188 Gao et al.

/ Front Inform Technol Electron Eng 2020 21(1):188-194

Frontiers of Information Technology & Electronic Engineering

www.jzus.zju.edu.cn; engineering.cae.cn; www.springerlink.com
ISSN 2095-9184 (print); ISSN 2095-9230 (online)
E-mail: jzus@zju.edu.cn

Correspondence:
5G evolution promoting innovation of antenna systems*

Feng GAO†‡1, Peng GAO1, Wen-tao ZHU1, Chen-xi ZHANG1, Xian-kun MENG1, Run-hong SHAN2
1
China Mobile Group Design Institute Co., Ltd., Beijing 100080, China
2
Copyright Protection Center of China, Beijing 100050, China
†
E-mail: gaofeng1@cmdi.chinamobile.com
Received Oct. 14, 2019; Revision accepted Jan. 4, 2020; Crosschecked Jan. 27, 2020

Abstract: We introduce the basic concept, background, and development of mobile communication systems from the first gen-
eration (1G) to the fifth generation (5G) including their antenna systems. We also describe the requirements for 5G networking and
optimization of antenna systems, and present the basic principle of three-dimensional array antennas. Weight optimization
methods of massive multiple-input multiple-output (MIMO) antennas are proposed and verified. Finally, several ideas are given to
solve the problem of power consumption of 5G antenna systems.

Key words: Fifth generation (5G); Massive multiple-input multiple-output (MIMO) antenna array; Power consumption; Weight
optimization
https://doi.org/10.1631/FITEE.1900561 CLC number: TN821

1 Introduction receiver end employed thin rod-based extendable

monopole antennas. In the second generation (2G),
In the past 30 years, wireless communication has the initial antenna technologies used in its base sta-
achieved a tremendous success in the global market. tions were single-polarized omnidirectional or direc-
Even after decades of fast growth, there is a steady tional antennas. Then, dual-polarized antennas with
increase in the number of cellular devices manufac- remote electrical downtilt were implemented. In 2000,
tured, which surpasses the population in some coun- the third generation (3G) was introduced. The smart
tries because of consumers’ need to stay connected antenna of the TD-SCDMA system was initially ap-
wirelessly. For instance, since China Mobile commer- plied in mobile communication. China Mobile was
cially launched Time Division-Synchronous Code Di- the only operator of the TD-SCDMA system in the
vision Multiple Access (TD-SCDMA) 3G in 2009 world, and it promoted the development of Time
and Time Division-Long Term Evolution (TD-LTE) Division Duplex (TDD) technology (Basavarajappa,
in 2013, statistics showed that there had been almost a 2018). Long-Term Evolution (LTE) is commonly
billion subscribers by October 2018 (Noble, 1962; referred to as advanced fourth generation (4G) or 4G
Donald, 2014). LTE, and the multiple-input multiple-output (MIMO)
The original, voice-only wireless telephone technology is one of the two key techniques of LTE.
technology is the first generation (1G). In 1G, the Based on the TD-LTE technology, China Mobile has
applied the 3D-MIMO technology, which is also
called pre-5G. To achieve 1000 times capacity gain
‡
Corresponding author and 10 Gb/s of speed in 5G, massive MIMO antenna
*
Project supported by the National Major Projects of China (No. systems have become the key in wireless systems
2018ZX03001022-001)
ORCID: Feng GAO, https://orcid.org/0000-0002-1882-1654
(Hoydis et al., 2013; Chen SZ et al., 2015). The evo-
© Zhejiang University and Springer-Verlag GmbH Germany, part of lution of the mobile communication antenna is illus-
Springer Nature 2020 trated in Fig. 1.
 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194 189

Fig. 1 Evolution of the mobile communication antenna (Chen LP, 2015)

BTS: base transceiver station; RET: remote electrical tilt

2 Requirements of 5G for antenna systems

China Mobile’s 5G networking modes are

standalone (SA) and non-standalone (NSA), which
coexisted in 2019, with SA focusing on meeting the
industry’s vertical application requirements. At first,
FDD1800 was considered to be an NSA anchor, fol-
lowed by TDD F band (Fig. 2).
2.1 5G networking to antenna systems
According to Shannon’s formula, there is

 S
C  B log 2 1   , (1)
 N
Fig. 2 5G network architecture
where C is the channel capacity, B the bandwidth (Hz), EPC: evolved packet core; LTE: long-term evolution; NGC:
S the signal power (W), and N the noise power (W). next-generation core; NG: next generation; CPE: customer
Shannon’s formula can be further expressed using the premise equipment; NR: new radio; SA: standalone; NSA:
non-standalone
formula shown in Fig. 3.
By the time when 5G encoding and decoding
processes would reach a capacity limit, antenna sys- improve the spectrum utilization (Nadeem et al.,
tems would be a key for improving the capacity of 5G 2018).
networks. Compared with beamforming to a single Because the speed of 5G is high, two systems are
user, most massive MIMO (MM) antennas have the used: one is the wide frequency band and the other is
ability to multiplex to multiple users, which can the MM antenna. Compared with 4G three-dimensional
190 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194

MIMO (3D-MIMO), the coverage ability of the signal processing of baseband and radio frequency
broadcast control channel of the 5G MM antenna is systems achieve the corresponding efficiencies.
stronger. Six narrow beams are used in a horizontal The core requirements for 5G MM antennas are
plane, and the coverage range is improved to 2–4 dB; miniaturization (space reduction), low profile, light
two wide beams are used to improve the coverage weight, multi-device integration (antenna on package,
range in a vertical plane compared with 4G AoP), and over-the-air self-calibration. Typical ex-
3D-MIMO. amples include the lens antenna array based on super-
material, magnetic dipole antenna, holographic
beamforming antenna, and solid-state plasma antenna
array (Fig. 4). Among them, the lens antennas with
super-materials are applied in 5G, studied in the
China Mobile R&D Center of Network Design and
Optimization.
2.2 5G massive MIMO antenna array
In a 3D MM antenna array, the antenna pattern is
not only determined by the weight vector of the array
elements, but also affected by the inductive signals on
each array element, which is different from the
common antenna array.
Fig. 3 Capacity factors of Shannon’s formula The matrix factor is given by

With the introduction of 5G communication E ( ,  )   amn exp  j (mdx  n)

systems, antenna systems transformed from passive m n (2)
to active, more intelligent, and more miniaturized.  k (u  u0 )  ndyk (v  v0 ) ,
The 5G MM antenna system not only has dipole an-
tennas and calibrated networks, but also includes where m and n are numbers of rows and columns of
radio frequency (RF) active devices such as power the antenna array respectively, θ is the angle between
amplifiers and filters. With this, more systematic and the direction of arrival and the z axis, ϕ is the angle
integrated antenna designs would be required. Inte- between the direction of arrival and the x axis, u and v
gration design of performances such as large-scale are the electric potentials in x and y directions re-
antenna arrays, 3D-MIMO, and low loss, can help the spectively, u0 and v0 are initial potentials in x and y

Fig. 4 New types of 5G massive MIMO antenna: (a) lens antenna with super-material; (b) integrated lens antenna;
(c) holographic beamforming antenna; (d) solid-state plasma antenna (Zhao et al., 2017)
 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194 191

directions respectively, and dx and dy are differentials optimized method is needed to realize the artificial
of x and y respectively. Δn=nd/(λcos θ) is the phase intelligence (AI) optimization for the 5G network.
difference caused by the space wave path difference, The targets of weight optimization include de-
where λ is the wavelength and d the array spacing. In creasing the interference and improving the
addition, k=2π/λ. For a 5G antenna array, the ampli- throughput and user satisfaction. First, the hot spots,
tude distribution amn can be separated. Assuming the e.g., the central business district (CBD), residential
induction signal component at each element is a sta- areas, and schools, are located through the AI tech-
tionary process with an average value of zero, then the nology, and different scenarios can be identified.
radiation pattern of the antenna array can be calcu- Second, 3D map grid is implemented based on the
lated by the conditional mathematical expectation of minimization of drive tests data of the 4G/5G shared
P(θ, ϕ, t) given the weight vector W. antenna and 5G beam data of user distribution. Third,
The 3D module of an MM antenna array can be the sample library of weights is determined using a
expressed below: machine learning algorithm, the K-nearest neighbor
(KNN) algorithm.
P( ,  )  E[ P( ,  , t )] When data accumulation reaches a certain pro-
portion, a more efficient differential evolution (DE)
 P 2 ( ,  )W H E[ S (t ) S H (t )]W (3)
E
iteration algorithm can be used to achieve unit-level
= P ( ,  )W H RW ,
E
2
joint weight adjustment. According to the user
equipment (UE) distribution in different scenarios,
where W is the conjugate of W and S(t) is composed the antenna weight self-optimizing configuration can
of m elements (i.e., s1, s2, …, sm). be adjusted dynamically to replace part of the work of
Given the correlation coefficients  and consid- traditional manual network planning and optimization.
ering model simplification, the correlation coefficient The DE algorithm is a stochastic optimization
matrix R can be expressed as algorithm based on population, using N d-
dimensional parameters (i=1, 2, …, N; j=1, 2, …, D)
to parallelly and directly search in the search space.
1  ...  
 Here, N is the population size and D is the number of
1 ...  
R  E[ S (t )  S H (t )]   . (4) decision variables. The basic operations of standard
   DE include mutation, crossover, and selection. To
 
  ... 1  verify the accuracy of the coverage optimization al-
gorithm for a 5G MM antenna array, the optimized
Then the pattern model of an MM antenna array target function is set as
can be simplified as follows:
 1 N
2
E  min   [ f (i )  fT (i )]2
 N
 N 1
P( ,  )  P ( ,  )  vi  (1   ) PE2 ( ,  )
2 (6)

E
N i 1
(5)   {f ( 0 )  max[f ( j )]} + E + E  ,
4

  N 2

 PE2 ( ,  ) 1   
  v i ( N  1)   .

 i 1  where , , , and  are weight factors constraining
the main beam, beam direction, sidelobe, and weight
efficiency respectively, and f(x1, x2, …, xn) is the
2.3 Weight optimization of a massive MIMO an-
nonlinear target function.
tenna array
Table 1 shows the suggestion of broadcast pat-
The MM antenna introduces more dimensionally terns in different scenarios. Generally, pattern 1 is
adjustable parameters to cover multiple scenarios, recommended when applying the scenario of a typical
making 5G optimization more complex. The tradi- 3-sector. If a large horizontal coverage is required,
tional way is no longer applicable. An intelligently scenario 8, 9, or 10 is recommended. In this case,
192 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194

higher beam gains can be obtained at the cell edge, to select appropriate time-frequency resources to
and the coverage range can be improved. If a fixed transmit narrow beams. Table 2 shows three different
interference source exists at the cell edge, scenario 2 configurations of broadcast beams. In the condition
or 3 can be used to narrow the horizontal coverage with eight beams, the horizontal half power beam
range and reduce the inference. width (HPBW) of the sub-beam series is [16, 10,
Given isolated buildings, scenario 4, 5, 6, or 7 is 10°, 10°, 10°, 10°, 10°, 16°], the direction of the
recommended to provide a small horizontal coverage. sub-beam series is [−29°, −20°, −12°, −4°, 4°, 12°,
These scenarios, however, are not suitable for con- 20°, 29°], and the sub-beam vertical HPBW is 6°.
tinuous networks. The eight-beam disposition is applied for com-
Either scenario 1 or 8 should be selected for mon scenarios for macro coverage and low-floor
low-rise buildings. For middle-rise buildings, one coverage. Two kinds of new radio (NR) broadcast
from scenarios 2, 4, and 9 can be selected, while for beam configurations are optimized and simulated
high-rise buildings, one from scenarios 3, 5, 6, 7, and using a commercial 3D electromagnetic field analysis
10 can be chosen. software ANSYS HFSS. Fig. 5 shows the coverage in
The purposes of weight optimization and beam the square scenario. The wide beams are used at the
management are to properly design narrow beams and cell center to ensure access, and the narrow beams are

Table 1 Broadcast beam patterns in different scenarios

Horizontal Vertical Azimuth Digital beam
Pattern Application scenario
HPBW (°) HPBW (°) (°) tilt (°)
1 65 6 [−10, +10] [−3, +12] Standard macro coverage and low-floor coverage
2 65 12 [−10, +10] [−3, +12] Standard macro coverage and middle-floor coverage
3 65 25 [−10, +10] [+9, +12] Standard macro coverage and high-floor coverage
4 45 12 [−20, +20] [−3, +12] Middle buildings and hotspot coverage
5 45 25 [−20, +20] [+9, +12] High buildings and hotspot coverage
6 30 25 [−30, +30] [+9, +12] High buildings and hotspot coverage
7 15 25 [−40, +40] [+9, +12] High buildings and hotspot coverage
8 90 6 0 [−3, +12] Standard macro coverage and low-floor coverage
9 90 12 0 [−3, +12] Standard macro coverage and middle-floor coverage
10 90 25 0 [+9, +12] Standard macro coverage and high-floor coverage
HPBW: half power beam width

Table 2 Different configurations of broadcast channel narrow beams

Broadcast beam subBeam index Azimuth (°) Downtilt (°) Horizontal HPBW (°) Vertical HPBW (°)
Single beam 1 0 6 65 6
0 −27 6 20 6
1 −9 6 20 6
Four beams
2 9 6 20 6
3 27 6 20 6
0 −29 6 16 6
1 −20 6 10 6
2 −12 6 10 6
3 −4 6 10 6
Eight beams
4 4 6 10 6
5 12 6 10 6
6 20 6 10 6
7 29 6 16 6
HPBW: half power beam width
 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194 193

used at the cell edge to improve the coverage range. evaluation reduces the optimized cell range and fo-
Fig. 6 shows the coverage in a high-rise building cuses on beam issues.
scenario. The beams with a wide vertical coverage are The key techniques of this system include
used to improve the vertical coverage range. weight library construction, user hotspot clustering,
Beam parameter optimization includes weight and spatial scene recognition. The spatial clustering
initialization and weight optimization. Fig. 7 shows technology enables automatic user hotspot discovery
the weight optimization flowchart. The input of the and automatic scene recognition.
system is network data, such as measure reports and
2.4 Problems in power consumption of 5G an-
electronic map data. The weight optimization threshold
tenna systems
Different factories have different equipment
forms and parameters. For 5G equipment, some
equipment bandwidth is 160 MHz and some is
100 MHz; some equipment power is 240 W and some
is 200 W. Its power consumption is far greater than
that of 4G devices in the pre-commercial test of China
Mobile.
The power consumption of the 5G base station in
the test network is about 2.5–3.5 times that of the 4G
base station. The full-load power of a 5G base station
is nearly 3700 W, which is necessary to plan the
power supply and supporting system in advance. It
not only challenges the distribution of power in the
communication workshop, but also puts forward high
requirements for the expansion of power supply.
Table 3 shows the power consumption of 4G and 5G
equipment under different load conditions.
The power consumption of 5G equipment comes
Fig. 5 New radio broadcast beams in square scenarios
mainly from the active antenna unit (AAU), and the
power amplifier consumes the most power in an AAU.

Fig. 6 New radio broadcast beams in high-rise buildings Fig. 7 Weight optimization flowchart
194 Gao et al. / Front Inform Technol Electron Eng 2020 21(1):188-194

Table 3 Power of 4G/5G under different conditions ommended. Weight optimization would be imple-
Power (W) Power ratio mented by the MM antenna array, and it would ob-
Load (%)
4G 5G (5G/4G) viously improve the quality of test network. However,
100 1044.72 3674.85 3.5 the improvement in the performance of the 5G system
50 995.06 2969.97 3.0 will also bring a significant increase in the power
30 949.22 2579.83 2.7 consumption. The 5G MM antenna arrays can poten-
Empty 837.21 2192.57 2.6 tially reduce uplink and downlink transmit powers
through coherent combination and an increased an-
Compared with a 4G remote radio unit (RRU), a 5G tenna aperture.
AAU has 10 times more channels and more than five
times larger bandwidth. Therefore, manufacturers Contributors
should improve the efficiency of the power amplifier Feng GAO designed the research. Feng GAO, Peng GAO,
and reduce the power consumption from the source by and Wen-tao ZHU processed the data. Feng GAO wrote the
first draft of the manuscript. Wen-tao ZHU, Chen-xi ZHANG,
improving the algorithm capability and adopting and Xian-kun MENG helped organize the manuscript. Feng
more advanced power amplifier technologies. GAO, Peng GAO, and Run-hong SHAN revised and edited the
Many factors such as RF devices, equipment, final version.
base stations, and network dimensions should be
studied to reduce the 5G power consumption. China Compliance with ethics guidelines
Mobile’s 2.6-GHz spectrum can be shared by 4G and Feng GAO, Peng GAO, Wen-tao ZHU, Chen-xi ZHANG,
5G equipment. The 4G system can be loaded in re- Xian-kun MENG, and Run-hong SHAN declare that they have
no conflict of interest.
verse at the 2.6 GHz frequency band of a 5G AAU
module. If the original 8-channel D-band module is
References
removed, the power consumption can be effectively Basavarajappa V, 2018. A Proposal of Antenna Topologies for
reduced for 5G base stations. Based on multi- 5G Communication Systems. PhD Thesis, Queen Mary,
frequency multi-network cooperative energy con- University of London, London, UK.
servation, when adopting intelligent energy-saving Chen LP, 2015, Evolution from traditional base station to
technologies such as symbol shutdown, carrier shut- active antenna system base station. Telecommun Eng,
55(8):942-946 (in Chinese).
down, channel shutdown, and deep dormancy for
Chen SZ, Sun SH, Wang YM, et al., 2015. A comprehensive
low-load scenarios, 2G/4G/5G cooperative energy- survey of TDD-based mobile communication systems
saving can be achieved. For example, the China Mo- from TD-SCDMA 3G to TD-LTE(A) 4G and 5G direc-
bile Network Department and Design Institute used tions. China Commun, 12(2):40-60.
intelligent power-saving technologies through the https://doi.org/10.1109/CC.2015.7084401
Donald VHM, 2014. Advanced mobile phone service: the
intelligent parameter management module on the
cellular concept. Bell Syst Techn J, 58(1):15-41.
network large-data platform. https://doi.org/10.1002/j.1538-7305.1979.tb02209.x
Hoydis J, ten Brink S, Debbah M, 2013. Massive MIMO in the
UL/DL of cellular networks: how many antennas do we
3 Conclusions need? IEEE J Sel Areas Commun, 31(2):160-171.
https://doi.org/10.1109/JSAC.2013.130205
Nadeem QUA, Kammoun A, Debbah M, et al., 2018. Design
We have reviewed the development of commu- of 5G full dimension massive MIMO systems. IEEE
nication systems from 1G to 5G and the evolution of Trans Commun, 66(2):726-740.
requirements of mobile communication systems for https://doi.org/10.1109/TCOMM.2017.2762685
antenna systems. Particularly, in the 5G era, MM Noble DE, 1962. The history of land-mobile radio communi-
antenna systems become the key technology. The cations. Proc IRE, 50(5):1405-1414.
https://doi.org/10.1109/JRPROC.1962.288119
AAU solution has been presented, and the basic Zhao JF, Tian S, Dai JW, et al., 2017. Discuss for 5G
principle and related theory of the 3D antenna array high-frequency antenna. Telecommun Technol, 2017(11):
have been elaborated in detail. New requirements of 125-128 (in Chinese).
5G mobile communication system networking and
optimization for the antenna system have been rec-