6G Vision PDF
6G Vision PDF
6G Vision PDF
The Next
Hyper Connected
Experience for All.
2 6G Services 13
Truly Immersive XR 13
High-Fidelity Mobile Hologram 14
Digital Replica 15
3 Requirements 17
Performance Requirements 18
Architectural Requirements 19
Trustworthiness Requirements 20
4 Candidate Technologies 22
Terahertz Technologies 22
Novel Antenna Technologies 24
Evolution of Duplex Technology 27
Evolution of Network Topology 28
Spectrum Sharing 30
Comprehensive AI 32
Split Computing 33
High-Precision Network 35
5 6G Timeline 37
6 Concluding Remarks 38
7 References 39
Preface
We expect that 6G will provide ultimate experience for all through hyper-connectivity
involving humans and everything. In this white paper, we aim to provide readers with a
comprehensive overview of various aspects related to 6G, including technical and societal
trends, services, requirements, and candidate technologies. The rest of this white paper is
organized as follows:
Section 2 discusses major services that have to be taken into account in developing 6G
technologies.
Section 3 describes requirements to realize the expected services for 6G. They consist
of performance requirements, architectural requirements, and trustworthiness re-
quirements.
Section 4 introduces candidate technologies that will be essential to satisfy the re-
quirements for 6G, which currently include support of the terahertz band, novel an-
tenna technologies, evolution of duplex technology, evolution of network topology,
spectrum sharing, comprehensive AI, split computing, and high-precision network.
Section 5 provides an initial expectation of the 6G timeline. We anticipate that the ear-
liest commercialization could occur as early as 2028 while massive commercialization
may emerge around 2030.
Preface 7
1 Megatrends
toward 6G
Connected Machines
Machine as a Main User
It is envisaged that the number of connected devices will reach 500 bil-
lion by 2030 [1], which is about 59 times larger than the expected world
population (8.5 billion [2]) by that time. Mobile devices will take various
form-factors, such as augmented reality (AR) glasses, virtual reality (VR)
headsets, and hologram devices. Increasingly, machines will need to be
connected by means of wireless communications. Examples of connect-
ed machines include vehicles, robots, drones, home appliances, displays,
smart sensors installed in various infrastructures, construction machiner-
ies, and factory equipment. Figure 1 illustrates this trend of mobile devices
and connected machines.
Megatrends toward 6G 9
Figure 1
Evolution of mobile devices and
connected machines.
Table 1
Human Machine
Comparison of the perception ca-
pability of humans and machines. Maximum 1/150°
Resolution (Smartphone display 290 ppi at 30 cm)
Latency
<100 ms
Perception
Audible Exceeds Human Limitations!
250-2,000 Hz
Frequency
Visible
280-780 nm
Wavelength
Viewing
Azimuth 200°, Zenith 130°
Angle
10
AI
New Tool for Wireless Communications
In recent years, the rise of AI has pervaded various areas such as finance,
health care, manufacturing, industry, and wireless communication sys-
tems. Application of AI in wireless communications holds the potential to
improve performance and reduce capital expenditure (CAPEX) and opera-
tional expenditure (OPEX). For example, AI can
Megatrends toward 6G 11
to provide an open and intelligent radio access network (RAN). Another
example is the open network automation platform (ONAP), which develops
a platform for network management and its automation through an open-
sourced shared architecture.
12
2 6G Services
Figure 2
Three key 6G services: truly im-
mersive XR, high-fidelity mobile
hologram, and digital replica.
Truly Immersive XR
XR is a new term that combines VR, AR, and mixed reality (MR). It has
attracted great attention and opened new horizons in various fields in-
cluding entertainment, medicine, science, education, and manufacturing
industries. Technical development to realize XR is still in progress, and new
6G Services 13
innovative technologies are constantly appearing. The critical obstacle
between the potential and reality of XR is hardware. In particular, these
technologies require advanced device form-factors, such as hand-held
components, to support mobile and active software content. Current mo-
bile devices lack sufficient stand-alone computing capability. Unfortunate-
ly, progress in hardware performance, especially mobile computing power
and battery capacity, cannot keep pace with what the boom of XR requires.
This discrepancy could severely deter market expansion. In our view, these
challenges can be overcome by offloading computing to more powerful
devices or servers.
Figure 3
Truly immersive XR.
14
will be essential. For example, 19.1 Gigapixel requires 1 terabits per second
(Tbps) [8]. A hologram display over a mobile device (one micro meter pixel
size on a 6.7 inch display, i.e., 11.1 Gigapixel) form-factor requires at least
0.58 Tbps. Moreover, support of a human-sized hologram requires a sig-
nificantly large number of pixels (e.g., requiring several Tbps) [9]. The peak
data rate of 5G is 20 Gbps. 5G cannot possibly support such an extremely
large volume of data as required for hologram media in real-time. To re-
duce the magnitude of data communication required for hologram displays
and realize it in the 6G era, AI can be leveraged to achieve efficient com-
pression, extraction, and rendering of the hologram data. The market size
for the hologram displays is expected to be $7.6 billion by year 2023 [10].
Figure 4
3D hologram display over mobile
devices.
Digital Replica
With the help of advanced sensors, AI, and communication technologies,
it will be possible to replicate physical entities, including people, devices,
objects, systems, and even places, in a virtual world. This digital replica
of a physical entity is called a digital twin. In a 6G environment, through
digital twins, users will be able to explore and monitor the reality in a vir-
tual world, without temporal or spatial constraints. Users will be able to
observe changes or detect problems remotely through the representation
offered by digital twins.
With the help of AI, digital replication, management of real world and
problem detection and mitigation can be done efficiently without the pres-
ence or even detailed supervision by a human being. For instance, if a prob-
6G Services 15
lem is detected in the digital twin representation, AI can invoke required
actions in the real world.
The technical challenges are significant. In order to, for example, dupli-
cate 1 m x 1 m area, we need a Tera-pixel, which requires 0.8 Tbps through-
put assuming periodic synchronization of 100 ms and a compression ratio
of 1/300. The expected market size for digital replica is estimated to be $26
billion in 2025 [11].
Figure 5
Digital replica: bridge the real
and virtual worlds.
16
3 Requirements
12
14
16
18
20
22
24
26
28
30
20
20
20
20
20
20
20
20
20
20
20
Requirements 17
In the rest of this section, we describe our view on 6G requirements for
the key performance indices, overall architecture, and trustworthiness.
Performance Requirements
In order to realize advanced multimedia services such as truly immersive
XR, mobile hologram, and digital replica, 6G needs to provide a much high-
er data rate than 5G. While 5G was designed to achieve 20 Gbps peak data
rate, in 6G, we aim to provide the peak data rate of 1,000 Gbps and a user
experienced data rate of 1 Gbps. To provide advanced multimedia services
to a large number of people, the overall network performance needs to be
improved, e.g., we can aim to have 2 times higher spectral efficiency than
5G.
Network coverage has always been important over past generations and
will remain very important in 6G. We aim to support larger coverage than
5G. The maximum supported speed of the mobile device improved from
350 km/h in 4G to 500 km/h in 5G. It may need to further improve in 6G de-
pending on the evolution of transportation systems. The explosive growth
in the number of connected machines will require 6G to support about 107
devices per square kilometer. This is ten times larger than the connection
density requirement of 5G.
18
networks should be minimized. We intend to improve the energy efficiency
of both devices and networks by at least two times.
10-5
10-6
10
-7
1x
Reliability 2x Spectral Efficiency
106 1
107
0.1
Connection Density Air Latency
(devices/km2) (ms)
Architectural Requirements
The architecture of 6G communication network should be developed so
that it can resolve the issues arising from the limited computation capa-
bility of mobile devices. A possible way to achieve this is to offload com-
putation tasks to more powerful devices or servers. In order to support
offloading of real-time intensive computation tasks, hyper-fast data rate
and extremely low latency communications are required. Understanding
that there must be practical limits on the achievable data rate and latency,
the communication network should be designed in a holistic manner, so as
to best utilize computation power that can be made available by various
entities in the network. We term this joint design “communications and
computing convergence.”
Requirements 19
gence of communications and computing so that an end user’s various de-
vices can seamlessly utilize the computing power available in the network.
5G
BS Edge Computing
Core UE
Cloud
6G
BS Communications and Computing
Convergence
Core UE
Cloud
Trustworthiness Requirements
As discussed as a megatrend prefiguring 6G, the use of open source
software and personal user information will increase the openness of
communication systems and hence increase the attack surface. This could
make the whole system more vulnerable to security and privacy threats as
described in the following examples. First, there may not be enough ver-
ification of open source software codes against possible security attacks.
Second, the service provider’s access to user information will expose attack
surfaces for leaking confidential user information and poses a severe threat
20
to user privacy. In addition, user devices can be hacked unless these devic-
es provide a sufficiently secure trusted environment. Compromised user
devices decrease the security of the whole telecommunication system and
the services accessed by users.
˗ Transparency to ensure that the system identifies how and when the
AI system accesses any code, training data, etc. related to personal
information as well as how securely the AI system operates against
adversarial machine learning
Requirements 21
4 Candidate Technologies
Terahertz Technologies
It is inspiring that in March 2019, the Federal Communications Commis-
sion (FCC) opened the spectrum between 95 GHz and 3,000 GHz for exper-
imental use and unlicensed applications to encourage the development
of new wireless communication technologies [13]. Moreover, discussions
on use cases and deployment scenarios for 5G new radio (NR) systems op-
erating at bands beyond 52.6 GHz have begun [14]. Following this trend,
it is inevitable that mobile communications will utilize the terahertz (THz)
bands (i.e., 0.1-10 THz [15]) in future wireless systems. The THz band in-
cludes enormous amount of available bandwidth, which will enable ex-
tremely wideband channels with tens of GHz-wide bandwidth. This could
potentially provide a means to meet the 6G requirement of Tbps data rate.
Considering the advance of related technologies, we expect that 6G would
need to be designed to utilize up to 3,000 GHz as shown in Figure 9.
Figure 9
5G 6G
Spectrum usage for different 4G
generations. 6 GHz 110 GHz 3,000 GHz
While the availability of wideband spectrum is the main driver for THz
communications, other benefits can also be realized. For example, the com-
munication in THz band can potentially provide high-precision positioning
capability for the following reasons: 1) Extremely wideband waveforms in
22
the THz band would enable accurate ranging between transmitter and re-
ceiver (possibly with sub-centimeter-scale accuracy) [16][17]. 2) Links be-
tween transmitter and receiver will most likely be line of sight (LoS), as
discussed further in detail later in this paper. This will greatly improve the
accuracy of distance-based positioning systems. 3) The use of pencil-point
sharp beams steered in both azimuth and elevation will greatly improve
angular resolution and triangulation accuracy of 3D position estimation.
THz Challenges
˗ Severe path-loss and atmospheric absorption: Free-space path-loss
is proportional to the square of the signal frequency. For example, a
link at 280 GHz has 20 dB additional path-loss compared to 28 GHz.
Nevertheless, the severe path-loss in THz band can be overcome, for
example, by utilizing very large antenna arrays at BSs, namely ul-
tra-massive multiple-input multiple-output (MIMO). In addition, the
effect of atmospheric absorption (i.e., absorption by molecules in air)
in the THz band is in general severer than in lower frequencies as the
absorption lines for oxygen and water are mostly located in the THz
band [18]. In order to design efficient THz communication systems in
practice, accurate yet tractable THz multipath channel models need
to be developed for both indoor and outdoor environments.
˗ RF front-end, photonics and data conversion: The THz band is often
referred to as the THz gap due mainly to the lack of existing efficient
devices, which can generate and detect signals in these frequencies.
In these bands, the device dimensions are significantly large relative
to the wavelength, and it results in high power loss or equivalently
low efficiency. On the positive side, during the last decade, research-
ers put great efforts for developing chip-scale THz technologies. As
a result, nowadays semiconductor technologies based on InP, GaAs,
SiGe, and even CMOS are capable of generating power in the mW
range with acceptable efficiency [19][20][21] at low THz band. How-
ever, further development of solid-state electronics is required for
operation in high THz band.
Candidate Technologies 23
other challenges: 1) transporting the signal within the integrated
system and to the antenna with low loss; 2) packaging of the inte-
grated system without significant loss, and maintaining proper heat
dissipation; 3) lowering the mixer phase-noise; 4) low power mul-
ti-Giga-samples-per-second analog-to-digital converters (ADCs)
and digital-to-analog convertors (DACs); and lastly 5) low power
digital input/output (IO) to DACs and ADCs to transfer data at Tbps
data rate with acceptable power consumption.
24
Metamaterial based Antenna and RF Front-End
A metamaterial is usually constructed by arranging multiple tunable el-
ements (PIN diodes, varactor diodes, etc.) in repeating patterns, at scales
that are smaller than the wavelengths [22]. Its precise shape, geometry,
size, orientation, and arrangement enable smart properties capable of ma-
nipulating electromagnetic waves, e.g., blocking, absorbing, enhancing, or
bending waves, to achieve benefits that go beyond what is possible with
conventional materials. In addition, each element constituting a metama-
terial can be controlled independently to achieve desirable characteristics
of the electromagnetic waves such as the direction of propagation and re-
flection. There are three outstanding approaches for utilizing metamaterial
as follows.
RF Processing
Figure 10 and
Metasurface lens. Analog
Beamforming
Metasurface Lens
Baseband RF-Chain
Processing 1
and
Digital
Beamforming RF-Chain
2
RF-Chain
N
Figure 11
Metamaterial antenna.
RF-Chain
Metamaterial Antenna
1
Baseband
Processing
and RF-Chain
Digital 2
Beamforming
RF-Chain
N
Candidate Technologies 25
˗ Reconfigurable intelligent surface (RIS) can be used to provide a
propagation path where no LoS link exists [25]. An example of signal
reflection via RIS is illustrated in Figure 12.
RIS-aided communication
between a BS and a mobile user,
where the LoS path is blocked.
Car
26
Figure 13 Spatially overlapped OAM modes
OAM 1
OAM 1 OAM 1
OAM 2 OAM
OAM 1 2
OAM
OAM 33 OAM
OAM 33
Free-space
Downlink
Figure 14 Cross-link
interference
Main obstacles in deviating from UE
the “mutually exclusive” principle. Self-interference
Uplink Downlink
BS Cross-link
interference
Uplink
UE UE UE UE
Allowing overlap between downlink and uplink over the entire time-fre-
quency resource (a.k.a. “full duplex”) can increase system capacity by two
times, in theory. The main obstacles encountered upon deviating from the
“mutually exclusive” principle include self-interference and cross-link in-
terference. Self-interference experienced by a BS receiver is illustrated in
Figure 14(a). The BS transmits downlink signal using the same time-fre-
quency resource as used for the uplink signal from UEs. Since the BS’s
Candidate Technologies 27
transmit and receive antennas are located in close proximity, self-interfer-
ence is much stronger than the desired signals from the UEs. Therefore, to
evolve duplex technology by departing from the “mutually exclusive” prin-
ciple, it is crucial to be able to remove self-interference. There has been
relevant research on self-interference cancellation (SIC) techniques, which
typically require both analog and digital domain cancellation [32][33].
Figure 15
Dynamic operation of duplex
modes.
Frequency
28
enable flexible network deployments, the mobile industry introduced sup-
port for network entities to connect to BSs via wireless connections, such
as relay in 4G and integrated access and backhaul (IAB) in 5G. In addition
to mobility support for individual mobile devices in the cellular network,
there has been interest in the concept of group mobility (also known as a
mobile relay or mobile BS) for efficient support of mobile devices that are
moving as a group on a bus, a train, or even an airplane.
Moving toward 6G, we expect that the technologies related to the above
trend will further advance to achieve the following.
Figure 16
Mesh type network topology.
Candidate Technologies 29
Another trend that continues to make progress in network topology evo-
lution is the use of non-terrestrial network (NTN) components, e.g., sat-
ellite and HAPS, to provide coverage even in locations where there is no
terrestrial network, as illustrated in Figure 17. Realization of NTN technol-
ogy necessitates consideration of new aspects absent from terrestrial net-
works, including support of moving cells, large cell sizes as great as hun-
dreds of kilometers, long propagation delays, large Doppler shift due to the
high speed of NTN components, and large path-loss. Additional aspects, as
yet unrecognized, may arise and need to be considered, since the mobile
industry is at the initial stage for developing technologies to support NTN.
As NTN components become widely deployed, investigation of technolo-
gies will proceed, to improve the overall performance of communications
involving NTN components and to provide tight integration of NTN compo-
nents in overall operation of mobile communication systems.
Figure 17
Inclusion of non-terrestrial
components in mobile commu- Satellite
nications.
HAPS
Desert Area
BS
Spectrum Sharing
Spectrum sharing technology enables the use of spectrum by multiple
entities. Exclusive licensees often underutilize licensed spectrum because
they do not actively utilize it all the time. Allowing opportunistic use of the
underutilized spectrum by others can make the best use of the limited and
30
precious spectrum resources, especially those at low frequency ranges,
e.g., below 6 GHz. These resources are critically important for guaranteeing
the seamless coverage of mobile communications, but are scarce. We also
observe that regulatory bodies begin to consider deviating from the tradi-
tional exclusive spectrum licensing approach to achieve better utilization
of the limited spectrum. Considering such trends, spectrum sharing tech-
nology is worth paying attention to.
In U.S., the Citizens Broadband Radio Service (CBRS) band (3.55-3.7 GHz)
has been opened for shared access by FCC. The sharing occurs according to
a unique three-tiered access model: incumbents, i.e., federal government
and fixed satellite service users, priority access licensees (PALs) and gen-
eral authorized access (GAA) users, in descending priority order [34]. The
Wireless Innovation Forum (WinnForum) [35] has defined the functionality
and architecture for Spectrum Access Systems (SAS). The SAS framework
provides access to a database and the ability to determine the availability
of CBRS channels at a given location. In addition, Environmental Sensing
Capability (ESC) is a functionality used to detect whether incumbent us-
ers occupy CBRS channels. Together, these two mechanisms maintain and
enforce the hierarchical use of the spectrum. WinnForum also defines the
interface between SAS and CBRS Devices (CBSDs), i.e., BSs, as well as the
framework for testing and certification. At the same time, the CBRS Alli-
ance has been developing the so-called Coexistence Manager (CxM) be-
tween SAS and GAA CBSDs to enable the sharing of the spectrum between
GAA CBSDs in a semi-static manner for the allowed spectrum indicated by
SAS [36]. In addition to the CBRS band, authorities consider making the 37-
37.6 GHz band available for coordinated co-primary shared access between
federal and non-federal users [37].
Candidate Technologies 31
The main challenge of the dynamic spectrum sharing is avoiding (or
minimizing) collision of spectrum usage among different entities while
allowing them to access spectrum in a dynamic manner. Theoretically, to
prevent such collisions, network operators could exchange all relevant
spectrum access information. In practice, however, this would not be possi-
ble because acquiring all required information for every entity in real time
would impose an enormous communication overhead. AI could avoid col-
lisions by predicting the spectrum usage of other entities with a limited
amount of information exchanged, as illustrated in Figure 18.
Figure 18
Intelligent spectrum sharing.
Scheduling Traffic UE
Info. Info. Info.
Information
AI engine
Collision Prediction
Comprehensive AI
AI receives much attention as a tool to solve problems that were previ-
ously deemed intractable due to their tremendous complexity or the lack
of the necessary model and algorithm. In this section, we discuss a com-
prehensive AI system to optimize the overall system performance and net-
work operation.
Figure 19
Comprehensive AI system model.
Joint AI Joint AI
End-to-End AI
Comprehensive AI
32
Local AI is implemented in each entity. An example is the use of AI for op-
timization of modulation, source coding, and channel coding [38][39]. Joint
AI can optimize the joint operation of UEs and BSs or the joint operation of
core networks and application servers. An example opportunity for joint
optimization is handover optimization based on prediction of future net-
work conditions in complex wireless environments [40]. E2E AI optimizes
the entire communication system. With the E2E AI, it becomes possible to
identify or predict anomalies in network operation and suggest corrective
actions [41].
Split Computing
Future applications, such as truly immersive XR, mobile holograms, and
digital replica, require extensive computation capabilities to deliver re-
al-time immersive user experience. However, it would be challenging to
meet such computational requirements solely with mobile devices, espe-
cially, given that many of future mobile devices will tend to become thinner
and lighter. For example, AR glasses should be as light, thin, and small as
regular glasses to meet the user’s expectations.
Candidate Technologies 33
Figure 20 Wireless communication
Low computing power @ mobile device
Split computing. High computing power @ BS
In order to realize the split computing concept, the following factors have
to be considered.
34
Figure 21
Examples of split computing by
avg pool
pool, /2
fc 1000
various devices.
image
Device 100% Edge 0%
avg pool
pool, /2
fc 1000
image
Device 70% Edge 30%
avg pool
pool, /2
fc 1000
image
Device 40% Cloud 60%
avg pool
pool, /2
fc 1000
image
High-Precision Network
To guarantee high QoE for interactive services with high data rate and
low latency requirements, it is important to maintain deterministic E2E
latency and to minimize jitter at the microsecond level. High-precision
network (HPN) is a solution to achieve this, when paired with massive con-
nectivity supported by both radio link protocols and protocols above radio
link. IEEE’s time-sensitive networking (TSN) defines mechanisms for the
transmission of time-sensitive data over Ethernet. Another solution for im-
plementing HPN is IETF’s deterministic networking (DetNet), which spec-
ifies a mechanism defined on Internet Protocol (IP) and transport layers.
These existing technologies have constraints, since TSN was not designed
for mobile networks and DetNet operates on top of TSN. Integrating them
with the mobile network is quite a difficult task due to fundamental mis-
matches between wireless and wired networks. For example, device mobil-
ity in mobile network causes changes in the data path far more frequently
than would be necessary in wired networks.
Candidate Technologies 35
simultaneously. Multi-homing requires support for multi-pathing. Non-IP
solutions such as information-centric networking (ICN) could be candi-
dates to provide these features and mitigate shortcomings of the present
IP suite. ICN changes the focus of internet architecture from host-centric
to data/content-centric. ICN, being content-centric, enables multi-pathing,
end-user mobility, and optimal usage of network bandwidth, since data
and content can be cached and served by intermediate routers.
36
5 6G Timeline
For 6G, we expect ITU-R will begin their work to define a 6G vision in
2021. Taking into account the trend of speeding up of development of tech-
nical standards for each new generation, we expect that the completion of
the 6G standard and its earliest commercialization could happen as early
as 2028, while massive commercialization may occur around 2030.
6G Timeline 37
6 Concluding Remarks
The mobile industry has achieved great successes, from 2G to 4G. While
it is still quite important to work to ensure commercial success of 5G in
coming years, we believe it is the right time to start preparing for 6G. Shap-
ing 6G will require many years, as we have seen with previous generations
in the past. In this spirit, we have presented our initial view of various as-
pects of 6G including the megatrends, services, requirements, candidate
technologies, and timeline for standardization and commercialization. Our
view will naturally be updated as we proceed with our research for 6G in
the future.
38
7 References
[1] Cisco, Cisco Edge-to-Enterprise IoT Analytics for Electric Utilities Solution Overview,
Available: https://www.cisco.com/c/en/us/solutions/collateral/data-center-virtual-
ization/big-data/solution-overview-c22-740248.html
[2] UN Projects World Population to Reach 8.5 Billion by 2030, Driven by Growth in
Developing Countries, Available: https://news.un.org/en/story/2015/07/505352-un-
projects-world-population-reach-85-billion-2030-driven-growth-developing
[3] GSMA, 2019 Mobile Industry Impact Report: Sustainable Development Goals, Sep.
2019, Available: https://www.gsmaintelligence.com/research/?file=a60d6541465e8
6561f37f0f77ebee0f7&download
[4] https://exponentialroadmap.org/wp-content/uploads/2019/09/ExponentialRoadm-
ap_1.5_20190919_Single-Pages.pdf
[8] Xuewu Xu et al., “3D Holographic Display and Its Data Transmission Requirement,” in
Proc. Int’l Conf. Info, Photonics and Optical Commun., pp. 1-4, Oct. 2011.
[11] GrandViewResearch, Digital Twin Market Size Worth $26.07 Billion By 2025 with
References 39
CAGR 38.2%, Available: https://www.grandviewresearch.com/press-release/glob-
al-digital-twin-market
[12] https://medium.com/@DAQRI/motion-to-photon-latency-in-mobile-ar-and-vr-
99f82c480926
[13] FCC Docket 18-21, “FCC Opens Spectrum Horizons for New Services and Technolo-
gies,” Mar. 2019.
[14] 3GPP TR 38.807, “Study on Requirements for NR beyond 52.6 GHz,” Mar. 2019.
[15] Roger D. Pollard, “Guest Editorial,” IEEE Transactions on Microwave Theory and Tech-
niques, vol. 48, no. 4, pp. 625-625, Apr. 2000.
[16] Eirini Karapistoli et al., “An Overview of the IEEE 802.15.4a Standard,” IEEE Commu-
nications Magazine, vol. 48, no. 1, pp. 47-53, Jan. 2010.
[19] Kang Ning et al., “A 140-GHz Power Amplifier in a 250-nm InP Process with 32%
PAE,” Available: https://www.src.org/library/publication/p098807/p098807.pdf
[20] Arda Simsek et al., “A 140 GHz MIMO Transceiver in 45 nm SOI CMOS,” in Proc. IEEE
BCICTS ‘18, pp. 231-234, Oct. 2018.
[21] Kaushik Sengupta et al., “Terahertz Integrated Electronic and Hybrid Electronic-Pho-
tonic Systems,” Nature Electronics, vol. 1, no. 12, pp. 622-635, Dec. 2018.
[22] Ricardo Marqués et al., Metamaterials with Negative Parameters: Theory, Design,
and Microwave Applications, John Wiley & Sons, 2007.
[23] John Brian Pendry, “Negative Refraction Makes a Perfect Lens,” Physical review let-
ters, vol. 85, no. 18, pp. 3966-3969, Oct. 2000.
[25] Chongwen Huang et al., “Reconfigurable Intelligent Surfaces for Energy Efficiency in
Wireless Communication,” IEEE Transactions on Wireless Communications, vol. 18,
no. 8, pp. 4157-4170, Aug. 2019.
[26] Alison M. Yao et al., “Orbital Angular Momentum: Origins, Behavior and Applica-
40
tions,” Advances in Optics and Photonics, vol. 3, no. 2, pp. 161-204, Jun. 2011.
[27] Ove Edfors et al., “Is Orbital Angular Momentum (OAM) Based Radio Communica-
tion an Unexploited Area?,” IEEE Transactions on Antennas and Propagation, vol.
60, no. 2, pp. 1126–1131, Feb. 2012.
[28] Doohwan Lee et al., “Orbital Angular Momentum (OAM) Multiplexing: An Enabler
of a New Era of Wireless Communications,” IEICE Transactions on Communications,
vol. 100, no. 7, pp. 1044-1063, Jul. 2017.
[30] Kenneth E. Kolodziej et al., “In-Band Full-Duplex Technology: Techniques and Sys-
tems Survey,” IEEE Transactions on Microwave Theory and Techniques, vol. 67, no. 7,
pp. 3025-3041, Jul. 2019.
[32] Min Soo Sim et al., “Nonlinear Self-Interference Cancellation for Full-Duplex Radios:
From Link-Level and System-Level Performance Perspectives,” IEEE Communica-
tions Magazine, vol. 55, no. 9, pp. 158-167, Sep. 2017.
[34] FCC GN Docket No. 17-258, “Promoting Investment in the 3550-3700 MHz Band,”
Oct. 2018.
[35] https://www.wirelessinnovation.org/
[37] FCC 16-89 Report and Order and Further Notice of Proposed Rulemaking, “Use of
Spectrum Band Above 24 GHz for Mobile Radio Services,” Jul. 2016.
[39] Timothy O’Shea et al., “An Introduction to Deep Learning for the Physical Layer,”
IEEE Transactions on Cognitive Communications and Networking, vol. 3, no. 4, pp.
563–575, Dec. 2017.
References 41
tions, vol. 19, no. 6, pp. 4038-4053, Jun. 2020.
[42] 3GPP TR 23.791, “Study of Enablers for Network Automation for 5G,” Jun. 2019.
[43] O-RAN Alliance White Paper, “O-RAN: Towards an Open and Smart RAN,” Oct. 2018.
42
research.samsung.com