Radar Conf.
analysis of beamforming aided uncorrelated mm-Wave MIMO system (RadarConf), Seattle, WA, USA, May 2017, pp. 378–382.
for IEEE 802.15.3c standard,” Phys. Commun., vol. 41, pp. 101–114, [37] W.-Q. Wang, “DM using frequency diverse array antenna for secure
Aug. 2020. transmission,” IET Microw., Antennas Propag., vol. 11, no. 3,
[12] M. Di Renzo, H. Haas, A. Ghrayeb, S. Sugiura, and L. Hanzo, “Spa- pp. 336–345, Apr. 2017.
tial modulation for generalized MIMO: Challenges, opportunities, and [38] S. Y. Nusenu, S. Huaizong, Y. Pan, and A. Basit, “Directional modula-
implementation,” Proc. IEEE, vol. 102, no. 1, pp. 56–103, Jan. 2014. tion with precise legitimate location using time-modulation retrodirective
[13] E. Basar, “Index modulation techniques for 5G wireless networks,” IEEE frequency diversity array for secure IoT communications,” IEEE Syst.
Commun. Mag., vol. 54, no. 7, pp. 168–175, Jul. 2016. J., pp. 1–11, 2020, doi: 10.1109/JSYST.2020.3010787.
[39] Y.-Q. Hu, H. Chen, S.-L. Ji, and W.-Q. Wang, “Ambient backscatter
[14] A. Younis, N. Serafimovski, R. Mesleh, and H. Haas, “Generalised
communication with frequency diverse array for enhanced channel
spatial modulation,” in Proc. Conf. Rec. 44th Asilomar Conf. Signals,
capacity and detection performance,” IEEE Sensors J., vol. 20, no. 18,
Syst. Comput., Pacific Grove, CA, USA, Nov. 2010, pp. 1498–1502.
pp. 10876–10885, Sep. 2020, doi: 10.1109/JSEN.2020.2993843.
[15] Y. Bian, X. Cheng, M. Wen, L. Yang, H. V. Poor, and B. Jiao, [40] B. Qiu, L. Wang, J. Xie, Z. Zhang, Y. Wang, and M. Tao, “Multi-beam
“Differential spatial modulation,” IEEE Trans. Veh. Technol., vol. 64, index modulation with cooperative legitimate users schemes based on
no. 7, pp. 3262–3268, Jul. 2015. frequency diverse array,” IEEE Trans. Veh. Technol., vol. 69, no. 10,
[16] R. Mesleh, S. S. Ikki, and H. M. Aggoune, “Quadrature spatial mod- pp. 11028–11041, Oct. 2020, doi: 10.1109/TVT.2020.3007003.
ulation,” IEEE Trans. Veh. Technol., vol. 64, no. 6, pp. 2738–2742, [41] S. Ji and W.-Q. Wang, “Physical-layer security for frequency diverse
Jun. 2015. array communication system over Nakagami-m fading channels,”
[17] R. Mesleh, H. Haas, C. W. Ahn, and S. Yun, “Spatial modulation—A IEEE Syst. J., vol. 14, no. 2, pp. 2370–2381, Jun. 2020, doi:
new low complexity spectral efficiency enhancing technique,” in Proc. 10.1109/JSYST.2019.2923103.
1st Int. Conf. Commun. Netw. China, Beijing, China, Oct. 2006, pp. 1–5. [42] X. Sun, W. Yang, Y. Cai, and M. Wang, “Secure mmWave UAV-
[18] O. S. Badarneh and R. Mesleh, “A comprehensive framework for enabled SWIPT networks based on random frequency diverse arrays,”
quadrature spatial modulation in generalized fading scenarios,” IEEE IEEE Internet Things J., vol. 8, no. 1, pp. 528–540, Jan. 2021, doi:
Trans. Commun., vol. 64, no. 7, pp. 2961–2970, Jul. 2016. 10.1109/JIOT.2020.3005984.
[19] N. R. Naidoo, H. J. Xu, and T. Al-Mumit Quazi, “Spatial modulation: [43] Y. Hong, X. Jing, H. Gao, and Y. He, “Fixed region beamforming using
Optimal detector asymptotic performance and multiple-stage detection,” frequency diverse subarray for secure mmWave wireless communica-
IET Commun., vol. 5, no. 10, pp. 1368–1376, Jul. 2011. tions,” IEEE Trans. Inf. Forensics Security, vol. 15, pp. 2706–2721,
[20] S. Sugiura, C. Xu, S. X. Ng, and L. Hanzo, “Reduced-complexity 2020, doi: 10.1109/TIFS.2020.2969576.
coherent versus non-coherent QAM-aided space-time shift keying,” [44] S. Y. Nusenu and W.-Q. Wang, “Range-dependent spatial modulation
IEEE Trans. Commun., vol. 59, no. 11, pp. 3090–3101, Nov. 2011. using frequency diverse array for OFDM wireless communications,”
[21] P. Yang, Y. Xiao, L. Li, Q. Tang, and S. Li, “An improved matched- IEEE Trans. Veh. Technol., vol. 67, no. 11, pp. 10886–10895, Nov. 2018,
filter based detection algorithm for space-time shift keying systems,” doi: 10.1109/TVT.2018.2870045.
IEEE Signal Process. Lett., vol. 19, no. 5, pp. 271–274, May 2012. [45] S. Y. Nusenu, S. Huaizong, Y. Pan, and A. Basit, “Space-frequency
[22] Q. Tang, Y. Xiao, P. Yang, Q. Yu, and S. Li, “A new low-complexity increment index modulation approach for fifth generation and beyond
near-ML detection algorithm for spatial modulation,” IEEE Wireless wireless communication systems,” IEEE Trans. Veh. Technol., vol. 69,
Commun. Lett., vol. 2, no. 1, pp. 90–93, Feb. 2013. no. 6, pp. 6286–6298, Jun. 2020, doi: 10.1109/TVT.2020.2984912.
[46] S. Y. Nusenu and E. Asare, “Butler matrix frequency diverse retrodi-
[23] R. Mesleh and A. Younis, “LOS millimeter-wave communication with
rective array beamforming: An energy-efficient technique for mmWave
quadrature spatial modulation,” in Proc. IEEE Int. Symp. Signal Process.
networks,” Wireless Commun. Mobile Comput., vol. 2020, Feb. 2020,
Inf. Technol. (ISSPIT), Limassol, Cyprus, Dec. 2016, pp. 109–113. Art. no. 9408519.
[24] Y. Wang, T. Zhang, W. Yang, J. Guo, Y. Liu, and X. Shang, “Secure [47] S. Y. Nusenu, “Development of frequency modulated array antennas for
transmission for differential quadrature spatial modulation with artificial millimeter-wave communications,” Wireless Commun. Mobile Comput.,
noise,” IEEE Access, vol. 7, pp. 7641–7650, 2019. vol. 2019, Apr. 2019, Art. no. 6940708.
[25] F. Chen, K. Yang, P. Xing, H. Zhang, and Y. Jiang, “Multiuser pre-coding [48] Y. Hong, X. Jing, Y. He, and J. Mu, “Dynamic rotated angular
aided quadrature spatial modulation for large-scale MIMO channels,” beamforming using frequency diverse phased-array for secure mmWave
China Commun., vol. 15, no. 11, pp. 62–69, Nov. 2018. wireless communications,” Electronics, vol. 9, no. 1, p. 10, Dec. 2019.
[26] R. Neelakandan, “Sub-optimal low-complexity detector for gener- [49] K. Chen, S. Yang, Y. Chen, and S.-W. Qu, “Accurate models of
alised quadrature spatial modulation,” Electron. Lett., vol. 54, no. 15, time-invariant beampatterns for frequency diverse arrays,” IEEE Trans.
pp. 941–943, Jul. 2018. Antennas Propag., vol. 67, no. 5, pp. 3022–3029, May 2019.
[27] L. Gou, J. Ge, and Y. Cao, “Low-complexity sphere decoding for [50] M. Steinbauer, A. F. Molisch, and E. Bonek, “The double-directional
quadrature spatial modulation,” in Proc. IEEE 87th Veh. Technol. Conf. radio channel,” IEEE Antennas Propag. Mag., vol. 43, no. 4, pp. 51–63,
(VTC Spring), Porto, Portugal, Jun. 2018, pp. 1–5. Aug. 2001.
[28] L. Xiao, P. Xiao, Y. Xiao, I. Hemadeh, A. Mohamed, and L. Hanzo, [51] A. Forenza, D. J. Love, and R. W. Heath, Jr., “Simplified spa-
“Bayesian compressive sensing assisted space–time block coded quadra- tial correlation models for clustered MIMO channels with different
ture spatial modulation,” IEEE Trans. Veh. Technol., vol. 67, no. 10, array configurations,” IEEE Trans. Veh. Technol., vol. 56, no. 4,
pp. 10044–10048, Oct. 2018. pp. 1924–1934, Jul. 2007.
[29] P. Antonik, M. C. Wicks, H. D. Griffiths, and C. J. Baker, “Fre- [52] M. K. Samimi and T. S. Rappaport, “3-D millimeter-wave statistical
quency diverse array radars,” in Proc. IEEE Conf. Radar, Apr. 2006, channel model for 5G wireless system design,” IEEE Trans. Microw.
pp. 215–217. Theory Techn., vol. 64, no. 7, pp. 2207–2225, Jul. 2016.
[30] P. Antonik, “An investigation of a frequency diverse array,” [53] J. Jeganathan, A. Ghrayeb, L. Szczecinski, and A. Ceron, “Space
Ph.D. dissertation, Univ. College London, London, U.K., 2009. shift keying modulation for MIMO channels,” IEEE Trans. Wireless
Commun., vol. 8, no. 7, pp. 3692–3703, Jul. 2009.
[31] A. Basit, W. Khan, S. Khan, and I. M. Qureshi, “Development of [54] J. G. Proakis, Digital communications, 4th ed. New York, NY, USA:
frequency diverse array radar technology: A review,” IET Radar, Sonar McGraw-Hill, 2001.
Navigat., vol. 12, no. 2, pp. 165–175, Feb. 2018. [55] H. Yonghong, W. Pichao, W. Xiang, Z. Xiaoming, and H. Chunping,
[32] W.-Q. Wang, “Frequency diverse array antenna: New opportunities,” “Ergodic capacity analysis of spatially modulated systems,” China
IEEE Antennas Propag. Mag., vol. 57, no. 2, pp. 145–152, Apr. 2015. Commun., vol. 10, no. 7, pp. 118–125, Jul. 2013.
[33] W.-Q. Wang, “Retrodirective frequency diverse array focusing for wire- [56] Y. Yang and B. Jiao, “Information-guided channel-hopping for high
less information and power transfer,” IEEE J. Sel. Areas Commun., data rate wireless communication,” IEEE Commun. Lett., vol. 12, no. 4,
vol. 37, no. 1, pp. 61–73, Jan. 2019. pp. 225–227, Apr. 2008.
[34] S. Y. Nusenu, W.-Q. Wang, and J. Xiong, “Time–modulated fre- [57] Y. Yang and B. Jiao, “On the capacity of information-guided channel-
quency diverse array for physical–layer security,” IET Microw., Antennas hopping in multi-antenna system,” in Proc. IEEE Conf. Comput. Com-
Propag., vol. 15, no. 3, pp. 336–345, Apr. 2017. mun. Workshops (INFOCOM), Apr. 2008, pp. 1–5.
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.
4584 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 20, NO. 7, JULY 2021
Abdul Basit received the M.S. and Ph.D. degrees Shaddrack Yaw Nusenu received the Ph.D. degree
from International Islamic University, Islamabad, in information and communication engineering from
Pakistan, in 2009 and 2016, respectively. He is the University of Electronic Science and Technology
currently a Post-Doctoral Research Fellow with of China (UESTC), Chengdu, China. He is currently
the School of Information and Communication with Koforidua Technical University, Koforidua,
Engineering, University of Electronic Science and Ghana, as a Lecturer with the Electrical/Electronic
Technology of China. He is also an Assistant Department. He is also a Post-Doctoral Fellow
Professor with International Islamic University, with the School of Information and Communication
Islamabad. His research interests include radar tech- Engineering, UESTC. His research interests include
nology, cognitive radar signal processing, frequency radar technology, cognitive radar signal processing,
diverse array signal processing, space modulation frequency diverse array signal processing, wireless
techniques, RADCOM techniques, and millimeter-wave communications. physical layer security, space modulation techniques, spectrum sharing
techniques, and millimeter-wave communications.
Wen-Qin Wang (Senior Member, IEEE) received
the B.S. degree in electrical engineering from Shan-
dong University, Shandong, China, in 2002, the
M.E. and Ph.D. degrees in information and com-
munication engineering from the University of Elec-
tronic Science and Technology of China (UESTC),
Chengdu, China, in 2005 and 2010, respectively.
From March 2005 to March 2007, he was with
the National Key Laboratory of Microwave Imaging
Technology, Chinese Academy of Sciences, Beijing,
China. Since September 2007, he has been with the
School of Communication and Information Engineering, UESTC, where he is Samad Wali received the B.Sc. degree in mathemat-
currently a Professor. From June 2011 to May 2012, he was a Visiting Scholar ics from Forman Christian College Lahore, Pakistan,
with the Stevens Institute of Technology, NJ, USA. From December 2012 to the M.Sc. degree in applied mathematics from The
December 2013, he was a Hong Kong Scholar with the City University of Islamia University of Bahawalpur, Pakistan, and the
Hong Kong, Hong Kong. From January 2014 to January 2016, he was a Marie Ph.D. degree in computational mathematics from
Curie Fellow with the Imperial College London, U.K. He has authored two Nankai University, Tianjin, China, in June 2018.
books Multi-Antenna Synthetic Aperture Radar (Springer) and Near-Space He is currently a Post-Doctoral Research Fellow
Remote Sensing (CRC Press). His research interests include array signal in image processing and medical image analysis
processing and circuit systems for radar, communications and microwave with the School of Information and Communication
remote sensing. Engineering, University of Electronic Science and
He is the editorial board members of four international journals. He was a Technology of China (UESTC), Chengdu. He is also
recipient of the Marie Curie International Incoming Fellowship, the National an Assistant Professor with the Namal Institute, Mainwali, Pakistan. His
Young Top-Notch Talent of the Ten-Thousand Talent Program Award, and the current research interests include image processing, variational methods, and
Hong Kong Scholar Fellowship. numerical solution to the partial differential equations.
Authorized licensed use limited to: Rajiv Gandhi Institute of Technology. Downloaded on December 07,2021 at 14:04:19 UTC from IEEE Xplore. Restrictions apply.