3592 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 37, NO.

14, JULY 15, 2019

5G Trial Services Demonstration: IFoF-Based

Distributed Antenna System in 28 GHz
Millimeter-Wave Supporting Gigabit Mobile Services
Minkyu Sung , Joonyoung Kim , Eon-Sang Kim, Seung-Hyun Cho, Young-Jun Won, Byoung-Chul Lim,
Sung-Yeop Pyun, Joon Ki Lee, and Jong Hyun Lee

Abstract—In this paper, we demonstrate fifth-generation (5G) I. INTRODUCTION

trial services using intermediate frequency-over-fiber (IFoF) based
OBILE communication has evolved beyond long term
distributed antenna system (DAS) implemented in a 28-GHz
millimeter-wave (mmWave) during Pyeongchang winter olympics.
To extend mmWave-based 5G wireless coverage, IFoF-based DAS
M evolution (LTE) based fourth-generation (4G) systems
toward fifth-generation (5G) systems. Commercial service for
with 2 × 2 multiple-input multiple-output configuration is pre- 5G is expected to be launched in the near future. 5G mobile com-
sented. Error vector magnitude (EVM) measurements of the 5G
signal on the IF and mmWave bands are performed for both down-
munication requires significantly ultra-fast data transfer, low la-
link and uplink transmissions over 5-km single mode fiber. The tency, and massive connectivity in wireless networks to support
experimental results confirm that the minimum requirement for an increasingly diverse set of services and applications such as
5G signals, defined by the technical specification of the 3G partner- mobile ultra-high definition video streaming, autonomous driv-
ship project, are satisfied by EVMs less than 9%. In addition, the ing, and virtual/augmented realities applications [1], [2]. The
developed IFoF-based DAS that supports 5G mobile service using
28-GHz mmWave was deployed in the city of Pyeongchang, during
architecture of the radio access network (RAN) has gradually
winter olympics. An ultra-high speed 5G data service was offered become important to satisfy 5G requirements [3], [4].
to the general public for 5G trial demonstration. The reference sig- As efficient approach to support RAN, a splitting the function-
nals received power and data throughput of the 5G handset are alities of evolved NodeB (eNB) into remote radio heads (RRHs)
measured while moving in a DAS section and 5G connected car and baseband units (BBUs) is used [5]. The RRH carries out
section. We confirm that the downlink throughput of ∼1 Gb/s and
uplink throughput of ∼200 Mb/s are achieved for the 5G trial ser-
simple roles that antenna and basic radio frequency (RF) func-
vice area in which an IFoF-based DAS is deployed. We also suc- tionalities, such as signal amplification and frequency conver-
cessfully demonstrate the 5G connected car service, which is one of sion. All functionalities related to digital signal processing from
the 5G key services, through deployed IFoF-based DAS. At the in- the physical layer and above are performed in the BBU. The
tersection of baseband unit and DAS deployed sections, a seamless BBUs are consolidated in a centralized location and share mul-
handover is performed to smoothly support 5G-based autonomous
driving service.
tiple distributed RRHs. This architecture was originally called
centralized/cloud RAN (C-RAN). The C-RAN provides not just
Index Terms—5G communication system, distributed antenna better cost optimization, but also better mobility performance.
system, millimeter wave, radio-over-fiber. Until now, mobile fronthaul, a transport network between cen-
tralized BBUs and distributed RRHs, was typically implemented
using digital optical transmission methods such as common pub-
lic radio interface (CPRI) [6]. The CPRI, which is most used by
RAN vendors, has several remarkable benefits such as simple
Manuscript received February 14, 2019; revised April 20, 2019; accepted hardware implementation and simplified control and manage-
May 14, 2019. Date of publication May 22, 2019; date of current version July 9, ment. However, the digitizing baseband mobile signal for CPRI-
2019. This work was supported by “The Cross-Ministry Giga KOREA Project” formatted frame requires very large transmission capacity over
grant funded by the Korea government (MSIT) (No. GK18N0300, Development
of Indoor DAS Technology based on IFoF for 5G Mobile Communication). the fronthaul link [7]. Thus, many research efforts have been
(Corresponding author: Minkyu Sung.) made to find a practical solution for 5G mobile fronthaul: there
M. Sung, J. Kim, E.-S. Kim, S.-H. Cho, J. K. Lee, and J. H. Lee are with are two well-knowns solutions [8]–[14]. The first method is the
the Optical Network Research Group, Electronics and Telecommunications
Research Institute, Daejeon 34129, South Korea (e-mail: [email protected]; BBU functional-split that separates a part of the medium ac-
[email protected]; [email protected]; [email protected]; [email protected]; cess control (MAC)-layer and physical-layer functions from the
[email protected]). centralized BBUs to distributed RRHs [15]. Therefore, the re-
Y.-J. Won and B.-C. Lim are with the Fiber Radio Technologies, Gyeonggi-do,
South Korea (e-mail: [email protected]; [email protected]). quired transmission capacity over the fronthaul link is reduced
S.-Y. Pyun is with the KT Cooperation, Gyeonggi-do, South Korea (e-mail: significantly. The other method is the analog radio-over-fiber
[email protected]). (RoF) technology based on the use of multiple carriers in the
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. intermediate frequency [16]. It aggregates and transmits mobile
Digital Object Identifier 10.1109/JLT.2019.2918322 signals over the fronthaul link, allowing significant mitigation in

0733-8724 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
SUNG et al.: 5G TRIAL SERVICES DEMONSTRATION: IFOF-BASED DISTRIBUTED ANTENNA SYSTEM 3593

the transmission capacity requirements while maintaining low- construct RoF-based DAS supporting mmWave 5G mobile ser-
latency characteristics [17]–[19]. In addition, data-compression vice could be reduced using low-cost DML. Despite its attractive
and passive optical network (PON)-based fronthauling advantages, the IFoF transmission system based on DML could
techniques have been studied [20]–[23]. Using these methods, experience severe performance degradation due to second-order
mobile fronthaul in RAN can meet 5G requirements, and net- distortions induced by interplay between fiber chromatic disper-
work equipment vendors can present their 5G network solutions sion and DML’s chirp. To resolve this problem, there have been
to the mobile operators. efforts such as digital signal processing (DSP) based second-
In recent years, there has been a growing interest in the spectra order distortion cancellation technique [32]. Besides, since the
of the millimeter wave (mmWave) range, which is largely un- frequency band of the second-order distortions is predictable, it
regulated, because it has a wide range of available bandwidth. is possible to derive an appropriate frequency plan avoiding the
This makes it ideal for 5G mobile communication [24]. Ow- distortions. By utilizing the frequency plan such as IF spacing
ing to its small wavelength, compared with ultra-high frequency optimization and octave mapping, we easily could avoid second-
bands (UHF) in the range of 300 MHz–3 GHz, it can incorpo- order distortions in IFoF-based transmission system [33], [34].
rate multiple antenna elements into the limited space at mmWave Using these approaches, IFoF-based transport systems have be-
transceivers. Therefore, using mmWave-based mobile commu- come more attractive on 5G networks that require huge amounts
nication, it is easy to implement beamforming techniques, which of data traffic.
provide antenna array gain that compensates for the frequency- Previously, we introduced a bandwidth-efficient IFoF-based
dependent path loss and reduces out-of-cell interference [25]. indoor DAS to increase network coverage [35]. We also exper-
In South Korea, 28-GHz spectrum was auctioned to mobile imentally confirmed the technical feasibility of the simultane-
operators; 5G commercial service will be launched using the ous transmission of 4G LTE-advanced (LTE-A) and 5G mobile
28-GHz mmWave frequency band for urban hotspots. However, signals in the form of an IF carrier [36]. In addition, we demon-
5G mobile networks that use mmWave frequency band would strated using a 5G prototype through IFoF-based analog optical
experience a greater link loss than 3G and 4G mobile networks link for 5G mobile communication in real-time [37]. However,
when penetrating low-emissivity windows, concrete walls, etc. previous studies have only carried out studies in the lab envi-
It is considerably difficult to cover indoor environments using ronment using prototypes. Before deploying a mmWave-based
outdoor 5G macro signals; therefore, a dedicated network is nec- 5G network infrastructure on a nationwide scale, it is necessary
essary for indoor environments. to verify that IFoF-based DAS technology can be applied to 5G
Distributed antenna system (DAS) are commonly used as a commercial service beyond simple technical validation.
solution to increase network coverage in indoor environments To the best of our knowledge, for the first time, we demon-
[26]. Many DAS solutions employ radio frequency-over-fiber strated 5G trial service using IFoF-based DAS in 28-GHz
(RFoF) technology, in which RF signals are converted to opti- mmWave frequency band [38]. We evaluated the transmission
cal signals in an analog waveform and transmitted to a remote performance of IFoF-based DAS for downlink by using error
antenna unit through optical fibers; then, they are radiated by vector magnitude (EVM). In this paper, we report the results
the antenna in the licensed frequency [27]. Therefore, RFoF- of 5G field demonstration using IFoF-based DAS. During Win-
based DAS could also accommodate multiple wireless signals ter Olympics, KT cooperation, a South Korean mobile operator,
simultaneously, such as wireless local area network (WLAN) provided 5G trial service in the city of Pyeongchang. To im-
and mobile signals (3G and LTE). By using simple direct modu- prove network coverage, we deployed IFoF-based DAS in the
lated laser (DML) and PIN photodiode (PD), the cost of im- Olympics area. By moving the 5G handset from the IFoF-based
plementation and complexity could be reduced significantly, DAS deployed area to 5G connected car section, the reference
making RFoF-based DAS an attractive solution for indoor envi- signals’ received power (RSRP) and throughput of the 5G hand-
ronments [28]. Despite these features, there are some limitations set are measured for the both downlink and uplink transmissions.
of implementing mmWave-based 5G services through RFoF- EVM of IFoF-based DAS are also experimentally investigated
based DAS. First, unlike 3G and 4G mobile services that use the for uplink.
UHF band in the range of 300 MHz–3 GHz, mmWave-based
5G mobile signals would experience power fading induced by
fiber dispersion, which significantly deteriorates transmission II. IFOF-BASED DAS SUPPORTING 5G MOBILE SERVICE IN
performance [29]. Next, a commercial DML does not support 28 GHZ mmWAVE FREQUENCY BAND
the mmWave frequency band. Therefore, an external modulator As mmWave-based 5G mobile communication system are go-
is required, which increases the implementation cost and com- ing to be commercialized, many stringent technical challenges
plexity. Another limitation is the loss of bandwidth efficiency. such as limited wireless coverage have been raised. The low pen-
Frequency bands outside the licensed frequency bands are left etration characteristics of mmWave frequency band make it diffi-
unused. A promising alternative is an intermediate frequency- cult to cover indoor environments using outdoor macro signals.
over-fiber (IFoF) technology, in which the carrier frequency is To support indoor environments and increase mmWave-based
converted to an intermediate frequency that is relatively lower 5G network coverage, we present the IFoF-based DAS in the
than the RF carrier frequency [30], [31]. The IFoF technol- 28-GH mmWave. Fig. 1 shows the developed IFoF-based DAS
ogy has flexible and efficient bandwidth utilization compared that provides a 1:8:32 network topology with a 2 × 2 MIMO
with RFoF technique. Furthermore, the capital expenditure to configuration. The IFoF-based DAS is composed of one main
3594 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 37, NO. 14, JULY 15, 2019

Fig. 1. Conceptual diagram of developed IFoF-based DAS for 5G trial services.

Fig. 3. Block diagram of distributed radio unit in IFoF based DAS.

Fig. 2. Block diagram of main hub unit in IFoF-based DAS.

and UL#2: 1530 nm, and two downlinks; DL#1: 1550 nm and
DL#2: 1570 nm) to support the 2 × 2 MIMO configuration.
hub unit (MHU), eight distributed radio units (DRU), and four In addition to mobile data transmission interval, auxiliary
mmWave radio units (mRU). links are required for transmitting reference clock and TDD-
As shown in Fig. 1, 5G BBU is directly connected to the MHU synchronization signal (T-sync). The MHU has a 10-MHz ref-
providing 5G signal, reference clock, and time division duplex erence clock and a T-sync from the 5G BBU. It then digitizes
(TDD) synchronization signal. The MHU has eight slots, each the signal along with the control and management (C&M) sig-
of which is connected to the DRUs via optical fibers. Each DRU nal using MHU digital unit. The digitized signals on the MHU
can accommodate four mRUs, and the DRU and mRU links are digital board are multiplexed using coarse wavelength division
configured with a few meters of coaxial cable. (de)multiplexer (CWDM) of the MAT along with the analog op-
In the proposed system, the MHU sends and receives 5G sig- tical signals that carry 5G mobile data. The MHU digital unit
nals in an IF domain in the range of 1.6–2.4 GHz as presented employs an O-band of 1270 nm and 1330 nm for uplink and
in Fig. 2. In the MHU IF unit, a switching functionality is also downlink transmissions, respectively.
provided to support dynamic TDD. RF amplification and atten- Using an analog optical link between MHU slots and DRUs,
uation of the received IF signal are performed in the MHU IF the transmitted 5G signals are detected by the PIN-PD of the
unit for the downlink transmission path after removing out-of- DRU analog optical transceiver (DAT), after being further am-
band signals using a band pass filter (BPF). The RF attenuator is plified by the two-stage RF amplifier to compensate for the link
located between the RF amplifiers to prevent saturation effects. loss due to optical transmission, as illustrated in Fig. 3. In the
Then, the MHU analog optical transceiver (MAT) in the MHU DRU IF unit, the IF signal carrying the mobile data is multi-
slots directly converts the electrical IF signal into an optical sig- plexed with the reference clock, T-sync, and C&M signals that
nal. Uplink transmission is carried out similar to the downlink are detected from the DRU digital unit. Then, the multiplexed
transmission; however, additional RF amplification and attenu- signal is distributed to four mRUs over a coaxial cable. In up-
ation functions are necessary in the MAT to compensate the loss link transmission, 5G signal from the mRUs is delivered to DAT,
of the analog optical link. Based on the up/downlinks configu- which then converts the 5G signal optical signal. The converted
ration described, each MAT contains a pair of transmitters and optical signal is then transmitted to the MHU through optical
receivers, which form four links (two uplinks; UL#1: 1510 nm fibers.
SUNG et al.: 5G TRIAL SERVICES DEMONSTRATION: IFOF-BASED DISTRIBUTED ANTENNA SYSTEM 3595

the mmWave band. For downlink transmission, the generated 5G

analog waveform in IF band is input into the MHU with an input
power of −10 dBm, which is equal to the output power of the
BBU used in the 5G trial service. In the MHU, an analog optical
waveform carrying 5G mobile data is generated by MAT which
is biased at 36 mA and has an output optical power of +6 dBm.
The generated optical signal is multiplexed with the digital op-
tical waveform that transmits reference clock, TDD, and C&M.
Subsequently, the multiplexed optical waveform is transmitted
to DRU over a 5-km span of single mode fiber (SMF). The
wavelengths of 1550 nm and 1570 nm are assigned to DL#1
and DL#2. To investigate the performance of analog optical link
with received optical power, we use a variable optical attenua-
Fig. 4. Block diagram of mmWave radio unit in IFoF based DAS.
tor (VOA) connected the MHU and DRU. The received optical
signal is detected to an IF signal in DAT consisting of PIN-PD
with a bandwidth of 6 GHz. The IF signal detected in the DAT
The mRU extracts 10 MHz clock, TDD-sync and C&M sig- is split into four paths, and further distributed to the mRU over
nal from distributed signal. Among them, the 10-MHz reference the coaxial cable along with 10-MHz clock, T-sync, and C&M
clock is retimed, reshaped, and re-amplified in a phase-locked signal. The 5G signal power applied to the mRU is approxi-
loop (PLL). And then the reference clock is frequency multiplied mately −10 dBm. The distributed IF signal is then up-converted
to desired mmWave frequency band, being further distributed to mmWave-band in 28 GHz, after being sent to the antenna. The
frequency mixers. As in Fig. 4, the mmWave unit converts re- uplink transmission is similar to the downlink transmission; it
ceived IF signal to mmWave frequency band using reference is performed at wavelengths of 1510 nm and 1530 nm and the
clock. The up-converted signals on a 28-GHz mmWave are ra- input power of mRU, which is set to be –50 dBm.
diated by the antenna after being amplified by a high-power
amplifier (HPA). In case of uplink transmission, the 5G signal
received at the antenna is amplified by a low noise amplifier, and B. EVM Performance of the IFoF-Based DAS
then down-converted into the IF domain. The down-converted In Figs. 6(a) to (d), we evaluated the transmission performance
5G signal is sent to the DRU along with the C&M signal that and measured RF spectra of 5G signal, which is composed of
contains the mRU status information. eight FAs, as each FA has a 100-MHz bandwidth. The frequency
To support commercial service, the TDD functions should be of the IF band was in the range of 1.6–2.4 GHz, and that of the
synchronized on MHU, DRU, and mRU. Otherwise, the TDD mmWave band was in the range of 27.5–28.3 GHz. First, we di-
timing is misaligned between devices. The TDD switching func- rectly inserted 5G signal from VSG into vector signal analyzer
tion synchronously operates at each position of the MHU, DRU, (VSA) to evaluated EVM performance of 5G source in Fig. 6(a).
and mRU. To compensate for the time delay owing to the dis- The figure shows that 5G source has EVMs of 0.7% to 1.1%.
tance between MHU, DRU, and mRU, the DAS provides a delay In Fig. 6(b), we also performed EVM performance measure-
measurement function. The measured delay information is pro- ments on the DRU output after SMF of 5-km to investigate the
vided to the BBU for the timing advance function. Further, the effect of optical link. It is well known that the analogue trans-
output powers of mRU’s antenna for downlink and input power mission experiences severe performance degradation owing to
of BBU from the MHU for the uplink should be constant. It is second-order intermodulation distortions caused by the interac-
also important to prevent damage to the system by high inci- tion between fiber chromatic dispersion and frequency chirp of
dent RF power. Therefore, the RF gain is constantly controlled DML [40], [41]. To avoid this, we set the center frequency of the
through the MAT and DAT as well as the MHU and DRU IF first FA to be greater than the total bandwidth occupied by the
units. 5G signal as the octave-confined frequency plan as in [34]. The
second-order distortion components of the received signal tend
III. EXPERIMENTAL SETUP AND RESULTS to further degrade the EVM performance in the high frequency
band, which is not shown in Fig. 6(b). In other word, the second-
A. Experimental Setup
order intermodulation falls in the out-of-signal band without in-
To investigate the transmission performance of the developed terference; in this band the performance degradation would be
IFoF-based DAS, we performed a series of experiments on the negligible. On the other hand, the RF power difference between
system described in Fig. 5. We modulated 5G signals consist- FAs #0 to #7 appears approximately 8 dB due to gain flatness in
ing of eight frequency allocation (FA) sub-bands, where each frequency response of MHU-DRU. A chain consisting of multi-
FA had a bandwidth of 100 MHz, based on Verizon 5G new ple RF devices, such as amplifier, filter, and attenuator in IF units
radio specification [39]. A vector signal generator (VSG) con- and optical module such as laser diodes and PIN-PD, makes the
verts the modulated 5G baseband signal to analog waveforms. frequency response non-flat. Due to that, EVM difference be-
The frequencies of the analog waveforms range from 1.6 GHz tween FAs in the IF band also occurs. An EVM degradation of
to 2.4 GHz for the IF band and from 27.5 GHz to 28.3 GHz for about 1% is additionally generated in FA #0 as compared with
3596 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 37, NO. 14, JULY 15, 2019

Fig. 5. Lab experimental setup.

Fig. 6. Measured spectra and EVM performances of IFoF-based DAS at received optical power of 0 dBm (a) Electrical 5G signal in IF band (b) after 5-km
transmission in IF band (c) Electrical 5G signal in 28 GHz mmWave (d) after 5-km transmission in 28 GHz mmWave.

FA #7. Nevertheless, we can confirm that the EVM degradation between FAs is flat relative to the output of the DRU. Because
less than 1.5% occurs in the IF band after 5-km transmission. The it is important to ensure the flatness of the frequency response
mRU of DAS radiates 5G signal having RF power of 20 dBm of each FA for reliable 5G commercial service, we designed
in the 28-GHz mmWave band. Therefore, the frequency con- the DAS to have a flat frequency response, in which the RF gain
version onto mmWave band and RF amplification (30 dB gain) deviation between FAs of 5G signal is less than ±3 dB. It is also
are performed in the mRU. When the 5G signal of IF domain is observed that the EVMs of 5G signal in 28 GHz mmWave band
converted to mmWave frequency band, the EVM performance are at less than 9%. On the other hand, the EVM degradation
would be degraded due to phase noise of reference clock. Fur- of the analogue optical link in the mmWave-band was measured
thermore, because the required output power at the antenna of within 1%, which is relatively low compared with that caused by
mRU is 20 dBm, the signal amplification would be operated in signal amplification and mmWave conversion at mRU. There-
saturation region of RF amplifier where a clipping phenomenon fore, the performance degradation due to frequency conversion
occurs. Therefore, the clipping distortions cause intermodula- and amplification in the mRU is more dominant than the perfor-
tion products, which result in inter-carrier interference, a high mance degradation in optical links. In addition to EVM, adjacent
out-of-band harmonic distortion power, and the degradation of channel power leakage ratio (ACLR) is also an important fig-
the performance of EVM. In Fig. 6(c), the 5G source of the ure of merit in RoF system [42], [44]. We obtained ACLR of
IF domain was directly input to mRU, and EVM measurement −32 dBc in 28 GHz mmWave band after 5-km transmission.
was performed to investigate the impact of frequency conver- We also investigated the downlink transmission performance
sion and RF amplification in mRU. Compared with the EVM as a function of the received optical power for the optical back-
performance of DRU output as presented in Fig. 6(a), the EVM to-back measurement chain and 5-km transmission for DL#1.
of the 28 GHz mmWave 5G signal clearly exhibits performance The results are presented in Figs. 7(a) and (b). Even if the re-
degradation of 4 ∼ 6% EVM, since RF amplification and fre- ceived optical power changes, the DAS controls the input power
quency conversion in the mRU significantly distort signal qual- of 5G signal to the mRU to be constant using an automatic gain
ity. In Fig. 6(d), when the DAS modules (MHU-DRU-mRU) are control function. In this experiment and field demonstration, the
fully connected, the RF spectra shows that frequency response input power of mRU having fixed 30 dB gain was maintained
SUNG et al.: 5G TRIAL SERVICES DEMONSTRATION: IFOF-BASED DISTRIBUTED ANTENNA SYSTEM 3597

Fig. 7. Measured performances and constellations of IFoF-based DAS in IF and mmWave bands for downlink (a) EVMs in optical BTB (b) EVMs after 5-km
SMF, (c)–(f) constellations of FA#7: IF signal (c) in optical BTB (d) after 5-km SMF, 28 GHz mmWave (e) in optical BTB (f) after 5-km SMF.

at −10 dBm because output power of mRU antenna should be

kept constant (+20 dBm) to provide a reliable mobile service.
In addition to mRU input, the input power of MAT is also set
to be −10 dBm. Therefore, an optical modulation index per
channel is calculated as about 0.07∼0.08. Here, dominant per-
formance limitations arise from nonlinear distortions as 3rd in-
termodulation distortion and clipping, rather than linear noise
components such as shot noise, relative intensity noise (RIN)
and receiver noise. [45], [46]. Similar to Fig. 6(b), FA#0 in the
IF band exhibits the worst EVM performance owing to the non-
flat frequency response of the analog optical link in both opti-
cal back-to-back and 5-km transmission. However, at 28 GHz,
FA#0 shows better EVM performance than FA#7. Furthermore,
EVM degradation in the mmWave band is negligible even if the
received optical power is reduced. This is because the perfor-
mance degradation caused by the RF amplification function and
frequency conversion in mRU is more dominant than the degra-
dation caused by the optical link. It should be noted that there is
little difference in the EVM performance between the 5-km
transmission and the optical back-to-back condition. The disper-
sion induced power fading could be ignored due to low frequency
of 5G signal in IF band (1.6 GHz to 2.4 GHz). Furthermore, be-
cause we utilize the frequency plan of octave-mapping, the over-
all transmission performance degradation caused by second-
order intermodulation distortion is also negligible in which the
distortion components did not interfere with the desired 5G sig-
nal. Figs. 7(c)–(f) show the constellations of seventh FA for IF
and mmWave bands. As in Figs. 7(c) and (d), the constellations
of 5G signal received in IF band are relatively clear. That is, the Fig. 8. Measured performances of IFoF-based DAS in IF and mmWave bands
effect of the optical BTB condition and the 5 km transmission for uplink (a) EVMs in optical BTB (b) EVMs after 5-km SMF.
are negligible; however, it can be confirmed that the constella-
tions spread after the frequency conversion onto mmWave band
as shown in Figs. 7(e) and (f). power of the uplink signal is considerably low, signal amplifi-
Fig. 8(a) and (b) show that the measured EVM performances cation in the mRU is essential, and in this experiment, the mRU
of the 5G signal for the UL#1 path in optical back-to-back con- provides a gain of 50 dB. To support high gain above 50 dB, the
dition and after SMF of 5 km. The input 5G signal of mRU for mRU uses two cascaded RF amplifiers, which inevitably wors-
the uplink has RF power of −45 dBm and ranges from 27.5– ens the signal-to-noise ratio of the uplink signal. In the optical
28.3 GHz. In Fig. 8, a clear degradation of the EVM performance back-to-back configuration, we can observe that the EVMs be-
of approximately 5% can be observed when the 5G signal was low 7% are measured at the received optical power of −6 dBm,
down-converted and amplified (DRU input). Because the input which refers EVM degradation of 1.5% occurs in MAT and DAT.
3598 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 37, NO. 14, JULY 15, 2019

Fig. 9. Demonstration of 5G trial service: (a) Field trial configuration (b) IFoF-based DAS deployment in Pyeongchang.

After the 5-km SMF, the additional performance degradation

is negligible compared with optical back-to-back condition be-
cause the 5G signal of the uplink transmission is located in the
frequency range of 1.6–2.4 GHz, which is identical to that of
downlink transmission without the second-order intermodula-
tion distortion. The experimental results for both downlink and
uplink transmissions show that the maximum achievable EVMs
are estimated to be less than 9%, while satisfying the minimum
requirement of the 5G signal with 64-quadrature amplitude mod-
ulation (QAM) defined by the technical specification of the 3rd
generation partnership project (3GPP) [47].

Fig. 10. Captured TDD synchronization signal provided by BBU and DAS.
C. Demonstration Setup of 5G Trial Service Using
IFoF-Based DAS in the Pyeongchang Winter Olympics
Fig. 10 shows the alignment status of the T-sync provided by
The 5G trial service using IFoF-based DAS in a 28 GHz
the BBU and DAS to ensure that dynamic TDD is operated prop-
mmWave band was provided in the Pyeongchang area shown in
erly. The T-sync of the DAS is well matched to the T-sync of the
Fig. 9. The 5G connected car section comprised multiple BBU
BBU, and it is confirmed that the 5G trial service is appropriately
racks, installed at 150–200 m intervals. In the BBU racks, radio
provided according to the DAS T-sync provided.
frequency unit (RFU) radiates 5G signal in 28 GHz mmWaves to
In Fig. 11(a), investigation results of the downlink through-
provide the 5G trial service. To increase network coverage and
put and RSRP of the 5G handset over the measured time by
support the 5G trial service, we deployed IFoF-based DAS be-
moving 5G handset from the Olympic stadium to the 5G con-
tween 5G connected car sections and the Olympic stadium. The
nected car section through the DAS deployed section are shown.
MHU was installed inside the BBU rack and directly connected
It is observed that the throughput decreases dramatically in the
to 5G BBU and RFU. The frequency conversion function of the
“C” and “E” areas. Because these areas are shaded regions be-
RFU was removed to receive the IF band signal. Two DRUs
tween mRUs or mRU and RFU, the RSRP was also measured
were installed on the road of the DAS section and connected to
poorly. The measurement results in the DAS section reveal that
MHU via optical fibers. On each DRU, a pair of mRUs is built
the achieved peak data-rate of 5G handset is up to 1 Gb/s. Be-
at separate poles (8 m) through the RF cable.
cause the 5G handset shows measurement results of one of the
two MIMO paths, we infer that the peak data-rate achieved in
the DAS section is approximately 2 Gb/s. In addition, the base
D. 5G Trial Service Demonstration Using IFoF-Based DAS station limited the modulation coding scheme (MCS) of the 5G
To verify the technical feasibility of IFoF-based DAS in 5G signal during the trial service to provide a reliable autonomous
trial services, we moved the 5G handset between the DAS de- driving service and 5G service. As a result, a theoretical max-
ployed area and the 5G connected car section in which multiple imum downlink speed of 2.5 Gb/s per path should have been
BBU are deployed. The movement speed of the 5G handset was achieved, but due to the limited MCS, the throughput provided
maintained at the 0 ∼ 40 km level. by the BBU per path is ∼1.1 Gb/s. In this demonstration, we
SUNG et al.: 5G TRIAL SERVICES DEMONSTRATION: IFOF-BASED DISTRIBUTED ANTENNA SYSTEM 3599

Fig. 11. Measured throughput and RSRP: (a) downlink (b) uplink.

have confirmed that the downlink throughput of the DAS sec- in Pyeongchang. A downlink throughput of 1 Gb/s and an up-
tion achieved over 98% of the downlink throughput of the 5G link throughput of 200 Mb/s for 5G trial services were achieved
BBU section. in the DAS deployment area where the 5G handset moved at a
We also performed an uplink test while moving in the oppo- speed of 40 km/h. The 5G-based autonomous driving services
site direction of the downlink measurement, the results of which were also successfully provided through a seamless handover
are illustrated in Fig. 11(b). Because the 5G handset only pro- between the DAS section and the 5G connected car section dur-
vides the RSRP of the signal received, we used downlink RSRP ing the Winter Olympics. We believe that our study will make a
with the same wireless channel to determine the conditions of significant contribution to the deployment mmWave-based 5G
the channel. Similar results to that of the downlink are seen; network infrastructure on a nationwide scale.
the uplink throughputs are reduced in the “C” and “E” areas.
We confirmed that the maximum uplink data rate measured in REFERENCES
DAS sections is approximately 200 Mb/s, and a seamless 5G [1] 5G Forum, “5G vision, requirements, and enabling technologies,” V.1.0,
trial service was provided in the DAS deployed area. Feb. 2015.
Using IFoF-based DAS, we also demonstrated the perfor- [2] P. Marsch et al., “5G radio access network architecture: Design guidelines
and key considerations,” IEEE Commun. Mag., vol. 54, no. 11, pp. 24–32,
mance of the 5G connected car service, one of the 5G key Nov. 2016.
services, during the Pyeongchang Winter Olympics. The 5G [3] D. Pliatsios, P. Sarigiannidis, S. Goudos, and G. K. Karagiannidis, “Realiz-
handset, an autonomous vehicle, featured its autonomous driv- ing 5G vision through cloud RAN: Technologies, challenges, and trends,”
EURASIP J. Wireless Comm., vol. 136, Dec. 2018.
ing service in the DAS section. A seamless handover was suc- [4] P. Chanclou, L. A. Neto, K. Grzybowski, Z. Tayq, F. Saliou, and N. Genay,
cessfully established at the intersection of the DAS and the 5G “Mobile fronthaul architecture and technologies: A RAN equipment as-
connected car sections. sessment,” J. Opt. Commun. Netw., vol. 10, pp. A1–A7, 2018.
[5] A. Checko et al., “Cloud RAN for mobile networks—A technology
overview,” IEEE Commun. Surv. Tut., vol. 17, no. 1, pp. 405–426, Mar.
2015.
IV. CONCLUSION [6] CPRI Consortium, “CPRI specifcation V7.0 common public radio inter-
face (CPRI); interface specification,” 2015.
In this study, we present an IFoF-based DAS that supports [7] S. H. Cho, H. Park, H. S. Chung, K. H. Doo, S. Lee, and J. H. Lee, “Cost-
5G mobile broadband services to expand 28 GHz mmWave- effective next generation mobile fronthaul architecture with multi-IF car-
based 5G wireless coverage. The IFoF-based DAS provides rier transmission scheme,” in Proc. Opt. Fiber Commun. Conf. Exhib., San
Francisco, CA, USA, 2014, Paper Tu2B.6.
2 × 2 MIMO configuration with 1 × 8 × 32 network topol- [8] L. M. P. Larsen, A. Checko, and H. L. Christiansen, “A survey of the func-
ogy. The experimental results showed that the EVM degradation tional splits proposed for 5G mobile crosshaul networks,” IEEE Commun.
caused by IFoF-based analog optical links were evaluated within Surv. Tut., vol. 21, no. 1, pp. 146–172, First Quarter 2019.
[9] K. Miyamoto, S. Kuwano, T. Shimizu, J. Terada, and A. Otaka, “Perfor-
1.5% in the IF band. Furthermore, the 5G signal on IFoF-based mance evaluation of ethernet-based mobile fronthaul and wireless CoMP
DAS link achieved an EVM performance of less than 9% after in split-PHY processing,” J. Opt. Commun. Netw., vol. 9, no. 1, pp. A46–
5-km SMF for both the uplink and downlink transmissions. We A54, Jan. 2017.
[10] Y. Alfadhli, M. Xu, S. Liu, F. Lu, P. Peng, and G. Chang, “Real-time
also reported on 5G trial services of an IFoF-based DAS in a demonstration of adaptive functional split in 5G flexible mobile fronthaul
28 GHz mmWave supporting Giga-bit mobile services that was networks,” in Proc. Opt. Fiber Commun. Conf. Exhib., San Diego, CA,
demonstrated during the Pyeongchang Winter Olympics. The USA, 2018, Paper Th2A–48.
[11] S. Ishimura, A. Bekkali, K. Tanaka, K. Nishimura, and M. Suzuki, “1.032-
IFoF-based DAS was deployed between the Olympic stadium Tb/s CPRI-equivalent rate IF-over-fiber transmission using a parallel
and the 5G connected car section where the BBUs were installed, IM/PM transmitter for high-capacity mobile fronthaul links,” J. Lightw.
thereby increasing the network coverage of the 5G trial service Technol., vol. 36, no. 8, pp. 1478–1484, Apr. 2018.
3600 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 37, NO. 14, JULY 15, 2019

[12] P. T. Dat, A. Kanno, N. Yamamoto, and T. Kawanishi, “Full-duplex trans- [34] M. Sung, S.-H. Cho, H. S. Chung, S. M. Kim, and J. H. Lee, “Investigation
mission of LTE-A carrier aggregation signal over a bidirectional seamless of transmission performance in multi-IFoF based mobile fronthaul with
fiber-millimeter-wave system,” J. Lightw. Technol., vol. 34, no. 2, pp. 691– dispersion-induced intermixing noise mitigation,” Opt. Express, vol. 25,
700, Jan. 2016. no. 8, pp. 9346–9357, Apr. 2017.
[13] B. G. Kim, S. H. Bae, H. Kim, and Y. C. Chung, “RoF-based mobile [35] J. Kim, M. Sung, E. S. Kim, S. H. Cho, and J. H. Lee, “Experimental
fronthaul networks implemented by using DML and EML for 5G wireless demonstration of analog IFoF-based seamless fiber-wireless interface for
communication systems,” J. Lightw. Technol., vol. 36, no. 14, pp. 2874– 5G indoor DAS supporting 8 FA and 2 × 2 MIMO configuration,” in
2881, Jul. 2018. Proc. Opt. Fiber Commun. Conf. Exhib., San Diego, CA, USA, 2018,
[14] R. Zhang et al., “An ultra-reliable MMW/FSO A-RoF system based on Paper W4B–4.
coordinated mapping and combining technique for 5G and beyond mobile [36] M. Sung, J. Kim, S. H. Cho, H. S. Chung, J. K. Lee, and J. H. Lee, “Exper-
fronthaul,” J. Lightw. Technol., vol. 36, no. 20, pp. 4952–4959, Oct. 2018. imental demonstration of bandwidth-efficient indoor distributed antenna
[15] I. Chih-Lin, H. Li, J. Korhonen, J. Huang, and L. Han, “RAN revolution system based on IFoF technology supporting 4G LTE-A and 5G mobile
with NGFI (xHaul) for 5G,” J. Lightw. Technol., vol. 36, no. 2, pp. 541–550, services,” in Proc. Opt. Fiber Commun. Conf., San Diego, CA, USA, 2018,
Jan. 2018. Paper Tu3J.3.
[16] ITU-T G-series Recommendations - Supplement 55, “Radio-over-fiber [37] M. Sung, S. H. Cho, J. Kim, J. K. Lee, J. H. Lee, and H. S. Chung,
(RoF) technologies and their applications,” Jul. 2015. “Demonstration of IFoF-based mobile fronthaul in 5G prototype with
[17] N. Argyris et al., “DSP enabled fiber-wireless IFoF/mmWave link for 5G 28-GHz millimeter wave,” J. Lightw. Technol., vol. 36, no. 2, pp. 601–
analog mobile fronthaul,” in Proc. IEEE 5G World Forum, Santa Clara, 609, Jan. 2018.
CA, USA, 2018, pp. 482–487. [38] M. Sung et al., “Demonstration of 5G trial service in 28 GHz millimeter
[18] H. Kim, “RoF-based optical fronthaul technology for 5G and beyond,” wave using IFoF-based analog distributed antenna system,” in Proc. Opt.
in Proc. Opt. Fiber Commun. Conf. Exhib., San Diego, CA, USA, 2018, Fiber Commun. Conf. Exhib., San Diego, CA, USA, 2019, Paper Th1F.2.
PaperTu3J.1. [39] “Verizon 5th generation radio access; physical channels and modulation
[19] C. Han, S. Cho, M. Sung, H. Chung, and J. Lee, “Clipping distortion sup- (release 1),” TS V5G.211v1.0’(2016-10), Verizon 5G TF, Oct. 2016.
pression of directly modulated multiIF-over-fiber mobile fronthaul links [40] E. E. Bergmann, C. Y. Kuo, and S. Y. Huang, “Dispersion-induced com-
using shunt diode predistorter,” ETRI J., vol. 38, no. 2, pp. 227–234, Apr. posite second-order distortion at 1.5 µm,” IEEE Photon. Technol. Lett.,
2016. vol. 3, no. 1, pp. 59–61, Jan. 1991.
[20] M. Xu, F. Lu, J. Wang, L. Cheng, D. Guidotti, and G.-K. Chang, “Key [41] N. S. Andŕe, H. Louchet, K. Habel, and A. Richter, “Analytical formulation
technologies for next-generation digital RoF mobile fronthaul with sta- for SNR prediction in IMDD OFDM-based access systems,” IEEE Photon.
tistical data compression and multiband modulation,” J. Lightw. Technol., Technol. Lett., vol. 26, no. 12, pp. 1255–1258, Jun. 2014.
vol. 35, no. 17, pp. 3671–3679, Sep. 2017. [42] R. Schuh and D. Wake, “Distortion of W-CDMA signals over optical fiber
[21] L. Zhang et al., “Spectrally efficient digitized radio-over-fiber system links,” in Proc. Int. Top. Meeting Microw. Photon. Tech. Dig., Melbourne,
with k-means clustering-based multidimensional quantization,” Opt. Lett., Australia, 1999, pp. 9–12.
vol. 43, no. 7, pp. 1546–1549, Apr. 2018. [43] D. Visani, G. Tartarini, L. Tarlazzi, and P. Faccin, “Transmission of UMTS
[22] D. Hisano et al., “TDM-PON for accommodating TDD-based fronthaul and WIMAX signals over cost-effective radio over fiber systems,” IEEE
and secondary services,” J. Lightw. Technol., vol. 35, no. 14, pp. 2788– Microw. Wireless Compon. Lett., vol. 19, no. 12, pp. 831–833, Dec. 2009.
2796, Jul. 2017. [44] D. Wake and R. E. Schuh, “Measurement and simulation of W-CDMA
[23] H. S. Chung, “Low latency PON and RoF for 5G wireless systems,” in signal transmission over optical fiber,” Electron. Lett., vol. 36, no. 10,
Proc. IEEE Photon. Conf., Reston, VA, USA, 2018, pp. 1–2. pp. 901–902, May. 2000.
[24] T. S. Rappaport, Y. Xing, G. R. MacCartney Jr., A. F. Molisch, E. Mellios, [45] M. Sung, H. S. Chung, S. Cho, J. Kim, J. K. Lee, and J. H. Lee, “Impact
and J. Zhang, “Overview of millimeter wave communications for fifth- of second-order intermodulation distortion on analogue optical link to
generation (5G) wireless networks—With a focus on propagation models,” bandwidth efficient multi-IF-over-fibre based mobile fronthaul,” in Proc.
IEEE Trans. Antennas Propag., vol. 65, no. 12, pp. 6213–6230, Dec. 2017. Eur. Conf. Opt. Commun., Gothenburg, Sweden, 2017, pp. 1–3.
[25] T. S. Rappaport, R. W. Heath Jr, R. C. Daniels, and J. N. Murdock, Millime- [46] E. S. Kim, J. Kim, M. Sung, S. H. Cho, J. H. Lee, and H. S. Chung, “Power
ter Wave Wireless Communications. Englewood Cliffs, NJ, USA: Prentice- budget improvement of multi-IFoF based analog indoor-distributed an-
Hall, Sep. 2014. tenna system (DAS) with laser over-modulation,” Opt. Commun., vol. 435,
[26] M. J. Crisp, S. Li, A. Wonfor, R. V. Penty, and I. H. White, “Demonstration no. 15, pp. 129–133, Mar. 2019
of radio over fibre distributed antenna network for combined in-building [47] “LTE; evolved universal terrestrial radio access (E-UTRA); base station
WLAN and 3G coverage,” in Proc. Conf. Opt. Fiber Commun. Nat. Fiber (bs) conformance testing,” Tech. Specification Group Radio Access Net-
Opt. Eng. Conf., Anaheim, CA, USA, 2007, Paper JThA81. work, 3GPP TS 36.141 v. 10.1.0, Rel. 10, Jan. 2011.
[27] A. E. Aighobahi and N. J. Gomes, “Capacity and error performance ver-
ification of multi-antenna schemes in radio-over-fiber distributed antenna
system,” J. Lightw. Technol., vol. 34, no. 20, pp. 4779–4785, Oct. 2016.
[28] Y. Yang, M. J. Crisp, R. V. Penty, and I. H. White, “Low-cost MIMO radio
over fiber system for multiservice DAS using double sideband frequency
translation,” J. Lightw. Technol., vol. 34, no. 16, pp. 3818–3824, Aug.
2016. Minkyu Sung received the B.S. degree in computer and communication engi-
[29] U. Gliese, S. Norskov, and T. N. Nielsen, “Chromatic dispersion in fiber- neering and the Ph.D. degree in computer and radio communication engineering
optic microwave and millimeter-wave links,” IEEE Trans. Microw. Theory from Korea University, Seoul, South Korea, in 2010 and 2015, respectively. Since
Techn., vol. 44, no. 10, pp. 1716–1724, Oct. 1996. 2015, he has been with Electronics and Telecommunication Research Institute
[30] S. H. Cho, C. Han, H. S. Chung, and J. H. Lee, “Demonstration of mobile (ETRI), Daejeon, South Korea, where he is currently a Senior Researcher. His
fronthaul test bed based on RoF technology supporting two frequency current research interests include optical OFDM, digital signal processing, mo-
assignments and 2 × 2 MIMO antennas,” ETRI J., vol. 37, no. 6, pp. 1055– bile fronthaul, indoor distributed antenna system, and radio-over-fiber.
1064, Dec. 2015.
[31] E. A. Bekkali, S. Ishimura, K. Tanaka, K. Nishimura, and M. Suzuki, “High
capacity mobile fronthaul using DP-MZM-based IF-over-fiber system with
1-Tbit/s CPRI-equivalent data rate,” in Proc. IEEE Int. Conf. Commun.,
2018, pp. 1–6.
[32] B. G. Kim, S. H. Bae, H. Kim, and Y. C. Chung, “DSP-based CSO can- Joonyoung Kim received the Ph.D. degree in electrical engineering from
cellation technique for RoF transmission system implemented by using the Korea Advanced Institute of Science and Technology, Daejeon, South
directly modulated laser,” Opt. Express, vol. 25, no. 11, pp. 12152–12160, Korea, in 2013. He was with the Optoelectronics Research Centre (ORC), Uni-
May 2017. versity of Southampton as a Research Fellow from 2013 to 2017. He joined
[33] C. Han, M. Sung, S. H. Cho, H. S. Chung, S. M. Kim, and J. H. Lee, the Electronics and Telecommunication Research Institute, Daejeon, South Ko-
“Performance improvement of multi-IFoF-based mobile fronthaul using rea, as a Senior Researcher in March 2017. His current research interests include
dispersion-induced distortion mitigation with IF optimization,” J. Lightw. amongst others: optical wireless convergence system, microwave photonics, and
Technol., vol. 34, no. 20, pp. 4772–4778, Oct. 2016. novel photonic integration technologies.
SUNG et al.: 5G TRIAL SERVICES DEMONSTRATION: IFOF-BASED DISTRIBUTED ANTENNA SYSTEM 3601

Eon-Sang Kim received the B.S. degree in electronic and information engi- Sung-Yeop Pyun received the M.S. and Ph.D. degrees in electrical engineering
neering from the Seoul National University of Science and Technology, Seoul, from the Korea Advanced Institute of Science and Technology, Daejeon, South
South Korea, in 2010 and the Ph.D. degree in information and communication Korea, in 2008 and 2012, respectively. Since 2012, he has been with the Korea
engineering from the University of Science and Technology, Daejeon, South Telecom, South Korea, where he is currently a Principal Research Engineer. His
Korea, in 2019. Since 2019, he has been with Electronics and Telecommuni- current research interests include 5G access network, 5G fronthaul network, and
cation Research Institute (ETRI), Daejeon, South Korea, as a Researcher. His 5G outdoor/indoor solutions.
current research interests include next-generation optical access network, mo-
bile fronthaul, indoor distributed antenna system, and radio-over-fiber.

Seung-Hyun Cho received the B.S. and M.S. degrees in electronic materials en-
gineering from Kwangwoon University, Seoul, South Korea, in 1997 and 1999,
respectively. He received the Ph.D. degree in materials science and engineering
from Hanyang University, Seoul, South Korea, in 2010. From 1999 to 2000, he
was with the Access Network Laboratory of Korea Telecom, Daejeon, South Joon Ki Lee received the M.S. degree in applied physics from the Gwangju
Korea. Since 2000, he has been with the Electronics and Telecommunication Institute of Science and Technology (GIST), Gwangju, South Korea, in 1997 and
Research Institute, Daejeon, where he is currently a Project Leader for THz the Ph.D. degree in electronics engineering from Chungnam National University,
short distance transmission technology based on photonics. His current research Daejeon, South Korea, in 2016. From 1997 to 2001, he was an engineer with
interests include next-generation optical access network, mobile fronthaul and Hanwha Information and Telecommunications Corporation. Since 2001, he has
DAS for 5G networks, and THz wireless delivery system. He is also a Chair of been with Electronics and Telecommunications Research Institute (ETRI), South
IEC TC103 (Transmitting equipment for radio communication) and Co-Editor Korea, where he is currently the Managing Director of Optical Networking
of G.9803 (Radio over Fiber Systems) in ITU-T SG15/Q2. Research Group. His research interests include high speed optical transmission
systems, optical transceivers, data center network, and optical-mobile converged
access network.

Young-Jun Won received the Master’s degree in IT convergence engineering

from Ajou University, Suwon, South Korea, in 2014. Since August 2000, he
has been with FRTek, Gyeonggi-do, South Korea, as a Senior Researcher. His
current research 5G DAS based on RoF technology, 5G DAS and 5G RF repeater
based on digital technology.

Jong Hyun Lee received the B.S., M.S. and Ph.D. degrees in electronics en-
gineering from Sungkyunkwan University, Suwon City, South Korea, in 1981,
1983, and 1993, respectively. Since 1983, he has been with the Electronics and
Byoung-Chul Lim received the Master’s degree in IT convergence engineer- Telecommunication Research Institute, Daejeon, South Korea, where he was the
ing from Ajou University, Suwon, South Korea, in 2014. From June 2000 to Director of Optical Communication Department as well as Research Strategy
November 2016, he was with NextLink as a Senior Researcher. Since October and Planning Department. He was also an Executive Director of Optical Inter-
2000, he has been with FRTek, Gyeonggi-do, South Korea, as an R&D Center net Research Department. His research interests include packet-circuit-optical
Director. His current interests include 5G DAS based on RoF technology, 5G converged switching systems, green internet data center, and optical access
DAS, and 5G RF repeater based on digital technology. networks.