ADAPTIVE RANDOM SPATIAL BASED CHANNEL ESTIMATION (ARSCE) FOR MILLIMETER WAVE MIMO SYSTEM

International Journal of Computer Networks & Communications (IJCNC) Vol.14, No.1, January 2022
DOI: 10.5121/ijcnc.2022.14106 85
ADAPTIVE RANDOM SPATIAL BASED CHANNEL
ESTIMATION (ARSCE) FOR MILLIMETER WAVE
MIMO SYSTEM
Rajkumar M Vadgave1
, Manjula S2
and T S Vishwanath3
1
Department of Electronics & Communication Engineering,
Veerappa Nisty Engineering College, Shorapur, India
2
Lingaraj Appa Engineering College, Bidar, India
3
Bheemanna Khandre Institute of Technology, Bhalki, India
ABSTRACT
Millimeter-wave and mMIMO communications are the most essential success systems for next-generation
wireless sensor networks to have enormous amounts of accessible throughput and spectrum. Through
installing huge antenna arrays at the base station and performing coherent transceiver processing,
mMIMO is a potential technology for enhancing the bandwidth efficiency of wireless sensor networks. The
use of mmWave frequencies for mMIMO systems solves the problem of high path-loss through offering
greater antenna gains. In this work, we provide a design with a random spatial sample structure that
incorporates a totally random step before the analogue is received. It contains a totally random step before
the analogue received signals are sent into the digital component of the HBF receiver. Adaptive random
spatial based channel estimation (ARSCE) is proposed for channel session measurement collection, and an
analogue combiner with valves has been used to estimate the signals at each receiving antenna. The
proposed optimization problem formulation attempts to discover the orientations and gains of wideband
channel routes. In addition, our proposed model has compared to various state-of-art techniques while
considering error minimization.
KEYWORDS
Millimeter-wave (mmWave), massive multi-input–multi-output (mMIMO), Hybrid Beamforming (HBF),
Channel Estimation, Error Minimization.
1. INTRODUCTION
Modern broadband cellular systems will have to keep up with the demand for high-speed data
capabilities, which is expected to skyrocket. Two major ways to meet these needs are large-
antenna-array multiple-input-multiple-output (MIMO) designs (also known as massive MIMO /
m-MIMO) and millimeter wave (mmWave) systems. The mMIMO is preferred when the
objective is to enhance spectral and energy efficiency, whereas the mmWave approach
maximises a broader bandwidth [1]. Many contemporary commercial mobile network systems
operate at frequency band below 6 GHz, whereas millimeter wave communications utilise the
spectrum from roughly 30 GHz to 300 GHz. The major benefit of owning at higher frequencies is
the possibility of significantly greater bandwidth channels [2].

86
The mmWave communications potential for cellular networks has been acknowledged in [3], as
well as its ability to offer extremely high throughput. Even in non-line-of-sight conditions, [4] [5]
claimed a transmission distance of far more than 200 m. For 5th generation (5G) cellular
networks and beyond, millimeter wave communications are of interest [6]. Modern signal
processing approach are required to make mmWave communication systems commercially viable
[4], [6]. There are several reasons behind this: new hardware restraints develop as a result of
operating at larger bandwidths and higher frequencies and, mmWave channel characteristics
change. Significant antenna arrays are required to achieve large beamforming gains, which brings
additional power consumption and circuit implementation issues.
Conventional digital precoding is too expensive to implement in millimeter-wave communication
systems. As a result, hybrid precoding is a superior option that works at mixes of digital and
analogue precoding. It's generally based on one of two types of structures: subconnected and
completely linked, whereas the completely linked structure has been intensively explored in the
academic world because it can reach the theoretical ideal spectrum efficiency. In [4], the spatial
sparsity of the mmWave channel was used to reconstruct the hybrid precoding problem. The
orthogonal matching pursuit (OMP) method was suggested to tackle the problem and greedy
algorithm is presented based on the OMP method in [7].
The antenna array shape is not taken into account by the method, which substantially decreases
computing complexity while maintaining high performance. By choosing an appropriate
threshold, an orthogonal matching pursuit method with performance similar to the optimum and
higher than the OMP algorithm is provided [5]. Even though OMP approach has good speed, it
has a rather high level of complexity, so as a result, [6] investigates a high-performance and real-
time precoding approach based on SVD (singular value decomposition). Overall performance of
this technique is comparable to the OMP algorithm, but the complexity is significantly lower. An
alternative minimization method based on the concept of manifold optimization is proposed in
[8], which reduces the program's complexity. Many researchers have been drawn to the
subconnected structure because of its features, which allow it to strike a proper balance between
cost and performance. However, unlike fully connected structure, which has a one-to-one
correlation between both the antenna and the RF chain, each RF chain in the subconnected
configuration is only linked to a portion of the antenna. As a result, in recent years, hybrid
precoding techniques using subconnection architectures has gotten a lot of interest.
With the aid of semidefinite relaxation, an alternative minimization method for the subconnected
structure is also constructed in [8]. In a similar vein, [9] proposes a novel divide-and-conquer
precoding method, this scheme's performance is comparable to SDR- alternative minimization,
that can save time and is resistant to potential saddle spots with underperformance. In [10], a
hybrid precoding method based on SIC (successive interference cancellation) is presented in
response to the hybrid precoding challenge of energy-saving subconnection design. It eliminates
the need for matrix inversion and SVD and it has a considerably low computational cost than
standard sparse reconstruction precoding techniques. The sparsity and low rank characteristics of
specific wireless channels were recently combined for effective channel state information (CSI)
estimation in [11] [13]. In [12], a two-step approach (one stage per channel attribute) was
developed for narrowband mmWave MIMO channel estimation with Hybrid Beamforming
(HBF) transceivers. In [13] proposed a matrix completion method that took advantage of these
features for the same transceivers and narrowband channels with simple switches. This approach
outperformed the approach of [12] in terms of estimate performance while needing just brief
channel sounding periods. In [14], wideband mmWave MIMO channels featuring frequency
selectivity were recently introduced, which offered a CSI estimation approach that used the
sparsity of the channel in frequency and time domains.

87
In this paper, we provide a unique analogue combining design with a random spatial sampling
layout, which includes a purely random stage before the analogue received signals are fed into
the HBF receiver's digital component. This design is utilized for channel session measurement
gathering, and an analogue combiner with valves was used to approximate the signals at
receiving antennas individually. The proposed model named as adaptive random spatial based
channel estimation (ARSCE) and this design major feature is the random choice of available
analogue receive beams during channel period. It combined with the proposed CSI estimation
approach, results in enhanced performance with short training periods. The matrix generated at
the HBF receiver following analogue processing of training signals containing multiple receiving
beams has the same rank as the wideband mmWave m-MIMO channel matrix. Based on this
discovery, we devise a CSI estimate optimization model that takes use of both the low-rank
feature of the sparsity of the channel matrix and the received training signal matrix in the
beamspace space. The suggested optimization problem formulation seeks to identify the
wideband channel routes' orientations and gains. The distinctions are in the mathematical
formulas. The low-rank feature of narrowband mmWave m-MIMO channels is used to determine
the measurement needs for them and the scalability of the technique in wideband mmWave m-
MIMO channels.
2. LITERATURE SURVEY
In m-MIMO systems, hybrid beamforming designs are developed to reduce hardware and
training overhead costs. The carrier frequency (mmWave), CSI (average or instantaneous), and
complexity of hybrid beamforming may all be categorized to reduce the switched complexity.
The method to use to achieve the optimal tradeoff between these factors is determined on the
application and channel characteristics [15], while obtaining specific and reliable CSI for Multi
user-mMIMO systems at millimeter waves is difficult because to the large number of BS
antennas. In mMIMO, a combined iterative approach based on step-length optimization may be
used to obtain accurate channel estimate [16]. In the context of faulty CSI and hardware
restrictions, a beamforming neural network (deep learning networks) for mmWavemMIMO
systems can improve the beamforming layout and is resilient with greater spectrum efficiency as
compared to standard beamforming methods [17]. In mMIMO systems, the ideal number of UEs
programmed concurrently for a specific number of BS antenna components offers maximum
spectrum efficiency. The frequency spectrum of downlink (DL) and uplink (UL) can be the same
(allowing combined network improvement), and it is also independent of UE locations at any
given time [18]. In comparison to existing methods, the multitask deep learning (MTDL)-based
HBF algorithm for mmWavemMIMO OFDM systems can provide good results in terms of run-
time and sum-rate [19].
Based on the arrival/departure angle and with the fewest number of pilots, an effective channel
estimate technique for the time-varying DL channel of mmWave MIMO systems is suggested
[20]. With flawless CSI and a low-complex single-cell DL mMIMO HBF system, the sum-rate
approaches optimal network throughput is shown in [21]. The performance difference between
hybrid and digital beamforming can be narrowed for a specific number of Relay nodes by
decrease the quantity of multiplexed symbols [22]. In comparison to OMA systems, clustering
and feedback-based hybrid beamforming for DL mmWave MIMO NOMA systems provide
highest sum-rates [23]. A multi user (MU) detection technique based on a ‘‘markov random
field" for model cluster sparsity and predict mMIMO channel outperforms systems which do not
take use of channel cluster sparsity [24]. In developing hybrid beamforming for broadband
mmWave MIMO systems, eigenvalue decomposition, manifold optimization, and OMP
algorithms give BER and spectral efficiency (SE) that is closer to completely digital BF solutions
[25]. When the number of RF chains in a mmWavemMIMO system increases, the best hybrid

88
combining and precoding approaches utilising the OMP algorithm to achieve total sum-rates that
are comparable to completely digital precoding [26].
Hybrid beamforming with user scheduling technique is described for DL to increase throughput
(near to completely digital beamforming) of mmWavemMIMO OFDM systems (users with
similar strongest beams). Digital beamforming is used to achieve the highest performance
increase (by minimising excess inter-user congestion) [27] and analogue beamforming vectors
are utilised to determine the ideal beam of each consumer. For mmWave MIMO uplink channels,
a low computational complexity hybrid beamforming system minimises inter-user interference
while performing similarly to a completely digital beamforming architecture. In
mmWavemMIMO systems based on Generalized Low Rank Approximation of Matrices, the
coordinated RF beamforming method requires just composite CSI rather than the entire physical
channel matrix.
In both FDD and TDD systems, this method gives a competitive option by addressing the
coordination between BS and UEs to get maximum array gain with no dimensionality limitation
[22]. HBF is used in mmWavemMIMO relay systems to improve sum-rates via lowering the sum
MSE between both the received signal of hybrid and digital BF schemes. The overall sum-rates
rise as the correctness of arrival/departure angles and the number of RF chains rise [28] and the
possibility of low-cost analogue optimal unconstrained precoder, RF hardware implementations,
and combiner methods for mmWavemMIMO systems are proposed. The suggested methods'
numerical findings demonstrated that the SE of mmWave systems with transceiver hardware
restrictions approaches the unrestrained performance limits [29]. In mmWavemMIMO systems,
low-complexity phased-ZF hybrid precoding is utilised in the RF domain to get substantial power
gains and low-dimensional ZF precoding has been used in the baseband domain allowing multi-
stream operations [30].
3. PROPOSED METHOD
A point-to-point large MIMO communication network functioning across wideband
mmWave channels is considered, where transmitter is having antennas and the receiver is
having antennas. Here, we assume that each of the antenna components of the transmitter are
linked to a separate RF chain, whereas the receiver antenna elements are associated to that is
less then at RF chains. Because of this hardware design, the transmitter may digitally precode
up to separate signals from a single devoted RF link. We assume, on the other hand, that
receiver is prepared with any of the known HBF designs [31] that enable both digital and
analogue compounding. The considered mmWave MIMO network comprising wireless
communication link of information that will be separate from data streams.
Initially, proposed the solution for large number of antennas by analog based HBF receiver at
wideband MIMO channel. The assumption has been made where the communication system
remains consistent throughout each frame but may change independently by one frame to the
next. Each frame is made up of blocks allocated to channel estimates, with the rest of the
frames allocated to data transmission. A greater , of course, improves channel prediction while
leaving less frame duration for actual data transfer. The -antenna transmitter uses the
training symbols' vector with each block with to approximate the desired
wideband mmWave MIMO channel. Ignoring the influence of Additive White Gaussian Noise
(AWGN) for the sake of clarity, the -dimensional acquired training signal may be represented
as:

89
(1)
This shows the convolution matrices of delay channel and training vectors of is
. similar to eq. (1), is shown as;
(2)
where indicates the -th column of and the h-th element of . We may
describe the inner convolution sum of the Toeplitz matrix by altering the summation order
in eq. (2). Specifically, by include the Toeplitz matrix , with its -th element
provided by:
(3)
Where, and , and above eq. (2) can be written as
(4)
Where, . The structure of are also
reorganized to grouped at transmitting antennas and equation eq. (4) can be written as;
(5)
Where, and . In addition,
the decomposition of beamspace at every delay path matrix is;
(6)
Where, and for the case of the wideband channel matrix , the
decomposition of eq. (6) is identical to the beamspace decomposition of eq. (7). When all of the
above information is combined, the matrix containing the received training symbols is provided
by;
(7)
The difference between eq. (6) and eq. (7) is the rightmost matrix , which includes the learning
symbols. In our method, the latter equation will be utilised to represent the incoming training data
in terms of the combined virtual channel gains .

90
We provide the proposed channel estimation process issue description and comprehensive
algorithmic solution. The sum of the low-rank matrix, which includes the learning signs
following passing over the wideband mmWave MIMO link, and the AWGN matrix yields the
received learning signal matrix at the outcome of the extended connector. In this paper, we using
the low-rank feature of to estimate the wideband channel matrix . To begin, the rank
characteristics of the later matrix is considered for the sake of simplicity.
We design the following dual purpose optimization process for the estimate of , taking use of
both the sparse composition of and the low rank characteristic of the learning symbols' matrix
:
(8)
Which subject to,
(9)
Where D's fission norm in the optimization process that imposes its low rank characteristic, and
-norm enforces its sparse form. The weighted parameters are often determined by
the number of unique mmWave MIMO channel propagation pathways. Because the noisy
matrix is uncertain, we replace the first requirement in eq. (8) with its least-squares
approximation in the follows.
The ADMM method is used to solve the process eq. (8) optimally and the sampling is done on
the channel matrix rather than the received learning signal, as indicated in the mathematical
equation eq. (8). To solve eq. (8), we will do the following; restate the desired proces in the
following equivalent form, we first incorporate the two input matrix variables
and :
(10)
This is equal to eq. (8), except the cost function has now been dissected into four variables:
. It is worth noting that the third component with in objective function now accounts
for the discretization error, while the fourth term accounts for the AWGN interference noise. The
optimization process of Lagrangian function readily written as:
(11)
where and are dual factors (the Lagrange multipliers) that add the
requirements of eq. (10) to the cost function, and indicates the ADMM stepsize. The following
distinct sub-problems must be addressed at the i-th algorithmic step, with , as per
the usual ADMM approach:
(12)

91
(13)
Because the first subproblem is concerned with optimization over through the variable , now let
express by retaining just the words that are relevant to it and rounding the square, i.e.,
(14)
The answer to eq. (14) is believed to be derived using the Singular Value Thresholding (SVT)
function as follows [32]:
(15)
where and contain the right and left singular vectors, of the matrix ( , and
with singular values denoted by . The set of equations must be solved, which
necessitates the inversion of the matrix. However, because this matrix is diagonal, the resultant
issue is identical to solving the mathematical model shown below;
(16)
Concerning the unidentified variable , that may be written correspondingly as; minimization of
by vectorization is similar to the accompanying sparse optimization issue. To address the
problem, we use the usual LASSO form [33], i.e.
(17)
Where, and a soft-thresholding function may be used to approximate in (17),
which is written as follows:
(18)
where and the sign operators and are used component-wise and it is
worth noting that the superscript will be employed in the suggested evolutionary
algorithm. The vector obtained in eq. (18) is then converted into parametric form as
.
4. RESULTS AND ANALYSIS
In this part, we explore a mmWave MIMO technology for various huge
and values, and we evaluate the performance of the suggested wideband channel estimation
approach for HBF transmission utilising the adaptive random spatial combining design.
MATLAB 2016b with system configuration intel i7 processor, 16GB RAM and windows 10
operating system was used to generate all computational simulation results. We particularly
simulated the Average Mean Square Error (MSE) performances, which were then averaged
across several Monte-Carlo realizations. We also examined the suggested approach's
convergence rate and computing expense in comparison to similar state-of-the-art approaches.

92
The channel estimation has been conducted in the time domain on receiver across F frames, using
block transmissions with thresholding applied to each sent frame. We expected that the
convergence rate of the channels is greater than the period of the F sessions necessary for channel
estimation and the learning signal is received by the HBF RX's antenna components, which
are linked to MR RF chains. Furthermore, we assumed frequency synchronization
and complete time between the transmitting and receiver sides, as well as fully linked proposed
systems at all HBF receiver under consideration. We simulated the Normalized Mean-Square-
Error (NMSE) criterion when evaluating the performance of the proposed algorithm and making
comparisons them to relevant state-of-the-art wideband mmWave MIMO channel estimation
approaches, that is defined as:
(19)
where symbolises the approximate channel matrix in the beamspace domain. The
benchmarking channel estimation approaches under consideration are based on matrix
completion and compressive sensing tools. NMSE, is analyzed to avoid bias towards model
overestimate / underestimate and giving an impression of the model performance at entire
network.
The singular value thresholding (SVT) [32] method is used to solve the rank constrained
optimization problem. Although iterative thresholding algorithms, such as Orthogonal Matching
Pursuit (OMP) [14], can efficiently solve the L1 minimization with low computational burden,
they only deliver excellent estimation accuracy for highly sparse vectors, i.e., when the
unknowable vector only has the few non-zero qualities. Message passing techniques, such as
Vector Approximate Message Passing (VAMP) [34], on the other hand, give more robust
estimate performance in terms of signals with reduced sparsity and also measurements with
increasing noise level. These approaches have considered for the comparison with proposed
approach.
Figure 1. NMSE efficacy as a proportion of training frame number

93
Considered parameter details for figure 1; 16 number of transmitting antennas, 32 number of the
receiving antenna, total number of clusters 4, channel realization 6, 4 receive RF chains, 5 to 40
training frames, and -15 SNR with 20 algorithmic iterations. The NMSE efficacy of the proposed
channel estimation approach versus the training frames number F is shown in figure 1, where the
training frame numbers; 5, 15, 25, 35 and 40 were considered for the analysis. At training frame
15, our proposed ARSCE approach has got 15.64% and 14.25% less NMSE as compared to SVT
and OMP approaches. Similarly at 40 training frames, proposed ARSCE approach has achieved
51.83% and 27.24% less NMSE as compared to SVT and OMP approaches. As per the figure 1,
it clearly visible that the NMSE values are decrease while increment in training frame numbers
and our proposed approach error decreasing w.r.t state-of-the-art methods.
Figure 2. NMSE efficacy as a proportion of RF chains number
Figure 3. NMSE efficacy as a proportion of channel delay taps number

94
receiving antenna, total number of clusters 4, channel realization 6, 16 receive RF chains, 5
training frames, and -5 SNR with 20 algorithmic iterations. The proposed channel estimation
approach NMSE efficacy versus the RF chains number is shown in figure 2, were RF chains
number i.e., 4, 8, 12, and 16 considered. At 8 RF chains number, proposed ARSCE approach has
achieved 16.63% and 33.98% less NMSE as compared to OMP and VAMP approaches. The
performance of considered approaches tends to increase while increase in RF chains number.
receiving antenna, 12 channel propagation paths, total number of clusters 4, channel realization 6,
4 receive RF chains, 25 training frames, and -10 SNR with 10 algorithmic iterations.
Figure 4. NMSE efficacy as a proportion of SNR
Figure 5. NMSE efficacy as a proportion of RF chains number

95
The NMSE efficacy for proposed channel estimation approach versus the channel delay taps
number is shown in figure 3, were channel delay taps number i.e., 2, 4, 6, 8, and 10 considered.
At 4 channel delay taps number, SVT, OMP and VAMP approaches has 3.93%, 21.59 and 29.7%
more NMSE as compared to proposed ARSCE approach. The performance is almost stable while
changing the channel delay taps number.
At figure 4 the parameter details; 4 number of transmitting antennas, 32 number of the receiving
antenna, total number of clusters 4, channel realization 100, 4 receive RF chains, 35 training
frames, and -15 to 15 SNR with 100 algorithmic iterations. Figure 4 shows the NMSE efficacy
w.r.t SNR, where SNR varies -15dB to 15dB with 3dB interval. At -15dB SNR, OMP and
VAMP approaches has got >1dB and 0.91dB which is 48.57%, and 43.63% more NMSE as
compared to proposed ARSCE approach. Similarly, at 6dB SNR, OMP and VAMP approaches
has got 0.37dB and 0.17dB which is 73%, and 41% more NMSE as compared to proposed
ARSCE approach. The performance is optimized at increasing the SNR.
Figure 6. NMSE efficacy as a proportion of channel paths number
Parameter details for figure 5; 4 to16 number of transmitting antennas, 32 number of the
receiving antenna, 12 channel propagation paths, total number of clusters 4, channel realization 6,
16 receive RF chains, 5 training frames, and -5dB SNR with 20 algorithmic iterations. At figure 5
shows NMSE w.r.t RF chains number, where at 12 RF chains, OMP approach has 38.2% more
NMSE as compared to proposed ARSCE approach. Parameter details for figure 6; 16 number of
transmitting antennas, 32 number of the receiving antenna, 48 channel propagation paths, total
number of clusters 4, channel realization 1, 4 receive RF chains, 30 training frames, and -5 SNR
with 10 algorithmic iterations. Whereas NMSE w.r.t channel paths number has shown in figure
4.6, SVT approach has shown more NMSE as compared to proposed ARSCE approach at various
channel paths number.

96
5. CONCLUSIONS
In this article, we proposed ARSCE method for wideband mmWave MIMO channel estimates
that take use of both the beamspace sparsity and channel's low rank to enable more accurate
channel restoration, particularly for short beam learning intervals. The received training signal
matrix at the termination of the extended connection is the low-rank matrix sum, that contains the
learning signs after passing across the wideband mmWave MIMO system. Extensive simulation
findings demonstrate that the suggested methods perform better in terms of MSE for channel
estimation when only short beam train lengths are used and while operating in severe noise
circumstances. Several approaches shown in above section has considered for the comparison
with proposed approach, and proposed ARSCE model has performed well at receiving optimized
channel estimation, which result better throughput and better user quality experience.
ACKNOWLEDGEMENTS
The authors would like to thank everyone, just everyone!
CONFLICTS OF INTEREST
The authors declare no conflict of interest.
REFERENCES
[1] F. Boccardi, R. W. Heath Jr., A. Lozano, T. L. Marzetta, and P. Popovski, “Five disruptive
technology directions for 5G,” IEEE Communication Mag., vol. 52, no. 2, pp. 74–80, Feb. 2014.
[2] T. S. Rappaport, G. R. MacCartney, Jr., M. K. Samimi, and S. Sun, “Wide-band millimeter-wave
propagation measurements and channel models for future wireless communication system design,”
Transactions on Wireless Communications, vol. 63, no. 9, pp. 3029–3056, Sept. 2015.
[3] Y. Niu, Y. Li, D. Jin, L. Su, and A. V. Vasilakos, “A survey of millimeter wave communications
(mmWave) for 5G: opportunities and challenges,” Wireless Netw., vol. 21, no. 8, pp. 2657–2676,
Nov. 2015.
[4] T. S. Rappaport, R. W. Heath Jr., R. C. Daniels, and J. N. Murdock, Millimeter Wave Wireless
Communications. Upper Saddle River, NJ: Prentice Hall, 2014.
[5] M. Samimi, K. Wang, Y. Azar, G. N. Wong, R. Mayzus, H. Zhao, J. K. Schulz, S. Sun, F. Gutierrez,
Jr., and T. S. Rappaport, “28 GHz angle of arrival and angle of departure analysis for outdoor cellular
communications using steerable beam antennas in New York City,” in Proc. IEEE 77th Veh. Technol.
Conf. (VTC 2013-Spring), Dresden, Germany, Jun. 2013, pp. 1–6.
[6] R. W. Heath Jr., N. Gonzalez-Prelcic, S. Rangan, W. Roh, and A. M. ´ Sayeed, “An overview of
signal processing techniques for millimeter wave MIMO systems,” IEEE J. Sel. Topics Signal
Process., vol. 10, no. 3, pp. 436–453, Apr. 2016.
[7] Y. Azar, G. N. Wong, K. Wang, R. Mayzus, J. K. Schulz, H. Zhao, F. Gutierrez, Jr., D. Hwang, and
T. S. Rappaport, “28 GHz propagation measurements for outdoor cellular communications using
steerable beam antennas in New York City,” in Proc. IEEE Int. Conf. Commun. (ICC), Budapest,
Hungary, Jun. 2013, pp. 5143–5147.
[8] L. Lu, G. Y. Li, A. L. Swindlehurst, A. Ashikhmin, and R. Zhang, “An overview of massive MIMO:
Benefits and challenges,” IEEE J. Sel. Topics Signal Processing, vol. 8, no. 5, pp. 742–758, Oct.
2014.
[9] C. B. Peel, B. M. Hochwald, and A. L. Swindlehurst, “A vector-perturbation technique for near-
capacity multiantenna multiuser communication—Part I: Channel inversion and regularization,”
IEEE Transactions on Wireless Communications, vol. 53, no. 1, pp. 195–202, Jan. 2005.
[10] Q. H. Spencer, A. L. Swindlehurst, and M. Haardt, “Zero-forcing methods for downlink spatial
multiplexing in multiuser MIMO channels,” IEEE Transactions Signal Process., vol. 52, no. 2, pp.
461–471, Feb. 2004

97
[11] D. Lee, S.-J. Kim, and G. B. Giannakis, “Channel gain cartography for cognitive radios leveraging
low rank and sparsity,” IEEE Transactions on Wireless Communications, vol. 16, no. 9, pp. 5953–
5966, Sep. 2017.
[12] X. Li, J. Fang, H. Li, and P. Wang, “Millimeter wave channel estimation via exploiting joint sparse
and low-rank structures,” IEEE Transactions on Wireless Communications, vol. 17, no. 2, pp. 1123–
1133, Feb. 2018.
[13] E. Vlachos, G. C. Alexandropoulos, and J. Thompson, “Massive MIMO channel estimation for
millimeter wave systems via matrix completion,” IEEE Signal Process. Lett., vol. 25, no. 11, pp.
1675–1679, Nov. 2018.
[14] K. Venugopal, A. Alkhateeb, N. Gonzalez-Prelcic, and R. W. Heath, Jr., ´ “Channel estimation for
hybrid architecture-based wideband millimeter wave systems,” IEEE J. Sel. Areas Commun., vol. 35,
no. 9, pp. 1996– 2009, Sep. 2017.
[15] Molisch, A. F., Ratnam, V. V., Han, S., Li, Z., Nguyen, S. L. H., Li, L., & Haneda, K. (2017). Hybrid
beamforming for massive MIMO: A survey. IEEE Communications Magazine, 55(9), 134–141.
[16] J. Du, J. Li, J. He, Y. Guan, & H. Lin, “Low-complexity joint channel estimation for multi-user
mmWave massive MIMO systems.” Electronics, 9, 301. 2020
[17] T. Lin, Y. Zhu, “Beamforming design for large-scale antenna arrays using deep learning. IEEE
Wireless Communications Letters, Volume 9, Issue 1, 2020.
[18] Bjornson, E., Larsson, E. G., & Debbah, M. “Massive MIMO for maximal spectral efficiency: How
many users and pilots should be allocated?” IEEE Transactions on Wireless Communications, 15(2),
1293–1308. 2016.
[19] J. Jiang, Y. Li, & C. Li, “Multitask deep learning-based multiuser hybrid beamforming for mm-wave
orthogonal frequency division multiple access systems.” Science China Information Sciences, 63,
180303:1-180303:11. 202
[20] Q. Qin, L. Gui, P. Cheng, & B. Gong, “Time-varying channel estimation for millimeter wave
multiuser MIMO systems.” IEEE Transactions on Vehicular Technology, 67(10), 9435–9448. 2018.
[21] X. Wu, D. Liu and F. Yin, "Hybrid Beamforming for Multi-User Massive MIMO Systems," in IEEE
Transactions on Communications, vol. 66, no. 9, pp. 3879-3891, Sept. 2018.
[22] T. E. Bogale and L. B. Le, "Beamforming for multiuser massive MIMO systems: Digital versus
hybrid analog-digital," 2014 IEEE Global Communications Conference, 2014, pp. 4066-4071.
[23] J. Jiang, M. Lei, & H. Hou, “Downlink multiuser hybrid beamforming for MmWave massive MIMO-
NOMA system with imperfect CSI.” International Journal of Antennas and Propagation, Article ID
9764958, p. 10. 2019.
[24] L. Chen and X. Yuan, "Blind Multiuser Detection in Massive MIMO Channels With Clustered
Sparsity," in IEEE Wireless Communications Letters, vol. 8, no. 4, pp. 1052-1055, Aug. 2019, doi:
10.1109/LWC.2019.2906168.
[25] T. Lin, J. Cong, Y. Zhu, J. Zhang and K. Ben Letaief, "Hybrid Beamforming for Millimeter Wave
Systems Using the MMSE Criterion," in IEEE Transactions on Communications, vol. 67, no. 5, pp.
3693-3708, May 2019.
[26] Y. Zhang, J. Du, Y. Chen, M. Han and X. Li, "Optimal Hybrid Beamforming Design for Millimeter-
Wave Massive Multi-User MIMO Relay Systems," in IEEE Access, vol. 7, pp. 157212-157225, 2019.
[27] J. Jiang, & D. Kong “Joint user scheduling and MU-MIMO hybrid beamforming algorithm for
mmwave FDMA massive MIMO system.” International Journal of Antennas and Propagation,
Article ID 4341068, p 10. 2016.
[28] X. Xue, T. E. Bogale, X. Wang, Y. Wang and B. L. Long, "Hybrid analog-digital beamforming for
multiuser MIMO millimeter wave relay systems," 2015 IEEE/CIC International Conference on
Communications in China (ICCC), 2015, pp. 1-7.
[29] O. E. Ayach, S. Rajagopal, S. Abu-Surra, Z. Pi and R. W. Heath, "Spatially Sparse Precoding in
Millimeter Wave MIMO Systems," in IEEE Transactions on Wireless Communications, vol. 13, no.
3, pp. 1499-1513, March 2014.
[30] L. Liang, W. Xu and X. Dong, "Low-Complexity Hybrid Precoding in Massive Multiuser MIMO
Systems," in IEEE Wireless Communications Letters, vol. 3, no. 6, pp. 653-656, Dec. 2014.
[31] A. F. Molisch, V. V. Ratnam, S. Han, Z. Li, S. L. H. Nguyen, L. Li, and K. Haneda, “Hybrid
beamforming for massive MIMO: A survey,” IEEE Communications Mag., vol. 55, no. 9, pp. 134–
141, Sep. 2017.
[32] J. F. Cai, E. J. Candes, and Z. Shen, “A singular value thresholding algorithm for matrix completion,”
SIAM J. Opt., vol. 20, no. 4, pp. 1956–1982, 2010.

98
[33] R. Tibshirani, “Regression shrinkage and selection via the Lasso,” J. Royal Stat. Society, Series B,
vol. 58, no. 1, pp. 267–288, 1996.
[34] P. Schniter, S. Rangan and A. K. Fletcher, "Vector approximate message passing for the generalized
linear model," 2016 50th Asilomar Conference on Signals, Systems and Computers, 2016, pp. 1525-
1529.
AUTHORS
Rajkumar M Vadgave
Received the BE in Electronics and Communication Engineering from VTU Belagavi
University in 2011 and M.Tech in Communication Systems from VTU Belagavi
University in 2013. He is pursuing Ph.D from VTU Belagavi University.He is currently
working as Assistant Professor in the Dept of E & CE Veerappa Nisty Engineering
College Shorapur from 2013 to till dated. He has authored or coauthored more than 3
refereed journals and conference papers. His research interest includes wireless sensor
network, communication system. He can be contacted at email:
rajkumarv074@gmail.com
Manjula S
Received the B.E in Electronics and Communication Engineering from Gulbarga
University in 1997 and M.Tech in Electronics and Communication Engineering from
RVD University in 2006 and Ph.D degree from Singhania University in 2014. She
became an Assistant professor in 1998, an Associate professor in 2007 and Professor of E
& CE dept in 2014.She has authored or coauthored more than 20 referred journal and
conference papers. Her research interest includes Synthesis characterization and
microwave properties of conducting polymer composites. Member of ISTE, IETE. She
can be contacted at email: manjulasidramshetty@gmail.com
Dr. T.S. Vishwanath
Ph.D in Control System, Institution: Dr MGR Educational and Research Institute
University Chennai India, M. E. in Control System, Institution: WCE Sangli, BE in
Electrical Engineering, Institution: PDA College of Engineering Gulbarga. Research
work at Robust H∞ control and analysis” application to electrical drives. Member of
Institute of Engneer (MIE), ISTE, IETE and IJEET. He can be contacted at email:
tsvrec1@rediffmail.com

ADAPTIVE RANDOM SPATIAL BASED CHANNEL ESTIMATION (ARSCE) FOR MILLIMETER WAVE MIMO SYSTEM

More Related Content

What's hot (18)

Similar to ADAPTIVE RANDOM SPATIAL BASED CHANNEL ESTIMATION (ARSCE) FOR MILLIMETER WAVE MIMO SYSTEM (20)

More from IJCNCJournal (20)

Recently uploaded (20)

ADAPTIVE RANDOM SPATIAL BASED CHANNEL ESTIMATION (ARSCE) FOR MILLIMETER WAVE MIMO SYSTEM