Page 137 - ITU Journal Future and evolving technologies – Volume 2 (2021), Issue 2

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 2

interference, needs to be ef iciently incorporated and han‑ cal issue in FD MIMO systems. The same holds for chan‑

dled into the next generation Non‑Orthogonal Multiple nel aging, which requires sophisticated channel tracking
Access (NOMA) techniques. The design of analog and digi‑ algorithms. To maximize the ergodic mutual informa‑
tal TX/RX beam codebooks for SI mitigation is another im‑ tion, which is a non‑convex function of the power alloca‑
portant research topic towards the practical deployment tion, ef icient techniques for optimizing the power/pilot‑
of FD MIMO transceivers. We next detail some of the most overhead allocation vectors need to be developed.
promising future directions, and their challenges, for FD Machine‑learning‑based con igurations: Arti icial
MIMO communication systems. Neural Networks (ANNs) are lately gaining substantial
mmWave and THz communications: Given the limited attention in wireless communications [44] as an ef icient
spectral resources at frequencies below 6GHz, the 5G New means to deal with hardware nonlinearities and to tame
Radio has adopted mmWave frequencies to accommodate wireless channel dynamics [45, 46]. It is appealing to
high data rate communications, and 6G is expected to sup‑ consider ef icient hardware implementations of ANNs

port wireless links at subTHz [1]. To compensate for the to handle the nonlinearities of the multiple power am‑
high path loss at those high frequency bands, highly di‑ pli iers in the FD MIMO architectures, especially when

rective A/D BF is necessary. Hence, combining FD MIMO considered for high‑frequency communications. This
radios with mmWave and THz technologies is a promis‑ will enable more ef icient analog SI cancelers, thus,

ing approach to reach higher spectral iciencies. Inter‑ increasing the lexibility of A/D TX/RX BF. In addition,

estingly, the STAR applications of the recent FD MIMO supervised and reinforcement learning techniques can
framework [39, 40] can be used for low latency analog be deployed for traf ic predictions in order to enable

beam tracking, paving the way for beyond 5G standalone dynamic scheduling of UL and DL users in multiple
mmWave communications. For example, the beam re ine‑ interfering FD massive MIMO connections. With such
ment for multiple users can take place in the UL simulta‑ approaches, the large overhead of multiple pilot‑assisted
neously with their DL data communications from the FD massive MIMO channel estimates will be avoided.

MIMO base station. However, the nonlinearities of the Massive metasurface antennas: Incorporating recon‑

power iers become the bottleneck of SI cancella‑ igurable metasurfaces in wireless networks has been

tion in mmWave and THz sfrequencies. The RF imper‑ recently advocated as a revolutionary means to trans‑

fections will not be suf iciently suppressed in the analog form any naturally passive wireless propagation environ‑
domain with reduced hardware complexity, and sophisti‑ ment into a dynamically programmable one [47, 48, 49].
cated digital SI cancellation will be needed possibly with This can be accomplished by deploying cost‑effective and
the aid of digital RX BF. easy to coat metasurfaces to the environment’s objects,

Integrated communications and sensing: The STAR thus, offering increased environmental intelligence for

applications [39, 40] constitute only a portion of the po‑ the scope of diverse wireless networking objectives [50].
tential of the FD MIMO technology for integrated commu‑ Metasurfaces are arti icial planar structures of multiple
nications and sensing applications. In fact, the FD opera‑ recon igurable radiating metamaterial elements that can
tion can enable joint communications and radar [42, 43], be programmed to re lect an incoming electromagnetic
offering re‑usability of the available resources and inte‑ ield in a recon igurable way (passive metasurfaces [51]),

gration of advanced sensing capabilities in future wire‑ or can be used as TX or RX antenna arrays (hybrid [52]

less systems. To this end, the recent extension [35] of or active [53] metasurfaces). Interesting directions of re‑

the presented FD MIMO framework can be exploited for search are the investigation of FD massive MIMO oper‑
realizing highly lexible multiple beams for both commu‑ ation in networks empowered be passive metasurfaces,
nications and radar, trading off the complexity of SI can‑ and the design of FD MIMO architectures with massive

cellation with that for the sensing resolution. Further‑ numbers of metasurface antennas. In the former direc‑

more, it is interesting to consider the presented reduced‑ tion, ef icient NOMA techniques to manage FD‑based re‑
complexity architectures and schemes for integrated ac‑ lections need to be devised, whereas in the latter, hybrid
cess and backhaul, which is of particular interest for 5G A/D TX/RX BF schemes will be required.
and beyond wireless systems [2].

Channel estimation schemes: Most FD MIMO tech‑

8. CONCLUSIONS
niques rely on the idealized simplifying assumption of the
availability of perfect CSI; however, this assumption is un‑ In this paper, we presented a detailed overview of two
realistic. Indeed, the CSI acquired at the FD nodes is im‑ recent SI mitigation schemes for FD MIMO systems with

perfect due to estimation errors induced by: i) nonlinear reduced hardware complexity. Each presented scheme
hardware components; ii) phase noise from the device’s includes a reduced complexity analog canceler architec‑
oscillators; and iii) channel aging caused by the mobility ture, one based on analog taps and the other on AUX TXs.

of the users. Those CSI errors will be exacerbated at the The main simpli ication of the multi‑tap canceler hard‑
FD node when massive numbers of TX and RX antennas ware was obtained via the use of MUXs/DEMUXs for sig‑

are used. In addition, as shown in [25], it is very hard to nal routing among the TX and RX RF chains and the re‑
completely cancel the SI signal in the practical case of im‑ duced number of taps, as well as the joint design of the tap
perfect CSI. Hence, CSI estimation is a particularly criti‑ values and MUXs/DEMUXs con iguration with the TX/RX

© International Telecommunication Union, 2021 123