Communications Network - Building a Fully Connected, Intelligent World - Huawei
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Communications
Network
Building a Fully Connected,
Intelligent World
Contents Industry Trends P01 Future Network Use Cases P05 Next-Generation Human-Machine Interaction Network: A Human-centric Hyperreal Experience Networks That Deliver a Consistent Experience for Homes, Offices, and Vehicles: The Third Space with the Same Broadband Experience Satellite Broadband Internet: Continuous Broadband Coverage from Ground to Air Industrial Internet: A New Type of Network for Intelligent Manufacturing and Human-Robot Collaboration Computing Power Network: Orienting Towards Machine Cognition and Connecting Massive Amounts of User Data and Computing Power Services at Multiple Levels Cognitive Network: Evolution Towards Advanced Levels of Intelligence Vision for Future Networks and Their Defining Features P19 Vision for Future Networks Defining Features20 Summary and Technology Outlook Recommendations P36 Appendix A: Acronyms and Abbreviations P37
Communications Network 2030
Industry Trends
G
oing intelligent has become the 90% of European SMEs will reach at least a
general direction that the world is basic level of "digital intensity". To achieve
heading in over the coming decade. these targets, the EU announced an increase
China, the EU, and the US have all published in investment into energy and digital
their new visions for this area. infrastructure.
In its Outline of the 14th Five-Year In its Vision 2030 report, the US
Plan (2021–2025) for National Economic National Science Board (NSB) recommends
and Social Development and the Long- increasing investment in data, software,
Range Objectives Through the Year 2035, computing, and networking capabilities over
China prioritizes industry intelligence as the next decade in order to help maintain
an important area of development, and the US's competitiveness in the digital
sets clear development goals for industries economy.
including manufacturing, energy, agriculture, The intelligent development of
healthcare, and education, as well as for industries first requires companies to
government management. upgrade their networks. In its Industrial
In its 2030 Digital Compass plan, the Internet Innovation and Development Action
EU articulates the following targets: By 2030, Plan (2021–2023), the Chinese government
75% of European enterprises will have taken put forward the following measures: (1)
up cloud computing services, big data, and Accelerate the network-based development
Artificial Intelligence (AI), and more than of industrial equipment, drive the upgrade
1
Communications Network 2030
of enterprise Intranet, and promote the edge computing.
integration of information technology (IT) • China Telecom has set out the goal
networks and operational technology (OT) of building an integrated cloud-
networks to build industrial Internet campus network architecture by 2030.
networks. (2) Explore the deployment of new • China Unicom published its CUBE-
technologies such as cloud-network synergy, Net 3.0 strategy, which articulates
deterministic networking, and Segment a new development direction that
Routing over IPv6 (SRv6). In its Digitising combines connectivity, computing,
European Industry platform plan, the EU and intelligence.
considers nanophotonics, AI, 5G, and Internet • In its outlook for 2030, Deutsche
of Things (IoT) to be key enablers of future Telekom aims to become the
industrial networks, and plans to increase leading digital enabler in the B2B
investment in these technologies in order to market, providing comprehensive
stay ahead in the future. network, IoT, cloud, and digital
As industries increasingly adopt services.
intelligent technologies, leading telecom A survey conducted by GSMA shows
carriers around the world are taking action that B2B, cloud, and IoT services that
and beginning to explore how they can fully target industry, finance, health, energy, and
unleash the potential of connectivity in this agriculture will be the most promising areas
process. For example: for carriers worldwide to fully unleash the
• China Mobile has unveiled a "5G potential of their connectivity portfolios.
+ AICDE" development strategy, In the world of 2030, many amazing
where AICDE stands for AI, IoT, things that we can only dream of today will
cloud computing, big data, and be a reality.
2
Communications Network 2030
With highly sensitive biosensors and such as collaborative robots, automated
intelligent hardware connected through mobile robots (AMRs), and digital labor, can
broadband networks, we can obtain and be adopted in tandem with the industrial
track the indicators of our physical health Internet to increase accuracy and decrease
in real time, and securely store massive costs throughout the whole process from
amounts of health data in the cloud. This demand to production and delivery,
will allow us to proactively manage our while also improving the resilience of the
own health and reduce our dependence manufacturing industry.
on doctors, thus improving our health and Energy IoT can be integrated into
quality of life. smart grids to form a green energy Internet
New technologies, such as home and fully digitalize all activities, including
broadband that supports speeds of over 10 generation, grid, load, and storage. Zero-
Gbit/s and holographic communications, carbon data centers and zero-carbon
will enable more intuitive human-machine communications sites may soon become
interactions. An air-ground cubic network a reality. We can also guarantee digital
will connect all means of transportation, security and trustworthiness by combining
facilitating easy, smart, and low-carbon blockchain, digital watermarking, AI-driven
travel. Sensing technology, 10-gigabit wired anti-counterfeiting, privacy-enhancing
and wireless broadband, inclusive AI, and computing, and endogenous network
applications that target numerous industries security.
will be available everywhere, allowing us In 2030, communications networks will
to build urban digital infrastructure that evolve from connecting billions of people to
improves the quality of city life. connecting hundreds of billions of things,
With Harmonized Communication and and face many challenges along the way.
Sensing (HCS), automation, and intelligence First, the scale of communications
technologies, we will be able to efficiently networks will continue expanding. This
protect our environment. New types of labor, means network management will become
3
Communications Network 2030
even more complex, so networks must these factors are increasingly becoming
become more intelligent. Over the next bottlenecks. Over the next decade, how can
decade, how can we innovate in software we innovate in fundamental technologies
technology to prevent operation & to build a green, low-carbon network and
maintenance (O&M) costs from rising in step increase network capacity by dozens of times
with the continuous expansion of network without increasing energy consumption? This
scale? This poses a daunting challenge. is another extremely challenging task that
Second, IoT scenarios such as lies ahead of us.
unattended operations in industrial and Communications networks are one of
agricultural settings and self-driving vehicles the major forces driving the world forward.
will require carriers to further improve the The development of communications
coverage, quality assurance, security, and networks kicked off during the first Industrial
trustworthiness of their networks. Over Revolution and, unlike traditional industries,
the next decade, how can we innovate in it still shows no signs of slowing down
protocols and algorithms to enable networks after nearly two centuries. In fact, the
to carry multiple types of services while pace of development of communications
meeting the requirements for high quality technologies has been particularly rapid in
and flexibility? This will be a very challenging recent decades. Both the evolution from 2G
task. to 5G and the shift from the asymmetric
Third, although Moore's law has held digital subscriber lines (ADSL) to gigabit
true for decades, the semiconductor industry optical home broadband took just 30 years.
is now struggling to maintain that pace of Over the next decade, we will witness the
improvement, and new technologies like emergence of new use cases and scenarios
quantum computing are not yet mature. for communications technologies and fully
Meanwhile, demand for computing embrace an intelligent world.
power, storage capacity, and network
energy efficiency continues to grow, and
4
Communications Network 2030
Future Network Use Cases
S
ince Samuel Morse invented the electric computing nodes, and cloud resources.
telegraph in 1837, communications The scope of network connections is
networks have come a long way, expanding, business needs are changing, and
moving from connecting individuals and the industry has reached a consensus that,
homes to connecting organizations. In today's over the next 10 years, networks will evolve
environment of diverse and rapidly changing from 5G to 5.5G/6G, from F5G to F5.5G/F6G,
services, it takes continuous innovation for and from IPv4/Multiprotocol Label Switching
communications networks to keep up with (MPLS) to IPv6+, and the autonomous driving
the needs of customers. To meet the rich network will evolve from L2 to L5. In addition,
and diverse business needs that will arise in new use cases will continue to emerge.
the intelligent world of the next 10 years,
communications networks will need to go Next-Generation Human-
beyond connecting individuals, to connect Machine Interaction
multiple perception, display, and computing Network: A Human-centric
resources related to each individual. In the near
Hyperreal Experience
future, networks will have to connect home
users as well as home appliances, vehicles, In a world of cold machines, it is up to
and content resources, while organizations will human beings to adapt to the machines. With the
expect networks to do more than just create wide use of the automobile, we learned to work
connections between employees – they must with pedals and a gearstick. In the PC era, we
also connect an organization's machines, edge learned to use the mouse and keyboard. In the
5
Communications Network 2030
smartphone era, we learned to use touchscreens. interaction. ( Figure 1 Hyperreal human-machine
However, with sufficiently advanced levels of interaction experience )
intelligence, it is possible to turn this paradigm on Over the course of the coming decade,
its head and have machines adapt to the needs communications networks must evolve
of their human users. Intelligent machines (e.g., to support brand-new human-machine
smart screens, smart home appliances, intelligent interaction experiences such as XR, naked-eye
vehicles, and smart exoskeletons) will be able to 3D display, digital touch, and digital smell.
understand natural language, gestures, and eye
XR: An Intuitive Interaction
movement, and even read human brain waves,
Experience Through a Perfect
enabling more intuitive integration between
the virtual and physical worlds and bringing a
Synthesis of the Virtual and
hyperreal sensory experience to human-machine Physical Worlds
Figure 1 Hyperreal human-machine interaction experience
6
Communications Network 2030
Table 1 Network requirements of XR services
Virtual Reality (VR) is about rendering which are expected to go to market in about
packaged digital visual and audio content. two years. With the wide adoption of 5G, Wi-Fi
Augmented Reality (AR) refers to the overlaying 6, and fiber broadband, all of which can deliver
of information or artificially generated content gigabit speeds, XR services are set to boom
onto the existing environment. Mixed Reality (MR) over the next decade. Huawei predicts that the
is an advanced form of AR that integrates virtual number of VR/AR users is expected to reach 1
elements into physical scenarios. eXtended Reality billion by 2030.
(XR), which covers VR, AR, and MR, is a catchall In its Virtual Reality/Augmented
term that refers to all real and virtual combined Reality White Paper, the China Academy of
environments and human-machine interactions Information and Communications Technology
generated by computer technology and wearables. (CAICT) divides the technical architecture of XR
Characterized by three-dimensional environments, into five parts: near-eye display, perception and
intuitive interactions, spatial computing, and other interaction, network transmission, rendering
features that set it apart from existing Internet processing, and content creation. The white
devices, XR is considered the next major platform paper also predicts the development stages
for personal interactions. of XR. The CAICT's conclusions have, to some
In 2020, due to the impact of social extent, been endorsed by the ICT industry. (
distancing caused by COVID-19, demand for Table 1 Network requirements of XR services )
VR games, virtual meetings, and AR-assisted Currently, XR is still at the stage of
temperature taking increased exponentially. partial immersion. Today, a typical XR
The number of active VR users on the US experience involves 2K monocular resolution,
video game digital distribution service Steam 100°–120° FOV, 100 Mbit/s bitrate, and
doubled. Some manufacturers have unveiled 20 ms motion-to-photon (MTP) latency. If
more portable AR-enabled contact lenses, all content is rendered in the cloud, 20 ms
7Communications Network 2030
of MTP latency is the threshold above which 3D display involves three major phases:
users start to report feelings of dizziness. the digitalization of 3D objects, network
We predict that XR will reach the stage of transmission, and optical or computational
full immersion by 2030, by which time it will be reconstruction and display.
supported by 8K monocular resolution, 200° There are two types of naked-eye
FOV, and a gigabit-level bitrate. If all rendering is 3D display technology: light field display
still conducted in the cloud, MTP latency will need (through lenslets) and the use of spatial light
to be kept below 5 ms. If technology is developed modulators (SLMs).
to support the local rendering of environment- Light field display leverages the binocular
related content that could easily make users parallax to create 3D visual effects. It uses
dizzy, the latency will be specifically linked to the parallax barriers, lenticular lenses, and
types of content. For content that requires only directional backlight, all of which impose fairly
weak interaction (such as a streamed video), 20 inflexible requirements in terms of viewing
ms of MTP latency is acceptable. For content that angles. Their large-scale adoption would
requires strong interaction like games, less than 5 require real-time capturing of user location
ms MTP latency will be needed. and dynamic adjustment.
Therefore, to support the development of An alternative approach would be to
XR services over the next 10 years, networks use SLMs. An interferometric method is used
must have bandwidth of at least 1 Gbit/s and to store all amplitude and phase information
latency of less than either 5 ms or 20 ms, of light waves scattered on the surface of a
depending on the scenario. 3D object in a recording medium. When the
hologram is irradiated with the same visible
Naked-eye 3D Display: A Brand-new
light, the original object light wave can be
Visual Experience Through Lifelike
reproduced thanks to diffraction, providing
Image Reproduction users with a lifelike visual experience. (Table 2
The implementation of naked-eye Network requirements of naked-eye 3D display)
Table 2 Network requirements of naked-eye 3D display
References: IEEE 1981.1 Tactile Internet and Digital Holography and 3D Display
8Communications Network 2030
Table 3 Network requirements of digital touch
Reference: IEEE 1981.1 Tactile Internet
In recent years, naked-eye 3D display products will require higher bandwidth (over 1
featuring light field display has developed Tbit/s), but we do not expect them to be ready
rapidly, in step with the development of for large-scale commercial deployment by
user location awareness and computing 2030.
technologies. Some manufacturers are already Therefore, the naked-eye 3D display
showcasing their products. We predict that a products coming to market over the next
large number of use cases will emerge in the decade will need to be supported by networks
entertainment and commercial sectors by 2025. capable of delivering 1–10 Gbit/s bandwidth
This type of 3D display requires gigabit-level per user, latency of 1–5 ms, and 99.999%
bandwidth and real-time interaction. In strong availability.
interaction scenarios, the network latency must
Digital Touch: Tactile Internet Made
be less than 5 ms, and commercial applications
Possible Through Multi-dimensional
will require network availability of 99.999%
(this means annual downtime must be less
Sensory Interaction
than 5 minutes and 15 seconds). In IEEE's tactile Internet architecture,
Over the past several years, digital tactile technology is divided into three
breakthroughs have also been made in layers: user layer, network layer, and avatar
holographic technology, which is based on layer. The user layer enters information such
optical reconstruction. Product prototypes as location, speed, force, and impedance.
have been developed with a thickness of 10 After being digitalized over the network,
cm and a projection size of around 100 cm2. the information is converted into instruction
We predict that these small-scale holographic data and provided to the avatar layer. The
products will become commercially available avatar layer then collects tactile, auditory, and
at exhibitions, for teaching purposes, and as proprioception data and provides the data to
personal portable devices over the next 10 the user layer through the Internet to inform
years. They will require bandwidth of around users' real-time decision making.
10 Gbit/s, latency of no more than 5 ms or as Digital tactile technology has two
low as 1 ms, and network availability of more interaction modes. The first is machine control.
than 99.999%, the same as that required in Use cases include remote driving and remote
commercial settings. True-to-life holographic control. The second is hyperfine interaction,
9Communications Network 2030
and use cases include electronic skin and Interaction
remote surgery. (Table 3 Network requirements Among our five senses, two of them –
of digital touch) touch and taste – require direct contact, while
Machine control has numerous use three – sight, hearing, and smell – do not. Of
cases in industrial settings, and has high the latter three, smell involves the deepest
requirements for network availability (above interaction.
99.999%). Some industries even require Digital smell includes three technical
availability to reach 99.99999%. The required phases: odor perception, network transmission,
bandwidth is generally less than 100 Mbit/s, and smell reproduction.
and the maximum permissible latency varies There have been some use cases for odor
from 1 to 10 ms, depending on the specific perception, such as using composite materials
circumstances. to form a barcode, which can generate
Electronic skin powered by flexible chemical reactions according to the odor and
electronics in hyperfine interactions has the create color changes. The relationship between
most development potential. Electronic skin the barcode and odor can then be identified
integrates a large number of high-precision through Deep Convolutional Neural Network
sensors such as pressure and temperature (DCNN) algorithms. Use cases can be found in
sensors. According to a study by the University specific scenarios like detection of dangerous
of Surrey in the UK, each square inch of goods and detection of food freshness.
electronic skin will require bandwidth of There are already some commercial odor
20 to 50 Mbit/s, meaning that an average reproduction products available in the industry,
hand would require bandwidth of 1 Gbit/ such as smelling generators for VR games,
s. The wearers of electronic skin won't all be which use five odor cartridges and selectively
humans; intelligent machines present another release odors from the cartridges. They emit
class of potential users. The user layer may scents such as the ocean, gunpowder, wood,
perform analysis, computing, and decision and soil, deepening the immersion of the
making based on the massive amounts of gaming experience. However, some research
data collected by the electronic skin on the reports suggest that the future of smell in
avatar layer to control the avatar layer. The VR won't rely on these odor cartridges, but
user layer can also be directly connected to will instead work through brain-computer
humans through brain-computer interfaces interfaces to enable people to sense odors
or myoelectric neural interfaces to deliver an more directly and accurately.
immersive remote interaction experience. We The combination of odor perception
predict that network bandwidth of 1 to 10 (using electronic noses) and odor reproduction
Gbit/s will be required in hyperfine interaction can help create an Internet that enables us
scenarios. to not only hear and see, but also smell. It is
Therefore, to support digital touch, not yet clear what kind of network bandwidth
networks will need to deliver 1–10 Gbit/s and latency this function will require, but the
bandwidth per user, availability greater than computing requirements are already relatively
99.999%, and latency below 10 ms, or as low well understood.
as 1 ms in certain use cases. In a nutshell, the next-generation human-
machine interaction network will support
Digital Smell: Internet That Enables
brand-new experiences including XR, naked-
Us to Smell Through Deep Sensory
eye 3D display, digital touch, and digital smell.
10Communications Network 2030
Making these technologies work will require and users will enjoy the same broadband
networks capable of delivering bandwidth of service experience in all three scenarios. (Table
10 Gbit/s and 99.999% availability, with latency 4 Network requirements for delivering a
as low as 1 ms for some use cases. consistent experience at home, in the office,
and on the go)
Networks That Deliver a Over the next decade, common home and
office services will include smart screens, multi-
Consistent Experience for
screen collaboration, 3D, holographic teaching,
Homes, Offices, and Vehicles:
and XR. As true-to-life holographic meetings will
The Third Space with the not be widely adopted by 2030, the mainstream
Same Broadband Experience broadband requirements of home and office
services will stand at 1 to 10 Gbit/s of bandwidth
When we envision the future of self- and less than 5 ms of latency. In the future,
driving cars, the most appealing feature for home and office networks will not only provide
many is that we will be able to enjoy the seamless broadband coverage, but also support
immersive entertainment, social, and work brand-new scenarios such as working from
experience we get at home while on the go. home, premise security, and robotics. Based
Multi-screen collaboration, 3D display, and on HCS capabilities, home networks will be
holograms will all be used both at home and able to sense user locations, indoor space, and
in cars. 8K and 16K smart screens will be environment security, and create a more user-
gradually adopted at home and MR will be friendly living and work environment for people.
widely used in cars. Services like multi-screen collaboration, 3D,
With 5G, F5G, and Wi-Fi 6, mobile and holographic teaching, and XR will also be available
fixed broadband basically enters the gigabit in our self-driving cars. Over the next decade, their
era at the same time, making it possible to key requirements for network bandwidth will be
deliver the same level of experience to users 1 to 10 Gbit/s, and latency requirements will be
regardless of whether they are at home, in the less than 5 ms. As autonomous driving will require
office, or on the go. vehicle-road collaboration, it will require network
In the future, self-driving cars will become availability greater than 99.999% and positioning
the "third space" beyond homes and offices, precision of 10 cm.
Table 4 Network requirements for delivering a consistent experience at home, in
the office, and on the go
11Communications Network 2030
If networks are to meet the needs broadband network from ground to air will
of these new technologies and provide a deliver new experiences to people and power
consistent experience across our three spaces the full intelligent transformation of industries.
(home, office, and self-driving cars), we will (Figure 2 Satellite broadband network)
need to build new network capabilities that Due to spectrum resource constraints
deliver the high bandwidth, high availability, and communications disruptions, the actual
and low latency required. peak capacity of a single LEO satellite is
about 10–20 Gbit/s. Suppose a global satellite
Satellite Broadband network is supported by 10,000 satellites
Internet: Continuous distributed on multiple orbital planes from
Broadband Coverage from very low earth orbits (VLEOs) to LEOs, and
each satellite maintains links with satellites
Ground to Air
around it in all directions using over 100 Gbit/
Over the next 10 years, connected drones s lasercom. The actual effective capacity of
will be more widely adopted, helping create the satellite network will be around 100 Tbit/
new markets worth tens of billions of US s, considering at least half of the areas passed
dollars. We will see an increase in intra-city over by satellites are areas where demand for
passenger transport in the skies above our broadband is minimal (e.g., seas and deserts).
cities, supported by tens of thousands of low- Outside of areas covered by cellular networks,
earth orbit (LEO) broadband satellites. Satellite satellite broadband providers can use multi-
broadband will be commercially available channel phased array antennas to deliver
on a large scale, which will help turn space hundred-megabit broadband to consumers
tourism and deep-sea exploration into popular and use dual parabolic antennas to deliver
leisure activities. Broadband will become an gigabit broadband to enterprise customers.
indispensable part of our life; the increased They could also transmit data over optical
diversification of our leisure activities and the inter-satellite links to hundreds of gateways
growing demand for unmanned operations around the world where they can connect to
in intelligent industry and agriculture mean the Internet. Such a satellite network would be
that broadband will be needed everywhere, equivalent to a quasi-4G network, providing
from land to sea and sky. A comprehensive three-dimensional global coverage and latency
Figure 2 Satellite broadband network
12Communications Network 2030
of less than 100 ms. the development of intelligent manufacturing
Currently, terminal antennas for LEO requires a new architecture that will facilitate
satellite broadband are large, meaning they are human-robot collaboration.
ill-suited to mobile scenarios for individuals. The new architecture will be built upon
Current satellite broadband networks mostly three equal elements – humans, robots, and an
serve homes in remote areas, enterprises, and intelligent platform (cloud/edge computing).
ships. Some carriers have combined satellite Private industrial communication buses will
broadband, which is used for backhaul, with be replaced by universal industrial networks
cellular networks and WLAN networks on and open data layers that support real-time
the ground to provide both broadband and data transmission. The intelligent platform
narrowband coverage for villages or enterprises will aggregate data collected from humans
in remote areas. With wider adoption of and robots for real-time analysis and decision
satellite broadband, we may see it applied to making and support effective collaboration
mobile scenarios (terminals) such as connected between humans and robots.
cars and small personal devices. Satellite Huawei predicts that the total number
broadband will deliver a seamless, continuous of global connections will reach 200 billion
broadband experience beyond home Wi-Fi and by 2030, including about 100 billion wireless
urban cellular networks, meeting the network (cellular) connections (including passive
requirements of people and things. cellular connections) and about 100 billion
wired, Wi-Fi, and short-range connections. In
Industrial Internet: A
industrial settings, the billions of connected
New Type of Network for devices will include not only pressure,
Intelligent Manufacturing and photoelectric, and temperature and humidity
Human-Robot Collaboration sensors, but also a large number of intelligent
cameras, intelligent cars, drones, and robots.
The industrial Internet is a new type of Industrial networks, currently characterized
infrastructure that deeply integrates ICT into by a fragmented landscape of different
the industrial economy and fully connects narrowband technologies, will adopt universal
people, machines, things, and systems. For broadband technologies.
industries, this means the birth of a brand- Universal industrial networks will erase the
new manufacturing and service system that technical boundaries between consumption,
covers entire industry value chains and paves office work, and production. These networks
the way for digitalization, network-based will support multiple types of services using
operations, and the intelligent transformation deterministic broadband networks and slicing
of all industries. An industrial Internet system technologies, such as 5G, Time Sensitive
consists of four key components: industrial Networking (TSN), IPv6+, and industrial optical
control, industrial software, industrial network, networks, allowing companies to connect any
and information security. The industrial workforce and migrate all consumption, office
network is the foundation of the entire system. work, and production elements to the cloud.
Traditional industrial networks are built Universal industrial networks will enable
based on the ISA-95 pyramid model. This on-demand data sharing and seamless
architecture was introduced more than 20 collaboration between office and production
years ago and is a manufacturing system systems within a company, between
centered on human management. However, different companies in the same industry,
13Communications Network 2030
and even between the related services of latency requirements of each service and
different vertical industries. They will support forecasts on the number of devices used by
broadband-based interconnectivity and multi- enterprises in 2030, we predict that a medium-
cloud data sharing of any workload. to large-sized enterprise will require network
Universal industrial networks will also be bandwidth of 100 Gbit/s and the maximum
smarter than ever, facilitating the movement bandwidth per user will reach 10 Gbit/s.
of data in boundary-free and mobile scenarios Acceptable latency will vary greatly from one
across industries and across clouds. They will use case to another, from as low as 1 ms to as
support intent-driven automated network high as 100 ms. In addition, it will be necessary
management and AI-based proactive security to ensure the security and trustworthiness of
and privacy protection, ensuring service industrial networks.
security and trustworthiness at any workplace.
An enterprise usually has multiple types Computing Power Network:
of services, so a universal industrial network Orienting Towards Machine
must ensure the availability, security, and
Cognition and Connecting
trustworthiness of services. For example,
Massive Amounts of User
smart healthcare involves services such as
remote diagnosis, monitoring & nursing, and Data and Computing Power
remote surgery; a smart grid involves video- Services at Multiple Levels
based inspection, grid control, and wireless
monitoring; and smart manufacturing The social value of communications
involves factory environment monitoring, networks is reflected in the services they
information collection, and operation control. support. In the past, networks helped establish
(Table 5 Network requirements of intelligent communications channels between people
enterprises) by providing communications services. Today,
Based on the typical bandwidth and with smart devices and the cloud connected
Table 5 Network requirements of intelligent enterprises
Reference: CAICT, Research Report on Industry SLA Requirements for 5G End-to-end Network Slicing
14Communications Network 2030
to networks, more diverse content services are vision, and self-driving vehicles will have
provided through communications networks. enhanced performance in four dimensions:
The networks we use today are designed • Cognitive capacity: Systems will be able to
for human cognition. For example, the frame capture objects in the physical world more
rate for motion video (typically 30 frames per finely, precisely, and in a multi-sensory
second [FPS]) is chosen based on the human manner. For instance, in manufacturing
ability to perceive motion, and the audio data monitoring systems, motion capture at
collected is compressed with mechanisms 120 FPS will detect anomalies that would
that take advantage of the masking effects otherwise be undetectable.
of the human cognitive system. For human • Response speed: Systems will be able
perception, such encoded audio and video can to respond to the status change of a
be considered high quality. However, for use controlled object within 10 ms.
cases that require beyond-human perception, • Scalability in computing: Systems will be
the level of quality may be far from enough. able to accommodate varying and uncertain
For example, robotic monitoring systems workload while achieving high resource
will need to detect anomalies by listening to utilization, through methods such as dynamic
sounds beyond the human audible frequency linear scaling of computing resources.
range. In addition, the average human • Energy efficiency: Energy efficiency can be
response speed upon seeing an event is about greatly improved if enterprises eliminate
100 ms. Therefore, many applications have on-premise computing resources and
been designed based on this latency. However, adopt a cloud-based model. Moreover,
for certain applications that are beyond human energy efficiency will be further improved
usage, such as emergency stop systems, with an event-driven approach where
shorter response time is required. a system is deployed on a serverless
The Innovative Optical and Wireless computing platform.
Network Global Forum Vision 2030 and Intelligent machines will create more
Technical Directions states that compared accurate data. For example, network clocks
with today's networks that are designed for and geolocation stamps can be used for
human cognition, future networks designed precise modeling of the physical world in a
for intelligent machines such as XR, machine digital twin system. This will lead to a shift in
15Communications Network 2030
data processing and computing, from today's encountered when manufacturing CPUs with
Internet platform-centric model to a data- more than 128 cores in smart devices. In
centric model, decoupling data, computing, addition, due to bandwidth costs and latency,
and communications. cloud data centers may not be able to satisfy
The network infrastructure designed for the massive amount of time-sensitive service
machine cognition should satisfy the following processing required by intelligent systems.
requirements: Machine cognition requires a new type of
• Accommodating the collection and network in which data analytics and processing
transmission of massive amounts of can be performed on the edge, and not all data
data, having an ultra-low latency, and needs to be transmitted to the central cloud.
supporting a very large number of In the future, the cloud, edge, and devices
subscribers. will be connected, and computing workloads
• Managing publishers' data generation and will be apportioned to one of three levels
injection based on the overall condition (distributed edge nodes in a city, regional data
of the system and the importance of the center clusters that cover multiple cities, or
data. backbone centralized data centers) in real time
• Supporting the storage and sharing based on their latency thresholds. In use cases
of data among communications and that can tolerate latency of about 100 ms, data
computing nodes in the network. may be sent to a centralized data center. In
• Supporting precision time and geolocation use cases with lower latency tolerance (from
stamping. 10 ms to as low as 1 ms), computing will be
• Providing strong protections for data performed in a regional data center cluster or at
security, privacy, and integrity. the edge. （Figure 3 Three levels of computing
• Providing a data brokerage between resources for machine data services）
IP and non-IP nodes, with the data Computing efficiency and reliability are
brokerage being accessible through correlated with network bandwidth, latency,
multiple networks. security, and isolation. Therefore, computing
As the miniaturization of chips approaches and networks should be coordinated. Major
its physical limits, the computing industry carriers have articulated a new business vision
can no longer rely on Moore's law for rapid for computing and network convergence
development. Economic bottlenecks are services based on a new concept of "computing
Figure 3 Three levels of computing resources for machine data services
16Communications Network 2030
power network". They aim to connect diverse Many carriers have incorporated the computing
computing power in the cloud, on the edge, power network into their 6G and future network
and across devices to implement on-demand research. The computing power network will
scheduling and sharing for efficient computing be a key scenario for communications network
power services at multiple levels. The computing evolution over the next 10 years.
power network represents a significant shift
in network design, from focusing on human Cognitive Network: Evolution
cognition to focusing on machine cognition. Towards Advanced Levels of
The Chinese government released the
Intelligence
Guiding Opinions on Accelerating the Construction
of Collaborative Innovation System of National In the academic community, technological
Integrated Big Data Centers, which states: "With advances are often personified to help us
the acceleration of digital transformation and understand them more easily. In 1877, German
upgrade in various industries, the total volume philosopher Ernst Kapp first put forward the
of data being created by society as a whole is concept and theory that "tools and instruments
growing explosively, and the requirements for data produced by human hands conform to already
resource storage, computing, and applications existing organic structures" in the Elements
are greatly increasing. Consequently, there is an of a Philosophy of Technology. In 1964, in his
urgent need to promote an appropriate data book Understanding Media, Marshall McLuhan
center layout, balance between supply and proposed that mechanical technologies
demand, green and centralized development, extended our bodies in space and that electric
and interconnectivity. We should build a new technology extended our central nervous
computing power network system that integrates system. In 1995, in The Global Brain Awakens,
data centers, cloud computing, and big data, Peter Russell stated that the various connections
in order to promote flows and application of were making the earth an embryonic brain and
data elements and achieve green and quality the earth is awakening. The analogies used to
development of data centers." In addition, the describe the tools and technologies humans
document proposed that "as data centers should have created have shifted from body to nerve
be developed on a large scale in a centralized and and to brain as the digital world has advanced
green manner, network transmission channels towards advanced levels of intelligence.
between national hubs and nodes should be Communications networks have existed in
further streamlined to accelerate the program one form or another for about two centuries.
of 'Eastern Data and Western Computing' Today, the telegraph and analog telephone
and improve cross-region computing power networks of the 19th century have long since
scheduling." been superseded by more advanced digital
To support proactive development of networks.
computing power network standards, ITU-T has Over the past 50 years, mobile
launched the Y.2500 series of computing power communications, optical communications, and
network standards, with Y.2501 (Computing data communications networks have evolved
Power Network – framework and architecture) continuously to remain relevant. These network
as the first standard. This series of standards will types, together with optical cables, equipment
be compatible with a raft of computing power rooms, and sites, form a robust network trunk.
network standards developed by the China The biggest evolutionary step taken in
Communications Standards Association (CCSA). the last decade is the development of what
17Communications Network 2030
the academic community could understand
as being the network's nervous system. The
human nervous system comprises basic
systems that can initiate automatic stress
responses and feature closed-loop control, as
well as more advanced systems that support
thinking, analysis, and cognition (i.e., the brain).
The current shift from the software-defined
network (SDN) to the autonomous driving
network (ADN) is analogous to the evolution
of a basic nervous system for networks.
Over the next 10 years, the network
nervous system will evolve along the following
two tracks. The first is HCS, e.g., wireless sensing, abnormal behaviors at the packet level.
Wi-Fi sensing, and optical sensing. The second is The concept of cognitive networks
the development of a "brain" – the digital twin is not new. Some renowned universities,
of the physical world that allows autonomous research institutes, and companies have been
inference and decision making. This is how doing related research for years, but few
networks will evolve towards advanced levels of breakthroughs have been made. Cognitive
intelligence and feature cognitive intelligence. technology was first used in wireless networks.
Cognitive intelligence is both an In 2004, the IEEE established the IEEE 802.22
engineering and a mathematical issue. To Working Group and developed the first wireless
qualify as having cognitive intelligence, a standard based on cognitive technology. Recent
system must be able to sense status changes years have seen AI breakthroughs in many
internally and externally in real time and industries. Self-driving cars have already driven
manage itself through autonomous analysis millions of miles in the real world, completely
and prediction. autonomously. In production quality control, AI
The construction of cognitive intelligence vision has significantly cut inspection times. In
consists of two dimensions: time and function. agriculture, the efficiency of intelligent apple
Time: A network may predict changes pickers is more than double that of human
in the future (T3) by learning historical workers. The communications industry is also
information (T1 and T2). For example, an exploring the use of AI in networks. We hope
L5 ADN will be able to accurately predict that over the next 10 years, the combination
performance degradation based on historical of AI and digital twin technology will lead
performance records and warnings. to breakthroughs in cognitive networks, and
Function: A network may predict changes prediction and judgment of network status will
in its functions (information C) by learning be significantly improved though analysis and
function-related information in different inference of multi-dimensional data.
environments (information A and information As part of the digital world that is about
B). For example, a cognitive wireless system to awaken, communications networks will
may predict user service changes by identifying have both harmonized sensing capabilities
changes in user locations and channels, and in and cognitive intelligence, evolving to a higher
cyber security, a network may predict changes form with a robust trunk, sensitive perceptions,
in the security situation based on detection of and an agile "brain".
18Communications Network 2030
Vision for Future Networks
and Their Defining Features
Vision for Future Networks supported by green and cubic broadband
networks that are AI-native, secure,
F
uture networks won't just connect trustworthy, and capable of providing
billions of people; they will connect deterministic experiences and HCS. （Figure
hundreds of billions of things. 4 Vision for the communications network of
We envision those connections as being 2030）
Figure 4 Vision for the communications network of 2030
19Communications Network 2030
Defining Features Today's gigabit access enabled by 5G, F5G,
and Wi-Fi 6 for homes, individuals, and
The communications networks of 2030 organizations will evolve toward 10 gigabit
will have six defining features enabled capacity enabled by 6G, F6G, and Wi-Fi 8.
by 15 key technologies, and each key Huawei predicts that the average monthly
technology will rely on research on multiple data use on wireless cellular networks
technological fronts. （Figure 5 Defining per person will increase by 40-fold to 600
features of the communications network of GB in 2030. In addition, gigabit or higher
2030） fiber broadband household penetration
and 10 gigabit fiber broadband household
Cubic Broadband Network
penetration are expected to reach 55% and
The coming decade will see continuous 23%, respectively, and the average monthly
improvement in network performance. fixed network data usage per household
Figure 5 Defining features of the communications networks of 2030
20Communications Network 2030
Figure 6 Cubic broadband network
is forecast to increase by 8-fold to 1.3 space networks. Before we can make that
TB. Network ports will be upgraded from happen, there is still a lot of research that
400G to 800G or even 1.6T, and single- needs to be done. For example, we need to
fiber capacity will exceed 100T. In terms of develop:
coverage, network construction up until now • New air interface technologies with deep
has focused on connectivity on the ground, fading, large latency, and highly dynamic
but in the future, we will see the construction performance
of integrated networks connecting the • Intra-satellite and inter-satellite
ground, air, and space. beamforming technologies that can
1) Space-Air-Ground Integrated evenly allocate active user equipment (UE)
Network: Seamless and Continuous to different beams for load balancing and
Broadband Experience more efficient resource utilization
Moving forward, broadband • Anti-interference technologies for higher
coverage will extend beyond the ground, spectrum multiplexing rates
encompassing the air and even space. • Frameworks that support fast decision
These networks will connect devices at making in response to massive volumes of
various heights, such as drones flying less switching requests and complex switching,
than 1 kilometer above the ground, aircraft as well as mobility management
10 kilometers above the ground, and LEO frameworks for limited numbers of
spacecraft hundreds of kilometers above the ground stations
ground. A cubic network will consist of small Inter-satellite transmission requires
cells with a coverage radius of 100 meters, satellites at different orbital heights to form
macro sites with a range of 1–10 kilometers, multi-layer constellations, with each layer
and LEO satellites with a range of 300–400 networking through inter-satellite links.
kilometers, which will provide users with Inter-satellite links are established between
continuous broadband experiences of 10 satellites in the same orbit, on the same
Gbit/s, 1 Gbit/s, and 100 Mbit/s, respectively. layer, and on adjacent layers, forming a
(Figure 6 Cubic broadband network) cubic space network. Inter-satellite links will
For satellite-terrestrial access, devices use lasercom and terahertz technologies to
must be able to easily access terrestrial and support bandwidth of more than 100 Gbit/
21Communications Network 2030
s. This will require research on adapting new radio (NR), covering all spectrum bands
industrial products to aerial settings, making for International Mobile Telecommunications
phased array antennas more compact, (IMT) between 450 MHz and 52.6 GHz.
and enabling dynamic laser tracking and Research for Release 17 is still underway,
pointing. and one important focus of this research has
The network management and control been the use of spectrum above 52.6 GHz
domain consists of an operation and control for 5G NR. This points to industry consensus
center, network management center, earth on fully utilizing spectrum below 100 GHz
station, and integrated core network. It for 5G.
performs the tasks of satellite network To make 10-gigabit campus networks
management, user management, and possible, more research is needed on next-
service support. In this domain, we need generation Wi-Fi technologies that support
to research new dynamic routing protocols millimeter-wave and high-density MIMO.
between ground-based earth stations and Theoretically, Wi-Fi 7 standards that are
constellation networks, and hyper-distributed currently being defined should be able
convergent core networks that support to support 10-gigabit user access. With
intelligent switching of space-air-ground wireless air interface technology approaching
integrated networks. Shannon's limit, further evolution of Wi-Fi
2) 10-Gigabit Connectivity for and mobile technologies will require more
Individuals, Homes, and Organizations spectral resources, which are scarce. This has
Fiber networks are expected to be widely prompted industry-wide discussions about
deployed globally over the next 10 years, the feasibility of converging Wi-Fi 8 and 6G.
transforming today's gigabit connections for 3) All-terabit Network: Access,
individuals, homes, and organizations into Backbone, and DCN
10-gigabit connections. Taking into account the growing
To deliver 10-gigabit home broadband, broadband requirements of individuals,
200G passive optical network (PON) homes, and enterprises, as well as the need
technology will likely be used for optical to connect people and things, future access
access. The coherent detection technology network equipment will need to support
typically used for wavelength division terabit-level interfaces. Backbone equipment
multiplexing (WDM) will be used in the PON will support 40–100 Tbit/s per slot and data
field, which will significantly improve receiver center equipment 400 Tbit/s per slot.
sensitivity and support modulation formats By 2030, there will be broadband
with higher spectral rates, such as quadrature networks that can achieve terabit-level
phase shift keying (QPSK) and 16-quadrature transmission speeds in many parts of the
amplitude modulation (16-QAM), to achieve network, from access and backbone to data
higher data rates. center networks. These will mostly serve the
To deliver 10-gigabit broadband for world's largest cities – those with populations
individual users, mobile network research of 10 million or higher.
needs to focus on flexible use of the sub- In the terabit era, datacom equipment
100 GHz spectrum bands and continuous will need to have Ethernet interface
evolution of massive multiple-input multiple- technology that supports speeds of 800
output (MIMO). 3GPP Release 16 has defined Gbit/s or even 1.6 Tbit/s to meet service
two frequency ranges, FR1 and FR2, for 5G development needs. Unlike 200G and
22Communications Network 2030
400G Ethernet, 800G Ethernet is a nascent Assurance for Differentiated Service
technology that has yet to be standardized. Requirements
From a technical standpoint, there are Over the course of this decade, the
two routes that will take us to 800G: Internet traffic model will undergo a
continuing evolution of existing pluggable fundamental shift from today's top-down
optics modules and the adoption of new content traffic generated primarily from
co-packaged optics (CPO) modules. Both online services, retail, and entertainment
module types will have a place in the future to bottom-up data traffic from pervasive
market, but pluggable optics modules with intelligent applications deployed across
a capacity of over 800G are expected to various industries. Intelligent machines will
encounter power and density problems, so generate massive amounts of data, and
CPO modules will likely become the preferred this data will need to be processed in data
choice. centers. This decade will see a push toward
In addition, enabling long-distance the coordinated development of electricity
transmission capacity of more than 100 and computing power to enable society-
Tbit/s per fiber will require technical wide green computing power. Therefore,
breakthroughs in backbone WDM equipment, the networks of the future will need to be
including materials science breakthroughs able to support more centralized operations
in high-baud electro-optic modulation and of data centers. That will entail meeting
the development of new optical amplifier differentiated latency requirements, with
technology that goes beyond C-band to the acceptable latency for backbone, inter-
L-band and S-band. city, and intra-city network services being
100 ms, 10 ms, and 1 ms, respectively. In
Deterministic Experience
addition, networks will need to schedule
The ability of communications networks resources in real time at the network layer
to provide deterministic experiences is key to based on service attributes in order to make
supporting online office and learning, as well computing power greener and more efficient.
as meeting the security and reliability needs In addition to meeting differentiated
of production environments. latency requirements at the network
1) 100 ms, 10 ms, and 1 ms Latency architecture and system levels, the
23Communications Network 2030
industry also needs to research end-to-end isolated from each other. This is a key area
deterministic latency. we can work on in order to serve vertical
Real-time wireless access services industries. End-to-end slicing is a network
require high instantaneous rates over the virtualization technology with Service Level
air interface. However, due to the spectrum Agreement (SLA) assurance. Through
constraints caused by the multiplexing of network slicing, different logical or physical
multiple pieces of UE on a single carrier, it is networks can be isolated from the network
difficult to guarantee real-time performance. infrastructure to meet the SLA requirements
Moving forward, multi-carrier aggregation of different industries and services. Types
technologies need to be developed so that of slicing include wireless slicing, transport
carrier configuration is decoupled from network slicing, and core network slicing.
transmission, improving the bandwidth of When a carrier provides a slice to a
services under latency constraints on multi- customer, the carrier also provides end-to-
band carriers. end management and services.
For cloud-based wireless core networks, Wireless slicing: It can be further
real-time operating systems (OSs) are classified into hard slicing and soft slicing.
needed to enhance deterministic scheduling Hard slicing is achieved through resource
frameworks and ensure real-time service isolation, such as through static resource
performance. block (RB) reservation and carrier isolation
The optical access networks we have for specific slices. Soft slicing is achieved
today feature PON technology, which is through resource preemption, such as
based on time division multiplexing (TDM). QoS-based scheduling and dynamic RB
PON uses uplink burst to prevent collisions, reservation. Currently, the bitrates of
making it ill-suited to scenarios requiring different network slices can be guaranteed
low latency. Frequency division multiplexing based on priorities. The next step in the
(FDMA) needs to be explored to allow development of network slicing is to explore
concurrency of multiple optical network the most appropriate wireless protocols for
terminals (ONTs) and guarantee low latency the PHY, MAC, Radio Link Control (RLC), and
by addressing fundamental issues. Packet Data Convergence Protocol (PDCP)
For wide area networks (WANs), the layers. For example, we could have a PHY
current best-effort forwarding mechanism layer with a low-latency coding scheme for
needs to be changed, protocols at the slices that support ultra-reliable low-latency
Physical (PHY) and Medium Access Control communication (URLLC) services, or a MAC
(MAC) layers need to be improved, and new layer with an optimized hybrid automatic
technologies such as time-sensitive networks repeat request (HARQ) mechanism.
(TSNs) and deterministic IPs need to be Transport network slicing: This is
integrated to ensure on-demand, end-to-end achieved through physical isolation or logical
latency. isolation. Physical isolation technologies
2) End-to-end Slicing: Logical can be optical-layer hard pipes, which
Private Networks and Services That Are carry different services through different
More Adaptable to Vertical Industries wavelengths or through the optical
End-to-end slicing provides vertical channel data unit-k (ODUk) within a single
industries with customized private network wavelength. Flexible Ethernet (FlexE) at the
services that run independently and are MAC layer is also used to isolate services
24You can also read
