Professional Documents
Culture Documents
Abstract—Wireless communications and sensing at terahertz the current physical-layer efficiency. As a result, the future
(THz) band are increasingly investigated as promising short- sixth-generation (6G) networks are expected to exploit the
range technologies because of the availability of high operational THz frequencies (0.3-10 THz) [3, 4], where hundreds of
bandwidth at THz. In order to address the extremely high
attenuation at THz, ultra-massive multiple-input multiple-output GHz bandwidth is available to facilitate Tb/s rates without
(MIMO) antenna systems have been proposed for THz communi- dramatic efficiency increase in the physical-layer [5]. There
cations to compensate propagation losses. However, the cost and is, therefore, considerable interest in developing THz wireless
power associated with fully digital beamformers of these huge an- solutions [6, 7].
tenna arrays are prohibitive. In this paper, we develop wideband Higher bandwidths also result in improved radar range
hybrid beamformers based on both model-based and model-free
techniques for a new group-of-subarrays (GoSA) ultra-massive resolution. At present, mmWave radars with a few GHz
MIMO structure in low-THz band. Further, driven by the recent bandwidths such as those used in automotive applications
developments to save the spectrum, we propose beamformers [8] at 24 and 77 GHz are unable to yield high-resolution
for a joint ultra-massive MIMO radar-communications system, images compared to the optical sensors [9]. Higher operating
wherein the base station serves multi-antenna user equipment frequency have smaller antenna apertures and microwave com-
(RX), and tracks radar targets by generating multiple beams
toward both RX and the targets. We formulate the GoSA ponents, which is beneficial for radar deployment on cost and
beamformer design as an optimization problem to provide a area-sensitive platforms such as drones and ground vehicles.
trade-off between the unconstrained communications beamform- At THz, the physical aperture of the antenna is expected to
ers and the desired radar beamformers. To mitigate the beam be very small and the availability of large transmit bandwidth
split effect at THz band arising from frequency-independent has the potential to offer image resolutions closer to that from
analog beamformers, we propose a phase correction technique
to align the beams of multiple subcarriers toward a single the optical sensors [10].
physical direction. Additionally, our design also exploits second- The precise definition of THz band varies among different
order channel statistics so that an infrequent channel feedback community members. Recent works in wireless communica-
from the RX is achieved with less channel overhead. To further tions generally define this band in the range 0.03-10 THz
decrease the ultra-massive MIMO computational complexity and [11] with an obvious overlap with the conventional mmWave
enhance robustness, we also implement deep learning solutions
to the proposed model-based hybrid beamformers. Numerical frequencies. For the radar, microwave, and remote sensing
experiments demonstrate that both techniques outperform the engineers, THz band starts at the upper-mmWave limit of 100
conventional approaches in terms of spectral efficiency and radar GHz and, in particular, low-THz term is used for the range
beampatterns, as well as exhibiting less hardware cost and 0.1-1 THz [12]. In optics, on the other hand, THz spectrum
computation time. is defined to end at 10 THz, beyond which frequencies are
Index Terms—Deep learning, hybrid beamforming, joint radar- considered far-infrared [13]. The Terahertz Technology and
communications, Terahertz, ultramassive MIMO. Applications Committee of the IEEE Microwave Theory and
Techniques Society (MTT-S) focuses on 0.3-3 THz range
I. I NTRODUCTION while the IEEE Transactions on Terahertz Science and Tech-
In recent years, millimeter-wave (mmWave) spectrum has nology journal targets 0.3-10 THz. In this paper, in order
been extensively studied to address the demands for high data to be consistent with the developments in THz radar and
rates in the fifth-generation (5G) wireless communications communications, our work is relevant largely to the low-THz
[1, 2]. The maximum mmWave bandwidth being tens of frequencies used by the remote sensing community.
GHz, it is not possible to achieve data rates of the order The low-THz spectrum lies in the atmospheric window -
of terabits-per-second (Tb/s) without significantly enhancing a region of local minimum attenuation [14, 15]. However,
the associated path losses are so high that low-THz radar
This work was supported in part by the ERC Project AGNOSTIC. applications have conventionally been limited to indoor en-
A. M. E. is with the Department of Electrical and Electronics Engineering,
Duzce University, Duzce, Turkey, and SnT, University of Luxembourg, vironments such as vital sign monitoring [16], food con-
Luxembourg City, Luxembourg (e-mail: ahmetmelbir@gmail.com). tamination detection [17], and airport security [18]. Recent
K. V. M. is with the United States Army Research Laboratory, Adelphi, characterizations of low-THz radars carried out in the outdoor
MD 20783 USA (e-mail: kumarvijay-mishra@uiowa.edu).
S. C. is with the SnT at the University of Luxembourg, Luxembourg (e- environments for automotive applications [19, 20] suggest
mail: symeon.chatzinotas@uni.lu). feasibility of operation at a range of up to 200 m with the
2
Fig. 1. A radar-communications system for a vehicle-to-vehicle (V2V) and vehicle-to-device (V2D) scenario, wherein a single THz radar-communications
unit, with a NT = NTx × NTy antenna array, is mounted onto a vehicle to simultaneously transmit toward both communications receiver and vehicular
targets.
specific attenuation (derivative of attenuation with respect to While considering beamforming at THz, following unique
range) of ~10 dB km−1 over 0.1-0.3 THz resulting in a path characteristics (see also Table I) differentiate the THz-band
loss of ~4 dB [21]. from mmWave:
For THz communications, high propagation losses and 1) The path loss in THz channels includes both spread-
power limitations are compensated by the beamforming gains ing loss and molecular absorption. The latter is more
obtained through deployment of extremely dense nano-antenna significant at THz than mmWave [46]. The severe path
arrays [22], which may be based on graphene plasmonics [23, loss is compensated by deploying much larger antenna
24] or metasurfaces [25]. Analogous to the developments in arrays that require more creative choices for subarray
mmWave communications, [26] proposed a THz ultramassive geometries.
multiple-input multiple-output (MIMO) architecture that em- 2) Both line-of-sight (LoS) and non-LoS (NLoS) paths are
ploys an array-of-subarrays (AoSA) of nano-transceivers to significant at mmWave. However, at THz, the NLoS paths
increase the coverage and improve the data rates. Various have insignificant contribution to the received power. This
follow-up works (see, e.g., [22] for an overview) showed leads to LoS-dominant and NLoS-assisted communica-
further ultra-massive MIMO enhancements through waveform tions scenario at THz [11, 34, 36, 41, 48].
design, beamforming, and resource allocation. 3) Significant attenuation implies shorter ranges for THz
With this recent rise of both radar and communications systems than mmWave [49]. Additionally, the THz-
applications at THz, it has been suggested [27] to integrate specific molecular absorption effect leads to the channel
radar sensing and communications functionalities in future bandwidth varying with the range [50]. This requires
wireless THz systems to facilitate spectrum sharing, enhance stricter limitations on the deployment and coverage of
pencil beamforming, save hardware cost, and improve resource THz communications and radar.
usage. This follows recent efforts in realizing such joint radar- 4) THz channels are extremely sparse [51] in the angular
communications (JRC) architectures at mmWave [2], includ- domain and have smaller angular spread than mmWave.
ing for ultrashort ranges [28], joint MIMO-radar-MIMO- Therefore, it is feasible to adopt subarrayed models
communications [29], and distributed MIMO JRC [30]. In this such as AoSA [32] and GoSA to bring down the
paper, we focus on a ultra-massive MIMO structure for JRC high-frequency hardware and computational complexi-
at THz band. ties. These subarrayed structures have been shown to
Analogous to their massive MIMO counterparts at overcome the limited communications range while also
mmWave, the ultra-massive MIMO architecture implies that retaining a reasonable spatial multiplexing gain.
fully digital beamforming is infeasible because of huge as- 5) THz channel exhibits peculiarities such as misalign-
sociated cost, area, and power. This necessitates the use of ment and phase uncertainties in phase shifters [31]. The
hybrid beamforming [33], wherein the signal is processed by frequency-independent analog beamformers largely used
both analog and digital beamformer. Although some recent in the broadband mmWave communications may lead to
works [45–47] investigate THz hybrid beamformer designs, beam split effect in THz channels: the generated beams
they do not examine it in the context of the recently proposed, split into different physical directions at each subcarrier
practically feasible ultra-massive MIMO, and more so, its due to ultra-wide bandwidth and large number of anten-
application in THz JRC. nas [41]. This phenomenon has also been called as beam
3
TABLE I
C OMPARISON OF MM -WAVE AND TH Z T RANSMISSION C HARACTERISTICS
squint in mm-Wave works [37, 38]. While both beam formance compared to the fully-connected array. Hence, we
squint and beam split pertain to a similar phenomenon, suggest a partially-connected with overlapped (PCO) GoSA
the latter has more severe achievable rate degradation. In structure for performance improvement. To design the hybrid
particular, the main lobes of the array gain corresponding beamformers based on the PCO structure, we propose a
to the lowest and highest subcarrier frequencies do not modified version of the manifold optimization (MO)-based
overlap at THz at all while there is a relatively small alternating minimization (AltMin) technique [43], which is
deviation in mmWave band (see Fig. 11) [40, 42]. While originally suggested to solve the beamformer design problem
the beam-squint depends on the array size, the beam split in fully-connected arrays. Our numerical experiments show
is a function of both wide bandwidth and large arrays that the proposed approach has much lower hardware com-
[42]. plexity than the state-of-the-art techniques, while maintaining
6) The THz-band has several other propagation and scatter- satisfactory radar and communications performance. In this
ing effects [31]. The specular scattering is less dominant work, our main contributions are:
and partially-diffused scattering is a more appropriate 1) THz ultra-massive MIMO JRC. Our proposed JRC
model for THz. The THz channel coherence time may approach based on ultra-massive MIMO is inspired by recent
be smaller than the symbol time and the channel may no advancements in THz technologies and is, therefore, closer
longer be considered time-invariant. Further, the molecu- to practical feasibility. It is particularly helpful for short-range
lar absorption is range-dependent leading to very different vehicular applications, wherein the ego vehicle simultaneously
noise models than mmWave. At much shorter distances, communicates with the user equipment and detect/track the
the THz bandwidth is also range-dependent and spherical radar targets with higher angular resolution due to high beam-
wave propagation must be accounted for. forming gain of using ultra-massive number of antennas.
2) Model-based THz hybrid beamforming. Previous re-
In this work, we consider the THz hybrid beamforming search [45–47] examined THz hybrid beamforming without
problem by incorporating the above-mentioned unique THz ultra-massive MIMO. Our optimization-based hybrid beam-
features except the last scattering characteristic, which is forming for ultra-massive MIMO relies on both CSI and
beyond the scope of this paper. Contrary to prior works, channel covariance matrix. While the former provides higher
we focus on the THz wideband hybrid beamforming for spectral efficiency, the latter has lower channel overhead at the
a ultra-massive MIMO JRC configuration. To this end, we cost of slight performance loss.
develop both model-based and model-free techniques that 3) Novel GoSA structure. We propose GoSA structure to
rely on both channel state information (CSI) and channel lower the hardware cost which could be high for THz systems
covariance matrix. To reduce the hardware complexity, we due to use of a large number of antennas. GoSA allows us to
propose a group of subarrays (GoSAs) structure, in which the employ fewer number of phase shifters as compared to AoSA.
antenna elements in the same subarray are connected to the We analyze the performance of GoSA with respect to several
same phase-shifter. In other words, GoSA forms an array of design parameters, such as the number of antennas and the
subarray-of-subarrays, which is different than the prior array- antenna spacing. To provide a trade-off between the hardware
of-subarray (AoSA) structures (see, e.g., Fig. 2). Thus, the complexity and the spectral efficiency, PCO-based analog
proposed structure employs even fewer phase-shifters than precoder is proposed based on modified manifold optimization
that of fully-connected arrays or partially-connected AoSA method.
structures [32, 49], while providing satisfactory radar and 4) Beam split correction. We present a hardware-efficient ap-
communications performance in terms of the beampattern and proach to correct the beam split effect in THz channels arising
the spectral efficiency, respectively. In order to improve the from their ultra-wide bandwidth. While prior works [40, 41]
radar performance, the higher degrees of freedom (DoF) are consider an additional time-delay network for this operation,
provided by using partially-connected GoSAs. Nevertheless, the proposed approach effectively mitigates the beam split
partially-connected structure has poor spectral efficiency per-
4
effect without requiring such a complex structure. Q generates a single beam [26]. This is done by connecting
5) Deep learning (DL) solutions. We design two learning the Q antennas in each subarray to a single phase-shifter to
models using convolutional neural networks (CNNs), one of lower the hardware complexity. Thus, the TX first applies
which is employed to estimate the direction of the radar subcarrier-dependent NRF × NS baseband precoder FBB [m].
targets, whereas the other is used to design the hybrid beam- The signal is, then, transformed to the time-domain via M -
formers. While DL-based beamforming techniques [52, 53] point inverse fast Fourier transform (IFFT). After adding the
have been proposed earlier, THz JRC hybrid beamformer cyclic prefix, the TX employs a subcarrier-independent RF
design remains unexamined in prior literature. precoder FRF ∈ CNT ×NRF by employing NT phase-shifters,
Throughout this paper, we denote the vectors and matrices as shown in Fig. 2c. In conventional fully-connected structures
by boldface lower and upper case symbols, respectively. In (see Fig. 2a), each antenna is connected to NRF RF chains
case of a vector a, [a]i represents its i-th element. For a matrix while the AoSA model in Fig. 2b has a partially-connected
A, [A]i,j denotes the (i, j)-th entry. The IN is the identity structure and it connects each RF chain to N̄ Q antennas in
matrix of size N ×N ; E{·} denotes the statistical expectation; each subarray, where N̄ = NNRFT
[32]. In this work, we propose
rank(·) denotes the rank of its matrix argument; k · kF is a GoSA architecture, as shown in Fig. 2c, in which NT Q
the Frobenius norm; (·)† denotes the Moore-Penrose pseudo- antennas are partitioned into NRF groups, each of which has
inverse; and ∠{·} denotes the angle of a complex scalar/vector. N̄ Q antennas. Also, each group consists of N̄ subarrays of
The Kronecker and element-wise Hadamard product are de- size Q. The main difference between AoSA and GoSA is that
noted by ⊗ and , respectively. The notation expressing a each RF chain is connected to N̄ Q phase-shifters in the former
convolutional layer with N filters/channels of size D × D is while the each RF chain is connected to only N̄ phase-shifters
given by N @D × D. in the latter. Hence, the number of phase-shifters in GoSA is Q
The rest of the paper is organized as follows. In the next times lower than that of AoSA. In GoSA, we assume that the
section, we describe the system and channel models of GoSA- antennas in each subarray are fed with the same phase shift
based ultra-massive MIMO JRC and formulate the beam- to reduce the hardware complexity and power consumption,
former design problem. Section III introduces the CSI- and which is critical in THz systems.
channel covariance matrix-based beamformer solutions along In the proposed GoSA model, the RF precoder has
1
with extension to broadband beamforming. We follow this in unit-modulus constraints, i.e., |[FRF ]i,j | = √N as i ∈
T
Section IV by DL-based solution. We validate our models {1, . . . , NT } and j ∈ {1, . . . , NRF }, since FRF is con-
and methods through numerical experiments in Section V and structed by using
P phase-shifters. Furthermore, we have power
conclude in Section VI. constrained m∈M kFRF FBB [m]kF = M NS . Thus, the
NT × 1 transmitted signal from the TX is given by x[m] =
II. S YSTEM M ODEL AND P ROBLEM F ORMULATION FRF FBB [m]s[m].
We consider a wideband ultra-massive MIMO architecture Assuming frequency-selective fading over multi-carrier
in the context of a JRC system for a vehicle to vehicle (V2V) transmission between the TX and RX [26], the received signal
and vehicle to device (V2D) scenario, in which the transmitter at the RX is given by
√
(TX) senses the environment via probing waveforms to the y[m] = ρH[m]FRF FBB [m]s[m] + n[m], (1)
targets and communicates with the receiver (RX), as illustrated
in Fig. 1. The antenna arrays at the TX and the RX employ where y[m] ∈ CNR is the output of NR subarrayed antennas at
graphene-based plasmonic nano-antennas, which are placed the RX, ρ is the received power and n[m] ∈ CNR denotes the
on a metallic surface layer, with a dielectric layer between additive white Gaussian noise (AWGN) vector with n[m] ∼
them [26, 32, 49]. The antennas form GoSA structure as CN (0, σn2 INR ). H[m] ∈ CNR ×NT denotes the THz channel
each subarray consists of Qx × Qy uniform rectangular arrays matrix between the TX and the RX.
(URAs) with Q = Qx Qy antennas, as shown in Fig. 1. Also, In THz transmission, the wireless channel H[m] can be
there are NT = NTx NTy and NR = NRx NRy subarrays represented by a single dominant LoS path with assisting a
of size Q at the TX and RX, respectively, which form an few NLoS paths, which are small due to large reflection loses,
NT Q × NR Q ultra-massive MIMO transceiver architecture. scattering and refraction [26, 31, 32]. Channel modeling at
In each Qx × Qy subarray, the antenna spacing along the x- THz-band has been a challenge largely because of lack of
and y-axis are δx , δy and the distance between each subarray realistic measurements. Very recently, measurement campaigns
are ∆x , ∆y , respectively. at 140 GHz have been reported [54, 55]. In particular, [55]
states that, while the delay/angular spread at 140 GHz and
lower frequencies are comparable, the correlation distance of
A. Communications Model shadow fading at the former is much shorter. The same study
In the downlink, the TX with NT subarrays, each of which mentions multiple NLoS and dominant LoS paths at low-THz.
has Q antenna elements, aims to transmit NS data streams This work is, however, not focused on only 140 GHz, and,
towards the RX in the form of s[m] = [s1 [m], . . . , sNS [m]]T therefore, employs the assumptions theorized for the entire
by using hybrid analog and digital beamformers with NRF RF upper-mmWave/low-THz region as in [31].
chains, where E{s[m]sH [m]} = INS and NS ≤ NRF . Here, While the ray-tracing techniques assume the channel to be
m ∈ M = {1, . . . , M } and M is the number of subcarriers. sparse and dominated by the LoS component for the graphene
Due to beamforming at subarray level, each subarray of size nano-transceivers [22], the other channel models such as the
5
Fig. 2. Hybrid beamforming-based transmitter structures for (a) fully connected, (b) partially-connected array of subarrays (AoSAs) and (c) partially-connected
groups of subarrays (GoSAs) architectures. While all the architectures employ NT Q antennas with NRF RF chains, each antenna is connected to each RF
chain via combiners in the fully connected model with NT QNRF phase-shifters. In partially-connected AoSA, the same RF chain is connected to N̄ Q
(N̄ = NNT ) antennas with NT Q phase-shifters totally. In partially-connected GoSA model, each RF chain is connected to N̄ Q antennas while each
RF
phase-shifter is connected only Q antennas, introducing NT group of subarrays with only NT phase-shifters.
NT ×Q
3GPP model [56, 57] are also popular for THz beamforming. AmT (Ψl ) ∈ C are the steering matrices corresponding to
In this work, we adopt the Saleh-Valenzuela (SV) THz channel the AoAs and AoDs of the GoSAs, respectively, and they are
model channel [26, 34], wherein H[m] is constructed by the defined as
superposition of a single LoS path and the contribution of
T
mT
Nclu cluster of Nray NLoS paths. Then, the NR × NT THz am (Θ l ) a (Ψ l )
ultra-massive MIMO channel matrix is given by [31] R,1. T,1.
Am
m
R (Θ l ) =
.
.
,
T A (Ψl ) =
.. (5)
T
H[m] = mT
amR,NR (Θ l ) aT,NT (Ψ l )
L
H
mH
X
γ α1,m Am m
αl,m Am
R (Θ1 )AT (Ψ1 )+ R (Θl )AT (Ψl ) , (2) where am m
R,nR (Θl ) (aT,nT (Ψl )) is Q × 1 steering vector cor-
l=2 responding to the antennas in the nR th (nT th) subarray for
nR ∈ {1 . . . , NR } (nT ∈ {1, . . . , NT }), respectively. The ith
q
where γ = NTLNR and L = 1 + Nclu Nray denotes the total
element of the transmit steering vector am T,nT (Ψl ) is given by
number of LoS and NLoS paths. Furthermore, αl,m represents
the channel gain of the lth path for mth subcarrier, and we
1 2π T
have [am
T,nT (Ψl )]i = √ exp{−j κ Ωl }, (6)
LoS NT λm nT ,i
αm , l=1
αl,m = NLoS , (3)
αrc,m , l > 1, l = (Nray − 1)c + r + 1 where λm = fcm0 is the wavelength for the subcarrier m with
for which l ∈ {1, . . . , L}, r ∈ {1, . . . , Nray } and c ∈ frequency of fm = fc + M B
(m−1− M2−1 ), where B denotes the
LoS
{1, . . . , Nclu }. αm ∈ C denotes channel gain of the LoS bandwidth. κnT ,i = [xnT ,i , ynT ,i , znT ,i ]T denotes the position
path and it is defined as of the ith antenna of the nT th subarray in Cartesian coordinate
c0 γ̄ 1 ¯ system and Ωl is a direction-dependent parameter defined
LoS
αm =( ¯ ) 2 e− 2 κ̄(fm )d , (4) as Ωl = [cos ϕl sin ϑl , sin ϕl sin ϑl , cos ϑl ]T . The structure
4πfm d
of am m
R (Θl ) is similar to that of aT (Ψl ). Without loss of
where c0 is the speed-of-light, κ̄(fm ) is frequency-dependent generality, we assume that the antennas are perfectly calibrated
molecular absorption coefficient and d¯ is the distance between against mutual coupling and gain/phase mismatches [26].
the TX and RX. In mmWave channels, the path loss exponent Finally, by exploiting the GoSA structure, the (nR , nT )th
γ̄ is around 2 while it has typical values between 3 and element of H[m] ∈ NR × NT is given as
4.5 in THz channels for dense urban environments [31].
NLoS NLoS j β̄m,rc
αm,rc = |αm,rc |e corresponds to the channel gain L
X ∗
of the rth NLoS path in the cth cluster and β̄m,rc is an [H[m]]nR ,nT = γ αl,m am m
R,nR (Θl ) aT,nT (Ψl ). (7)
independent uniformly distributed phase shift over [0, 2π) [31]. l=1
The notations Θl = {φl , θl } and Ψl = {ϕl , ϑl } denote
the azimuth/elevation angle-of-arrival (AoA) and angle-of- By connecting the Q antennas in the subarrays to a single
departure (AoD) of the received/transmitted paths at the RX phase-shifter, we are able to construct an NT × NRF , (instead
NR ×Q
and the TX, respectively. The matrices Am R (Θl ) ∈ C and of NT Q × NRF as in Fig. 2a and Fig. 2b) RF precoder, as
6
illustrated in Fig. 2c. Using partially-connected GoSA, the corresponds to the mutual information for subcarrier m.
associated RF precoder has the form of We note here that the maximization of (9) is provided
by exploiting the similarity between the hybrid beamformer
u1 0 · · · 0 FRF FBB [m] and the optimal unconstrained beamformer
0 u2 · · · 0 FC [m] ∈ CNT ×NS . The latter is obtained from the right
FRF = . ∈ CNT ×NRF , (8)
. . . singular matrix of the channel matrix H[m] [1, 43]. The
. 0 . 0
0 0 · · · uNRF singular value decomposition of the channel matrix is H[m] =
UH [m]Π[m]VH H
[m], where UH [m] ∈ CNR ×rank(H[m]) and
NT ×rank(H[m])
where ui ∈ CN̄ represents a portion NT × 1 phase-shifter val- VH [m] ∈ C are the left and the right singular
ues with indices {(i−1)N̄ +1, . . . , iN̄ } for i ∈ {1, . . . , NRF }, value matrices of the channel matrix, respectively, and Π[m]
where N̄ = NNRF T
. Each entry of ui is then applied to Q is rank(H[m])×rank(H[m]) matrix composed of the singular
antennas in NT subarrays to steer the transmitted beams (see, values of H[m] in descending order. By decomposing Π[m]
e.g., Fig. 2c) so that a total of NT Q antennas are fed. and VH [m] as Π[m] = diag{Π[m], e Π[m]}, VH [m] =
To address the performance degradation due to GoSA, the [VH [m], VH [m]], where VH [m] ∈ CNT ×NS , the uncon-
e e
columns of FRF is designed with overlapping terms [58, 59]. strained precoder is readily obtained as FC [m] = V e H [m] [1].
Assume ūi ∈ CM̄ to include the overlapped phase-shifter Then, the maximization of (9) is achieved by minimizing the
terms, where M̄ ∈ [N̄ , NT − NRF + 1], for which M̄ = N̄ Euclidean distance between FC [m] and FRF FBB [m] as
provides non-overlapped partially-connected structure as in (8)
1 X
while M̄ = NT − NRF + 1 provides maximum overlap among min kFRF FBB [m] − FC [m]kF
FRF ,{FBB [m]}m∈M M
the phase-shifters. In this case, the performance improvement m∈M
is at the cost of using more phase-shifters. Nevertheless, it
X
s. t. : kFRF FBB [m]kF = M NS ,
still has lower number of phase-shifters as compared to the m∈M
partially non-overlapped case in conventional AoSA (see, e.g., 1
Fig. 4). The use of partially-connected/PCO GoSA structure |[FRF ]i,j | = √ , ∀i, j. (11)
NT
provides higher DoF as compared to the simple phased-array
MIMO radar structure, for which NRF = 1 [58] and we
have a fully-connected MIMO structure when NRF = NT .
While MIMO radar outperforms the phased-array in terms B. Radar Model
of angular resolution and DoF for parameter estimation and
parameter identification, phase-array provides higher coherent The goal of radar processing is to achieve the highest
processing gain and lower computation and hardware com- possible SNR gain towards the direction of interest. The radar
plexity [60]. This complexity is further reduced by using the first transmits an omni-directional waveforms to detect the
GoSA structure by feeding each of Q antennas with the same unknown targets within the angular space of interest in the
phase shift. Thus, the partially-connected GoSA provides a search phase, then it generates directional beams towards to
trade-off between the DoF and the hardware complexity, both the targets for tracking purposes [60]. We assume a subarrayed
of which increase as NRF → NT . MIMO radar architecture with GoSAs, wherein each GoSA
In communications-only systems, the aim is to design the is used to coherently transmit waveforms that are orthogonal
hybrid precoders such that the spectral efficiency at the TX to the ones generated by other GoSAs [58], thereby, coher-
is maximized [1, 43, 52], while there are also other re- ent processing gain is achieved. To this end, the transmit
lated performance metrics, such as energy-efficiency [61] and waveform of the kth GoSA (k ∈ {1, . . . , K}) is designed
minimum mean-squared-error (MMSE) [62]. By decoupling as wk (t) = W (t)ej2πk∆f t , 0 < t < T0 , where W (t) is
the beamformer design problem at the TX and the RX, the pulse shape with duration of T0 , so that the orthogo-
the mutual information at the TX is maximized instead of nality of wk (t) is satisfied for a variety of time delays and
spectral efficiency, for which a perfect combiner is assumed Doppler shifts if the frequency increment among the GoSA
at the receiver [1]. Once the transmitter is designed, the waveforms satisfies ∆f = |fk+1 − fk | 1/T0 [58]. Denote
receive beamforming design is done by using the MMSE {Φ1 , . . . , ΦK } as the set of target directions (Φk = (ϕ̄k , ϑ̄k )),
as performance metric as in [1, 43, 52]. Then the mutual then, the NT × K GoSA-MIMO radar-only beamformer is
information of the communications system is given by modeled as FR = blkdiag{v1 , . . . , vK } similar to (8), where
vk ∈ CK̄ denotes the values of the transmit steering vector
M
1 X aT (Φk ) ∈ CNT with indices {(k − 1)K̄ + 1, . . . , k K̄} for
I= I(FRF , FBB [m]), (9) k = 1 . . . , K and K̄ = NKT . It is possible to construct FR via
M m=1
overlapped GoSA with v̄k ∈ CNT −K+1 for k ∈ {1, . . . , K}.
where The estimation of the target directions {Φk }K k=1 is per-
ρ formed in the search phase of the radar. This is achieved via
I(FRF , FBB [m]) = log2 INR + H[m]FRF FBB [m] both: model-based methods, such as MUSIC (multiple signal
NS σn2
classification) algorithm [63], and model-free techniques based
H H H on DL [64]. In this work, we assume that search operation
× FBB [m]FRF H [m], (10)
is completed and the direction information of the targets is
7
C. Problem Formulation f (FRF , FBB [m], P[m]) = kη[FRF FBB [m] − FC [m]]kF
+ kη̄[FRF FBB [m] − FR P[m]]kF . (16)
The aim of this work is designing the hybrid beamformer
Then, using triangle inequality, we get
FRF FBB [m] to simultaneously maximize the spectral effi-
ciency of the communications link and provide as much f (FRF , FBB [m], P[m]) ≥
SNR as possible towards the radar targets by forming the
kηFRF FBB [m] − ηFC [m] + η̄FRF FBB [m] − η̄FR P[m]kF
beampattern of the transmit antenna array. To jointly solve
the problems in (11) and (14), we formulate the following = kFRF FBB [m] − ηFC [m] − η̄FR P[m]kF . (17)
8
Define FCR [m] ∈ CNT ×NS as the JRC beamformer as due to the constraint (21c). Thus, we propose a MMO-based
solution to account for (21c) in the following.
FCR [m] = ηFC [m] + η̄FR P[m], (18) Assume that F e BB and Pe are fixed, then (21) is written in
and define the new cost function f˜(FRF , FBB [m], P[m]) as vectorized form as
min kGfRF − fCR kF
f˜(FRF , FBB [m], P[m])=kFRF FBB [m] − FCR [m]kF , (19) fRF
1
where we have f˜(FRF , FBB [m], P[m]) ≤ s. t. : |[fRF ]i | = √ , ∀i ∈ V,
NT
f (FRF , FBB [m], P[m]) due to (17). Then, we rewrite
|[fRF ]i | = 0, ∀i ∈ V̄, (23)
the optimization problem (15) as
1 X where G = (F e T ) ⊗ IN ∈ CNT M NS ×NT NRF , fRF =
BB T
min kFRF FBB [m] − FCR [m]kF vec{FRF } ∈ CNT NRF and fCR = vec{F e CR } ∈ CNT M NS .
FRF ,{FBB [m],P[m]}m∈M M
m∈M
X V and V̄ denote the set of non-zero and zero terms in fRF ,
s. t. : kFRF FBB [m]kF = M NS , respectively. The sizes of V and V̄ depend on the selection
m∈M of M̄ . As an example, for NT = 100, NRF = 10 and
1 M̄ = NRF (NT − NRF + 1), we have |V| = 910 and
|[FRF ]i,j | = √ , ∀i, j ∈ S,
NT |V̄| = 90. Now, the aim is to exclude the portion of G and fRF
|[FRF ]i,j | = 0, ∀i, j ∈ S̄, corresponding to V̄ and find the portion of fRF corresponding
P[m]PH [m] = INS . (20) to V so that we employ manifold optimization accordingly and
all the elements of the unknown vector will obey unit-modulus
The optimization problem in (20) can be written in a constraints.
compact form as
Lemma 1. Define GV ∈ CNT M NS ×T and fRFV ∈ CT as
min kFRF F
e BB − F
e CR kF the portion of G and fRF corresponding to V, where T
FRF ,F
e BB ,P
e is the number of remaining columns (entries) of G (fRF )
s. t. : kFRF Fe BB kF = M NS , (21a) after excluding the terms related to V̄. Then, the optimization
1 problem in (23) is equivalent to
|[FRF ]i,j | = √ , ∀i, j ∈ S, (21b)
NT min kGV fRFV − fCR kF
fRFV
|[FRF ]i,j | = 0, ∀i, j ∈ S̄, (21c)
1
P
ePe H = IM N , (21d) s. t. : |[fRFV ]i | = √ , ∀i ∈ V. (24)
S
NT
where P e = [P[1], · · · , P[M ]] is a K × M NS matrix, Proof: The cost of problem (23) is
and we have F e CR = [FCR [1], FCR [2], · · · , FCR [M ]] ∈
fRFV
C NT ×M NS
and Fe BB = [FBB [1], FBB [2], · · · , FBB [M ]] con- k[GV ,GV̄ ] − fCR kF
fRFV̄
taining the beamformers for all subcarriers. F e CR corresponds
= kGV fRFV + GV̄ fRFV̄ − fCR kF
to the unconstrained radar-communications beamformer as
F
e CR = η F e C + η̄FR P,
e where F e C = [FC [1], · · · , FC [M ]] ∈ = kGV fRFV − fCR kF . (25)
CNT ×M NS . Since fRFV̄ = 0NT NRF −T ×1 , we have GV̄ fRFV̄ =
Now, the problem (21) looks similar to the communications- 0NT M NS ×(NT NRF −T ) , and thus (23) is equivalent to (24).
only problem in (11), and is solved via alternating minimiza- Using Lemma 1, we solve (24) via MO, for which the
tion techniques suggested to solve (11), e.g., [1, 43]. In this search space is regarded as a Riemannian submanifold M of
case, FRF , F e BB and Pe are estimated one-by-one while the
complex plane CT since fRFV ∈ CT forms a complex circle
others are fixed. By fixing FRF and F e BB , P
e is found via the
manifold, i.e., MTcc = {fRFV ∈ CT : |[fRFV ]1 | = |[fRFV ]2 | =
SVD of the matrix FH F F
R RF BB
e , and for the mth subcarrier, · · · = |[fRFV ]T | = √N 1
} due to unit-modulus constraint.
we have T
Define the Riemannian gradient at fRFV , gradf (fRFV ) as the
P[m] = U[m]I
e K×NS V[m],
e (22) orthogonal projection of the Euclidean gradient ∇f (fRFV )
onto the tangent space of fRFV , i.e.,
for which U[m]Σ[m]
e Ve H [m] = FH FRF FBB [m] and
R ∗
gradf (fRFV ) = ∇f (fRFV ) − Re{∇f (fRFV ) fRF } fRFV ,
IK×NS = [IK , 0K×(NS −K) ] [1]. Similarly, when FRF and V
P
e are fixed, F e BB is calculated. For the mth subcarrier, where the Euclidean gradient of the cost function in (24) is
†
FBB [m]
√
= FRF FCR [m], and it is normalized as FBB [m] = given by
NS
kFRF FBB [m]kF FBB [m]. The main challenge in (21) is the ∇f (fRFV ) = −2GH
V [fCR − GV fRFV ]. (26)
estimation of FRF due to unit-modulus constraints. In fully-
connected case, FRF is found via MO-based techniques and After defining the cost function and the gradient as in (24)
the optimal solution is readily obtained for the partially- and (26), the remaining part of the solution is similar to the
connected structure via phase-rotation [43, 65]. However, the conventional manifold optimization algorithm [71]. This is
design of FRF for the overlapped case is not straightforward done via the conjugate gradient descent technique iteratively
9
Algorithm 1 Hybrid beamforming for joint ultra-massive The hybrid beamformer design problem in (21) should be
MIMO radar-communications solved for every η. Furthermore, P[m] changes even if FR
Input: η, F e C , FR . is kept fixed since it matches the hybrid beamformer to the
Output: FRF , FBB [m] and FcBB [m] for m ∈ M. unconstrained radar-only beamformer. As a special case, where
1: Initialize with random FRF ∈ CNT ×NRF , FBB [m] ∈ P[m] = IK , i.e., K = NS , the following lemma shows that
CNRF ×NS and P[m] ∈ CK×NS . the solution of (21) is obtained from the linear combination
2: F
e CR = η F e C + η̄FR P.
e of the solutions of communications- and radar-only problems
3: Construct S and S̄ depending on the structure of FRF . in (11) and (14), respectively.
4: while
Lemma 2. Denote F̌C ∈ CNT ×M NS and F̌R ∈ CNT ×M NS
5: Construct P e = [P[1], . . . , P[M ]], where
e H [m] = as the hybrid beamforming solutions of (11) and (14), respec-
P[m] = U[m]IK×NS V[m]
e e and U[m]Σ[m]
e V
H tively. Then, the solution of (21) is
FR FRF FBB [m] for m ∈ M.
6: Compute FBB [m] as FBB [m] √= F†RF FCR [m] and F̌CR = η F̌C + (1 − η)F̌R , (28)
NS
normalize as FBB [m] = kFRF FBB [m]kF FBB [m]. if K = NS and P[m] = IK for m ∈ M.
7: Use S and S̄ and find FRF with the MMO algorithm
e R = [FR , · · · , FR ] ∈
Proof: If P = IK , we have F
in (20) and (23). | {z }
8: until convergence M
9: Beam split correction: FcBB [m] = CNT ×M K , and the solution of (21) involves the alternations
† c only between FRF and F e BB ∈ CNRF ×M NS . Then, rewrite the
FRF FRF [m]FBB [m] for m ∈ M as in (42).
cost function of (21) as
f (F̌C , F̌R ) = ηkF̌C −F
e C kF + (1 − η)kF̌R −F
e R kF , (29)
(k+1)
such that fRFV at the kth iteration is obtained with the update from which it is clear that F̌C and F̌R correspond to the
rule communications- and radar-only hybrid beamforming solu-
(k)
tions in (11) and (14) for η = 1 and η = 0, respectively.
(k+1) (fRFV + αk d(k) ) Using the triangle inequality expression in (17), f (F̌C , F̌R )
fRFV = , (27)
(k) (k)
|(fRFV + αk fRFV )| in (29) is lower-bounded by f¯(F̌C , F̌R ) as
kη F̌C − FC + (1 − η) F̌R − F e R kF ≤ f (F̌C , F̌R ), (30)
where αk is Armijo backtracking line search step size [71] | {z }
∆
and dk denotes the direction of decrease, which is defined =f¯(F̌C ,F̌R )
(k) (k)
as dk = −gradf (fRFV ) + βk d̄(k−1) , for which gradf (fRFV )
where f¯(F̌C , F̌R ) can be rewritten as
denotes the Riemannian gradient at the kth iteration and βk
is the Polak-Ribiere parameter. d̄(k) is the vector transport of f¯(F̌C , F̌R ) = kη F̌C + (1 − η)F̌R − ηFC − η̄ F
e R kF
conjugate direction d(k) , which is defined as d̄(k) = d(k) − = kF̌CR − (1 − η)FC − (1 − η)Fe R kF
(k+1) (k+1) (k+1)
Re{d(k) fRFV } fRFV , where fRFV is directly obtained
(0) = kF̌CR − FCR kF , (31)
from (27) and d0 = −gradf (fRFV ). The optimization process
(0)
is initialized from a random point, i.e., [fRFV ]t = ej θ̄t where which gives (28) as the linear combination of F̌C and F̌R by
θ̄t ∼ uniform([0, 2π)), t = 1, . . . , T . Once the non-zero depending on η.
entries of fRF , i.e., fRFV is optimized, then RF precoder This analysis allows us to design the JRC hybrid beam-
FRF ∈ CNT ×NRF is reconstructed from fRFV according to former as a function of η after solving the communications-
the index sets V, V̄, S and S̄. and radar-only problems (11) and (14), respectively, instead
In (21), the convergence to an optimum solution is guar- of solving (21) for every η.
anteed such that the Euclidean distance between the radar-
communications beamformer F e CR and the hybrid beamformer B. Hybrid Beamformer Design With Channel Covariance Ma-
FRF FBB is minimized [43]. We present the algorithmic steps
e trix
of the CSI-based hybrid beamformer design in Algorithm 1, Instead of designing the hybrid beamformer FRF FBB [m]
which also includes beam split correction procedure as de- with respect to H[m], the usage of the channel covariance
scribed in Appendix A. The implementation of the iterative matrix provides lower channel overhead via infrequent updates
algorithm takes no more than 10 iterations while the MMO of the THz channel information between the RX and the TX.
steps requires approximately 20 sub-iterations for the settings However, this approach has the cost of slight performance
NT = 1024, Q = 9 and NRF = 10. Furthermore, the loss in the spectral efficiency due to long-term statistics of
complexity order of the MMO algorithm is the same as the the channel information. The usage of channel covariance
conventional manifold optimization algorithm [43], and it is matrix is particularly helpful in THz transmission compared
mainly due to the computation of the conjugate gradient in to the mm-Wave case due to smaller number of LoS/NLoS
(26) which is O(Niter NT M NS T 2 ), where Niter denotes the signal components, which reduces the angular spread of the
number of iterations and T is the number of non-zero entries received signals [22]. To exploit the structure of channel co-
of FRF [43, 65, 71]. variance matrix-based hybrid beamforming, we first introduce
10
where Bm R = [Am m
R (Θ1 ), . . . , AR (ΘL )] ∈ CNR ×QL
and BT m
= [AT (Ψ1 ), . . . , AT (ΨL )] ∈ CNT ×QL
m m
where the expectation is performed over H[m]. Taking statis- Once the unconstrained statistical beamformer is obtained,
tical expectation over the AoA/AoD angles and the channel the next task is to determine the analog precoder FRF via the
gains, respectively, (33) becomes following optimization problem, i.e.,
H H H
m m
C[m] = γ 2 EΨ {Bm EΘ {Bm m m
ē
T Eα {Γ R BR }Γ }BT }, (34) min e BB − F
FRF F CR F
FRF ,F
e BB ,P
e
H m H 1
for which we have EΘ {Bm m
R BR } = IQL and Eα {Γ Γm } = s. t. : |[FRF ]i,j | = √ ,
m NT
e = blkdiag{σ 2 IQ , . . . , σ 2
Γ α1,m αL,m IQ } due to the indepen-
dent zero-mean channel gains. Thus, we finally get kFRF F
e BB kF = M NS ,
H
P
ePe = IM N , (38)
e m mH
S
C[m] =γ 2 Bm
T Γ BT
L NT ×M NS
where F CR = [F̄CR [1], · · · , F̄CR [M ]] ∈ C . The
ē
X H
= γ2 σα2 l,m E{Am m
T (Ψl )AT (Ψl )}. (35) channel covariance matrix-based problem in (38) is similar
l=1
to (21) and the solution is obtained by following the similar
Since the path gains of the NLoS paths (i.e., l = 2, . . . , L) procedure presented in Algorithm 1 by replacing F e C with F
ē .
C
are significantly smaller as compared to the LoS path in
THz channels, the channel covariance matrix in (35) can be
approximated as IV. M ODEL -F REE H YBRID B EAMFORMING
50
V. N UMERICAL E XPERIMENTS
In this section, we evaluate the performance of the proposed 40
hybrid beamforming approach for different array structures.
30
The communications performance of the algorithms is evalu-
ated in terms of spectral efficiency while the radar performance 20
is presented with the beampattern analysis of the hybrid
beamformers. Furthermore, we analyze the trade-off between 10
realizations with the aforementioned channel statistics and 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
K = 3 radar target directions, which are generated uniform
randomly from the interval [−50◦ , 50◦ ] with 1◦ resolution for
ZR = 104 realizations. Once the input data Ξi is prepared
Fig. 6. Spectral efficiency versus η for SNR= 10 dB.
as described in Section IV, the optimization problem in (21)
is solved for each input data, then the corresponding output
30
label, i.e., ξ i is computed for i = 1, . . . , D in an offline
20
manner. As a result, the resulting dataset is comprised of
10
D = ZC ZR = 106 samples of size NT × (NR + K) × 2.
0
The cost function for DeepBF is the MSE between L(Ξi , θ)
-10
and ξ i . The DeepMUSIC model is constructed as described 0 20 40 60 80 100 120 140 160 180
10
Fig. 4 shows the number of phase-shifters with respect
0
to NT and Q for different array structures, i.e., AoSA and
-10
GoSA, respectively. The fully-connected structures employ 0 20 40 60 80 100 120 140 160 180
60
50
40
30
20
10
Fig. 8. Beampattern in 2-D plane for η = 0.5. The radar targets are located Fig. 10. Spectral efficiency versus SNR for channel covariance matrix-based
at {(80◦ , 85◦ ), (110◦ , 80◦ ), (140◦ , 85◦ )} and the LoS communication path beamforming when η = 0.5.
is received from (−50◦ , 80◦ ), which are illustrated with red and pink colors,
respectively.
RX and radar targets as compared to the other algorithms. The
2-D angular distribution is illustrated in Fig. 8 for η = 0.5,
45 where we present the beampattern corresponding to the GoSA
with PCO. We observe that the proposed hybrid beamforming
40 approach accurately generates beams towards both targets and
RX paths in 2-D angular space. This is provided with the 2-D
structure of the antenna array in two dimensions.
35
While the design of the antenna array is straightforward
in AoSA case by selecting the antenna spacing as λ/2, the
30
selection of δ is critical for the GoSA structure illustrated
in Fig. 1. In Fig. 9, we present the spectral efficiency per-
25 formance with respect to d, ¯ where d¯ = λ . As d¯ increases,
δ
we reduce the antenna element spacing in the subarrays of
20 GoSA. Specifically, when d¯ = ∞, we have δ = 0, thus the
Qx × Qy antennas in each subarray become co-located. We
2 4 6 8 10 12 14 16 18 20
infer this from Fig. 9 as the performance of AoSA approaches
to that of GoSA as d¯ increases. While slight performance
loss is observed from AoSA with partially-connected array,
the performance of the fully-connected structure and the fully
Fig. 9. Spectral efficiency versus d¯ for SNR= 10 dB when η = 0.5. digital beamformers significantly reduce as d¯ increases. As a
result, this figure is helpful when designing the GoSA because
of the improvement in the spectral efficiency by changing d¯
tracking the targets or when there is no target. Conversely, while the lower limit is d¯ = 2 (i.e., δ = λ/2) to avoid spatial
lower η is selected if the radar task demands more resources, aliasing among the antennas.
e.g., more transmit power is required depending on the range Fig. 10 shows the spectral efficiency of the competing
of the radar targets. algorithms for both CSI- and channel covariance matrix-based
We illustrate the beampattern of the designed hybrid beam- beamforming when η = 0.5. A slight performance loss is
formers in Fig. 7 and Fig. 8 for 1-D and 2-D angle distribu- observed for all channel covariance matrix-based approaches
tions, respectively. In Fig. 7, the beampatterns are presented for due to loss of precision in the angle and path gain information
η = 0, η = 0.5 and η = 1, where we assume that the all of the while less channel overhead is involved in channel covariance
paths have the elevation angle of 90◦ . The ideal beampatterns matrix-based beamforming. In particular, in the CSI-based
correspond to the radar-only beamformer FR for AoSA and approach, the RX should feedback NR × NT (81 × 1024)
GoSA structures. We note that for η = 0 (η = 1) all the beams channel matrix whereas only the angle and path information
are generated towards the radar targets (the RX), respectively, needs to be sent to the TX in the channel covariance matrix
while FRF generates K + 1 beams towards both targets and so that C[m] is constructed as in (36).
the RX when η = 0.5. The proposed GoSA PCO structure Fig. 11 shows the effect of frequency on the generated
provides lower side lobes and narrower beams towards both beams for (a) sub-6 GHz, (b) mmWave and (c) THz MIMO
14
0.7 0.4
0.75 19 14
0.65 17 10
0.55 0.1
0.5 0.6 0
0.4 0.5 0.6 0.4 0.5 0.6 0.4 0.5 0.6 10-2 16 8
0 10 20 -20 0 20 -20 0 20
Fig. 11. Normalized array gain with respect to physical direction θ at lowest, Fig. 13. (a) Radar target direction RMSE versus spectral efficiency, and the
center and highest subcarrier frequencies (i.e., f1 , fc and fM ) for (a) fc = cost function for (b) communications- and (c) radar- only beamformer.
3.5 GHz, B = 0.1 GHz; (b) fc = 28 GHz, B = 2 GHz; and (c) fc = 300
GHz, B = 30 GHz when M = 128.
Ω
e l = Ωl ∆m , (40)
50
0
F̂cRF [m] = F̂RF (Ω∆m ), (41)
-20 -15 -10 -5 0 5 10 15 20 25 30
where F̂cRF [m] points to Ω for all m ∈ M. The beam split
correction operation in (41) by simply multiplying the phase
values of F̂RF (Ω) by ∆m . However, the implementation of
Fig. 14. Spectral efficiency for model-based and model-free hybrid beam-
former design when η = 0.5. F̂cRF [m] is not efficient since it requires M phase shifter net-
work of size NT NRF . To mitigate this, the effect of ∆m can be
conveyed to the frequency-dependent baseband beamformers.
free hybrid beamforming techniques. To lower the hardware The modified baseband beamformer at subcarrier m becomes
complexity critical in THz systems, we proposed GoSA ultra- †
massive MIMO architecture. We developed hybrid beamform- F̂cBB [m] = F̂RF (Ω) F̂cRF [m]F̂BB [m]. (42)
ing via PCO structures to provide a trade-off between higher
Finally, the beam split corrected hybrid beamformer can be
spectral efficiency and hardware complexity in terms of the
realized as F̂RF (Ω)F̂cBB [m].
number of phase-shifters. The hybrid beamformers for THz
JRC system are designed relying on both CSI and channel
covariance matrix of the wireless channel information between ACKNOWLEDGEMENTS
the TX and the RX. The computation times for beamformer The authors are sincerely grateful to the guest editor
design could be prohibitively high for ultra-massive MIMO Prof. Robert W. Heath, Jr. and three anonymous reviewers
THz systems. We addressed this by suggesting a model-free whose valuable comments greatly helped in improving the
DL-based approach. manuscript.
We evaluated the performance of the proposed methods
in terms of spectral efficiency and radar beampattern. We
demonstrated that GoSA provides less hardware complexity R EFERENCES
compared to full array and AoSA structures. To mitigate the [1] R. W. Heath, N. Gonzalez-Prelcic, S. Rangan, W. Roh, and A. M.
beam split effect, we also introduce hardware-efficient ap- Sayeed, “An overview of signal processing techniques for millimeter
wave MIMO systems,” IEEE J. Sel. Topics Signal Process., vol. 10,
proach by correcting the phases of the frequency-independent no. 3, pp. 436–453, 2016.
beamformers. Compared to CSI-based beamforming, channel [2] K. V. Mishra, M. R. Bhavani Shankar, V. Koivunen, B. Ottersten, and
covariance matrix-based approach has a slight performance S. A. Vorobyov, “Toward millimeter wave joint radar-communications:
A signal processing perspective,” IEEE Signal Process. Mag., vol. 36,
loss, while the latter enjoys less channel overhead. The model- no. 5, pp. 100–114, 2019.
free method is advantageous in terms of computational com- [3] N. Rajatheva, I. Atzeni, E. Bjornson, A. Bourdoux, S. Buzzi, J.-B.
plexity and exhibits approximately 500 times lower compu- Dore, S. Erkucuk, M. Fuentes, K. Guan, Y. Hu et al., “White paper
on broadband connectivity in 6G,” arXiv preprint arXiv:2004.14247,
tation time as compared to the MO-based approaches, while 2020.
maintaining spectral efficiency performance close to that of [4] H.-J. Song and T. Nagatsuma, “Present and future of terahertz commu-
channel covariance matrix-based technique. nications,” IEEE Trans. THz Sci. Technol., vol. 1, no. 1, pp. 256–263,
2011.
[5] I. F. Akyildiz and J. M. Jornet, “Realizing ultra-massive MIMO (1024×
A PPENDIX A 1024) communication in the (0.06-10) terahertz band,” Nano Commu-
B EAM S PLIT C ORRECTION nication Networks, vol. 8, pp. 46–54, 2016.
[6] T. Kürner and S. Priebe, “Towards THz communications-status in
In mmWave wideband hybrid beamforming, the analog research, standardization and regulation,” Journal of Infrared, Millimeter,
beamformers are usually designed with respect to a single and Terahertz Waves, vol. 35, no. 1, pp. 53–62, 2014.
frequency, (i.e., fc ) while the baseband beamformers are [7] K. Tekbıyık, A. R. Ekti, G. K. Kurt, and A. Görçin, “Terahertz
band communication systems: Challenges, novelties and standardization
frequency-dependent. Hence, the analog beamformers may efforts,” Physical Communication, vol. 35, p. 100700, 2019.
point in different directions at different subcarriers because [8] S. H. Dokhanchi, B. S. Mysore, K. V. Mishra, and B. Ottersten, “A
of ultra-wide bandwidth and large number of antennas. In mmWave automotive joint radar-communications system,” IEEE Trans.
Aerosp. Electron. Syst., vol. 55, no. 3, pp. 1241–1260, 2019.
particular, let f (Ωl ) be the beamforming vector corresponding [9] G. Wang and K. V. Mishra, “Displaced sensor automotive radar imag-
to the 3 × 1 spatial direction vector Ωl as defined in (6), then ing,” arXiv preprint arXiv:2010.04085, 2020.
16
[10] E. Marchetti, R. Du, B. Willetts, F. Norouzian, E. G. Hoare, T. Y. Tran, Wideband Ultra-Massive MIMO Terahertz Communications,” arXiv,
N. Clarke, M. Cherniakov, and M. Gashinova, “Radar cross-section of Apr 2021. [Online]. Available: https://arxiv.org/abs/2104.11054v2
pedestrians in the low-THz band,” IET Radar, Sonar & Navigation, [32] C. Lin and G. Y. L. Li, “Terahertz Communications: An Array-of-
vol. 12, no. 10, pp. 1104–1113, 2018. Subarrays Solution,” IEEE Commun. Mag., vol. 54, no. 12, pp. 124–131,
[11] A. Faisal, H. Sarieddeen, H. Dahrouj, T. Y. Al-Naffouri, and M.-S. Dec 2016.
Alouini, “Ultramassive MIMO systems at Terahertz bands: Prospects [33] A. Alkhateeb, O. E. Ayach, G. Leus, and R. W. Heath, “Hybrid
and challenges,” IEEE Veh. Technol. Mag., vol. 15, no. 4, pp. 33–42, precoding for millimeter wave cellular systems with partial channel
2020. knowledge,” in IEEE Inf. Th. Appl. Workshop, 2013, pp. 1–5.
[12] F. Norouzian, R. Du, E. G. Hoare, P. Gardner, C. Constantinou, [34] H. Yuan, N. Yang, K. Yang, C. Han, and J. An, “Hybrid Beamforming
M. Cherniakov, and M. Gashinova, “Low-THz transmission through for Terahertz Multi-Carrier Systems Over Frequency Selective Fading,”
water-containing contaminants on antenna radome,” IEEE Trans. THz IEEE Trans. Commun., vol. 68, no. 10, pp. 6186–6199, Jul 2020.
Sci. Technol., vol. 8, no. 1, pp. 63–75, 2017. [35] K. Dovelos, M. Matthaiou, H. Q. Ngo, and B. Bellalta, “Channel
[13] C. A. Schmuttenmaer, “Exploring dynamics in the far-infrared with Estimation and Hybrid Combining for Wideband Terahertz Massive
terahertz spectroscopy,” Chemical Reviews, vol. 104, no. 4, pp. 1759– MIMO Systems,” arXiv, Feb 2021. [Online]. Available: https:
1780, 2004. //arxiv.org/abs/2102.06772v1
[14] J. Sun, F. Hu, and S. Lucyszyn, “Predicting atmospheric attenuation [36] F. Gao, B. Wang, C. Xing, J. An, and G. Y. Li, “Wideband Beamforming
under pristine conditions between 0.1 and 100 THz,” IEEE Access, for Hybrid Massive MIMO Terahertz Communications,” IEEE J. Sel.
vol. 4, pp. 9377–9399, 2016. Areas Commun., vol. 39, no. 6, pp. 1725–1740, Apr 2021.
[15] R. Piesiewicz, C. Jansen, D. Mittleman, T. Kleine-Ostmann, M. Koch, [37] B. Wang, F. Gao, S. Jin, H. Lin, G. Y. Li, S. Sun, and T. S. Rappa-
and T. Kurner, “Scattering analysis for the modeling of THz commu- port, “Spatial-Wideband Effect in Massive MIMO with Application in
nication systems,” IEEE Trans. Antennas Propag., vol. 55, no. 11, pp. mmWave Systems,” IEEE Commun. Mag., vol. 56, no. 12, pp. 134–141,
3002–3009, 2007. Aug 2018.
[16] Z. Xu and T. Liu, “Vital sign sensing method based on EMD in terahertz [38] M. Wang, F. Gao, N. Shlezinger, M. F. Flanagan, and Y. C. Eldar, “A
band,” EURASIP Journal on Advances in Signal Processing, vol. 2014, Block Sparsity Based Estimator for mmWave Massive MIMO Channels
no. 1, pp. 1–8, 2014. With Beam Squint,” IEEE Trans. Signal Process., vol. 68, pp. 49–64,
[17] G. Ok, K. Park, H. J. Kim, H. S. Chun, and S.-W. Choi, “High- Nov 2019.
speed terahertz imaging toward food quality inspection,” Applied Optics, [39] H. Hashemi, T.-s. Chu, and J. Roderick, “Integrated true-time-delay-
vol. 53, no. 7, pp. 1406–1412, 2014. based ultra-wideband array processing,” IEEE Commun. Mag., vol. 46,
[18] R. Appleby and H. B. Wallace, “Standoff detection of weapons and no. 9, pp. 162–172, Sep 2008.
contraband in the 100 GHz to 1 THz region,” IEEE Trans. Antennas [40] J. Tan and L. Dai, “Delay-Phase Precoding for THz Massive MIMO
Propag., vol. 55, no. 11, pp. 2944–2956, 2007. with Beam Split,” in 2019 IEEE Global Communications Conference
[19] F. Norouzian, E. Marchetti, M. Gashinova, E. Hoare, C. Constantinou, (GLOBECOM). IEEE, Dec 2019, pp. 1–6.
P. Gardner, and M. Cherniakov, “Rain attenuation at millimeter wave and [41] ——, “Wideband Beam Tracking in THz Massive MIMO Systems,”
low-THz frequencies,” IEEE Trans. Antennas Propag., vol. 68, no. 1, IEEE J. Sel. Areas Commun., vol. 39, no. 6, pp. 1693–1710, Apr 2021.
pp. 421–431, 2019. [42] L. Dai, J. Tan, and H. V. Poor, “Delay-Phase Precoding for
[20] F. Norouzian, E. Marchetti, E. Hoare, M. Gashinova, C. Constantinou, Wideband THz Massive MIMO,” arXiv, Feb 2021. [Online]. Available:
P. Gardner, and M. Cherniakov, “Experimental study on low-THz https://arxiv.org/abs/2102.05211v1
automotive radar signal attenuation during snowfall,” IET Radar, Sonar [43] X. Yu, J. Shen, J. Zhang, and K. B. Letaief, “Alternating Minimization
& Navigation, vol. 13, no. 9, pp. 1421–1427, 2019. Algorithms for Hybrid Precoding in Millimeter Wave MIMO Systems,”
[21] R. Appleby and R. N. Anderton, “Millimeter-wave and submillimeter- IEEE J. Sel. Topics Signal Process., vol. 10, no. 3, pp. 485–500, April
wave imaging for security and surveillance,” Proc. IEEE, vol. 95, no. 8, 2016.
pp. 1683–1690, 2007. [44] C. Han, L. Yan, and J. Yuan, “Hybrid Beamforming for Terahertz Wire-
[22] H. Sarieddeen, M.-S. Alouini, and T. Y. Al-Naffouri, “An overview of less Communications: Challenges, Architectures, and Open Problems,”
signal processing techniques for terahertz communications,” arXiv, May arXiv, Jan 2021.
2020. [45] S. A. Busari, K. M. S. Huq, S. Mumtaz, J. Rodriguez, Y. Fang,
[23] L. Ju, B. Geng, J. Horng, C. Girit, M. Martin, Z. Hao, H. A. Bechtel, D. C. Sicker, S. Al-Rubaye, and A. Tsourdos, “Generalized hybrid
X. Liang, A. Zettl, Y. R. Shen et al., “Graphene plasmonics for tunable beamforming for vehicular connectivity using THz massive MIMO,”
terahertz metamaterials,” Nature nanotechnology, vol. 6, no. 10, pp. 630– IEEE Trans. Veh. Technol., vol. 68, no. 9, pp. 8372–8383, 2019.
634, 2011. [46] H. Yuan, N. Yang, K. Yang, C. Han, and J. An, “Hybrid beamforming for
[24] Z. Xu, X. Dong, and J. Bornemann, “Design of a reconfigurable MIMO terahertz multi-carrier systems over frequency selective fading,” IEEE
system for THz communications based on graphene antennas,” IEEE Trans. Commun., vol. 68, no. 10, pp. 6186–6199, 2020.
Trans. THz Sci. Technol., vol. 4, no. 5, pp. 609–617, 2014. [47] B. Ning, Z. Chen, W. Chen, Y. Du, and J. Fang, “Terahertz multi-user
[25] M. Moccia, C. Koral, G. P. Papari, S. Liu, L. Zhang, R. Y. Wu, massive MIMO with intelligent reflecting surface: Beam training and
G. Castaldi, T. J. Cui, V. Galdi, and A. Andreone, “Suboptimal coding hybrid beamforming,” IEEE Trans. Veh. Technol., 2021, in press.
metasurfaces for terahertz diffuse scattering,” Scientific reports, vol. 8, [48] B. Ning, Z. Chen, W. Chen, Y. Du, and J. Fang, “Terahertz Multi-User
no. 1, pp. 1–9, 2018. Massive MIMO With Intelligent Reflecting Surface: Beam Training and
[26] H. Sarieddeen, M.-S. Alouini, and T. Y. Al-Naffouri, “Terahertz-band Hybrid Beamforming,” IEEE Trans. Veh. Technol., vol. 70, no. 2, pp.
ultra-massive spatial modulation MIMO,” IEEE J. Sel. Areas Commun., 1376–1393, Jan 2021.
vol. 37, no. 9, pp. 2040–2052, 2019. [49] C. Han, J. M. Jornet, and I. Akyildiz, “Ultra-Massive MIMO Chan-
[27] C. Chaccour, M. N. Soorki, W. Saad, M. Bennis, P. Popovski, and nel Modeling for Graphene-Enabled Terahertz-Band Communications,”
M. Debbah, “Seven defining features of terahertz (THz) wireless sys- 2018 IEEE 87th Vehicular Technology Conference (VTC Spring), pp.
tems: A fellowship of communication and sensing,” arXiv preprint 1–5, Jun 2018.
arXiv:2102.07668, 2021. [50] Z. Hossain and J. M. Jornet, “Hierarchical bandwidth modulation
[28] G. Duggal, S. Vishwakarma, K. V. Mishra, and S. S. Ram, “Doppler- for ultra-broadband terahertz communications,” in IEEE International
resilient 802.11ad-based ultrashort range automotive joint radar- Conference on Communications, 2019, pp. 1–7.
communications system,” IEEE Trans. Aerosp. Electron. Syst., vol. 56, [51] C. Han, L. Yan, and J. Yuan, “Hybrid beamforming for terahertz wireless
no. 5, pp. 4035–4048, 2020. communications: Challenges, architectures, and open problems,” arXiv
[29] S. H. Dokhanchi, M. R. Bhavani Shankar, K. V. Mishra, and B. Ot- preprint arXiv:2101.08469, 2021.
tersten, “Multi-constraint spectral co-design for colocated MIMO radar [52] A. M. Elbir and K. V. Mishra, “Joint antenna selection and hybrid
and MIMO communications,” in IEEE International Conference on beamformer design using unquantized and quantized deep learning
Acoustics, Speech and Signal Processing, 2020, pp. 4567–4571. networks,” IEEE Trans. Wireless Commun., vol. 19, no. 3, pp. 1677–
[30] J. Liu, K. V. Mishra, and M. Saquib, “Co-designing statistical MIMO 1688, March 2020.
radar and in-band full-duplex multi-user MIMO communications,” arXiv [53] A. M. Elbir, K. V. Mishra, M. R. B. Shankar, and B. Ottersten, “A Family
preprint arXiv:2006.14774, 2020. of Deep Learning Architectures for Channel Estimation and Hybrid
[31] S. Tarboush, H. Sarieddeen, H. Chen, M. H. Loukil, H. Jemaa, M. S. Beamforming in Multi-Carrier mm-Wave Massive MIMO,” arXiv, Dec
Alouini, and T. Y. Al-Naffouri, “TeraMIMO: A Channel Simulator for 2019.
17
[54] Y. Xing and T. S. Rappaport, “Propagation measurement system and [65] F. Liu and C. Masouros, “Hybrid Beamforming with Sub-arrayed MIMO
approach at 140 GHz - Moving to 6G and above 100 GHz,” in IEEE Radar: Enabling Joint Sensing and Communication at mmWave Band,”
Global Communications Conference, 2018, pp. 1–6. ICASSP 2019 - 2019 IEEE International Conference on Acoustics,
[55] S. Ju and T. S. Rappaport, “140 GHz urban microcell propaga- Speech and Signal Processing (ICASSP), pp. 7770–7774, Dec 2017.
tion measurements for spatial consistency modeling,” arXiv preprint [66] A. M. Elbir, “A Deep Learning Framework for Hybrid Beamforming
arXiv:2103.05496, 2021. Without Instantaneous CSI Feedback,” IEEE Trans. Veh. Technol., pp.
[56] K. M. S. Huq, S. A. Busari, J. Rodriguez, V. Frascolla, W. Bazzi, and 1–1, 2020.
D. C. Sicker, “Terahertz-Enabled Wireless System for Beyond-5G Ultra- [67] A. Brighente, M. Cerutti, M. Nicoli, S. Tomasin, and U. Spagnolini,
Fast Networks: A Brief Survey,” IEEE Network, vol. 33, no. 4, pp. 89– “Estimation of Wideband Dynamic mmWave and THz Channels for 5G
95, Jul 2019. Systems and Beyond,” IEEE J. Sel. Areas Commun., vol. 38, no. 9, pp.
[57] Y. Lu and L. Dai, “Reconfigurable Intelligent Surface Based Hybrid 2026–2040, Jun 2020.
Precoding for THz Communications,” arXiv, Dec 2020. [68] A. M. Elbir and S. Coleri, “Federated Learning for Channel Estimation
[58] A. Hassanien and S. A. Vorobyov, “Phased-MIMO Radar: A Tradeoff in Conventional and IRS-Assisted Massive MIMO,” arXiv preprint
Between Phased-Array and MIMO Radars,” IEEE Trans. Signal Pro- arXiv:2008.10846, 2020.
cess., vol. 58, no. 6, pp. 3137–3151, Feb 2010. [69] H. Xie, F. Gao, S. Jin, J. Fang, and Y. Liang, “Channel estimation
[59] N. Song, T. Yang, and H. Sun, “Overlapped Subarray Based Hybrid for tdd/fdd massive mimo systems with channel covariance computing,”
Beamforming for Millimeter Wave Multiuser Massive MIMO,” IEEE IEEE Trans. Wireless Commun., vol. 17, no. 6, pp. 4206–4218, June
Signal Process. Lett., vol. 24, no. 5, pp. 550–554, May 2017. 2018.
[60] F. Liu, C. Masouros, A. P. Petropulu, H. Griffiths, and L. Hanzo, “Joint [70] W. Xu, F. Gao, J. Zhang, X. Tao, and A. Alkhateeb, “Deep Learning
Radar and Communication Design: Applications, State-of-the-Art, and Based Channel Covariance Matrix Estimation with User Location and
the Road Ahead,” IEEE Trans. Commun., vol. 68, no. 6, pp. 3834–3862, Scene Images,” arXiv, Jan 2021.
Jun 2020. [71] N. Boumal, B. Mishra, P.-A. Absil, and R. Sepulchre, “Manopt, a Matlab
[61] A. Kaushik, J. Thompson, E. Vlachos, C. Tsinos, and S. Chatzinotas, Toolbox for Optimization on Manifolds,” Journal of Machine Learning
“Dynamic RF Chain Selection for Energy Efficient and Low Complexity Research, vol. 15, pp. 1455–1459, 2014.
Hybrid Beamforming in Millimeter Wave MIMO Systems,” IEEE Trans. [72] S. Park, J. Park, A. Yazdan, and R. W. Heath, “Exploiting spatial
Green Commun. Networking, vol. 3, no. 4, pp. 886–900, Jul 2019. channel covariance for hybrid precoding in massive MIMO systems,”
[62] T. Lin, J. Cong, Y. Zhu, J. Zhang, and K. B. Letaief, “Hybrid Beamform- IEEE Trans. Signal Process., vol. 65, no. 14, pp. 3818–3832, 2017.
ing for Millimeter Wave Systems Using the MMSE Criterion,” IEEE [73] Y. Xiao, F. Norouzian, E. G. Hoare, E. Marchetti, M. Gashinova, and
Trans. Commun., vol. 67, no. 5, pp. 3693–3708, Jan 2019. M. Cherniakov, “Modeling and Experiment Verification of Transmis-
[63] R. Schmidt, “Multiple emitter location and signal parameter estimation,” sivity of Low-THz Radar Signal Through Vehicle Infrastructure,” IEEE
IEEE Trans. Antennas Propag., vol. 34, no. 3, pp. 276–280, 1986. Sens. J., vol. 20, no. 15, pp. 8483–8496, Mar 2020.
[64] A. M. Elbir, “DeepMUSIC: Multiple Signal Classification via Deep [74] A. M. Elbir and K. V. Mishra, “Sparse Array Selection Across Arbitrary
Learning,” IEEE Sensors Letters, vol. 4, no. 4, pp. 1–4, 2020. Sensor Geometries With Deep Transfer Learning,” IEEE Trans. Cognit.
Commun. Networking, vol. 7, no. 1, pp. 255–264, Jun 2020.