Received: 17 June 2020 Revised: 12 January 2021 Accepted: 6 March 2021 IET Communications

DOI: 10.1049/cmu2.12180

ORIGINAL RESEARCH PAPER

Massive MIMO systems for 5G: A systematic mapping study on

antenna design challenges and channel estimation open issues

Mohamed Benzaghta1 Khaled M. Rabie2

1
Department of Electrical and Electronics Abstract
Engineering, Atilim University, Kizilcasar Mahallesi,
The next generation of mobile networks (5G) is expected to achieve high data rates, reduce
Ankara 06830, Turkey
2
latency, as well as improve the spectral and energy efficiency of wireless communication
School of Engineering, Manchester Metropolitan
University, Manchester, UK
systems. Several technologies are being explored to be used in 5G systems. One of the
main promising technologies that is seen to be the enabler of 5G is massive multiple-
Correspondence input multiple-output (mMIMO) systems. Numerous studies have indicated the utility of
Mohamed Benzaghta, Department of Electrical and mMIMO in upcoming wireless networks. However, there are several challenges that needs
Electronics Engineering, Atilim University, Kizilcasar
to be unravelled. In this paper, the latest progress of research on challenges in mMIMO
Mahallesi, Ankara, Turkey.
Email: [email protected] systems is tracked, in the context of mutual coupling, antenna selection, pilot contamina-
tion and feedback overhead. The results of a systematic mapping study performed on 63
selected primary studies, published between the year 2017 till the second quarter of 2020,
are presented. The main objective of this secondary study is to identify the challenges
regarding antenna design and channel estimation, give an overview on the state-of-the-art
solutions proposed in the literature, and finally, discuss emerging open research issues that
need to be considered before the implementation of mMIMO systems in 5G networks.

1 INTRODUCTION main key for moving toward 5G. In mMIMO systems, the
base station (BS) is deployed by using hundreds of antenna ele-
The next-generation of mobile networks, also known as the ments. The deployment of such massive number of antennas
fifth-generation (5G) is foreseen to deliver a solution for provides significant enhancement to system performance (i.e.,
bandwidth-hungry application, due to the dramatic increase in system capacity, link robustness, energy efficiency and spec-
traffic demand of users. 5G aims to meet much higher data trum efficiency) [3]. For a BS having an array of many active
rates with fast connectivity, more robust reliability, spectral effi- antennas, it will be possible to communicate with the user-
ciency (SE) and energy efficiency (EE) [1]. However, there are equipment (UE), on the same time-frequency resource by spa-
technical requirements and challenging issues that 5G needs tial multiplexing [4]. Furthermore, by upgrading the BS hard-
to overcome so that it can be implementable. One of the ware with this massive number of antennas instead of deploy-
major challenges is that the new generation must support much ing new BS sites, reduction of required transmission energy can
higher data rate (approximately 1000× compared to legacy net- be possible by using the beam-forming gain techniques devel-
works), which makes it essential to go to higher frequencies, 30– oped for these mMIMO systems [5]. Additionally, mMIMO can
300 GHz (Mm-Wave) spectrum, to allow more bandwidth. The improve SE by controlling the available space resources using
use of the mm-wave spectrum requires the use of many anten- the excess degrees of freedom (DoF) resulting from using large
nas allowing more throughput gain in the spatial dimension to size antenna array elements [2].
overcome high path loss. Furthermore, smaller cells provided by Although the potential of using mMIMO systems is excit-
the use of these many antenna are needed to avoid interference ing, the challenges that these systems will face in march toward
[2]. the next-generation of mobile networks era are enormous and
Recently, massive multiple-input multiple-output (mMIMO) hardly understood. The major challenges related to antenna
has become the most promising wireless technology and the design and channel estimations in 5G mMIMO systems are

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is
properly cited.
© 2021 The Authors. IET Communications published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology

IET Commun. 2021;15:1677–1690. wileyonlinelibrary.com/iet-com 1677

 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1678 BENZAGHTA AND RABIE

mutual coupling effect, antenna selection in large antenna formance of communication systems in 5G, the use of a mas-
arrays, pilot contamination in time division duplex (TDD) sive number of antenna comes with several challenges that need
systems, and feedback overhead in frequency division duplex to be taken into consideration.
(FDD) systems. Different types of antenna array configurations (i.e. planar,
Detailed theoretical explanation of the mentioned challenges circular and cylindrical array), result in different channel char-
and open research issues along with state-of-the-art solutions acteristics, and thus will have a huge impact on the overall per-
overcoming some of these challenges are given in this study. formance of the system. General configuration of array systems
In light of those challenges, this paper can indeed be very use- at the BS, such as planar and circular configuration exhibit a
ful for researchers who try to use mMIMO systems for 5G. To huge deficiency, where the beam can only be adjusted in the
the best of the authors’ knowledge, there has not been a system- horizontal dimension. Also, these configurations do not satisfy
atic mapping study of research on antenna design and channel the increasing capacity needs. To overcome this deficiency, 3D
estimation challenges regarding mMIMO implementation in 5G massive array configurations, such as cylindrical and spherical
systems that would allow to review the maturity in general and array configurations are recommended [6, 7]. The cylindrical
identify research gaps, trends, and future directions. array configuration, provides high directivity, and narrow beams
Throughout the systematic mapping study, 63 primary stud- pointing at any space direction. However, these cylindrical con-
ies (among 105 primary and secondary studies) from the year figurations are considered to be limited to direct visibility, and
2017 to the second quarter of 2020 were selected. Afterward, we cannot be used for high buildings. A solution for this, is to use
introduced a classification framework to compare the state-of- spherical array configuration, where this type of antenna array
the-art approaches given in the literature. Finally, we discussed significantly enhances the data rate, link reliability and provide
the emerging challenges and research open issues that need to full 3D scanning [8].
be considered in future mMIMO systems approaches for 5G. One of the main challenges in effectively implementing large-
The remainder of this paper is structured as follows: Sec- scale arrays for mMIMO systems in 5G, is how to reduce the
tion 2 describes background literature explaining the mentioned mutual coupling effect among the antenna elements of an array.
challenges. Section 3 identifies the main research questions. The Strong coupling effects will result in high correlation and low
research methodology is described in Section 4. In Section 5, the isolation between the antenna elements, which significantly limit
analysis of the mapping study and discussion of the findings are the system performance, in terms of SE and EE [9].Also, mutual
presented. Future challenges and research directions are given coupling effect can distort the power amplifier linearity, which
in Section 6. Finally, in Section 7, the conclusion is presented. causes interference to adjacent-channels, which is known as out-
of-band (OOB) emission issue [10]. Since, mMIMO systems
require that a large number of antenna elements of a large-scale
2 BACKGROUND array have to be placed in a device with limited space, reduc-
ing the mutual coupling effect by increasing the spacing distance
Despite having huge benefits, there are major issues that must between antenna elements, is not possible [11]. Therefore, tradi-
be resolved for mMIMO before the implementation of such tional techniques for mutual coupling suppression are not suit-
systems is possible for 5G. Theoretical background related to able in such crowded capacity of antenna elements.
antenna design challenges and channel estimation issues are dis-
cussed in this section of the paper.
2.2 Antenna allocation for BSs, and
hardware complexity
2.1 Massive antenna array design
challenges Adequate adjusting of BS’s antennas is an important issue that
mMIMO systems need to overcome, in order for this technol-
In 5G systems, high directivity is a necessity, which can be ogy to be implemented in 5G. Since the number of antenna ele-
achieved through the implementation of massive antenna arrays. ments in mMIMO is significantly large, existing traditional allo-
The main motivation behind mMIMO is to be able to construct cation schemes fail to apply on these systems. There are many
a massive number of antenna arrays, while controlling its pat- reasons for this; firstly, in mMIMO for each additional antenna
tern, for long distance communication over mmWave spectrum. the data-rate gain is not constant, which means that adding two
The distance separating the antenna elements, and the excita- antennas does not guarantee that the transmission rate is dou-
tion amplitude and phase of these different elements, along with bled. Secondly, in mMIMO the performance of the link depends
the array configuration are the main parameters effecting the on both sides’ antennas rather than just a single BS, which makes
pattern control in mMIMO antenna arrays. A more advanced the resource constraints more complicated, and a challenging
design for antenna array consists of using phased arrays. Using problem to solve [12]. This opens the research field to come
this method, the radiation pattern in a desired direction is maxi- up with brand new antenna scheme assignment that solve the
mized by designing the feeding mechanism in a way that differ- antenna allocation problem in mMIMO.
ent antenna elements use different relative phases [6]. As the number of antenna increases in mMIMO systems, the
Although the use of massive antenna arrays in mMIMO sys- processing cost and hardware complexity at the BS increase as
tems brings many advantages, especially in enhancing the per- well. The reason is that the number of radio frequency chains
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1679

(RFC), is associated with the number of antennas in the system overhead in FDD mMIMO systems are needed to achieve the
[13]. To solve this problem, antenna selection methods are pro- promising high data rate for 5G.
posed in the literature, which can be used to implement fewer
RFC than the number of available antennas, by carefully select-
ing a number of transmitting/receiving antenna from the large- 3 RESEARCH QUESTIONS
antenna array [14]. With the use of this technique, the operating
cost can be significantly reduced for mMIMO systems; how- After choosing the topic that the systematic mapping study is
ever, it is impracticable to obtain the optimum antenna selec- going to be mapped to, the first step to start such a study is
tion, due to high computational complexity [15]. Therefore, to identify the research questions. Our research investigation is
more research on new antenna selection techniques is needed designed mainly on five research questions:
for mMIMO in order for them to be implementable in 5G
systems. ∙ RQ1. How are the publications on mMIMO systems in 5G,
distributed over the last years?
∙ RQ2. What are the challenges regarding antenna design and
2.3 Pilot contamination in time division channel estimation that mMIMO systems face, toward its
duplex (TDD) systems implementation in 5G?
∙ RQ3. What are the existing solution approaches concerning
TDD systems make use of the same frequency block for uplink those challenges?
and downlink, while the separation happens in time. The chan- ∙ RQ4. Regarding the state-of-the art solutions to the men-
nel state information (CSI) is acquired for the uplink when the tioned open research issues, which are the most cited?
users send their mutually orthogonal pilot to the BS. Follow- ∙ RQ5. What are the future research agenda?
ing this the BS correlate the received pilot with the already
known pilot to estimate the channel. Due to the channel reci- To answer the mentioned research questions, we decided to
procity in TDD systems, the same uplink estimated channel is conduct a systematic mapping study on the area of antenna
used for the downlink channel estimation by the BS. In a multi- design and channel estimation challenges in mMIMO. In the
cellular system, where the available frequency spectrum is lim- literature, there exist some work that review the concept of
ited, the cells in such systems have to reuse the exact same time- mMIMO and its implementation in 5G in general; however,
frequency blocks. This makes it impossible to allocate orthog- there is no secondary study that is up to date and reviews all
onal sequences of pilots for each user in all cells, which means the challenges that are associated with antenna design com-
all of the mutually orthogonal pilots have to be reused by those bined with review of all open research issues related to chan-
users. When a user in a neighbouring cell sends its same orthog- nel estimation. More specifically, in [16], the authors gave an
onal pilot sequences, the channel estimation is contaminated by overview of mMIMO advances based on published literature
those pilots, this is known as the effect of pilot contamination before 2017. In [2], the survey was done on the evolution of
[16]. This challenge in TDD systems of mMIMO is consid- mmWave mMIMO systems before the year 2018, and its pos-
ered to be a major limitation toward its use in 5G. Considerable sible implementation in 5G. The author in [19] reviewed a spe-
research had been done in this area to decrease the effect of cific type of mMIMO, which is based on printed multi-antenna
pilot contamination; however, more is needed before mMIMO systems. Review on mMIMO hardware impairments and chal-
can be fully implanted in 5G systems. lenges was done in both [20] and [21]. In [22], the authors
reviewed the issue of limited feed-back in mMIMO. The sec-
ondary studies mentioned do not concentrate on the research
2.4 CSI pilot overhead in frequency division efforts on antenna design and channel estimation challenges.
duplex (FDD) Therefore, there is an essential need for a systematic mapping
study to identify and classify research challenges in the men-
The use of FDD systems in mMIMO is a promising approach tioned areas of mMIMO implementation in 5G, compare the
toward the implementation of 5G, as they increase the system existing state-of-the art solutions, and to give future research
capacity and robustness of the link [3]. In FDD mMIMO sys- directions.
tems, the same time block is used by the uplink and down-
link, where they are separated only in frequency domain. In this
case, the reciprocity of the channel as in TDD systems can- 4 RESEARCH METHODOLOGY
not be exploited, since there is a difference in frequency car-
rier between the uplink and downlink [17]. Thus, the CSI for In this section, we describe the research instruments used for
the downlink is estimated by each user and fed-back to the BS. conducting the review, and the characterization framework used
Having a massive number of antennas at the BS, means that to classify the chosen primary studies is presented. To choose
users have to estimate a large number of channels associated the relevant papers related to our research questions, the fol-
with each antenna, which results in an enormous feedback pilot lowing inclusion criteria (IC), and exclusion criteria (EC) were
overhead [18]. Therefore, methods of reducing the high pilot performed:
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1680 BENZAGHTA AND RABIE

FIGURE 2 Papers published per year regarding massive MIMO

FIGURE 1 Study distribution of mMIMO antenna design and channel

estimation challenges

IC1: Papers that contain state-of-the art solutions and/or

general studies on mMIMO systems, regarding antenna design
challenges, and channel estimation issues.
EC1: Papers that present other open research issues toward
mMIMO implementations in 5G that are not in the scope of FIGURE 3 Papers published per year regarding challenges and open
research issues on massive MIMO implementation in 5G
our study.
EC2: Articles that are summary of conference, poster sum-
mary, keynote talks, editorial and Ph. D. thesis. - Pilot contamination in TDD mMIMO systems.
After these IC and EC were applied, the final reference - CSI feedback overhead in FDD mMIMO systems.
list contained 63 primary studies that are relevant and directly
related to the research questions asked in the previous section.
In this research we analysed the mMIMO domain with a 5 ANALYSIS OF THE LITERATURE
focus on antenna design challenges and channel estimation SURVEY
issues, and then identified specific parameters with respect to
research questions, RQ2 and RQ3. In order to systematize those The key terms extracted from the studies, which are based
parameters, we categorize them into two main clusters, built on on the asked research questions mentioned previously, are
the two main related challenges that answer the research ques- addressed. The distribution of publications on mMIMO sys-
tions mentioned early. The subheadings were identified accord- tems in general, and the implementation challenges for those
ing to the scan of the selected primary studies after the filtering systems in 5G over the years of 2014 till 2019, are shown in
round. Figures 2 and 3, respectively. It is seen that with only a few pub-
The key terms which were extracted from the selected stud- lished studies in 2014 regarding the concept of mMIMO in gen-
ies, helped to categorize them in terms of specific challenges eral and the challenges of implementing mMIMO systems in 5G
and open research issues, regarding antenna design and channel in specific, there has been a dramatic increase in the following
estimation. Figure 1 gives the terms extracted that give an under- years, which shows a significant interest in this research field.
standing of key research concerns regarding those challenges in In this section, we report the latest progress of research on
mMIMO implementation for 5G systems. challenges in mMIMO systems, in the context of mutual cou-
The main clusters and their subheadings that construct our pling, antenna selection, pilot contamination and feedback over-
comparison framework are as follows: head, by focusing on the recent published literature between the
year 2017 and the second quarter of the year 2020.
∙ Massive antenna array design challenges: Tables 1 and 2 represent mMIMO antenna design challenges
- Mutual coupling effect and choosing of the separation dis- and the proposed solution extracted from the selected studies
tance between the antenna elements in massive antenna between 2017 and second quarter of 2020. The main two chal-
arrays. lenges are mutual coupling effect and antenna selection of mas-
- The need for new antenna selection schemes to support sive antenna arrays.
mMIMO implementation in 5G systems to reduce hard- In [23], a non-uniform antenna array design is proposed,
ware complexity. where the intra-array elements are packed close to each other
∙ Perfect channel estimation challenges: to fit in the limited space of the BS, while the sub-arrays are
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1681

TABLE 1 State-of-the-art proposed solutions for mutual coupling

Number of antenna
Ref. Design challenge Proposed solution elements Operating frequency Results approach

[11], 2017 Mutual coupling of Array-antenna 8-element array 2.45 GHz Experimental along
massive antenna decoupling surface antenna with simulation
array elements results
[23], 2017 Mutual coupling of Non-uniform linear Dividing the 2 GHz Simulation results
massive antenna antenna array design antenna-array to
array elements sub-arrays with
uniform spacing
[24], 2018 Mutual coupling of Use of metal mender 16-element array 3.2–5 GHz Simulation results
massive antenna line strips antenna
array elements
[25], 2018 Mutual coupling of Three-port 6-element array 2.574−2.647 GHz Experimental along
massive antenna polarization and antenna with simulation
array elements pattern diversity results
antenna design
[9], 2019 Mutual coupling of Graphene-based 4-element array 1.1–1.7 THz Simulation results
massive antenna frequency selective antenna
array elements surface (FSS)
[10], 2019 Mutual coupling of Decoupling ground 16-element array 4.8–5 GHz Simulation results
massive antenna technique antenna
array elements
[26], 2019 Mutual coupling of A transmission-line- 16-element array 2.45 GHz Experimental along
massive antenna based decoupling antenna with simulation
array elements technique results
[27], 2019 Mutual coupling of Molecule fractal 4-element array 2.4–10.6 GHz Experimental along
massive antenna structure antenna antenna with simulation
array elements design results
[28], 2019 Mutual coupling of Frequency-selective 4-element array 0.69–0.96 GHz and Experimental along
massive antenna surface design antenna 3.5–4.9 GHz with simulation
array elements approach results

well separated. Due to the reduction of correlation between decoupling network is presented in [26]. The decoupling net-
the sub-arrays signals, the inter-user interference cancellation work is realized in a single layer, with compact size and low
is increased, and the systems performance is enhanced. To insertion loss, which can be used for phased array mMIMO
reduce mutual coupling effect between array-antenna elements, systems. To enhance the isolation of antenna array elements
an array-antenna decoupling surface (ADS) is proposed in [11]. in mMIMO, a decupling ground concept is introduced in [10].
An ADS consists of a thin surface placed in front of the By adjusting the ground plane shape under each antenna ele-
array antenna, the surface is composed of small metal patches, ment, the isolation is improved by making the mutual cou-
which can be easily implemented for large-scale antenna arrays. pling from the ground plane and the free space out of phase.
Decoupling is achieved in [24] by embedding metal meander An isolation improvement using molecule fractal structure is
line strips horizontally and vertically between the antenna ele- given in [27]. The isolation is achieved by orthogonal orienta-
ments. By the use of those metal line strips, an additional cou- tion of the antenna perpendicular elements. This method shows
pling path is provided to reduce the coupling effect between the an impedance bandwidth (S11 < −10 dB) in the range between
ports. A three-port antenna design, exhibiting pattern diversity 2.4 and 10.6 GHz, achieving an isolation of better than −20 dB
is proposed in [25]. In the proposed method, the radiating ele- over the entire UWB range. The proposed approach was com-
ments comprises of a ring which is loaded with periodical inter- pared with 11 different approaches which were reported in lit-
digital capacitors. The selective excitation properties of the three erature ([27], Table 2). It is found that the proposed method is
ports of the proposed antenna enables dual-polarized and omni- compact in terms of size and provides high isolation without a
directional broadside radiation patterns. Using this method, the decoupling structure. In [28], frequency selective surface (FSS)
mutual coupling between any pairs of ports in the three-port is used as a decoupling technique. An FSS is introduced between
proposed antenna system is found to be less than −20 dB. The the two layers of the antenna, where the high frequency antenna
mentioned design showed to have low mutual coupling effect elements are placed above the FSS layer, and the lower oper-
and uncorrelated radiation patterns, which makes it a good can- ating frequency antennas are placed under the FSS. The differ-
didate for implementation in mMIMO systems for 5G appli- ent band antennas mutual coupling was effectively suppressed
cations. A decoupling technique using transmission-line-based by the use of FSS technique. Also, in [9], reducing the mutual
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1682 BENZAGHTA AND RABIE

TABLE 2 State-of-the-art proposed solutions for antenna selection

Number of antennas
Design at transmitter and
Ref. challenge Proposed solution receiver (Tx × Rx)

[30], 2017 Antenna Capacity-based 64 × 10

selection reduced-complexity
exhaustive
search(CRCES) and
norm-based
reduced-complexity
exhaustive search
(NRCES) technique
[31], 2017 Antenna Joint antenna and user Several scenarios
selection (AU) algorithm including FIGURE 4 The three most cited solution approach of mutual coupling
16 × 10 to 16 × 24 for mMIMO
32 × 10 to 32 × 24
[14], 2018 Antenna Group layer 32 × 32
selection MU-MIMO scheme
[32], 2018 Antenna Channel norm 64 × 64
selection statistics based
selection method
[13], 2019 Antenna Use of user clustering Several scenarios
selection techniques including
40 × 160
60 × 160
[15], 2019 Antenna Use of algorithm based 125 × 3
selection on calculation of the
highest achievable
rate of an individual
antenna element in
an antenna array FIGURE 5 The three most cited solution approach for antenna selection
in mMIMO

coupling effect in dense antenna-arrays for multi-band mMIMO didate antennas according to user scheduling techniques, and
systems, was achievable by using graphene-based FSS. In this search for the highest achievable channel gain. The favourite
proposed technique, by placing the FSS structure in between antennas are sought from limited candidate antennas which are
the array elements, the mutual coupling effects were reduced. favourable to the related users. Then by using the proposed
The isolation is improved further by inserting the FSS decou- joint antenna and user (AU) contribution algorithm, selection
pling structure into the substrate as well. By doing so, the elec- of antennas and scheduling of users, is performed for multi-
tromagnetic wave propagation is blocked in the substrate. The band multi-user scenario. In [32], an mMIMO transceiver sys-
reported results show that both the coupled electric and mag- tem design for 5G applications is proposed. The antenna selec-
netic fields were eliminated, using the FSS decoupling struc- tion process in the mentioned design, was based on channel
ture. A coprime cubic array (CCA) design configuration is pro- norm statistics. A group layer multi-user MIMO scheme with
posed in [29]. The CCA consists of two uniform cubic subar- antenna selection algorithm is proposed by [14]. The best sub-
rays, which can extend the antenna element spacing in a given set of antennas are selected according to their channel condi-
array, by selection of three pairs of coprime integers. The pro- tion. The proposed scheme achieved significant reduction in RF
posed CCA configuration, can effectively decrease the mutual chain, which can be used for 5G mMIMO systems. In [15], the
coupling effect, and increase the array aperture compared to tra- antenna selection algorithm is based on calculating the achiev-
ditional uniform cubic array (UCA) design configurations. able rate. The antennas with the highest transmission rates are
A transmit-antenna selection scheme for mMIMO sys- chosen at the BS. The proposed technique provided good qual-
tems is proposed in [30]. The scheme consists of two meth- ity of service for the whole uplink of the mMIMO network.
ods, which are capacity-based reduced-complexity exhaustive An antenna selection algorithm based on user clustering is pro-
search (CRCES) and norm based reduced-complexity exhaus- posed in [13]. The clusters are based on the channel conditions
tive search (NRCES). In the proposed method, the criteria and the number of accessible RF chains. The antenna selection
of selection of candidate transmit-antenna subset, is based on algorithm then follows minimum rate and power constraints to
SNR maximization. It is found that the CRCES method is well effectively select the candidate antennas.
suited for low SNR values, and has the capability to improve Figures 4 and 5 show the most cited solutions for mutual
the system performance in terms of capacity. In [31], antenna coupling and antenna selection issues, based on google scholar
selection is performed by using an algorithm which selects can- number of citations till the second quarter of 2020.
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1683

Tables 3 and 4, represent mMIMO channel estimation estimator is an efficient solution to overcome pilot contami-
challenges and the proposed solution extracted from the nation, this method performance is dependent on the num-
selected studies between 2017 and second quarter of 2020. ber of BS antennas and the accuracy of the sample covariance
The main two challenges are pilot contamination in TDD matrix. It is found that the rate loss is proportion to the BS
mMIMO systems and feedback overhead in FDD mMIMO antenna, and when those antennas are finite, there is always a
systems. non-zero rate loss in an mMIMO system using the eigenvalue-
In [33], the authors introduced superimposed pilots scheme decomposition based estimator. Also in [40], time-shifted pilot
for uplink channel estimation as a solution to pilot contami- scheme is used to combat pilot contamination. The channel esti-
nation in mMIMO. By using this scheme, it is found that the mator presented is divided to two stages. The interface from
channel estimation is improved due to the reduction of interfer- other cells is reduced by introducing null symbol periods deter-
ence, since the pilots are transmitted alongside with the trans- mined using the beamforming vectors after that the channel
mitting data. A channel estimator for multi-cell mMIMO TDD estimation is performed. Simulation results show that the pro-
system is presented in [34]. The proposed scheme for the uplink posed scheme can effectively reduce the pilot interference and
training process is based on the use of Zadoff-Chu (ZC) pilot improve SE in high SNR region.
sequence. It was found that the proposed estimator performs as In [41], a channel estimation method for OFDM mMIMO
well as the minimum mean square error (MSE) estimator and is systems is proposed using least squares channel estimation
able to perform under strong pilot contamination without prior (LSCE). Simulation results show that the channel estimation is
knowledge of noise power and inter-cell large scale fading coef- enhanced by overcoming pilot contamination, and the perfor-
ficients. Authors in [35], proposed an estimation algorithm for mance of the system is improved in terms of Bit Error Rate
Rician fading channels based on statistical channel information (BER). In [42], graph colouring-based pilot assignment (VGC-
and contaminated CSI. The line-of-sight (LOS) component of PA) algorithm in combination with the existing post-processing
the BS users are estimated based on the angle of arrival (AoA) discrete Fourier transform (DFT) filtering is proposed for chan-
and the contaminated CSI, then using those LOS components nel estimation to enhance the system capacity and mitigate
the data are detected, which are used to update the CSI esti- inter-cell interference between users sharing the same pilots.
mate through an iterative process. Simulation results show that A channel estimation algorithm based on joint singular value
in coherence-time-limited systems high SE is achieved, which decomposition (SVD) and iterative least square with projection
makes the proposed scheme a good channel estimator can- (SVD-ILSP) is proposed in [43]. The proposed algorithm over-
didate for mMIMO systems using Rician fading channels. In comes the disadvantage of finite sample data assumption of the
[36], a channel estimation scheme is based on assigning spe- covariance matrix that exist in the SVD-based semi-blind chan-
cific orthogonal variable spreading factor (OVSF) code row for nel estimation structure. The optimization method is to com-
each BS, and ZC sequences are then used for uplink pilot train- bine both the SVD and the ILSP algorithms, through apply-
ing. Element-wise multiplication of the ZC sequence at each BS ing the principle of iteration. Simulation results show that the
with its OSVF code row is then performed to generate orthog- proposed algorithm can mitigate pilot contamination impact
onal pilot sequences among the neighbouring cells. The pro- and accurately obtain CSI with low computational complexity.
posed scheme is able to eradicate pilot contamination, result- Authors in [44], proposed a channel estimation scheme using
ing in interference-free downlink transmission and enhanced factor analysis which decomposes the space supported by the
sum-rate performance. For broadband mMIMO systems using received signal convenience matrix into three subspaces. Simula-
OFDM modulation, a joint angle-delay subspace channel esti- tion results show that by exploiting the spatial correlation using
mator is used in [37]. Based on the parametric channel model, factor analysis, the proposed channel estimation scheme can
the concept of joint angle-delay subspace is used, along with mitigate the impact of both pilot contamination and noise inter-
low-complexity and low-rank adaptive filtering algorithm. Due ference. A block-diagonal Grassmannian line packing (BDGLP)
to the reason that channel statistics are typically unknown, a approach to mitigate pilot contamination is given in [45]. Spe-
robust MMSE estimator is developed. This estimator takes into cific sequences for cells are designed using GLP, and then
consideration the worst precondition of pilot decontamination, extended to form the superimposed pilot matrices for the users.
bearing in mind that fully overlap occurs in the joint angle-delay Then a channel estimation method based on Tikhonov regular-
subspaces of the interfering users. A semi-blind channel estima- ization (TR) is used. Numerical results show that the proposed
tion algorithm is given in [38]. The proposed estimation algo- technique improves the accuracy of channel estimation and fur-
rithm requires neither pilot scheduling nor co-operation among ther improve the SE. In [46], the authors propose a semi-blind
the cells. The proposed signal is first projected onto the sub- uplink interference suppression scheme for multi-cell mMIMO.
space with minimal interference, and an initial channel estima- The proposed method employs a blind interference suppression
tion is made based on the small number of pilot symbols. After scheme which is a constant modulus algorithm (CMA) and uses
that, the data symbols are detected and the channel estimation MMSE post coding weight as a primary weight of the CMA.
is distinguished alternatively. The proposed algorithm has low Although MMSE weight cannot suppress interference due to
complexity and is able to effectively reduce the pilot contami- pilot contamination, it can be beneficial to the CMA to cap-
nation effect. An eigenvalue-decomposition based channel esti- ture the desired signals that is needed for the CMA to effec-
mator is presented in [39] as a solution to pilot contamination. tively operate. In the next step, a decision feedback channel
Even though that the eigenvalue-decomposition based channel estimation (DFCE) is then presented to further enhance the CSI
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1684 BENZAGHTA AND RABIE

TABLE 3 State-of-the-art proposed solutions for pilot contamination

Performance
Ref. Channel estimation issue Channel type Proposed channel estimation strategy metric

[33], 2017 Pilot contamination Additive Non-iterative scheme based on superimposed pilots BER, sum rate
White Gaussian Noise and
(AWGN) approximate
rate per user
[34], 2017 Pilot contamination Multipath channel ZC sequences and minimum-variance unbiased MSE
estimator (MVUE) method
[35], 2017 Pilot contamination Rician fading channels LOS angles of arrivals (AOAs) BER and SE
[36], 2017 Pilot contamination Fast fading Orthogonal variable spreading factor (OVSF) code row and a set of Sum rate
Zadoff–Chu (ZC) sequences
[37], 2017 Pilot contamination Multipath channel Joint angle-delay NMSE
subspace
[38], 2017 Pilot contamination AWGN Semi-blind channel estimation algorithm NMSE and BER
[39], 2017 Pilot contamination Fast-fading channel Eigenvalue Achievable rate
decomposition and the symbol
error rate
[40], 2017 Pilot contamination Rayleigh fading Time-shifted pilot scheme NMSE and SE
[41], 2020 Pilot contamination Rayleigh fading Least squares channel estimation (LSCE) Bit error
rate(BER)
[42], 2018 Pilot contamination Multi-path channel Vertex Average MSE
graph colouring-based pilot assignment (VGC-PA) algorithm
[43], 2018 Pilot contamination Small-scale and large-scale Joint singular value decomposition (SVD) MSE and SE
fading and iterative least square with projection (SVD-ILSP)
[44], 2018 Pilot contamination AWGN Factor analysis technique Estimation error
variance
[45], 2018 Pilot contamination AWGN Block-diagonal Grassmannian line packing (BDGLP) NMSE and SE
approach
[46], 2018 Pilot contamination AWGN Blind interference suppression scheme BER and SINR
[47], 2019 Pilot contamination Small-scale and large-scale Hybrid pilot-aided channel Achievable data
fading estimation technique based on time-multiplexed (TM) pilot and rate
time-superimposed (TS) pilot
[48], 2019 Pilot contamination AWGN Coordinated pilot sequence DESIGN MSE
[49], 2019 Pilot contamination Block-fading Pilot assignment strategy and NMSE
pilot design-based channel estimation with Zadoff-Chu (ZC)
sequences
[50], 2019 Pilot contamination AWGN Channel estimation scheme based on spatial spectrum analysis MSE
[51], 2019 Pilot contamination Multipath channel CS based algorithm with non-zero neighbourhood (NZN) structure Average pilot
scheme length and sum
rate
[52], 2020 Pilot contamination AWGN Least-squares (LS) estimation using deep neural network (DNN) Normalized mean
square
error(NMSE)
[53], 2020 Pilot contamination Rayleigh fading Orthogonal pilotreuse sequence Average data rate
[54], 2020 Pilot contamination Block-fading Use of superimposed (SP) pilots in combination Mean square error
with time-multiplexed (TM) pilots (MSE) and SE
[55], 2020 Pilot contamination Small-scale and large-scale Space-alternating generalized BER,MSE and SE
fading expectation-maximization (SAGE) based semi-blind estimator
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1685

TABLE 4 State-of-the-art proposed solutions for feedback overhead

Performance
Ref. Channel estimation issue Channel type Proposed channel estimation strategy metric

[56], 2017 CSI feedback overhead AWGN Robust closed-loop pilot MSE
and CSIT feedback resource adaptation framework
[57], 2017 CSI feedback overhead Rayleigh fading Low-rank covariance-assisted downlink training NMSE
[58], 2017 CSI feedback overhead WINNERII Exploiting the angle domain channel sparsity MSE
[59], 2017 CSI feedback overhead Time-varying block Rician Estimation of the Average
fading channel strongest AoAs at both the BSand the users achievable rate
per user
[60], 2017 CSI feedback overhead Block-fading Threshold-based estimation MSE
[61], 2017 CSI feedback overhead Spatio correlation channel Beam-blocked sparsity Achievable sum
rate and NMSE
[62], 2017 CSI feedback overhead AWGN Spatial downlink channel estimation scheme Sum rate and
BER
[63], 2018 CSI feedback overhead AWGN Single-bit sparse NMSE,
maximum-likelihood estimation (MLE) and single-bit compressed beamforming
sensing gain and sum
capacity
[64], 2018 CSI feedback overhead AWGN Beamformed channel state information reference signalCSI-RS SE
transmission mechanism
[65], 2018 CSI feedback overhead AWGN Estimation algorithm MSE
based on projection methods in an infinite-dimensional Hilbert
space
[66], 2018 CSI feedback overhead Quasi-static channel Sparse adaptivematching pursuit (SAMP) algorithm NMSE
[67], 2018 CSI feedback overhead One-ring scattering model Interference alignment and soft-space-reuse (IA-SSR) Sum rate and
MSE
[68], 2018 CSI feedback overhead Frequency selective channel Block Bayesian matching pursuit (BBMP) NMSE
[69], 2018 CSI feedback overhead Time-varying channel Combination of CS, block Iterative-support-detection (block-ISD), NMSE
angle-of-departure
(AoD) and structured compressive sampling matching pursuit
(S-CoSaMP) algorithms
[3], 2019 CSI feedback overhead AWGN Deep learning with superimposed coding (SC) BER and NMSE
[17], 2019 CSI feedback overhead One-ring channel model, Joint pre-coding and scheduling algorithm Cell average
AWGN throughput
[18], 2019 CSI feedback overhead FDD, fasttime-varying Structured compressive sensing (SCS) Normalized mean
squared error
[70], 2019 CSI feedback overhead Multipath channel Angular scattering function (ASF) with sparsifying precoder Sum rate
[71], 2019 CSI feedback overhead Multipath channel Newtonized NMSE and sum
orthogonal matching pursuit (eNOMP) algorithm rate
[72], 2019 CSI feedback overhead AWGN CS scheme NMSE and BER
[73], 2019 CSI feedback overhead Small-scale and large-scale DNN approach for non-linear CSI structure Sum rate
fading
[74], 2019 CSI feedback overhead Channel model is formulated Pilot-data superposition Sum rate
by the authors using third
order tensor
[75], 2020 CSI feedback overhead Rayleigh fading Angle reciprocity of multipath components estimation technique MSE and SE
[76], 2020 CSI feedback overhead Quasi-static frequency Quantized partially joint orthogonal matching NMSE and BER
selective fadingchannel pursuit (Q-PJOMP) and quantized partially joint iterative hard
thresholding (Q-PJIHT) compressed sensing (CS) algorithms
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1686 BENZAGHTA AND RABIE

estimation. Simulation results reveal that the proposed scheme

can reduce pilot contamination and improve outage probability.
A hybrid pilot-aided technique for channel estimation of
multi-cell multi-user mMIMO systems is proposed in [47]. The
hybrid pilot-aided estimation technique consists of both time-
superimposed (TS) pilot and time-multiplexed (TM) pilot, and
thus allow a better solution to pilot contamination compared
with conventional pilot schemes. Theoretical and numerical
simulation results show that by using the proposed scheme
a higher UL rate can be achieved. A matrix fractional pro-
gramming (FP) approach used in coordinating the uplink pilots
across multiple cells is proposed in [48] to overcome pilot
contamination. The proposed algorithm guarantees reduction
FIGURE 6 The three most cited solution approach for pilot
of the weighted sum MSE of channel estimation and numer- contamination challenge in mMIMO
ical results show that it can outperform conventional pilot
reuse methods, especially for the cell-edge users. In [49], a
pilot design-based channel estimation scheme is proposed by the interference is reduced using the proposed method, and
employing perfect auto-correlation property of Chu sequences, thus the SE of mMIMO is improved. A space-alternating gen-
aiming to obtain accurate CSI by overcoming pilot contamina- eralized expectation-maximization (SAGE) based semi-blind
tion. Simulation results show that the proposed pilot assign- channel estimator is proposed in [55]. The proposed chan-
ing strategy has low computational complexity and can be nel estimation method is divided into two stages. First, a
applied to mMIMO systems. A channel estimation method for linear minimum mean squared error (LMMSE) pilot-aided esti-
2D mMIMO systems with uniform rectangular array (URA) mator is initially used for channel estimation. Secondly, the
antenna is proposed in [50]. The spatial spectrum is derived initial estimation is then iteratively updated using the SAGE
using the AoA and the spatial frequency. Then using spectrum algorithm. Simulation results show that the proposed algo-
analysis, the channel gains can be estimated. Also, zero padding rithm has improved the CSI accuracy and enhanced the SE of
and interpolation methods are used to improve the computa- mMIMO systems.
tional resolution of the proposed method. Simulation results Figure 6 shows the most cited solutions for pilot contamina-
show that the accuracy of channel estimation is improved by tion issues in TDD mMIMO, based on google scholar number
using this scheme in mMIMO systems. A compressed sensing of citations till the second quarter of 2020.
(CS) based channel estimation algorithm is proposed in [51]. In [56], a multi-user mMIMO system with closed-loop pilot
To eliminate pilot contamination, a non-zero neighbourhood and CSI feedback resource adaptation framework is proposed.
(NZN) pilot assignment scheme is used. Based on this scheme The framework includes pilot resource adaptation, joint com-
the interference region of each user is defined. Simulations show pressive CSI recovery and CSI quality estimation. From simula-
that the proposed scheme achieves good performance in terms tion results, the framework proposed enhanced CSI estimation
of average pilot length and sum rate. performance, and improved robustness against dynamic chan-
A combination of deep neural network channel estimator fol- nel sparsity. To reduce downlink training overhead exploiting of
lowed by conventional LS estimation technique is given in [52]. low-rank structure of the channel covariance matrix is done in
The proposed scheme is found to be more suitable in dealing [57]. The behaviour of the MMSE estimator is studied when
with pilot contamination and co-channel interference compared the covariance matrix has a low-rank structure. Then to esti-
to traditional methods based on LS alone without the use of mate the covariance matrix, a training free scheme is proposed.
deep learning. A channel estimation scheme applying orthogo- Results show that the estimated covariance matrix used for
nal pilot reuse sequence for edge users is proposed in [53]. This MMSE estimator achieves better performance compared with
is performed based on large-scale fading. The scheme is eval- the CS method, and is robust to angular spread estimation error
uated by using zero-forcing precoding maximum ratio trans- and against AoA distribution mismatch. An angular domain
mission techniques. Interfering users in neighbouring cells are pilot design and channel estimation scheme is given in [58]. An
recognized based on an assessment of large-scale fading. These index calibration algorithm is used to estimate the DL dominant
interfering users are then included in the joint channel process- angular set from UL channel. For pilot design partial orthogo-
ing. By assigning orthogonal pilot reuse sequences to the centre nal pilot and complete orthogonal pilot designs are proposed.
user and the edge users based on their levels of pilot contamina- Simulations demonstrate that the given scheme can achieve
tion, the channel quality of users is enhanced. Simulation results larger DL throughput and reduce feedback overhead provid-
indicate that the channel approximation and the systems perfor- ing good MSE performance. A channel estimation algorithm
mance in the downlink are improved as well as enhancing the that does not require CSI feedback from the users is proposed
achievable data rate. In [54], superimposed (SP) pilots combined in [59]. The proposed algorithm aims to estimate the strongest
with TM pilots are used to estimate the CSI. The TM pilots AoAs at both the UE and BS. Then users send orthogonal
reduces the interference, which was a critical problem in using pilot symbols based on the estimated strongest AoAs. Analytical
SP pilots based channel estimators. Numerical results show that and simulation results prove that the proposed scheme can be
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1687

applicable for both sparse and non-sparse mmWave channel estimator is proposed. The cooperative transmission structure
environments. A threshold-based method for channel estima- under a two-stage precoding framework, is based on the IA-
tion is proposed in [60]. The optimal threshold is obtained SSR scheme. Explicitly, the cell-edge users and the cell-centre
as a function of sparsity, channel variance, and noise variance. ones, are treated separately to fully exploit the spatial degrees
The simplified threshold is a function of noise variance only. of freedoms. Later on, to maximize the capacity of the net-
The proposed technique has good estimation accuracy, with work, the optimal power allocation policy is established. The
reduced feedback overhead. In [61], a beam-blocked compres- cooperative transmission structure under a two-stage precod-
sive scheme for channel estimation is proposed. Pilot overhead ing framework, is based on the IA-SSR scheme. Explicitly, the
is reduced by taking benefit of inherent block sparsity struc- cell-edge users and the cell-centre ones, are treated separately to
ture of channel matrix in beam-space. Then, by using an opti- fully exploit the spatial degrees of freedoms. Later on, to maxi-
mal block orthogonal matching pursuit algorithm the CSI can mize the capacity of the network, the optimal power allocation
be accurately estimated at the BS using the limited number of policy is established. A variety of numerical results, represent-
pilots. Depending on statistical spatial correlation channel clus- ing the sum-rate per cell-edge and cell-centre clusters, as well as
ters of the UL and DL, a channel estimation scheme is proposed the MSE performance of the proposed channel training scheme
in [62] for FDD mMIMO systems. Given an arbitrary frequency are presented to demonstrate the efficiency of the proposed
band gap between the channels, the UL and DL channels, this algorithm. A channel estimation approach using block Bayesian
method construct a transformation matrix that precode the UL matching pursuit (BBMP) is given in [68]. In this approach,
channel based on the dominant estimated angles of departure. the channel estimation issue is formulated as a block sparse
The proposed scheme is found to be able to accurately estimate recovery problem, in which the prior probability for block sup-
the CSI at the BS without the cost of user feedback overhead, port set is derived accordingly. Based on the equivalent sensing
as well as without prior knowledge of the channel statistics. matrix, the block index is selected based on a selection met-
Authors in [63] proposed limited feedback algorithms to ric, by exploiting the maximum likelihood criterion. Addition-
solve feedback overhead in 5G mMIMO systems. The feed- ally, by applying a matching pursuit method, the update criterion
back framework uses dictionary-based sparse channel estima- is derived for the selection metric once the dominant support
tion algorithms that have low computational complexity. The set augments with the iteration. A combination of compressed-
used dictionaries are related with both the angle of departure sensing (CS), structured compressive sampling matching pur-
and the angle of arrival that precisely comprise the antenna suit (S-CoSaMP), angle-of-departure (AoD) and block iterative-
directivity patterns at both ends of the link. Simulations reveal support-detection (Block-ISD) algorithms are given in [69] as
that the proposed algorithms is able to accurately estimate the a solution to reduce feedback overhead of channel estimation.
channel using small number of feedback bits. A CSI acquisi- Simulation results show that the proposed technique reduces
tion scheme for mMIMO FDD systems is proposed in [64] the pilot feedback overhead by a significant percentage which
that uses limited feedback algorithms utilizing beam-formed improves the overall system SE and EE.
CSI reference signal transmission mechanism. The algorithms In [70], the authors presented an estimation that exploits the
proposed is divided into stages that consist of multiple wide- reciprocity of the angular scattering function, which is used to
band beamforming vectors selection and sub-band combina- estimate the users’ DL covariance matrix from UL pilots. Simu-
tion coefficients calculation. Simulations show that the pro- lations show that the proposed method can perform better than
hibitive amount of feedback is avoided, and the SE of the system CS-based estimation methods as it can be used even if the avail-
is enhanced. An estimation algorithm that exploit properties able DL pilot dimension is less than the inherent dimension of
of channel reciprocity in the angular domain, using projection the channel vectors. An estimation framework developed using
methods in an infinite-dimensional Hilbert space is proposed enhanced Newtonized orthogonal matching pursuit (eNOMP)
in [65]. Evaluation of the proposed scheme was done using algorithm is given in [71]. The algorithm is used to extract from
Monte Carlo simulations, and results prove that higher accu- the uplink the frequency-independent parameters, which can
racy of channel estimation was achieved with low computational be used to develop the downlink training scheme. Numerical
complexity. In [66], modified sparse adaptive matching pursuit results show that the proposed framework can reconstruct the
(SAMP) algorithm is introduced to reduce feedback overhead CSI with small amount of feedback and training overhead. In
in FDD mMIMO systems. The SAMP algorithm is presented [72] the authors proposed a CS estimation scheme that improves
to handle the issue when the massive MIMO channel sparsity the coding sampling matching pursuit (CoSaMP) algorithm, for
is unknown. To overcome the limitations of the SAMP algo- FDD mMIMO systems by proposing structured sparse adaptive
rithm such as fixed step size and too much iterations, the mod- (CoSaMP) algorithm. Simulation results show that the proposed
ified SAMP (M-SAMP) is presented. To reconstruct the signal, algorithm has good performance in low SNR and can reduce the
this approach combines the initial sparsity estimating, the signal pilot overhead, which provides better SE. Instead of exploit-
segmenting, and the variable step size. Simulation results show ing linear CSI structures, to attain dimensionality reduction, a
that proposed algorithm achieves better performance in terms deep neural network (DNN) approach is proposed in [73]. Case
of computation time and accuracy compared to subspace track- studies based on ray-tracing simulation show that the proposed
ing (SP), orthogonal matching pursuit (OMP) algorithms. In approach can accurately estimate the CSI, reduce pilot feedback
[67], a cooperative transmission scheme based on interference- overhead and thus enhance the wireless system performance. In
alignment and soft-space-reuse (IA-SSR) with low-cost channel [3], the authors combined deep learning DL with superimposed
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1688 BENZAGHTA AND RABIE

coding (SC) as a solution for CSI feedback overhead. In this

approach, the downlink CSI is spread and then superimposed
on the uplink user data sequences. A multi-task neural network
architecture is then proposed at the BS, by unfolding two itera-
tions of the MMSE criterion-based interference reduction, the
downlink CSI and the uplink user data sequences are recovered.
Additionally, a subnet-by-subnet technique is exploited to facili-
tate the parameter tuning and expedite the convergence rate for
the network training. Simulations show that the estimation of
the CSI is improved for 5G mMIMO systems without much
occupation of UL bandwidth resource. Authors in [18] propose
a non-orthogonal pilot design with CS based channel estima-
tion algorithm. By taking advantage of the spatial and tempo-
FIGURE 7 The three most cited solution approach for feedback
ral common sparsity in delay domain of mMIMO, results show overhead challenge in mMIMO
that fewer pilot overhead is achieved for accurate CSI estima-
tion. To cope with the pilot feedback overhead issue in FDD
mMIMO systems, authors in [17] proposed a design consist- limited number of antenna elements used. Therefore, extra
ing of joint precoding and scheduling algorithm for multi-cell research effort is needed to investigate the possibility to use
mMIMO systems. To further reduce the computational com- such techniques for higher frequency bands, as well as to test
plexity of the proposed algorithm, a greedy scheduling algo- their validity for higher number of antennas in terms of several
rithm that is compatible with proposed framework is intro- tens to hundreds.
duced. Simulations verify that reduction of feedback overhead For the accurate acquisition of the CSI, pilot contamination
is achieved. The intrinsic tensor feature of the FD-MIMO chan- and feedback overhead need to be solved, before mMIMO can
nel is explored in [74], to enhance CSI estimation. The proposed be fully integrated into future 5G wireless systems. Although
estimation algorithm is based on the expectation-maximization good progress has recently been made in both directions, there
framework via tensor as the processing data structure. More- is still a need for more research to be conducted on this area.
over, for evaluation, the Cramér-Rao lower bound is utilized
as a metric. Simulation results show an increase in CSI estima- ∙ Research done in [18] on SCS, needs to be extended so that
tions, providing flexibility in realizing trade-off points between the proposed scheme can be suitable for the multi-cell sce-
UL and DL throughput using pilot-data superposition strategy. narios.
In [75], multipath components angle reciprocity in the uplink ∙ Low computational complexity algorithms that offer optimal
and downlink is used, so that the overhead CSI in FDD performance is needed for the iterative data-aided channel
mMIMO systems scales only with the served users, instead of estimation scheme proposed in [33].
the number of antennas. Simulation results show that the pro- ∙ The OVSF code row and the set ZC sequences used by [36],
posed estimation technique outperforms traditional gradient- as a solution to pilot contamination multiplies OVSF code
descent and subspace-based techniques. A CS based algorithm row at each BS with the ZC sequences. This requires that the
is used in [76] to recover the CSI at the BS from limited and number of multiplications is equal to the square of the num-
quantized feedback. By preserving one bit per dimension direc- ber of pilot sequences. Therefore, reduction of this complex-
tion information along with the partial amplitude information, ity using pilot assignment technique is needed.
the received compressed pilots are quantized. Then, this infor- ∙ The SVD-ILSP technique proposed in [43], can be consid-
mation that is fed back to the BS, is used to recover the CSI by ered for FDD mMIMO scenarios, by using mmWave fre-
employing the CS algorithms proposed. Results show that the quency bands.
proposed algorithm overcome the problem of training and feed- ∙ The BDGLP approach used in [45] is based on flat-fading
back overhead in FDD mMIMO system; however, this tech- channel model. The performance of this approach needs to
nique has a cost of high complexity. be further improved by considering frequency selective fad-
Figure 7 shows the most cited solutions for feedback over- ing model.
head issues in FDD mMIMO systems, based on google scholar ∙ Observing the performance of LS estimation technique using
number of citations till the second quarter of 2020. DNN, proposed in [52] for high mobility mm-Wave channels,
will be necessary to apply this technique in future 5G systems.
∙ The LSCE technique proposed in [53], can be further
6 FUTURE DIRECTIONS improved by focusing on a regular pilot, where the data esti-
mates can be preserved to enhance the channel estimates.
To be able to benefit from mMIMO systems in the upcom- ∙ The ASF used for channel estimation in [70] needs to be gen-
ing 5G mobile network, its use with mm-wave spectrum is eralized to cover a wider category of array antenna geome-
essential. Most of the aforementioned techniques to overcome tries.
mutual coupling issue in massive antenna array design are mainly ∙ For the CS channel estimation scheme given in [72], the prob-
developed based on sub-6 GHz applications and for a certain lem of sparse structuring in the virtual angler domain need
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BENZAGHTA AND RABIE 1689

to be explored to improve the applicability of the proposed 14. Al-Hussaibi, W.A., Ali, F.H.: Extending the user capacity of MU-MIMO
method. systems with low detection complexity and receive diversity. Wireless Net-
works 24, 2237–2249 (2018)
15. Gao, H., et al.: Antenna selection and power allocation design for
5G massive MIMO uplink networks. China Communications 16, 1–15
7 CONCLUSION (2019)
16. Jameel, F., et al.: Massive MIMO: A survey of recent advances, research
Massive MIMO is considered to be one of the enabling tech- issues and future directions. In: 2017 International Symposium on Recent
Advances in Electrical Engineering (RAEE). Islamabad, Pakistan, pp. 1–6
nologies for the next generation of wireless systems (5G). With
(2017)
large number of antennas at the BS, simultaneously communi- 17. Han, B., et al.: Joint precoding and scheduling algorithm for mas-
cation with multiple users using the same resources is possi- sive MIMO in FDD multi-cell network. Wireless networks 25, 75–85
ble. This allows higher spectral and energy efficiencies. How- (2019)
ever, this technology faces some challenges which need to be 18. Waseem, A., et al.: Compressive sensing based channel estimation for mas-
sive MIMO communication systems. Wirel. Commun. Mob. Com. 2019,
addressed before its possible implantation in 5G systems. In this
6374764 (2019)
paper, we provide a systematic mapping study on the state-of- 19. Sharawi, M.S.: Current misuses and future prospects for printed multiple-
the-art research efforts regarding antenna design challenges and input, multiple-output antenna systems [wireless corner]. IEEE Antennas
channel estimation issues. Although there are several secondary Propag. Mag. 59, 162–170 (2017)
studies in the literature that covers massive MIMO challenges 20. Mokhtari, Z., et al.: A survey on massive mimo systems in presence of
channel and hardware impairments. Sensors 19, 164 (2019)
in general, our paper provides a clear classification framework
21. Zhang, J., et al.: On low-resolution ADCs in practical 5G millimeter-wave
on the proposed solution techniques given in literature, with massive MIMO systems. IEEE Commun. Mag. 56, 205–211 (2018)
specific focus on the topics of mutual coupling, antenna selec- 22. Behjati, M., et al.: What is the value of limited feedback for next generation
tion, pilot contamination and feedback overhead. The paper is of cellular systems?. Wireless Personal Communications 110, 1127–1142
expected to be a road map for researchers in this field. (2020)
23. Ahmed, S., et al.: Hybrid analog and digital beamforming for space-
constrained and energy-efficient massive MIMO wireless systems. In: 2017
40th International Conference on Telecommunications and Signal Pro-
REFERENCES cessing (TSP). Barcelona, Spain, pp. 186–189 (2017)
1. Ahmed, I., et al.: A survey on hybrid beamforming techniques in 5G: 24. Qin, Y., et al.: A new decoupling method for massive MIMO antennas.
Architecture and system model perspectives. IEEE Commun. Surv. Tutor. In: 2018 IEEE International Symposium on Antennas and Propagation
20, 3060–3097 (2018) & USNC/URSI National Radio Science Meeting. Boston, MA, pp. 1577–
2. Busari, S.A., et al.: Millimeter-wave massive MIMO communication for 1578 (2018)
future wireless systems: A survey. IEEE Commun. Surv. Tutor. 20, 836– 25. Saurav, K., et al.: A three-port polarization and pattern diversity ring
869 (2017) antenna. IEEE Antennas Wirel. Propag. Lett. 17, 1324–1328 (2018)
3. Qing, C., et al.: Deep learning for CSI feedback based on superimposed 26. Zhang, Y.-M., et al.: A transmission-line-based decoupling method for
coding. IEEE Access 7, 93723–93733 (2019) MIMO antenna arrays. IEEE Trans. Antennas Propag. 67, 3117–3131
4. Sanguinetti, L., Björnson, E., Hoydis, J.: Towards massive MIMO 2.0: (2019)
Understanding spatial correlation, interference suppression, and pilot con- 27. Rajkumar, S., Amala, A.A., Selvan, K.T.: Isolation improvement of UWB
tamination. IEEE Trans. Commun. 68(1), 232–257 (2019) MIMO antenna utilising molecule fractal structure. Electron. Lett. 55,
5. Björnson, E., Sanguinetti, L.: Making cell-free massive MIMO competi- 576–579 (2019)
tive with MMSE processing and centralized implementation. IEEE Trans. 28. Zhu, Y., et al.: Decoupling and low-profile design of dual-band dual-
Wireless Commun. 19(1), 77–90 (2019) polarized base station antennas using frequency-selective surface. IEEE
6. El Misilmani, H., El-Hajj, A.: Massive MIMO design for 5G networks: An Trans. Antennas Propag. 67, 5272–5281 (2019)
overview on alternative antenna configurations and channel model chal- 29. Gong, P., Ahmed, T., Li, J.: Three-dimensional coprime array for massive
lenges. In: 2017 International Conference on High Performance Comput- MIMO: Array configuration design and 2D DOA estimation. Wirel. Com-
ing & Simulation (HPCS). Genoa, Italy, pp. 288–294 (2017) mun. Mob. Com. 2020, 2686257 (2020).
7. Kamali, M., Cherif, A.: Adaptive cylindrical antenna array for massive 30. Lee, B.-J., et al.: Enhanced transmit-antenna selection schemes for mul-
MIMO in 5G. Int. j. comput. sci. netw. 18, 178–183 (2018) tiuser massive MIMO systems. Wirel. Commun. Mob. Com. 2017, 3463950
8. Kamali, M., Cherif, A.: Improved design of an adaptive massive MIMO (2017)
spherical antenna array. Int. J. Adv. Comput. Sci. Appl. 9, 417–423 (2018) 31. Liu, X., et al.: Highly efficient 3-D resource allocation techniques in 5G for
9. Zhang, B., et al.: Mutual coupling reduction for ultra-dense multi-band NOMA-enabled massive MIMO and relaying systems. IEEE J. Sel. Areas
plasmonic nano-antenna arrays using graphene-based frequency selective Commun. 35, 2785–2797 (2017)
surface. IEEE Access 7, 33214–33225 (2019) 32. Neduncheran, A.V., et al.: Design of a TAS-STBC-ESM (F) transceiver
10. Zhang, S., et al.: Mutual coupling suppression with decoupling ground for and performance analysis for 20 bpcu. IEEE Access 6, 17982–17995
massive MIMO antenna arrays. IEEE Trans. Veh. Technol. 68, 7273–7282 (2018)
(2019) 33. Upadhya, K., Vorobyov, S.A., Vehkapera, M.: Superimposed pilots are
11. Wu, K.-L., et al.: Array-antenna decoupling surface. IEEE Trans. Antennas superior for mitigating pilot contamination in massive MIMO. IEEE
Propag. 65, 6728–6738 (2017) Trans. Signal Process. 65, 2917–2932 (2017)
12. Kuo, W.-H., Chen, C.-H.: Backhaul antenna allocation scheme in massive 34. De Figueiredo, F.A., et al.: Channel estimation for massive MIMO TDD
MIMO cellular networks. In: 2019 IEEE 89th Vehicular Technology Con- systems assuming pilot contamination and frequency selective fading.
ference (VTC2019-Spring). Kuala Lumpur, Malaysia, pp. 1–5 (2019) IEEE Access 5, 17733–17741 (2017)
13. Al-Hussaibi, W.A., Ali, F.H.: Efficient user clustering, receive antenna 35. Wu, L., et al.: Channel estimation for multicell multiuser massive MIMO
selection, and power allocation algorithms for massive MIMO-NOMA uplink over Rician fading channels. IEEE Trans. Veh. Technol. 66, 8872–
systems. IEEE Access 7, 31865–31882 (2019) 8882 (2017)
 17518636, 2021, 13, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cmu2.12180 by Readcube (Labtiva Inc.), Wiley Online Library on [24/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1690 BENZAGHTA AND RABIE

36. Ali, S., Chen, Z., Yin, F.: Eradication of pilot contamination and zero forc- 57. Fang, J., et al.: Low-rank covariance-assisted downlink training and chan-
ing precoding in the multi-cell TDD massive MIMO systems. IET Com- nel estimation for FDD massive MIMO systems. IEEE Trans. Wireless
mun. 11, 2027–2034 (2017) Commun. 16, 1935–1947 (2017)
37. Zhang, Y., et al.: Channel estimation for massive MIMO-OFDM systems 58. Zhao, P., Wang, Z., Sun, C.: Angular domain pilot design and channel esti-
by tracking the joint angle-delay subspace. IEEE Access 4, 10166–10179 mation for FDD massive MIMO networks. In: 2017 IEEE International
(2016) Conference on Communications (ICC). Paris, France, pp. 1–6 (2017)
38. Wu, C.-Y., Huang, W.-J., Chung, W.-H.: Low-complexity semiblind channel 59. Zhao, L., Ng, D.W.K., Yuan, J.: Multiuser precoding and channel estimation
estimation in massive MU-MIMO systems. IEEE Trans. Wireless Com- for hybrid millimeter wave MIMO systems. In: 2017 IEEE International
mun. 16, 6279–6290 (2017) Conference on Communications (ICC). Paris, France, pp. 1–7 (2017)
39. Xu, W., et al.: Downlink performance of massive-MIMO systems using 60. Xiong, X., et al.: Beam-domain channel estimation for FDD massive
EVD-based channel estimation. IEEE Trans. Veh. Technol. 66, 3045– MIMO systems with optimal thresholds. IEEE Trans. Wireless Commun.
3058 (2016) 16, 4669–4682 (2017)
40. Wu, L., et al.: Enhanced time-shifted pilot based channel estimation in 61. Huang, W., et al.: Beam-blocked channel estimation for FDD massive
massive MIMO systems with finite number of antennas. In: 2017 IEEE MIMO with compressed feedback. IEEE Access 5, 11791–11804 (2017)
International Conference on Communications Workshops (ICC Work- 62. Esswie, A.A., et al.: A novel FDD massive MIMO system based on down-
shops). Paris, France, pp. 222–227 (2017) link spatial channel estimation without CSIT. In: 2017 IEEE International
41. Riadi, A., Boulouird, M., Hassani, M.M.R.: Least squares channel esti- Conference on Communications (ICC). Paris, France, pp. 1–6 (2017)
mation of an OFDM massive MIMO system for 5G wireless communi- 63. Alevizos, P.N., et al.: Limited feedback channel estimation in massive
cations. In: Proceedings of International Conference on the Sciences of MIMO with non-uniform directional dictionaries. IEEE Trans. Signal Pro-
Electronics, Technologies of Information and Telecommunications. Vol.2. cess. 66, 5127–5141 (2018)
SETIT 2018. Smart Innovation, Systems and Technologies, vol 147. pp. 64. Zhang, F., et al.: Enhanced CSI acquisition for FDD multi-user massive
440–450.Springer, Cham (2018) MIMO systems. IEEE Access 6, 23034–23042 (2018)
42. Dao, H.T., Kim, S.: Vertex graph-coloring-based pilot assignment with 65. Miretti, L., Cavalcante, R.L.G., Stanczak, S.: FDD massive MIMO channel
location-based channel estimation for massive MIMO systems. IEEE spatial covariance conversion using projection methods. In: 2018 IEEE
Access 6, 4599–4607 (2018) International Conference on Acoustics, Speech and Signal Processing
43. Bangash, K., et al.: A joint approach for low-complexity channel estimation (ICASSP). Calgary, Alberta, pp. 3609–3613 (2018)
in 5G massive MIMO systems. Electronics 7, 218 (2018) 66. Liao, Y., et al.: M-SAMP: A low-complexity modified SAMP algorithm for
44. Wei, X., et al.: Uplink channel estimation in massive MIMO systems using massive MIMO CSI feedback. In: 2018 IEEE/CIC International Confer-
factor analysis. IEEE Commun. Lett. 22, 1620–1623 (2018) ence on Communications in China (ICCC). Beijing, pp. 547–552 (2018)
45. Jing, X., et al.: Superimposed pilot optimization design and channel esti- 67. Ma, J., et al.: Interference-alignmen and soft-space-reuse based cooperative
mation for multiuser massive MIMO systems. IEEE Trans. Veh. Technol. transmission for multi-cell massive MIMO networks. IEEE Trans. Wire-
67, 11818–11832 (2018) less Commun. 17, 1907–1922 (2018)
46. Maruta, K., Ahn, C.-J.: Uplink interference suppression by semi-blind 68. Zhang, R., et al.: Block Bayesian matchin pursuit based channel estimation
adaptive array with decision feedback channel estimation on multicell mas- for FDD massive MIMO system. AEU-International Journal of Electron-
sive MIMO systems. IEEE Trans. Commun. 66, 6123–6134 (2018) ics and Communications 93, 296–304 (2018)
47. Li, J., et al.: On hybrid pilot for channel estimation in massive MIMO 69. Han, I., et al.: Spectral and energy efficient low-overhead uplink and down-
uplink. IEEE Trans. Veh. Technol. 68, 6670–6685 (2019) link channel estimation for 5G massive MIMO systems. Entropy 20, 92
48. Shen, K., Eldar, Y.C., Yu, W.: Coordinated pilot design for massive MIMO. (2018)
In: ICASSP 2019-2019 IEEE International Conference on Acoustics, 70. Khalilsarai, M.B., et al.: FDD massive MIMO via UL/DL channel covari-
Speech and Signal Processing (ICASSP). Brighton, UK, pp. 4509–4513 ance extrapolation and active channel sparsification. IEEE Trans. Wireless
(2019) Commun. 18, 121–135 (2018)
49. Zhao, J., et al.: Pilot contamination reduction in TDD-based massive 71. Han, Y., et al.: FDD massive MIMO based on efficient downlink channel
MIMO systems. IET Commun. 13, 1425–1432 (2019) reconstruction. IEEE Trans. Commun. 67, 4020–4034 (2019)
50. Wang, A., Yin, R., Zhong, C.: Channel estimation for uniform rectangular 72. Saraereh, O.A., et al.: Low-complexity channel estimation in 5G massive
array based massive MIMO systems with low complexity. IEEE Trans. MIMO-OFDM systems. Symmetry 11, 713 (2019)
Veh. Technol. 68, 2545–2556 (2019) 73. Jiang, Z., et al.: Exploiting wireless channel state information structures
51. Kenarsari, S.R., Naeiny, M.F.: Adaptive pilot reuse scheme for nonzero beyond linear correlations: A deep learning approach. IEEE Commun.
neighborhood structured downlink channel in massive MIMO. AEU- Mag. 57, 28–34 (2019)
International Journal of Electronics and Communications 99, 48–58 74. Zhou, Z., Liu, L., Zhang, J.: FD-MIMO via pilot-data superposition:
(2019) tensor-based DOA estimation and system performance. IEEE Journal of
52. Balevi, E., Doshi, A., Andrews, J.G.: Massive mimo channel estimation Selected Topics in Signal Processing 13, 931–946 (2019)
with an untrained deep neural network. IEEE Trans. Wireless Commun. 75. Abdallah, A., Mansour, M.M.: Efficient angle-domain processing for
19, 2079–2090 (2020) FDD-based cell-free massive MIMO systems. IEEE Trans. Commun. 68,
53. Salh, A., et al: Mitigating pilot contamination for channel estimation in 2188–2203 (2020)
multi-cell massive MIMO systems. Wireless Personal Communications 76. Kulsoom, F., et al.: Joint sparse channel recovery with quantized feed-
112, 1643–1658 (2020) back for multi-user massive MIMO systems. IEEE Access 8, 11046–11060
54. Lago, L.A., et al.: Pilot decontamination based on superimposed pilots (2020)
assisted by time-multiplexed pilots in massive MIMO networks. IEEE
Trans. Veh. Technol. 69(1), 405–417 (2019)
55. Mawatwal, K., Sen, D., Roy, R.: performance analysis of a SAGE-based
semi-blind channel estimator for pilot contaminated MU massive MIMO How to cite this article: Benzaghta, M., Rabie, K.M.:
systems. IEEE Access 8, 46682–46700 (2020) IET Commun. 15, 1677–1690 (2021).
56. Liu, A., Zhu, F., Lau, V.K.: Closed-loop autonomous pilot and compressive
CSIT feedback resource adaptation in multi-user FDD massive MIMO
https://doi.org/10.1049/cmu2.12180
systems. IEEE Trans. Signal Process. 65, 173–183 (2016)