COMMUNICATIONS THEORIES & SYSTEMS

5G Features from Operation Perspective and

Fundamental Performance Validation by Field Trial
Guangyi Liu*, Yuhong Huang, Fei Wang, Jianjun Liu, Qixing Wang
China Mobile Research Institute, Beijing 100053, China
* The corresponding author, email: liuguangyi@chinamobile.com

Abstract: The fast deployment and penetra- cilitating daily work and life. The rapid devel-
tion of 4G has cultivated human behaviors on opment of 4G has stimulated the flourishing of
mobile data consumption, leading to explosive Internet applications, such as instant messag-
growth in mobile traffic and stimulating new ing, online shopping, navigation, online pay-
requirements on the capabilities of mobile ment, and videos. Smart phones are not just
networks. To meet the requirements of mobile communication tools but also social and per-
networks toward year 2020, the next genera- sonal platforms. With a smart phone in hand,
tion of mobile networks (termed as IMT-2020, we can shop both online and offline shopping
or 5G) is designed to support 100 Mbps–1 via mobile payments, making our daily lives
Gbps user-experienced data rate, 1 ms radio more convenient than ever and changing our
transmission latency, and 1 million connec- consumption and daily behaviors.
tions per square kilometer. Recalling the vision The rapid penetration of 4G has cultivated
and requirements of 5G targeting for commer- new mobile services and applications, which
cial launch in 2020, this article overviews the requires mobile communication networks to
key features of 5G and compares with those feature new capabilities. The breakthrough of
of 4G, and reports the world first field trials consumer electronics is expected to promote
conducted to validate the key performance of many new applications to be popular in our
5G radio interface in 3.5GHz band. The trial daily lives, such as Augmented Reality (AR),
results show that a 1 ms transmission latency Virtual Reality (VR), Ultra High Definition
and 1 Gbps data rate are achievable. (UHD) video (4K/8K), and glass-free 3D
Key words: 5G; IMT-2020; 3D-MIMO; user display. These applications demand higher
centric network; software defined network; capacities and efficiency from mobile com-
Network Function Virtualization; field trial munication networks; industrial Internet and
automotive demands also require much lower
I. INTRODUCTION transmission latencies and higher reliability.
As the human-to-human communication
Fourth-generation mobile communication market may be projected to become saturat-
(4G, also called Long-Term Evolution or LTE) ed within a few years, the ecosystem must
opened the door for mobile broadband com- cultivate new markets and business models
Received: Apr. 4, 2018
munication and enabled more-efficient societal by expanding services to vertical industries. Revised: Jun. 15, 2018
operations and management, especially in fa- With more-efficient communication tools and Editor: Xiaohu Ge

China Communications • November 2018 33

 capabilities, all of society may be forecasted whole ecosystem, extremely low-cost 5G is
This article overviews to connect to the network. Compared with expected, i.e., 1000-fold improvement in cost
the key features of 5G traditional human-to-human communication, per bit.
and compares with Machine-Type Communication (MTC) fea- Following the requirements of a 5G mo-
those of 4G, and re-
tures extremely low cost, very low power con- bile network, 3rd Generation Partner Project
ports the world first
field trials conducted
sumption, massive connections, and extremely (3GPP) has initiated a feasibility study of the
to validate the key deep coverage. In particular, telemedicine 5G New Radio (NR) system in its Radio Ac-
performance of 5G ra- and the remote control of industrial operation cess Network (RAN) working group in 2015
dio interface in 3.5GHz and smart grid demand Ultra-Reliable and ul- [4] and is planning to publish the first 5G NR
band. tra-Low Latency Communication (URLLC). release in June 2018 and a second release in
The commercial deployment of 4G globally late 2019; 3GPP Service Architecture Working
has also learn some lessons in service pro- group 2 has begun specification work on the
viding when exploring the enterprise market network architecture earlier than that of 5G
of human communication. First, the service NR [5].
and content should be deployed as closely as Along with 3GPP 5G standardization work,
possible to the subscribers to reduce delays in many countries leading in 5G technology
service fetching and improve user experience. have announced their commercial plans for
Second, the deployment of new services and the service. South Korea has revealed plans
applications should be accomplished quickly to provide 5G service demonstrations in the
and at low cost. Third, the network should 2018 Winter Olympics, which has been held
support flexible and reconfigurable deploy- in South Korea, based on none-3GPP specifi-
ment and meet the requirements of differenti- cations (Open Test Specification Association,
ate usage scenarios. OTSA), while Japan has announced plans to
The capabilities necessary to address the commercially launch 5G in the 2020 Summer
above issues call for a new generation of mo- Olympics to be held in Tokyo. China has also
bile communication systems that extend be- publicized plans to commercially launch 5G in
yond 4G. The 5th Generation mobile communi- 2020. Verizon has claimed that it will deploy
cation system (5G) is targeted for commercial 5G in 2018. The work on 5G is on track to-
launch in 2020 and also called International wards its wide commercial launch by 2020.
Mobile Telecommunications 2020 (IMT- The rest of this article is organized as fol-
2020) [1] in International Telecommunication lows: The vision and requirements of 5G are
Union Radio Bureau (ITU-R) Working Party recalled in Section II, and the key features of
5D (WP5D). Based on research from 2013, 5G system are overviewed and compared with
the mobile industry has built up a consensus those of 4G in Section III and IV, respectively.
on the 5G vision and usage scenarios, which The pre-commercial system of 5G NR and
include enhanced Mobile Broadband (eMBB), field trial are introduced in Section V, and con-
massive MTC with low cost and low power clusions are drawn in Section VI.
consumption (mMTC), and URLLC [2]. The
corresponding performance requirements have II. VISION AND REQUIREMENTS
been specified in detail [3] for the above usage
scenarios, e.g., 20 Gbps peak data rate, 1 ms Mobile communications have profoundly
transmission latency of the radio interface, changed our daily life, and the desire for
3-fold improvements in spectrum efficiency higher-performing mobile communications is
compared with that of 4G, a density of mil- never-ending. Moving forward, several trends
lions of connections per square kilometer, 10 [6] are on a path to exhaust the capabilities of
Tbps traffic density per square kilometer, and existing 4G networks, which are:
100-fold improvements in power efficiency. - Explosive growth of data traffic;
To enable the sustainable development of - Massive increase in the number of intercon-

34 China Communications • November 2018

 nected devices; and service offerings, the usage scenarios
- Continuous emergence of new services and of 5G could be categorized into three types:
application scenarios. eMBB, mMTC, and URLLC [2]. The key
5G mobile communications systems will performance requirements for 5G are derived
emerge to meet new and unprecedented de- for each scenario according to the predicted
mands beyond the capability of previous distribution of users, percentage of different
generations. As shown in figure 1, 5G will not services, and service requirements, such as
only serve for human’s communication, but data rate and latency.
also fulfill the communication requirements 5G systems must dramatically outperform
of the vertical communities, e.g. industrial, previous generations. 5G should support [6]:
agricultural, medical, traffic, financial and en- - User-experienced data rate: 0.1–1 Gbps
vironment. So 5G will penetrate every element - Connection density: 1 million connections
of future society and create an all-dimensional, per square kilometer
user-centric information ecosystem. 5G will - End-to-end latency: millisecond level
break through the limitation of time and space - Traffic volume density: tens of terabytes
to enable an immersive and interactive user per second per square kilometer
experience. 5G will also shorten the distance - Mobility: higher than 500 Km/hour
between humans and things and provide easy - Peak data rate: tens of Gbps
and smart interconnections between people Among these requirements, user-experi-
and all things. enced data rate, connection density, and end-
5G will provide users with fiber-like access to-end latency are the three most-fundamental
data rate and a zero-latency user experience. items. 5G must significantly improve the effi-
5G can connect 100 billion devices and be ciency of network deployment and operations.
able to deliver a consistent experience across Compared with 4G, 5G must show 3–5-fold
a variety of scenarios, including cases of improvements in spectrum efficiency and over
ultra-high traffic volume density, ultra-high 100-fold improvements in energy and cost ef-
connection density, and ultra-high mobility. ficiency.
5G will also be able to provide intelligent op- The performance and efficiency require-
timization based on services and user aware- ments of the technology jointly define the key
ness, and improve energy and cost efficiency capabilities of 5G, which can be illustrated as
by over a hundred of times, thereby realizing a “blooming flower,” as shown below, with
the vision of 5G: “Information a finger away, petals and leaves that rely on each other. The
everything in touch.”.
5G will touch many aspects of life in the
future, such as residence, work, leisure, and
transportation. The 5G scenarios include at
least dense residential areas, office towers,
stadiums, open-air gatherings, subways, high-
ways, high-speed railways, and wide-area
coverage. These scenarios, which are charac-
terized by ultra-high traffic volume density,
ultra-high connection density, or ultra-high
mobility, may be quite challenging for 5G.
Some typical services, such as AR, VR,
UHD videos, cloud storage, vehicle-to-vehi-
cle, smart home, and Over-The-Top (OTT)
services, will take place in these scenarios.
According to the features of the deployment Fig. 1. General vision of a 5G system.

China Communications • November 2018 35

 petals represent the six key capabilities of 5G ciency improvement;
in terms of performance and can fulfill the - Software-defined Air Interface (AI): To
diverse requirements of future services and meet the requirements of different deploy-
scenarios. The top of each petal represents the ment scenarios by a unified AI;
maximum value of the corresponding capabil- - User-centric network: To guarantee the best
ity. The leaves represent the three key capabil- user experience in the network according to
ities in terms of efficiency and can guarantee user demands and the network situation;
the sustainable development of 5G. - Network automation: To replace the manual
Besides the subjective key performance in- operation of network planning, network op-
dicators defined above, the 5G is also expected timization, and maintenance by automatic
to enable the sustainable development of the mechanisms;
entire mobile industry; thus, the cost efficien- - Software Defined Network (SDN) and Net-
cy of 5G systems must be 100 times better work Function Virtualization (NFV) based
than that of 4G systems. To address cost effi- core network: To provide flexible network
ciency, the complexity of the network should slicing to meet diverse customer demands
be reduced, and network automation should and enable a short time to market and elas-
be enabled to simplify the network’s planning, tic capacity expansion.
deployment, optimization and maintenance. More details of the key technologies above
will be introduced below.
III. KEY FEATURES OF 5G SYSTEM
3.1 3D-MIMO with massive
antennas
5G has been defined in 3GPP Release 15 by
as a group of key features and their corre- To achieve higher spectrum efficiency, MIMO
sponding parameters. Besides the technologies and its future enhancement will play a key role
and protocol architecture inherited from 4G in 5G.
system, the following technologies will be the MIMO was successfully applied in LTE
corner stones for the 5G system: systems as a cornerstone technology, where
- Three-Dimensional Multiple Input and the spectrum efficiency was improved by 2–3
Multiple Output (3D-MIMO) [7]: To ad- times compared with that of 3 rd generation
dress the need for 3–5-fold spectrum effi- mobile system release 6 via the flexible usage
of 8 transmission modes of MIMO [8]. In fur-
ther evolutions of LTE technology (LTE-Ad-
vanced), MIMO technology was extended
to Coordinated Multiple-Point transmission
(CoMP) by advanced baseband processing
based on MIMO among multiple base stations
[9]. For example, in Time Division Duplex
(TDD) LTE-Advanced (TD-LTE-Advanced),
Joint Transmission (JT) and Joint Reception
(JR) of CoMP could be implemented to com-
press inter-cell interferences by the channel
reciprocity of TDD, and spectrum efficiency
could be improved by more over 30% and
40%, respectively, compared with Single-User
MIMO (SU-MIMO) [9]. Therefore, MIMO
applications are becoming increasingly im-
portant to mitigate interference and improve
Fig. 2. Key capabilities of 5G. channel capacity.

36 China Communications • November 2018

 However, due to the bottleneck of present MO over that of conventional 2D-MIMO
channel modeling and the implementation is illustrated by system level simulations in
skills of the antenna array and base station downlink and the simulation parameters giv-
[10], the existing MIMO is only constrained to en in [7]. From the results in figure 4, mul-
Two-Dimensional (2D) MIMO, where spatial tiuser MIMO in downlink can achieve very
processing is limited to the horizontal plane promising performance gains. As more users
with fixed down-tilting in the vertical plane are multiplexed together, higher gains are
for network coverage optimization, as shown achieved. In the case of 8 users, 64 antennas
in figure 3. will achieve 3.3-fold performance gains over
In traditional passive antennas, one antenna that achieved by 8 antennas at the base station.
is composed of 8–10 elements stacked in a If the terminals in a practical dense urban area
column, and limited phase adjustment among are distributed in a 3D manner, as in figure
these elements is applied in a fixed or semi-dy- 4, vertical beamforming will further contrib-
namic manner to offer a down-tilting angle ute more gains in performance due to better
(adjustable down-tilting) for network coverage inner-cell interference compression. Similar
optimization. In this case, the same RF signals performance gains can be observed from the
are fed to each element of one antenna with cell-edge spectrum. Thus, 3D-MIMO can be
predefined phase adjustments, and the vertical concluded to greatly boost the capacity of 5G.
beam pattern is fixed for the whole bandwidth.
Dynamic spatial processing is only applied
among independent antennas in the horizontal
plane.
Naturally, from figure 3, the current linear TXRU
TXRU

antenna array in the horizontal domain can be TXRU

TXRU
TXRU

extended to an antenna matrix by regarding

TXRU
TXRU
TXRU

stacked elements belonging to one antenna as TXRU

TXRU
TXRU

independent antenna elements; thus, baseband TXRU

TXRU

spatial processing can be extended to three Di-

TXRU
TXRU
TXRU

mensions (3D) by feeding the antenna matrix TXRU

TXRU
TXRU

with different weights in both the horizontal

3D-MIMO antenna
and vertical domains. This type of MIMO
extends the original concept to 3D-MIMO.
With 3D-MIMO, signal transmission can be
Fig. 3. Illustration of 3D-MIMO.
controlled in both the horizontal and vertical
planes, and interference can be compressed or
avoided in the full spatial domain, thereby im-
proving spectrum efficiency and interference 3D-MIMO gain
mitigation in cellular mobile communications. 4 3.3
2.9
Gain over 8 antenna

With this principle, further improvement of UMA Umi 2.5 2.6

3
2.1 1.84
the spectrum efficiency of 5G by 3D-MIMO
2
can be realized without changing the antenna 1 1
array size of the base station and the antenna 1
number of the terminal. 0
The performance gain contributed by a 8 16 32 64
antenna number
large antenna number at the base station will
boost either the coverage or multiuser capacity
in both the uplink and downlink scenarios.
The potential performance gain of 3D-MI- Fig. 4. Average spectral efficiency of 3D-MIMO [7].

China Communications • November 2018 37

 3.2 Software-defined air interface To provide the experience of a user-centric
network, the 5G network should support the
5G is designed to support multiple usage sce-
following functions:
narios and broad scope of spectrum with one
(1) User context awareness
AI. To address the requirement of differenti-
(2) Mobile Edge Computing (MEC)
ated performance optimization for different
(3) Multiple connections
usage scenarios and different spectrum bands,
(4) Flexible deployment
different system parameters, procedures, and
mechanisms could be implemented. To enlarge 3.3.1 User context awareness
the market scale and achieve low-cost system In principle, the network can serve the cus-
deployment, a single system for multiple us- tomer better when more information or knowl-
age scenarios is highly recommended. edge of the customer and network is available.
To achieve a software-defined AI, a general This information or knowledge could include
and unified framework for different usage sce- service information, user behavior, and net-
narios must be designed comprehensively, and work background. The service information
differentiated capabilities could be implement- could be an attribute of the requested service
ed by different sets of parameters. or application, service content, etc. User be-
3.3 User-centric network haviors could include moving speed, direction
of movement or route, user preference, and
Besides higher data rates and lower latency, user background. The network background
user experience enhancement will also be tak- includes the interference attribute, traffic load,
en as a key goal of 5G systems. The objective radio resource configuration, and radio propa-
of 5G is to build a user-centric environment gation characteristics of the serving and neigh-
where the network can adjust the end-to-end boring cells.
network configuration to adapt to the context As smart phones become more powerful in
of the customers. The best connections can be the 5G era, they can facilitate an awareness
built to serve the customer according to the of the user’s context. Smart phones are typi-
service attributes, contents, network status, cally embedded with many sensors, besides
and customer preference; thus, the best user featuring powerful computing and storage. As
experience is guaranteed for the customer. such, these phones may be able to sense user
behaviors, e.g., direction of movement and po-
sition. The private assistant offered by smart
phones could also collect the user’s daily
SON&OAM
HSS/AAA/PCC consumption, entertainment information, and
CN Entity Internet/App
behaviors; it can also communicate with the
service provider and request the corresponding
service. These statistics could be used as the
General Network
Controller
Radio Data Center subscriber’s preference and background and
(RDC)
utilized for optimization of the service offer-
(GNC)

RAN Entity ing from the network perspective. By tracking

Signalling Data and predicting a subscriber’s direction of
Macro signalling
movement, the handover procedure could be
Local Data Access
(data) layer
shortened by pre-handover operation.
During service transmission, traffic attri-
5G 4G WiFi
butes, service contents, and service provider
D2D information are also useful for transmission
optimization. Given information on service
Fig. 5. Radio network architecture. attributes and providers, the device could be

38 China Communications • November 2018

steered to access the data center closest to the complicated computations of the service or
subscriber to offer the shortest latency and application could be deployed in the RDC, and
enable correct billing according to one’s sub- processing delays could be avoided by fast
scription. Given information on service con- local exchanges of the computing task and the
tents, cross-layer scheduling and optimization corresponding results between the subscriber’s
could be done to provide a higher data rate and device and the RDC. In this manner, a sub-
robust service experience, e.g., avoiding slow scriber’s device can be greatly simplified for
triggering of TCP. With information on service applications, such as AR/VR and smart robots,
attributes, simultaneous services could be di- and the requirements for these commercial
rected to different radio networks or layers of products can be significantly relaxed. As such,
the network according the Quality of Service this type of MEC deployment will speed up
(QoS) requirements, traffic load distribution commercialization of cloud-based AR/VR and
and interference situation; thus, the multiple smart robots for 5G.
simultaneous services of one subscriber could Given location information of two or multi-
be guaranteed with the respective QoS. ple subscribers connected by one base station,
Information on the network background, local switching could be enabled to allow data
e.g., traffic load, interference, and spatial exchange among these subscribers and achieve
propagation characteristics of a subscriber’s the lowest latency and backhaul capacity de-
location, can enable configuration of the best mand by shortened routing. The local switch-
radio transmission mode in the base station ing deployed by RDC guarantees 1 ms-level
and corresponding device and guarantee the latency for local data exchange and facilitate
most sufficient transmission. the provision of services for the URLLC sce-
nario.
3.3.2 MEC
3.3.3 Multiple connections
In the 5G network, MEC is the most important
function to explore the differentiate enterprise With the breakthroughs in devices and
and vertical markets. MEC is a kind of net- chipsets, multiple connections from one
work deployment, where the network func- mobile device to different network or base
tions are deployed as required to facilitate fast stations will not be a challenge in the 5G era.
and efficient service launching. These func- Even today, WiFi and 4G are supported simul-
tions include data routing, billing, security, taneously by smart phones, and service data
separation in transportation and core network can be shared between these two links. When
perspective, user awareness-based cross layer multiple connections occur in one network, the
scheduling of the base station, and computing ensuing connections are called homogeneous
and storage resources. multiple connections; when multiple connec-
User awareness enables native support on tions occur between different networks, het-
the MEC, which facilitates service offerings erogeneous multiple connections are formed.
to enterprise subscribers and vertical markets. Two special cases are noted for homogeneous
MEC enables the network to deploy the ser- multiple connections. In the first case, multi-
vice as closely as possible to the subscriber, ple connections occur in different frequencies
thus reducing the latency of service fetching of one network, where different base stations
and capacity demands for backhaul. With the of different frequencies provide both control
logical network element called Radio Data and data channels. In the second case, the data
Center (RDC), as shown in figure 6, caching channel is from one base station and the con-
and cloud computing can be deployed locally trol channel is from another base station.
on demand. An extreme case is illustrated in figure 7,
Given mobile computing capabilities close where the signaling is obtained from a base
to the network edge (e.g., deployed at RDC), station that can provide much wider coverage

China Communications • November 2018 39

 than another base station and data can be pro- experience. For example, the voice service is
vided by several base stations, such as 5G, 4G, scheduled to 900 MHz LTE for optimal cover-
or WiFi. age and seamless mobility, while the data ser-
Multiple connections help improve the vice is scheduled to the 5G network to achieve
user-experienced data rate by connecting to the highest data rate.
multiple points of the network and aggregat- Multiple connections also help shorten ser-
ing different radio resources from different vice breaks during the handover procedure.
transmission points for one device. Especially The handover can be predicted according to
in homogeneous multiple connections, the the user behavior and network background,
spatial directional transmission of the signals and multiple connections can be used to im-
of different users help reduce inter-cell inter- plement a pre-handover while the service is
ferences to users who are allocated the same maintained by one connection and handover
radio resource. preparation is conducted by another connec-
Multiple connections enable multiple si- tion before it actually happens. By this way,
multaneous services to be steered toward the user will not experience service breaks
different base stations or networks according during the handover.
to the network background and service expe-
3.3.4 Flexible deployment
rience requirements. For example, voice and
data services usually occur simultaneously in To build a sustainable 5G ecosystem, the cost
one device. The voice service is very sensitive efficiency of 5G networks should be improved
to the service break and thus requires a seam- by 100-fold in comparison with that of 4G
less experience. While a data service is usual- networks. Thus, the Capital Expenditures
ly less sensitive to service breaks than voice (CAPEX) and Operating Expense (OPEX) of
and requires high data rates. Without multiple the 5G networks should be optimized.
connections, the device could be scheduled to To reduce the CAPEX of 5G network de-
a network with coverage, e.g., 900 MHz LTE; ployment, a cost-sufficient base station may
unfortunately, the bandwidth of LTE is quite be considered. As shown in figure 6, diverse
limited, which means the data rate could be low-cost access points are to be supported in
quite low. A user in this case may be happy 5G network deployment; some of these access
with the voice service provided but unhappy points include coverage access points, which
with the rate of data service. When multiple guarantee coverage, hotspot access points,
connections are supported by the device, dif- which fulfill high-throughput requirements
ferent services can be scheduled to the best in certain areas, full-stack access points, and
suitable network to guarantee the best service semi-stack access points. According to traffic
statistics and predictions, dynamic network
deployment could be realized by network vir-
tualization and compromise between the cost
Coverage
access point and traffic demand. Such flexible access points
could be dynamically switched on or off ac-
cording to the traffic demand.
Hot spot
access point For example, regional traffic distribution
RRC/IP
PDCP could be analyzed at the RDC, and traffic
RLC RLC and user attributes could be collected. Then,
MAC MAC
PHY PHY
hotspot access points could be deployed or
PHY
RF RF RF switched on according to the attributes above
to meet the local throughput and latency re-
quirements. To enable such flexible deploy-
Fig. 6. Illustration of flexible network deployment. ment, the supporting of the plug-and play-op-

40 China Communications • November 2018

eration is required. and the role it played in helping operators de-
To implement flexible base station types, ploy and manage LTE networks, considering
protocol splitting has been considered in dif- SON in 5G is necessary. Of course, as SON
ferent ways, as in figure 7, where eight func- helps optimize other functionalities, it is,
tion splits for a base station are illustrated. therefore, not a stand-alone feature–the princi-
Different options may be suited to different ples of the features supported by SON must be
deployment scenarios. known. However, building a system with SON
Software and hardware decoupling of net- in mind will help avoid limitations that must
work equipment is fundamental to flexible be overcome later. The big data generated in
functionality deployment, and radio network the network could also be used to perform net-
virtualization is necessary to achieve this goal. work planning and self-optimization, as well
Based on a general processor unit, a different as provide network value promotion by, for
software is deployed according to the func- example, opening some capabilities to a third
tionality demand decided by the General Net- party. Some SON features, e.g., Minimum
work Controller (GNC), and the GNC adapts Driving Test (MDT), Automatic Neighboring
different partial protocols to produce a full Relationship (ANR), and Random Access
protocol stack and interoperate with other net- Channel (RACH) optimization, have been uti-
work elements. For example, the Multimedia lized by operators for big data collection. The
Broadcast and Multicast Service (MBMS) or NR study in Rel.14 defined the cornerstones of
Device to Device (D2D) functionality module the new system, including some early require-
could be flexibly deployed in an area with ments for SON functionality. The features of
stringent broadcasting or public safety require- 5G are now going to be specified for Rel.15.
ments. All SON solutions, including SON-related
ones, played important roles in LTE develop-
3.4 Network automation
ment. A SON study for NR should focus on
To reduce the OPEX of 5G networks, network objectives based on those LTE SON use cases
automation is expected to avoid the manual that proved helpful in the early stages of the
operation of traditional network optimization LTE development. These objectives, which
and maintenance. Besides the Self-Organizing are deduced from LTE SON use cases, shall
Networks (SON) function [11], which was take the 5G deployments and 5G functions
successfully introduced to 4G networks, more into consideration. Such analysis will also en-
advanced functions targeting network automa- able studying of the impact of the function on
tion are highly expected. the 5G network and the necessary changes. If
Bearing in mind the history of SON in LTE work on the objectives leads to new use cases

High- Low- High- Low- High-

RRC PDCP Low-PHY RF
RLC RLC MAC MAC PHY

Data Option 1 Option 2 Option 3 Option 4 Option 5 Option 6 Option 7 Option 8

High- Low- High- Low- High-

RRC PDCP Low-PHY RF
RLC RLC MAC MAC PHY

Data

Fig. 7. Illustration of a base station’s function split.

China Communications • November 2018 41

 for 5G, these shall also be addressed. data based network optimization and mainte-
1. To identify IDs in 5G that may be the most nance. The same process could be designed to
burdensome to allocate manually (e.g., collect network big data with a higher efficien-
non-global unique physical cell ID) and to cy.
study methods to automate allocation of Big data-based network operation systems
these IDs without conflict (especially in ul- [12] include traffic sensing, user capability
tra-dense and small cells deployments). sensing, centralized and personalized func-
2. To identify the 5G scenarios that are the tionality management and distribution, and
most prone to failures due to sub-optimal uniformed wireless platforms.
network configuration and to study methods The system is divided into two parts: the
that will help find the cause of and correct decision-making domain and the implemen-
problems (including CU-DU split, if found tation domain. The decision-making domain
relevant). is mainly responsible for the collection and
3. To determine if resource allocation in 5G management of user information, as well as
may benefit from load balancing and to service and terminal state, network upgrade,
study methods to enable it (including load selection, configuration, and unified manage-
information exchange on network interfac- ment.
es, if needed). The implementation domain is responsible
4. To identify the required big data for net- for user, terminal, and network status report-
work optimization, user experience im- ing, network configuration, and dynamic
provement, and network value promotion deployment. The decision-making domain
and to study methods to collect these data provides the basic configuration via big data
(e.g., via MDT) analysis to the initial network. Then, the im-
RAN network slicing should be considered, plementation domain chooses and deploys the
and its possible impact should be identified functionalities and access points according to
when studying the abovementioned SON sce- the command of decision making domain.
narios. It is a significant challenge to analyze data,
In 4G systems, Deep-Packet Interception since the amount of data in the mobile com-
(DPI) has been deployed at the base station munication network may be eventually be-
to collect the necessary information used for come very large. At the end of 2015, the data
network optimization and maintenance via big to be analyzed in the network of China Mobile
was about 6.8 PB per day. In 5G, this number
is going to be much larger.

3.5 SDN/NFV-based core network

to provide network slicing
Capability openness to the third party

Network Function Virtualization (NFV) [13]

aims to decouple software from dedicated
DATA collection & strategy server hardware by migrating the telecom infra-
N1interface:
X interface: report Y interface: report structure from a dedicated platform to general
Report NAS and AS info of UE
info of Uu/N2/gNB info of N2/AMF,
Configure strategy from server
Configure strategy Configure strategy COTS servers. With NFV, different network
from server from server Z interface: report
info of N3/CPF, elements are decoupled as independent ap-
Configure strategy
from server plications, which can be flexibly deployed in
AMF
Uu interface
a unified platform constructed by a standard
UE gNB
CPF
server, storage, and switcher. When the soft-
DATA Strategy
collection configuration ware is decoupled from the hardware, the ca-
pacity of each application can be expanded or
Fig. 8. Illustration of a big data-based network operation. shrunk by adding or deleting virtual resources.

42 China Communications • November 2018

 NFV defines a general platform to support cultivation of new services
the virtualization of different types of net- (5) Automatic deployment shortens the pe-
works. The architecture of NFV is composed riod for service launch
of three layers: NFV Infrastructure (NVFI), SDN [14] provides flexible traffic con-
virtualized network, and OSS/BSS. trol and a platform with which to facilitate
(1) NFVI: NVFI is a hardware resource innovation in networks and applications by
pool that includes computing, storage, and splitting the control and forwarding planes of
network resource. The NFVI can be construct- the network. Separation of the control and for-
ed by multiple geographically distributed warding planes brings several benefits to 5G
data centers connected by a high-speed com- network architecture:
munication network. A Hypervisor abstracts (1) Improved forwarding performance
the NFVI as virtual resources (i.e., virtual (2) Improved network reliability
computing resource, virtual storage resource, (3) Flat network deployment
and virtual network resource), which can be (4) Facilitation of innovations in services
allocated to different customers per request. and applications
(2) Virtualized network layer: The virtu- NFV and SDN are two independent con-
alized network layer is composed of virtual cepts but complementary each other when they
network elements and the connections among are applied. The objective of NFV is to reduce
them. The virtual resources required by virtual the TCO of network deployment and operation
network elements are decomposed into virtual by decoupling the software and hardware and
computing resources, virtual storage resourc- enabling on-demand service deployment and
es, and virtual network resources, all of which scheduling of shared virtualized resources. The
are allocated and provided by NFVI. The vir- SDN enables a flexible and dynamic network
tual network elements are managed by NMS topology, service routing, and scheduling by
and EMS. separating the control and forwarding planes
(3) Operating and supporting layer: The and centralizing the control plane to decide
maintenance and management system of oper- the routing strategy for data forwarding. The
ators is OSS/BSS. SDN controller further provides an open north
Horizontally, the NFV network includes interface and enables flexible network traffic
two parts: the service and network domain scheduling via third-party software, which is
and the Management and organization domain beneficial for service innovation.
(MANO). MANO is responsible for the orga- SDN and NFV integration can provide
nization and management of NFVI resources, services in a customer-defined manner, adapt
mapping and connecting between the service/ to the network status, and create value-added
network and NFVI, and lifecycle management services according to the characteristics of
of the service and network VNF. The features customers and services.
of NFV, e.g., decoupled software and hard- SDN/NFV-based networks in 5G provide
ware, virtualized hardware resources, and the differentiated services, fast service launch-
general platform for scheduling and managing, ing, and elastic capacity expansion. Based
serves the 5G network architecture very well, on an SDN/NFV system architecture, the 5G
bringing with it the following benefits: network can schedule and separate network
(1) IT-based hardware reduces infrastruc- resources on demand and abstract the network
ture costs as multiple separated logical networks accord-
(2) Generalized hardware reduces the TCO ing to the different resources demanded by
(3) Software-based functions enable flexi- different types of services. This type of sepa-
ble service deployment rated logical network is called a network slice,
(4) A set of components as a service facili- and different network slices share network
tates the openness of network capabilities and resources but serves different services and

China Communications • November 2018 43

 customers. according to their service demand and facil-
The basic 5G network architecture is shown itates innovations in services and business
in figure 9, where the whole network is split models.
into three clouds: one for the radio access
network, one for the control plane of the core IV. COMPARISON BETWEEN 4G AND 5G
network, and one for data forwarding. As
shown in figure 10, the whole 5G network is With the guidance of ITU-R on the technical
deployed on unified physical hardware, but requirements and evaluation methodology,
it can be separated into different slices, e.g., 3GPP began 5G work in 2015. The brand-new
eMBB and mMTC. Network slicing enables air interface of 5G in 3GPP is called New Ra-
the 5G network to serve different customers dio (NR). The channel models for bands above
6 GHz, the requirements and scenarios, and a
feasibility study on the technical framework
Control Cloud have been completed. The detailed specifica-
Infrastructure Pipe abilities
Traffic control
Value added
codec
Data&Info
services
tion is in progress. The first release of 5G NR
VPN
Network
will be completed as Release 15 by June 2018,
IDC MMM DPI Subscriber

SW/HW AAA FW Network capability

exposure
Network
Control
MM AAA
while the second release of 5G NR will be
CM MANO
function
Plug and play
mesh
Network controller Network completed by 2020.
Resource
D-RAN
RAN
CDN
Service
enabler
management
The brand-new 5G system is composed of
M-CDN NR and New Core (Next-Generation Core net-
Wi-Fi
Wi Fi C-RAN
RAN IDC
C
work, NGC). For NR deployment, two major
options are defined as Standalone (SA) and
Distributed forwardin
forwarding

Access Cloud Forwarding Cloud

None-Standalone (NSA) [16].
As illustrated in figure 11, NSA is expected
Fig. 9. General 5G network architecture. to add the NR as an extra data pipe to the LTE
network via multiple connections without in-
troduction of a new core network as the first
Slice 1: eMBB
step of 5G deployment. The device is anchored
Always online
Data forwarding
to the LTE system, and NR can be utilized as
High mobility and Baseband Data forwarding conversation
Mobility
(billing) (billing) management
high throughput Slice 1 management

Slice 2: mMTC
Low power and
Baseband
(Simplified)
Data forwarding
(billing)
Lower power
conversation an extra data pipe when NR coverage is avail-
Massive connections Slice 2 management

…… Diverse and customized able. When outside of NR coverage, the NSA

network slice
Slice N
UE only works as an LTE UE. The mobility
Shared infrastructure management of NSA depends completely on
the LTE system, which provides a better user
Virtual platform Virtual platform

Backhaul络 IP Network

Access DC Core DC
experience during interworking between LTE
and NR. When service data transmission is
Fig. 10. Illustration of network slicing. accomplished by LTE and NR simultaneous-
ly, the data can be directly offloaded to the
NR base station from the LTE base station or
from the EPC to the NR base station. For base
EPC EPC NG Core
station offloading, the hardware of the legacy
LTE base station must be upgraded to support
higher throughput. EPC offloading will not
LTE NR eLTE NR require hardware upgrades to the legacy LTE
base station.
Step 1 Step 2 NSA requires the terminal to support mul-
tiple connections, where LTE and NR radio
Fig. 11. Basic NSA deployment. transmission chains must be maintained simul-

44 China Communications • November 2018

taneously. Sharing of the total transmission the new features of the new core network, e.g.,
power of a terminal between the two trans- network slicing, while LTE UE connects to the
mission chains leads to less coverage than the LTE core network.
LTE system in uplink and a shorter battery In SA deployment, the 5G system enables
life. completion of end-to-end 5G capabilities, e.g.,
As the step 2, the new core network of low latency on initial access and service trans-
5G will be introduced, the LTE is evolved as mission and end-to-end network slicing to
eLTE and connected to both EPC and new enable 5G networks to support MEC, which is
core network, while NR is updated and con- essential to explore the enterprise and vertical
nected to the new core network. The new ca- markets.
pability of the new core network, e.g. network To fulfill market demand on the early de-
slicing, can be supported by 5G system. For a ployment of 5G, the 5G specification work in
LTE user or a legacy 5G user of step 1 deploy- 3GPP was accelerated since Oct 2016, and a
ment, it anchors in EPC and the legacy service preliminary version of NSA, together with the
is supported. For a 5G user with the latest common portions of Layer 1 (L1)/Layer 2(L2)
features introduced by step 2, it anchors in of NSA/SA, were completed at the end of
new core network and is enabled with the new 2017. The remaining portions of SA and NGC
capability of new core network, e.g. network will be completed by June 2018.
slicing. For 5G user in step 2 deployment, the The major technical differences between
mobility management and control is offered 5G NR and 4G are summarized below.
by eLTE, while NR still works as a data pipe. Features such as unlicensed spectrum NR,
For operators who are not willing to intro- D2D, SON, self-backhauling, and two-lev-
duce the new core network for initial NR de- el mobility have been postponed to Release
ployment, NSA is a suitable compromise that 16. To ensure the completion of the NR SA
affords additional capacity and high peak data of Release 16 on time, further down scoping
rate. When necessary, the NR can be evolved of specifications could be performed before
to SA by introducing the new core network 2020.
and upgrading the base station. To address the different evolution tracks of
Considering that LTE has been deployed 5G NR, 3GPP has decided to submit two sepa-
by operators in the large scale and can provide rate sets of 5G proposals to ITU-R WP5D.
better coverage than 5G NR at initial deploy- - LTE as a Radio Interface Technology
ment, NSA is very attractive to those operators (RIT) plus NR RIT
who are unwilling to deploy a large-scale 5G - NR as a RIT
NR network at the early stages of its develop-
ment. V. 5G FIELD TRIALS
As illustrated in figure 12, SA is a tradi-
tional option to deploy a brand-new system; Towards year 2020’s commercial launch,
here, both the NR and new core network are the time left for the production of end to end
deployed at initial deployment. A special inter- 5G system is quite limited. To really achieve
face between the EPC and new core network,
called Nx, is introduced to provide service
interoperation. UE anchors to the 5G system EPC NG Core EPC NG Core

when 5G coverage is available and hand-overs Nx

to LTE system when it is out of the 5G net-
work coverage. LTE NR eLTE NR
As a further evolution, LTE bases station
could be upgraded to eLTE and support new
core network; Thus eLTE UE can also support Fig. 12. SA Deployment.

China Communications • November 2018 45

Table I. Technical differences between 4G and 5G. the commercialization of 5G by 2020, the
Item 4G 5G standardization, industrialization and service
OFDM in downlink and OFDM in downlink, and SC-FD- cultivation of 5G system should be conduct-
Waveform
SC-FDMA in uplink MA or OFDM in uplink ed in parallel. Thus the time to market of 5G
UL/DL: 256QAM as baseline network would be shorten greatly compared to
Modulation UL: 64QAM
New pi/2 BPSK for UL what has happened in 4G.
Data channel: Turbo To accelerate the industrialization of 5G,
Data channel: LDPC
channel coding Control channel: Tail biting
Control channel: Polar China Mobile has started its 5G trials together
convolutional coding
with its partners since 2015 [17]. The trials are
- 15/30/60/120 KHz,
Numerology
-15 KHz subcarrier
- shorter TTI, e.g., for 30 KHz, 0.5
composed of 4 phases: individual key tech-
-1 ms TTI nology validation, system concept trial, large
ms TTI
Frame structure Fixed Semi-static and dynamic scale trial and commercial launch. To validate
Carrier bandwidth 1.4/5/10/15/20MHz 1.4/5/10/15/20/40/50/80/100MHz the entire 5G system’s performance and pre-
Reference signal CRS-based CRS-Free, DMRS-based pare for the pre-commercial products’ speci-
DMRS resource Fixed Sharing with PDSCH fication, the field trial of 5G system concept
PBCH/SS Wide Beam for broadcasting Narrow beam with beam-sweeping have been conducted in 2017.
CRS-based cell specific The objective of the trials in this article is
PDCCH DMRS based UE specific weight
weight to validate the hardware architecture of the
PDCCH resource Symbol level Symbol and frequency 3D-MIMO with 192 antenna elements, and
ACK/NACK delay Minimal N+4 Minimal N+0 understand the fundamental performance
UE power Max 23 dBm Max 26 dBm of the 5G NR air interface for eMBB and
1T2R baseline 2T4R baseline URLLC scenario, which are the potential early
UE capability
Max 20 MHz/carrier Max 100 MHz/carrier market of 5G. So only part of the key features,

Frame
Periodicity
#0 #1 #2 #3 #4 #5 #6 #7 #8 #9

DL DL DL GP UL DL DL DL GP UL

Subframe
(1ms))
(1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Slot 0 Slot 1

Fig. 13. TDD frame structure for eMBB.

Frame
Periodicity
#0 #1 #2 #3 #4 #5 #6 #7 #8 #9
G G G G G G G G G G
DL UL DL UL DL UL DL UL DL UL DL UL DL UL DL UL DL UL DL UL
P P P P P P P P P P

Subframe
(1ms))
(1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Slot 0 Slot 1

Fig. 14. TDD frame structure for low latency.

46 China Communications • November 2018

like 3D-MIMO and software defined air inter- Table II. System parameters of 5G NR for trial.
face are implemented within the framework of System parameters Value
5G NR air interface. band 3.4-3.6 GHz
The base stations used in the field trial Component carrier bandwidth 100 MHz
have followed the technical guide of 3.5GHz Subcarrier spacing 30 kHz
5G NR [18] which was published during the Guard band ratio <10%, e.g., 2.08%
Mobile World Congress in Barcelona in Feb- Total subcarrier bandwidth >90 MHz, e.g., 97.92 MHz
ruary 2017. The key technical parameters are Total available number of subcarriers >3000; recommended, 3264
detailed below. OFDM symbol length (w/o CP) 33.4 us
The typical frame structure used for the CP overhead 6.67%
5G system concept’s performance test are as FFT Size 4096
figure 14 and figure 15, where the downlink Sampling rate 122.88 MHz
subframe, uplink subframe and guard period >=175 Mbps (2 layers, 64QAM);
Peak data rate in UL
between them are defined respectively for >=350 Mbps (4 layers, 64QAM)
eMBB and low latency use cases. >=1.3 Gbps (4 layers, 256QAM);
Peak data rate in DL
Based on the technical guide provided >= 2 Gbps (8 layers, 64QAM)
above, field trials on the system concept of 5G C-plane: 20 ms; U-plane: <= 4 ms (for a single
Latency
3.5GHz NR were conducted in the university cycle of uplink and downlink)

city of Guangzhou of China. The trial scenario Transmission direction of time resources and
Frame structure downlink-to-uplink switch-point periodicity
is a typical dense urban. The site map of the
can be dynamically configured
trial network is referred as figure 15. The trial
Orthogonal multiple access; non-orthogonal
network is composed of 7 sites, where one site multiple access
multiple access (optional)
is surrounded by 6 sites and certain inter-cell Wave form OFDM, new waveform (optional)
interference is produced to show the practical LDPC for eMBB data channel;
performance. Each red dot is a site, which is Channel coding
polar code for control channels
configured with 1~3 sectors as needed. The 2Tx and 4Rx with single-user multi-stream
Device antennas
base station and the CPE used in the field trial beam-forming; 4Tx and 8Rx are optional
are shown as figure 16. Device Tx power 23dBm+23dBm for 2 Tx
For the eMBB scenario, the peak data rate Antenna element 192
and throughput are the key performance indi- Antenna port of 3D-MIMO 64Tx/64Rx
cator. So the general frame structure presented
in figure 14 is configured for the eMBB field
trial.
For the field trial, the QPSK, 64QAM and
256QAM with code rate of 1/3, 1/2, 3/4, 5/6
are supported in the prototype.
Table 3 shows single-user downlink
throughput results, which illustrate that 2 Gbps
data rates could be achieved by 8 layers and
64QAM. The trial results are quite close to the
corresponding theoretical values. Compared
to 110Mbps/20MHz of LTE TDD system, 5G
NR can achieve 4-fold peak data rate when 8
layers and 64QAM are configured.
Table 4 shows multi-user throughput results Fig. 15. Site map of the 5G trial network.
of 16 UEs in downlink. Using different mul-
tiplexing layers and modulation orders, dif- Compared to the typical cell throughput of
ferent peak cell throughputs can be achieved. 40Mbps/20MHz for LTE TDD system, 20-fold

China Communications • November 2018 47

 for 5G by 3D-MIMO.
Considering that the uplink is the coverage
bottleneck due to the much smaller transmit
power than that of downlink, more transmis-
sion antenna at the UE is expected to boost
single user data rate and coverage in uplink.
5G CPE To validate the exact performance of different
numbers of transmission antenna at UE, 2
and 4 antennas at UE are selected as two cas-
es for trial test. Figure 17 reveals that nearly
2-fold increases in data rate of uplink can be
Fig. 16. 5G base station and CPE.
achieved by two transmission antennas at UE.
As such, multiple transmission antennas at UE
are proposed to enhance uplink coverage and
user-experienced data rates.
0.7
For the low latency case of URLLC scenar-
io, the system is configured as figure 14 and
0.6
one direction transmission latency (defined as
0.5 the air interface processing delay from physi-
DL
0.4
cal layer to packet data convergence protocol
UL
ms

of data reception) is tested in the field trial.

0.3
For the test, the data packet with random size
0.2 of FTP service is transmitted and the transmis-
0.1
sion latency of each data packet is tested and
averaged in a duration of 1 second.
0
Vendor 1 Vendor 2 Vendor 3
From the test results shown in figure 18, a
latency of less than 1 ms can be achieved by
the current implementation.
Fig. 17. One-direction latency of air interface.
VI. CONCLUSION
Table III. Single-user throughput results in downlink.
Maximum Theoretical Trial To meet the market demands beyond 2020,
Item Modulation
Streams THP Results
a 5G system is expected to be offer 1 Gbps
Vendor 1 8 64QAM 2.1Gbps 1.86Gbps
user-experienced data rates, 1 ms radio inter-
Vendor 2 4 64QAM 1.05Gbps 0.9Gbps
face latency, and 100 million connections per
Vendor 3 4 256QAM 1.36Gbps 1.22Gbps
square kilometer. To support the differentiate
usage scenarios and deployment scenarios,
Table IV. Multi-user throughput results in downlink.
5G system is enabled by key features, e.g.,
Maximum Theoretical
Item Modulation Cell Peak THP 3D-MIMO, software-defined radio interface,
Layers THP/layer
user-centric network, and SDN/NFV-based
Vendor 1 24 64QAM 5.91 0.26
new core network. Field trials on 5G system
Vendor 2 28 64QAM 7.57 0.26
concept in the 3.5 GHz band are conducted to
Vendor 3 16 256QAM 4.31 0.34
validate the hardware architecture of 3D-MI-
MO and understand the fundamental perfor-
throughput gain could be achieved by 3D-MI- mance of the entire air interface of 5G NR in
MO when the bandwidth is normalized. So 2017. Trial results have shown that the goals
compared to that of 4G, 3~5 times improve- of the 5G system design, e.g., 2 Gbps data
ment on the spectrum efficiency is achievable rate, 20-fold improvement in cell throughput

48 China Communications • November 2018

 200

180

160

140

120
Thpt(Mbps)

100

80

60

40

20

0
131
141
156
163
169
183
191
208
212
216
217
228
244
265
273
274
277
279
281
282
288
302
311
319
328
347
358
359
360
363
365
367
367
368
370
371
373
373
374
Distance (m)

Uplink with 2Tx UE Uplink with 1Tx UE

Fig. 18. Throughput of different transmit antenna number at UE in uplink.

compared with that of 4G, and 1 ms one-direc- References

tion latency of the air interface, are achievable. [1] ITU-R WP5D TD-0591, Naming for International
To fulfill all the technical demands from Mobile Telecommunications[R], 2015.
the different vertical industries, the 5G NR [2] ITU-R WP5D TD-0625, IMT Vision–Framework
and Overall Objectives of the Future Develop-
still has a long way to go. First, the standard ment of IMT for 2020 and Beyond[R], 2015
should be further enhanced to cover fully ul- [3] Report ITU-R M.2411-0, Requirements, eval-
tra-low latency and high reliability, boost the uation criteria and submission templates for
data rate at the cell edge, and support the net- development of IMT-2020, 2017.
[4] 3GPP RP-161596, New SID Proposal: Study on
work automation, V2X [19], etc. Second, the New Radio Access Technology, 2016.
products of end to end 5G system need further [5] 3GPP TSG SA meeting#70, SP-150853, study on
optimization on the hardware efficiency and Architecture for Next Generation System.
cost efficiency and support large scale deploy- [6] Vision and requirements of IMT-2020, http://
www.imt-2020.cn
ment. Third, the new service/application and [7] G. Liu, X. Hou, F. Wang, et al, “Achieving
business model should be cultivated to serve 3D-MIMO with Massive Antennas from The-
for the future enterprise and vertical markets. ory to Practice with Evaluation and Field Trial
The entire eco-system is encouraged to col- Results”,[J] IEEE Systems Journal, vol. 11, no. 1,
2016, pp. 62-71.
laborate together to make everything ready for [8] A. Scrase, “3gpp standards update”, www.3gpp.
year 2020’s commercial deployment and en- org/ftp /.../ 2010 _11_Africa_Com_%20SCRASE.
able 5G to serve us a fully connected society. ppt , Nov. 2010.
[9] S. Brueck, L. Zhao, J. Giese, et al., “Centralized
scheduling for joint transmission coordinated
ACKNOWLEDGEMENT
multi-point in LTE-advanced”, IEEE International
ITG WSA, 2010, pp. 281-302.
This work is supported in part by national Key [10] J. Zhang, “Review of Wideband MIMO Channel
Project (2016ZX03001021). The author would Measurement and Modeling for IMT-Advanced
like thank all colleagues who took part in this Systems”, [J]. Chinese Science Bulletin, vol. 57,
no. 19, 2012, pp. 2387-2400.
projects from Ericsson, Huawei, Nokia, and
[11] Chih-Lin I, Yunlu Liu, Shuangfeng Han, et al, “On
all the colleagues who have joined the 5G tri- big data analytics for greeer and softer RAN”,
als in Guangzhou from ZTE. IEEE Access, vol. 3, no. 94, 2015, pp. 3068-3075.

China Communications • November 2018 49

 [12] Ramiro, Juan; Hamied, Khalid, “Self-Organizing technology, and customer and market research. She
Networks”, John Wiley & Sons,2012 has participated in several important projects, such
[13] Network function virtualization (NFV) man- as technical and strategic research, specification, test
agement and orchestration: GS NFV-MAN 001 and trail of GSM900/1800, GPRS/EDGE, CMNet,
V0.3.14[S]. 2014. WLAN, 3G, and LTE. She also leads the research and
[14] Software-defined networking: the new norm for standardization work of LTE evolution and 5G. She
networks [EB/OL].[2016-03-15]. https://www. served as the Vice-Chair of the 3GPP SA. She serves
opennetworking.org/images/stories/down- as a Steering Committee Member of the NGMN and
loads/sdn-resources/white-papers/wp-sdn- the Secretary-General of the Global TD-LTE Initiative.
newnorm.pdf.
[15] Report ITU-R M.2412-0, Guidelines for evalu- Fei Wang, received the B.S.
ation of radio interface technologies for IMT- and M.S. degrees in electronic
2020, 2017. engineering from the Beijing
[16] 3GPP TSG RAN R3-161809, Analysis of migra- University of Posts and Tele-
tion paths towards RAN for new RAT, CMCC, communications in 2008 and
2016. 2011, respectively. In 2011, he
[17] http://www.imt-2020.cn joined the China Mobile Re-
[18] Technical guideline for 5G 3.5GHz NR PoC and search Institute, where he is
trials, China Mobile, Feb. 2017, MWC Barcelona. currently involved in the research, prototyping, and
[19] Xiaohu Ge, Zipeng Li, Shikuan Li, “5G Software standardization of multiple antenna techniques in
Defined Vehicular Networks”, IEEE Communica- 5G. His research interests include multiple antenna
tions Magazine, Vol. 55, no. 7, 2017, pp. 87-93. techniques, coordination schemes, and advanced re-
ceivers for next generation systems.
Biographies
Guangyi Liu, received the Jianjun Liu, received Ph.D. de-
Ph.D. degree in circuit and sys- gree from the Graduated
tem from Beijing University of School of Chinese Academy of
Posts and Telecommunication Sciences in 2008. He is current-
(BUPT) in 2006. He is the CTO ly with the China Mobile Re-
of wireless and device depart- search Institute and his re-
ment of China Mobile research search interests include
institute. He has published LTE-Advanced and 5G NR, such
more than 100 articles in referred journals and con- as massive MIMO, small cell enhancement, and inter-
ferences and filed more than 100 patents. He was ference coordination, etc.
awarded “national innovation award in science and
technology of 2016” of Chinese government due to Qixing Wang, received the B.S.
his contribution to TD-LTE standardization and in- and Ph.D. degrees from the
dustrialization globally in last 10 years. In 2009 and Beijing University of Posts and
2013, he received two “excellent patent award” from Telecommunications in 2003
the Chinese Intellectual property office; His current and 2008, respectively. He is
work focuses on LTE evolution, 5G research/stan- currently with the China Mobile
dardization and industrialization. Research Institute. His research
interests include massive
Yuhong Huang, received the MIMO, multi-user beamforming, and interference
degree from the Beijing Uni- cancellation.
versity of Post and Telecom.
She joined the China Mobile
Research Institute in 1996,
where she is currently the Dep-
uty General Manager, where
she is responsible for the work
of wireless technology, terminal technology, test

50 China Communications • November 2018