RF Front End for Mobile 2023 射频前端报告
RF Front End for Mobile 2023 射频前端报告
RF Front-End for
Mobile 2023
Market and Technology Report
AR Augmented Reality GNSS Global Navigation Satellite Service NR-U New Radio – Unlicensed
ASP Average Selling Price GSM Global System for Mobile communications NSA Non-Stand-Alone
AWS Advanced Wireless Service GSMA GSM Association OEM Original Equipment Manufacturer
BAW Bulk Acoustic Wave HBT Heterojunction Bipolar Transistor PAE Power-Added Efficiency
BPSK Binary Phase-Shift Keying HSPA High-Speed Packet Access PAD Power Amplifier with Duplexer
BRS Broadband Radio Service IDT Inter Digital Transducer PAM Power Amplifier Module
BT Bluetooth IHP SAW Incredible High-Performance Surface Acoustic Wave PAPR Peak to Average Power Ratio
CA Carrier Aggregation IIP3 Input Interception Point 3 PCS Personal Communications Service
CAGR Compound Annual Growth Rate IMD Intermodulation Distortion PHEMT Pseudomorphic High-Electron-Mobility Transistor
CBRS Citizen Broadband Radio Service InGaP Indium Gallium Phosphide POI Piezo-On-Insulator
CMOS Complementary Metal Oxide Semiconductor IPD Integrated Passive Device QAM Quadrature Amplitude Modulation
CP-OFDM Cyclic Prefix – Orthogonal Frequency Division Multiplexing LAA Licensed Assisted Access RCS RICH Content Service
Drx FEM Diversity Receive Front-End Module LTE Long-Term Evolution SMR Solid Mounted Resonator
EBS Educational Broadband Service LWA LTE - WLAN Aggregation SOI Silicon-On-Insulator
EDGE Enhanced Data GSM Environment M2M Machine-to-Machine SUL Supplementary Uplink
EVM Error Vector Magnitude MB Mid Band TAM Total Available Market
FBAR Film Bulk Acoustic Resonator Mbps Megabits Per Second TD-SCDMA Time Division Synchronous Code Division Multiple Access
FBW Fractional Bandwidth MIC Ministry of Internal Affairs and Communications (Japan) UHB Ultra-High Band
4
RF Front-End for Mobile 2023 | Report | www.yolegroup.com
METHODOLOGY AND DEFINITIONS
Preexisting
information
Market
Volume (in Munits)
ASP (in $)
Revenue (in $M)
Information
aggregation
Nour Popoff works for Yole Développement (Yole) as a technology Cédric Malaquin is the Team Lead Analyst for RF activity within
and market analyst for RF devices and technologies. the Power and Wireless Division at Yole Développement (Yole).
Nour is involved in producing technology and market reports for By contributing to multiple reports and custom analyses
the consumer segment, as well as custom consulting studies. dedicated to the automotive, telecom infrastructure, and
consumer markets, Cédric is actively shaping Yole’s products,
Prior to Yole, Nour was engaged in a program management
such as the RF Market Monitor, while also helping to structure
position, defining silicon qualification strategy for RF technology
Yole’s RF industry intelligence activities.
platforms developed at STMicroelectronics’ Crolles site (RFSOI,
FDSOI, SiGe Bi-CMOS). As such, she coordinated technology and Cédric also supports Yole’s financial activities with his technical
IP qualification activities through maturity milestones until the expertise and market knowledge in RF.
industrialization phase. During her twelve years at ST, she also
Prior to Yole, Cédric served as a process integration engineer and
worked as a digital circuit design technical leader and
then as an electrical characterization engineer at Soitec (France).
participated in several technical task forces. She led a one-year
He has (co-)authored several international papers and patents.
collaborative project, CEA-LETI/ST, developing CoolCube FDSOI.
Cédric graduated from Polytech Lille (France) with an
Nour graduated from IAE de Lyon with an International MBA. She
engineering degree in microelectronics and material sciences
holds a research master’s degree in Integrated Circuits from INSA
and holds a DEA in microwave and microtechnology from the
& Ecole Centrale de Lyon and an Electronics Engineering degree
University of Lille.
from CPE de Lyon.
Contact: [email protected]
Contact: [email protected]
Active Semi, AGC, Airoha, Akoustis, Anhui YUNTA Electronic, Apple, ASE, Asus, AT&T, AwinIC,AXT, Broadcom,
CanaanTek, Cavendish Kinetics, China Mobile, China Telecom, China Unicom, ChipBetter, CoolPad, Corning,
Cypress Semiconductor, DB-HiTek, Dowa, EE, Elisa, Epic MEMS, Ericsson, Etisalat, EtraSemi, Ferfics,
Freiberger, GlobalFoundries, Global Wafer, Google, HH Grace, HiSilicon, HMD Global, Honor, HTC, Huawei,
Huntersun-MEMS, Infineon, Intel, IQE, Jio, JRC, KDDI, KT, Kyocera, Lansus, Lenovo, LG, LG Siltron, LG U+,
Maxscend, MediaTek, Meizu, Menlo Micro, Microgate, Murata, NationZ, Nokia, NSI, NTT Docomo, NXP,
Okmetic, Omnivision, ON Semiconductor, OnePlus, Onmicro, Oppo, Orange, Pinnacle Microwave, pSemi,
Qorvo, Qualcomm, Radrocktech, Realme, Resonant, Richwave, RoFS microsystems, Sappland
Microelectronics, Samsung, Samsung Electro Mechanical, Sana, SAWNICS, Shin-Etsu, Shoulder, Siltronic, SK
Telecom, Skyworks, Smarter Micro, SMIC, SoftBank, Soitec, Sony, Sprint, ST Microelectronics, Sumco,
Sumitomo Electric, Sumitomo Metal Mining, Sunrise, Sunway Communication, SWI, Swisscom, Taiyo Yuden,
TCL, TDK EPCOS, Telefonica, Telia, Telstra, TIM, T-Mobile, Toshiba, Tower Semi, Tpsco, TSMC, UMC, Unisoc, USI,
Vanchip, Verizon, Vivo, Vodafone, WaveTek, WillSemi, WIN Semiconductors, WIPAM, Wisol, Xfab, Xiaomi,
Xinpletek, Xpeedic, Yamaju Ceramics, Yuzhen IC, ZTE and more.
*non-exhaustive list
• 5G is improving the quality of service delivered to the user by providing larger network capacity, better radio efficiency, and scalable latency.
This is key to securing the long-term viability of current use cases, such as video streaming, while opening the door to new, use cases, such as
cloud gaming. There is growing interest in new use cases at the MNO level. While immersive AR/VR applications are progressing in the
industrial segment, there is still no killer application in the consumer segment. Nevertheless, in the US, operators are pushing for the
adoption of AR and VR to transform the fan experience in stadiums and venues. In China, technology demonstrations from government-led
projects can be seen. Holographic calls have been demonstrated during MWC for the last two years running. These demonstrations show off
5G’s potential and capabilities.
• Besides mobile and consumer applications, Fixed Wireless Access (FWA) has become a strong driver for 5G penetration. FWA offers an
interesting alternative to fiber, especially in rural or suburban areas. For more information on the FWA market, please check our FWA 2023
report.
• MNO strategies for 5G differ between markets, including where they involve spectrum allocation/auctions. MNOs are investing in spectrum
to extend or secure their current assets, thus determining their strategy in the radio network rollout. This is the key driver of the growing
complexity in RF front-end that OEMs must cope with. Besides spectrum expansion, MNOs are also looking to transform their core network
architecture, moving from 5G NSA to 5G SA. However, this aspect does not strongly impact the RF front end, instead having more
implications for the baseband.
• From a regional standpoint, both China and the US rapidly implemented 5G in 2020, though with different strategies.
• China has established a strong rollout policy using sub-7GHz bands: 700MHz, 2.5GHz, 3.5GHz, and 4.8GHz. The next step is the use of the
licensed 6GHz band for cellular 5G, which will likely delay adoption of mmWave in the country.
• In the USA, mmWave was the first 5G platform, explaining the country’s first position in this technology. Aside from this, spectrum refarming
on the low and mid-band has enabled country-wide coverage. The long-awaited release of C-band spectrum is the latest development
receiving attention. As a result, the momentum for mmWave has decreased, though we estimate investments in the technology will
continue.
• Japan and South Korea are very advanced markets, with most of the population covered by the 5G sub-7GHz spectrum. Japan was the first
country where all operators deployed 5G mmWave base stations, whereas Korean MNOs were far behind their objective, with less than 1,000
mmWave BTS at the end of 2021.
• Europe employs low- and ultra-high band spectrum, while interest in mmWave is predicted in some markets such as Finland and Italy.
• India is the last big market to enter 5G, using low and ultra-high band spectrum while eyeing mmWave.
• Finally, 3GPP Release 17 has extended the mmWave spectrum up to 71GHz, providing more available spectrum. This signifies that the
industry will mature to prepare for new use cases. Additionally, new bands have been made available for Non-Terrestrial Networks (NTN).
• The cell phone market recovered in 2021 following the dip caused by the Covid-19 pandemic in 2020. However, pre-
Covid-19 levels have not been reached due to chip supply shortages.
• In 2022, the smartphone industry was seriously impacted following a global macroeconomic downturn: a market decline
with high inflation caused by geopolitical tensions such as the Russia-Ukraine war and tensions between China and
Taiwan. This downturn resulted in consumer hesitancy in purchasing new phones, thus pushing OEMs to enter an
inventory correction phase. The Zero-Covid policy in China further destabilized the smartphone manufacturing industry.
• Despite these challenging conditions, 5G phone production reached parity with 4G phone production in 2022, though
we believe the market penetration of 5G would have been higher without the aforementioned external factors. 5G
resilience can be explained by the fact that OEMs have prioritized their premium segments.
• We expect the mobile market to be flattish in 2023, as we don’t see any short-term improvement in the global economic
situation and, thus, no impetus for recovery.
• In the longer term, growth should come from emerging countries such as India, which might offset the Chinese market
decline over the next few years.
• The RF front-end market made a leap forward in 2021 as an effect of post-Covid-19 recovery and 5G penetration.
• CY2022 ended up flat following the smartphone market decline associated with lower-than-expected 5G penetration.
Consequently, the bill-of-material growth engine has been in low gear.
• During the last two years, 5G has evolved, bringing increasing complexity from, for instance, the need for MIMO UL/DL
support, an increasing number of CA combinations, and more coexistence scenarios to manage. Consequently, the bill of
materials has increased.
• The 4G semiconductor technology portfolio has evolved to accommodate 5G specifications. For instance, GaAs-based
PAs are still used for mobile phones, RFSOI remains the technology of choice for switches, LNAs, and antenna tuners,
and SAW/BAW technologies remain the filter standard. What is noticeable is the increase in the number of variations of
such technology. Take SAW-type filters, for example, for which bonding techniques are used to improve RF performance
while minimizing size.
• New technologies are now emerging, such as the long-awaited MEMS solution for antenna tuning that hit the market in
mid-2022, introduced by Qorvo, and FDSOI, which is becoming mainstream for mmWave transceivers.
• Meanwhile, other technologies are phasing out, such as BST for antenna tuners, while other technologies at the
development stage are trying to position themselves to tackle the market. Some examples are porous silicon, which
would provide improved switch linearity performance to the RF front-end industry, or GaN/Si PAs with state-of-the-art
power-added efficiency and wideband operation capability.
• The RF front-end market is still led by large players, while some competition is emerging from China. However, all
leading players have faced reductions in their revenue from the smartphone market.
• On the competition from China, the growth of these fabless/fablite players is mainly coming from the local market pull
at this stage.
• According to the Ericsson Mobility report, mobile data traffic continues to grow exponentially and is expected to follow
the same trajectory over the next five years.
• 5G has been developed as a technology capable of absorbing this data traffic as it is the only technology capable of
meeting efficiency requirements in growing the radio network capacity. At some point, 5G simply will not be a question
anymore.
Global mobile network data traffic (EB per month) Global mobile network data traffic and year-on-year growth (EB per month)
• Wireless connectivity will continue to play a vital role in the way people live, and businesses operate.
According to the GSMA, 5G will represent more than 70% of MNOs’ investments in all regions except sub-
Saharan Africa.
• 5G has been adopted quickly in pioneer markets such as Asia Pacific, North America and Europe. A new wave
of 5G rollouts is starting in emerging markets such as India, Indonesia, and Latin America.
• OEMs have adapted their 5G connectivity product offerings to these different markets.
Mobile Network
Operators will
invest massively
in 5G over the
next three
years.
Consequently,
5G’s footprint
will continue to
expand.
1600
1429
1384 1365 1380 1360
1400 1350
1313 1300 1300
1276 1250
1230 1220
1200 1116
Mobile phone volume (Munits)
1082
200 116
78 58
31 45 31 20
0 14 10 8 6 4
0
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
3G 116 78 58 45 31 20 14 10 8 6 4
4G 1313 1276 992 770 595 495 440 396 346 292 240
5G 0 31 250 550 603 706 796 894 996 1082 1116
All 1429 1384 1300 1365 1230 1220 1250 1300 1350 1380 1360
<$0.1B
$0.4B
<$0.1B
mmWave
$1.5B $23B
$1.5B
$4.1B
$13.2B
CAGR22-28: 5.8%
$0.9B
2022 $1B
$19.2B
$12.2B $1.9B
$0.5B $0.7B
$0.9B
$1.4B $1.2B
$8.7B
$4.5B
$2.5B
$3.1B
$1.3B
CAGR22-28: 5.8%
$1B
• Due to the increasing complexity of integrating 5G components along with 4G and other legacy radio components,
complex modules such as LPADs are gaining ground in the market. They offer a compact form factor while packing all
necessary components: PAs, LNAs, filters, and switches.
• However, simpler modules such as MMMB PAs will still be used in combination with antenna switch FEMs, mainly for
LTE handsets.
• In high-end 5G phones, a 2x2 MIMO UL is required, meaning a transmit replication.
• Another driver for complex module growth is the increased output power requested by mobile network operators.
Indeed, in some cases, MNOs request PC 1.5 compliance (i.e., 29dBm at the antenna port) which is not achievable with a
single PA using a conventional technology like GaAs HBT. For this reason, two coherent PC2 PAs will be used instead, as
2x26dBm gives the desired 29dBm.
• LNA front-end module (LFEM) growth will continue being pulled by 4x4 MIMO DL requirements for all phone categories
on HB and UHB frequencies. Two approaches implementing this feature will coexist. The first consists of duplicating the
number of modules to ensure four receive paths for the relevant bands. The second one, likely pulled by a desire for
densification, is the integration of at least two receive paths in a single module connected to two different antennae.
• AiP and mmWave FEM growth will be limited to some geographical regions, i.e., North America, Japan, South Korea,
and a few other countries. Further growth will happen only if another large country beyond the USA goes in this
direction. We estimate that China will not take that path, and Europe is not positioned as an early adopter. Therefore,
growth might come from India, which has just launched its 5G mmWave network.
23
RF FRONT-END FORECAST: DISCRETE COMPONENTS
Outlook
• This segment can be viewed as a commodity market with large volume and relatively low ASP. Discrete components
are easier to procure than modules and available from many vendors. In addition, discrete components offer flexibility to
the OEM in the board design, mainly for entry and mid-tier OEMs that don’t require full-coverage devices supporting
multiple regional bands.
• Filters will remain the dominant component in this segment. The market will increase due to the release of new
frequencies to support the transition to 5G and, consequently, new generations of filters such as BAW SMR, FBAR, and
XBAR resonators.
• In premium phones, the use of antennaplexing, meaning the multiplexing of different radio signals on a single antenna,
will increase as radio standards that accommodate cellular, GPS, WiFi, and UWB along with the growing band
combinations continue to proliferate. As an effect of the increasing use of carrier aggregation and dual connectivity, the
need for multiplexing will become more immediate.
• Growth in LNAs, switches and antenna tuners will be pulled by similar drivers.
• Growth in PMIC envelope tracking and power management circuits will be driven by the need for PA power
consumption optimization over a wider bandwidth with a greater supply voltage. We assume that this function will
continue to be implemented in discrete components as an IC that is utilized for multiple modules, rather than being
integrated into modules that are already crowded with RF components.
• Multi-radio support with MIMO/CA with limited space for antennae GNSS/LB div
WiFi 2.4G/MHB MIMO
• Increased power level: PC3 -> PC2 (+PC 1.5) MHB div
WiFi 5G/UHB
LNA
> 100GHz InP
PA
FR2-1 RFSOI
MMIC
20-52GHz CMOS FDSOI
FR3
GaN/Si
PA
7-20GHz
BAW SMR/FBAR
XBAR
Filter
ML-SAW
SAW/TC SAW
BST MEMS
Sub-7GHz
Switch Tuner
CMOS/RFSOI
FR1
P-Si?
GaAs CMOS/RFSOI
SiGe BiCMOS
LNA
GaAs CMOS/RFSOI
GaN/Si?
PA
GaAs
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 27
TECHNOLOGY PLATFORM
What are the next disruptions to expect?
• GaAs HBT and Bi-HEMT will remain mainstream for sub-7GHz cellular PAs. Both technologies ensure good linearity, high efficiency, low
leakage, and output power requirements for HPUE compliance: PC2/26dBm at the antenna port. MNOs’ demands for even higher output
power – PC1.5/29dBm – will be achieved using two coherent PAs operating at 26dBm. Therefore, there is no need for higher output power
technology. However, related to the dual-stream 2x2 MIMO up-link trend, some OEMs might find any solution that provides much wider
bandwidth than GaAs appealing in order to merge the main bands in the UHB frequency range – N77/N78 and N79 – onto a single PA.
• RF SOI is the mainstream technology for switches and tuners, with RF CMOS as an alternative. Among the emerging technologies, MEMS is
the most promising, especially for antenna tuning. We believe there is a sweet spot for MEMS tuners for high-end antenna tuning cases.
However, manufacturing MEMS tuners at high volumes remains a pain point. PCM switches promise to drastically improve RonxCoff
performance, which has not been sufficient for the technology to penetrate this market so far.
• While SiGe Bi-CMOS is a great fit for high-performance LNAs, RFSOI will take most of the growth due to the attractiveness of its switch
integration capability. In some cases, advanced CMOS remains a competitive alternative.
• High-performance filter technology such as BAW SMR, FBAR, or XBAR will keep gaining momentum as multi-radio coexistence will become
more of a concern. However, legacy SAW technology will maintain a high take-rate from OEMs seeking only a ‘good enough’ solution. On
the other hand, emerging ML SAW technology will continue to gain momentum as it offers thermal compensation along with multi-band
filtering integration in a small form factor. MLC and IPD filters will still be used in the long run due to their cost advantages.
• CMOS MMICs have been the fastest and most cost-effective way to bring 5G mmWave AiP to market. However, the technology has some
drawbacks, such as output power limitation and power consumption linked to efficiency. One piece of evidence of such limitations is the
spectacular disaggregation of commercial AiP done by Apple. Putting the MMIC in a FEM attached to a passive and implementing a larger
antenna has enabled a better radio link budget due to the higher antenna gain. Very recently, we saw FDSOI emerging in this market in 22
and 28nm, and we are expecting this technology to further penetrate the market over the forecast period. Despite the momentum in
FDSOI, we believe that Apple, which is likely to introduce its own mmWave chipset by 2024, will remain conservative and stick to advanced
RF CMOS, though it could be using RFSOI.
• The RF front-end market is currently navigating fluctuations following the market recovery from Covid-19 in 2021 and
the market decrease in 2022. Fortunately, many wafer suppliers learned from the recent supply shortage and created
long-term customer agreements.
• Another parameter impacting the wafer start forecast is the move to larger diameters, which is underway and ongoing
for several RF front-end technologies.
• GaAs, RF SOI, silicon, LT and LN are the main substrates needed for RF front-end, driving most of the volume.
• For filters, a new category of substrates is emerging, requiring bonding or layer transfer. Multiple material stacks can be
found on the market, such as LT on LT, LT/LN on silicon, LT/LN on sapphire, and LT/LN on glass. Therefore, in this report,
we group these different engineered substrates in the multi-layer SAW (ML SAW) segment.
• The use of FDSOI substrates is increasing significantly in order to serve MMIC demand for mmWave FEM and AiP
applications.
IDM
PA LNA/Switch Tuner Filter SoC CS Silicon Specialty Module and Packaging Module Discrete
• The RF front-end market reached over $19B in 2021 but flattened in 2022 following demand weakness and inventory
correction.
• In 2021, Qualcomm led the market with its end-to-end approach, followed by Broadcom’s custom PA module offering.
Skyworks and Qorvo have similar company profiles with a broad RF portfolio serving all market segments, although
Skyworks has been more exposed to increased Chinese competition. Murata’s revenue was low, suffering from growing
competition on its filters and restructuring its module portfolio.
• In 2022, traditional players were impacted by the deterioration in the macroeconomic environment. Except for
Broadcom, revenues generated by companies in RFFE-related segments have decreased.
• The Chinese RF front-end ecosystem has been growing, with fabless companies mainly emerging, most taking a share
of their local market. Maxscend, Vanchip, and Smarter Micro are the most relevant examples, though Maxscend recently
invested in upgrading to a fablite business model. There is a long list of companies with great ambitions and access to
financial capital through public offerings on the STAR market. However, not all initiatives will succeed, and we expect
consolidation to happen in the mid-term.
• To sum up, Chinese RF front-end players capture a limited fraction of the market, as OEMs still rely on the leading
players for their premium products, but the largest Chinese companies are on their way to catching up with the leaders.
4G LTE
FR 5G NR
4 Refarming
92
E
V N262
FR2
N259 & N260
Ka
N257 & N261
N258
20
Ku
FR3
• In our last annual report edition, 2022 vs. 2021 smartphone production forecast (Munits)
published in June 2021, we anticipated
market recovery in 2021 after the majority 1500
of the Covid-19 lockdowns, followed by
inventory correction in 2022.
• In fact, actual smartphone production 1450
numbers were lower in 2021 compared to Inventory
our expectation, mainly due to the impact correction
of the semiconductor component
shortage. Still, it was the first growth Shortages
1400
phase since 2017.
• The inventory correction phase
anticipated for 2022 happened, but
external factors exacerbated its scale. 1350
Geopolitics
• Geopolitical tensions, the war in Ukraine, + Inflation
and tensions in Taiwan led to an + Covid-19
unprecedented rise in energy prices,
1300
reinforcing the inflation that was just
starting to become a problem.
• On top of this, recurring Covid-19 waves
amid the Zero-Covid policy in China 1250
caused plant lockdowns.
• Following the actions of the central
banks, inflation is expected to decrease in
1200
2023 and 2024, which has prompted our
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029
optimistic forecast.
Smartphones 2022 Smartphones 2021
iPhone SE iPhone 12
2020, 2022 iPhone 13
iPhone 14
1600
1400
1600
1400
1200
Mobile phone volume (Munits)
1000
5G penetration will
continue through
to 2028, surpassing 800
1Bunits. 3G and 4G
support will
consequently 600
decrease.
400
200
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
(2022-2028)
5G 0 31 250 550 603 706 796 894 996 1082 1116 12%
4G 1313 1276 992 770 595 495 440 396 346 292 240 -13%
3G 116 78 58 45 31 20 14 10 8 6 4 -29%
1400
1200
Mobile phone volume (Munits)
1000
During the 2021
recovery phase,
OEMs prioritized 800
their mid-tier and Entry level
premium models.
600 Mid Tier
OEMs continued
this strategy in Premium
2022 to secure 400
profitability.
200
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 (2022-
2028)
Entry level 738 710 647 640 559 536 542 560 579 584 558 1%
Mid Tier 369 366 353 373 346 344 357 377 398 413 411 4%
Premium 321 309 299 352 325 340 351 363 373 383 391 3%
1600
1429
1384 1365 1380 1360
1400 1350
1313 1300 1300
1276 1250
1230 1220
1200 1116
Mobile phone volume (Munits)
1082
200 116
78 58
31 45 31 20
0 14 10 8 6 4
0
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
3G 116 78 58 45 31 20 14 10 8 6 4
4G 1313 1276 992 770 595 495 440 396 346 292 240
5G 0 31 250 550 603 706 796 894 996 1082 1116
All 1429 1384 1300 1365 1230 1220 1250 1300 1350 1380 1360
• The so-called mid-band (sub-7GHz) 2018-2028 mobile phone volume forecast per air standard (Munits)
was the most-used spectrum in 2022. 1600
GSMA recommends making 2GHz of
bandwidth available in the sub-7GHz 1400
spectrum in the next decade to serve a
wide range of use cases cost- 1200
• The cell phone market recovered in 2021 following the dip caused by the Covid-19 pandemic in 2020. However, pre-
Covid-19 levels have not been reached due to chip supply shortages.
• In 2022, the smartphone industry was seriously impacted following a global macroeconomic downturn: a market decline
with high inflation caused by geopolitical tensions such as the Russia-Ukraine war and tensions between China and
Taiwan. This downturn resulted in consumer hesitancy in purchasing new phones, thus pushing OEMs to enter an
inventory correction phase. The Zero-Covid policy in China further destabilized the smartphone manufacturing industry.
• Despite these challenging conditions, 5G phone production reached parity with 4G phone production in 2022, though
we believe the market penetration of 5G would have been higher without the aforementioned external factors. 5G
resilience can be explained by the fact that OEMs have prioritized their premium segments.
• We expect the mobile market to be flattish in 2023, as we don’t see any short-term improvement in the global economic
situation and, thus, no impetus for recovery.
• In the longer term, growth should come from emerging countries such as India, which might offset the Chinese market
decline over the next few years.
RF signal amplification,
filtering and rooting
Digital Analog up/down up/down to the
signal coding conversion antennae
This report
focuses on
the cellular Cellular Scope of this report
Baseband
Cellular RF RF
RF front-end Transceiver Front-End
market, with SoC
some insights
into
baseband
and RF
transceivers.
Wi-Fi and
UWB are not
covered.
Connectivity Connectivity Featured in the
SoC RF smartphone RF
monitor.
• For better visibility, we have added 5G 2018-2028 overall market forecast by segment ($M)
mmWave as a new segment, including $60,000
mmWave RF front-end and IF
transceivers.
• The overall market was boosted in 2021 $50,000
following a strong 5G uptake. 2022 has
been flat due to weakness in the
smartphone market. In the mid-term, $40,000
we foresee a mid- to high-single-digit
growth before the market reaches a
Market ($M)
plateau once 5G is established. $30,000
Entry Level , 6
55 phones
Premium, 30
Mid Tier , 19
Yole System
Plus has
analyzed over
50 phones
OnePlus; 1
across various Google; 2Nokia ; 1
OEMs and Nokia ; 2
Samsung; 12
market Asus; 2
segments. Motorola ; 2
Huawei ; 3
Apple; 3
55 phones
Oppo; 7
ZTE; 4
Vivo ; 4 Xiaomi;6
Realme; 6
iPhone 13 Find X3 Pro X60 Pro S22 ultra S22+ Z Flip3 12 Pro X70 Pro Rog 5 Find X5 Pro
Phone
teardowns
are selected Reno 6 5G 9 Pro 5G Pixel6 Pro
across major ZenFone 8 P50 Pro
A52 Black shark 4 Redmi Note11 A72
A32 5G Poco F4 GT
OEMs and
phone
segments,
spanning a
wide bill-of- X70 Pro 5G F19 Pro+ 5G
Mi Mix4
material X100 5G
range. Axon 30 5G Nubia 6S Pro
GT Neo 2 Nova 9 Pro
A21 In Note 1
Non-exhaustive list of smartphones analyzed RF Front-End for Mobile 2023 | Report | www.yolegroup.com 52
RF FRONT-END BILL OF MATERIALS
A wide range across air standard and phone segments
Switch
PMIC Switch PMIC Switch PMIC Switch
Tuner
Filter
Filter Tuner
Filter Filter Tuner Filter Tuner
Filter Tuner Filter Tuner Filter Tuner
PAM Filter Tuner PAM Filter Tuner PAM Filter Tuner
• We define the cellular RF front-end as the bill-of-materials for components between the transceiver and the antenna.
• In the cellular RF front-end bill-of-materials, we include modules and discrete components intended for amplifying,
routing, filtering, or tuning radio signals.
• Modules PAM MMMB PA PAD LPAD
• Transmit modules or PA modules PMIC PMIC PMIC
Switch
Switch
PMIC Filter Filter
Switch
Switch
Switch
Switch
Switch
LNA
Filter Filter
Filter Filter
PA PA PA Filter PA Filter
PA PA PA
Filter Filter
Switch
Switch
Filter
LNA
Filter
Filter Filter
Filter Filter
• Discrete components
• Filters: Single Tx or Rx filters, duplexers, multiplexers, diplexers, antennaplexers
• LNA: Single band LNAs, LNA banks
• Switch: Cross-switches, DPD switches, antenna switches, SRS switches
• Tuner: Aperture tuners, impedance tuners
• PMIC: Envelope trackers, average power trackers
5G mmWave support
MMIC, PMIC (FEM eventually)
Multi radio support: 3G and 4G, 4G and 5G and integrated antenna
PA, switches, filter, duplexer, multiplexer (and LNA for LPAD)
Optional
Antennas
Duplexer PMIC
LNA FEM
Multi-radio support PA
Switch
MMIC
2G and 3G, 3G and 4G LNA
es
PA
Switch
Filter
Single radio support No filter function PMIC
es
Switches
PA
2G, 3G, 4G or 5G Filter PMIC Filter
PMIC
Switches
PMIC
PMIC
Switches
PA
PA
PMIC
antennas
External
MMIC
Filter
Switches
Switches
LNA
Filter
5G mmWave support
MMIC, PMIC
Market ($M)
penetration stabilized, while 5G sub-
$15,000
7GHz penetration has been lower than
our expectations.
• We estimate the RF front-end market $10,000
air standard, i.e., 6G, to drive revenue 5G mmW $- $29 $501 $1,429 $1,485 $1,782 $1,814 $1,901 $1,976 $2,121 $2,195
2028
7%
growth. 5G sub-7 $- $674 $5,179 $11,893 $13,224 $15,487 $17,181 $19,404 $21,056 $22,426 $22,976 10%
4G $10,719 $9,883 $6,973 $5,294 $4,124 $3,431 $2,997 $2,671 $2,294 $1,914 $1,539 -15%
3G $73 $49 $38 $30 $22 $14 $10 $7 $5 $4 $3 -27%
2G $453 $434 $406 $476 $363 $320 $297 $272 $238 $208 $191 -10%
Report 2021 $12,019 $12,417 $14,202 $17,391 $18,742 $19,595 $20,409 $21,346
Note: Discrepancies between current data and the 2021 report for CY 2018, 2019, and 2020 are due to the removal of
feature phones from the analysis to comply with the segmentations used in this version of the report.
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 58
MARKET SEGMENTATION
Module, die and discrete
Switches
PA
technologies (i.e., GaAs, SAW, BAW, or RF SOI), enabling various
functions (i.e., PAs, LNAs, switches, filters, etc.) within the module. It is a PMIC Filter
System-in-Package (SiP), leveraging advanced packaging techniques to
merge all necessary components – both dies and passives – and Module = multiple dies with different functions
To accurately implement appropriate shielding methods to minimize potential
estimate interference.
market size, we
analyze the • A die refers to a piece of circuit derived from a wafer after dicing Die
go down to the
bonding (bare die) or in a flip-chip configuration (packaged die). In this
die/discrete Die
Switch n
Switch n
Antenna
PAn
Tx/Rx
Tx Filter 1
Tx/Rx n
Switch n
Antenna
PAn
Tx/Rx
Switch
Tx Filter 1
Switch n
Cross Switch n
Antenna
PA1 PAn Tx Filter n
Mplexer n PAn
Switch
PA1
Tx/Rx
Main
Main
RF front-end
PA1 Mplexer n Tx Filter 1
devices come in MplexerMplexer
1 n PA1
various forms in Mplexer 1
Switch
mobile phones. Mplexer 1
Switch
Cross
Switch n
Antenna
Integrated Duplexer n
Switch
Cross
architectures Duplexer n
Switch 1
Duplexer 1
Tx/Rx
Switch 1
Antenna
enable the best Duplexer n
Tx/Rx 1
LNA Duplexer 1 Duplexer n
Tx/Rx
Antenna 1
Antenna
Transceiver
Transceiver
Switch
performance and
Switch 1
LNA Duplexer 1 Mplexer n
Switch
Switch 1
board optimization,
LNA Duplexer 1
as well as Rx Filter n Mplexer 1
simplifying Rx Filter n
assembly for the Rx FilterRx1 Filter n
Rx Filter 1
Switch
Rx Filter n
Cross
OEM, while the
discrete approach Rx Filter 1
is the path to take Rx Filter 1
for bill-of-material
optimization and
supply chain
Diversity
Diversity
Rx Filter n
/MIMO
/MIMO
flexibility.
LNA
Rx Switch
Antenna
Rx Filter 1
Rx Switch
Switch
Antenna
Switch
Rx Switch
Rx Filter n
Antenna
LNA
Switch
Rx Filter n
Rx Switch
LNA
Antenna
Rx Filter 1
Switch
LNA Rx Filter
Rx Filter 1 n Rx Filter n
LNA
Rx Filter 1 Rx Filter 1
As the RF
$25,000
component
density
continues to $20,000
increase, the
Market ($M)
trend towards
in-module $15,000
integration will
keep increasing.
On the other $10,000
hand, discrete
components
will remain a $5,000
significant
alternative over
the forecast $-
CAGR
period. 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
Discrete $4,179 $4,623 $5,050 $6,396 $6,333 $7,001 $7,492 $8,162 $8,628 $8,995 $8,944 6%
Module $7,066 $6,447 $8,047 $12,726 $12,885 $14,032 $14,808 $16,092 $16,942 $17,678 $17,960 6%
Report 2021 $12,019 $12,417 $14,202 $17,391 $18,742 $19,595 $20,409 $21,346
Note: Discrepancies between current data and the 2021 report for CY 2018, 2019, and 2020 are due to the removal of feature phones
from the analysis to comply with the segmentations used in this version of the report.
• LPAD-type components are the most 2018-2028 RF front-end market value forecast breakdown by module type ($M)
complex SiP module among the sub- $20,000
7GHz components. It will exhibit 7%
$18,000
CAGR growth from 2022 to 2028.
LFEM-type components will strongly $16,000
Market ($M)
growing complexity in RF front-end
$10,000
devices.
$8,000
• Regarding mmWave, AiP is the main
format and will exhibit the highest $6,000
expand into more large markets such 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2022-
2028
as North America. mmWave FEM $- $- $98 $353 $375 $406 $400 $406 $419 $433 $433 2%
AiP $- $23 $259 $646 $643 $689 $686 $721 $754 $828 $874 5%
• As 5G penetrates and becomes LFEM $2,088 $1,840 $1,821 $2,626 $2,669 $2,946 $3,154 $3,625 $3,868 $4,087 $4,154 8%
FEM $528 $578 $581 $510 $441 $397 $390 $365 $361 $347 $341 -4%
established, RF front-end market LPAD $1,882 $1,493 $3,331 $6,926 $7,417 $8,380 $9,004 $9,831 $10,436 $10,930 $11,175 7%
growth will slow down. PAD $998 $985 $636 $367 $321 $286 $277 $278 $268 $249 $217 -6%
PAM $453 $434 $406 $476 $363 $320 $297 $272 $238 $208 $191 -10%
MMMB PA $1,116 $1,094 $915 $822 $656 $607 $599 $595 $597 $595 $575 -2%
10000
• Due to the increasing complexity of integrating 5G components along with 4G and other legacy radio components,
complex modules such as LPADs are gaining ground in the market. They offer a compact form factor while packing all
necessary components: PAs, LNAs, filters, and switches.
• However, simpler modules such as MMMB PAs will still be used in combination with antenna switch FEMs, mainly for
LTE handsets.
• In high-end 5G phones, a 2x2 MIMO UL is required, meaning a transmit replication.
• Another driver for complex module growth is the increased output power requested by mobile network operators.
Indeed, in some cases, MNOs request PC 1.5 compliance (i.e., 29dBm at the antenna port) which is not achievable with a
single PA using a conventional technology like GaAs HBT. For this reason, two coherent PC2 PAs will be used instead, as
2x26dBm gives the desired 29dBm.
• LNA front-end module (LFEM) growth will continue being pulled by 4x4 MIMO DL requirements for all phone categories
on HB and UHB frequencies. Two approaches implementing this feature will coexist. The first consists of duplicating the
number of modules to ensure four receive paths for the relevant bands. The second one, likely pulled by a desire for
densification, is the integration of at least two receive paths in a single module connected to two different antennae.
• AiP and mmWave FEM growth will be limited to some geographical regions, i.e., North America, Japan, South Korea,
and a few other countries. Further growth will happen only if another large country beyond the USA goes in this
direction. We estimate that China will not take that path, and Europe is not positioned as an early adopter. Therefore,
growth might come from India, which has just launched its 5G mmWave network.
64
RF FRONT-END DISCRETE MARKET VALUE FORECAST
Breakdown by type of discrete component
$7,000
$-
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2022-
2028
IF RxTx $- $6 $145 $430 $467 $686 $728 $773 $803 $860 $888 11%
PMIC $237 $267 $412 $678 $711 $803 $883 $952 $988 $1,007 $994 6%
Tuner $747 $837 $989 $1,440 $1,417 $1,507 $1,599 $1,744 $1,816 $1,871 $1,881 5%
Switch $337 $416 $472 $561 $552 $614 $690 $771 $859 $928 $927 9%
LNA/switch $330 $428 $460 $511 $536 $626 $733 $846 $966 $1,072 $1,066 12%
LNA $90 $101 $103 $124 $124 $133 $140 $146 $153 $158 $161 4%
Filter $2,437 $2,567 $2,470 $2,652 $2,527 $2,632 $2,718 $2,929 $3,044 $3,099 $3,027 3%
50000
front-end devices,
in-module 30000
integration is not
the only way to
go. We foresee 20000
consolidation of
the discrete
10000
component
market, with
standalone filters, 0
CAGR
switches and 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
LNAs, and, of IF RxTx 0 1.5 32 90 96 102 110 120 129 143 153 8%
course, tuners. PMIC 1245 1301 1673 2417 2377 2506 2631 2786 2922 3023 3046 4%
Tuner 4567 5292 6029 8155 7819 8248 8689 9500 9961 10308 10332 5%
Switch 5099 6016 6594 7319 6861 7312 7876 8601 9339 9911 9880 6%
LNA/switch 1534 1863 1827 1814 1682 1739 1862 2006 2136 2232 2183 4%
LNA 2214 2623 2684 2887 2730 2781 2845 2923 3015 3067 3080 2%
Filter 22384 23792 23422 25248 23756 24671 25866 27891 29321 30215 29610 4%
• This segment can be viewed as a commodity market with large volume and relatively low ASP. Discrete components
are indeed easiest to procure than modules and available from many vendors. In addition, discrete components offer
flexibility to the OEM in the board design, mainly for Entry and Mid-Tier OEM segment, that don’t require full coverage
devices supporting multiple regional bands.
• Filters will remain the dominant component in this segment. It will increase due to new frequencies to support with the
transition to 5G and consequently new generations of filters such as BAW SMR, FBAR and XBAR resonators.
• In premium phones, the use of antennaplexing, meaning the multiplexing of different radio signals on a single antenna,
will increase as the radio standard to accommodate (cellular, GPS, WiFi, UWB) along with the growing band
combinations will keep exploding. As an effect of an increasing use of carrier aggregation and dual connectivity, the
need for multiplexing will grow as well.
• LNAs, switches and antenna tuners growth will be pulled by similar drivers.
• The growth in PMIC (envelope tracking and power management circuits) will be driven by the need for PA power
consumption optimization over a wider bandwidth while having a greater supply voltage. We assume that this function
will continue to be implemented in discrete components as an IC that is utilized for multiple modules, rather than being
integrated into modules that are already crowded with RF components.
• The volumes given in the following forecasts are based on physical dies, even though a physical die often has multiple
functions, for instance, dual-band PAs, multiple-in-one filters, etc.
• We adopted this approach to be better able to analyze the size of the wafer-level market.
• Component classification is linked to die function; e.g., the PA segment is defined as dies intended for signal power
amplification.
• The MMIC category includes integrated mmWave transceivers/front-end components featured in AiP or in mmWave
FEM devices.
• Certain components have been merged into a single category to reduce the number of component classes:
• RFIC is the sum of MIPI controllers, impedance matching circuits, couplers, and PMICs.
• ET is the sum of envelope trackers and average power trackers intended for power consumption management
or power boosting.
$2,000
cellular PAs as it
offers the best
tradeoff
$1,000
between
linearity, output
power, and
power
consumption. $500
$-
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
GaAs Bi-HEMT $70 $47 $47 $94 $99 $102 $103 $106 $107 $107 $109 2%
GaAs HBT $1,165 $1,073 $1,207 $1,725 $1,644 $1,696 $1,757 $1,837 $1,909 $1,965 $1,961 3%
8000
The PA volume
produced 7000
substantially
Component volume (Munits)
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
GaAs Bi-HEMT 306 246 222 450 472 492 500 519 528 536 545 2%
GaAs HBT 4911 4701 5076 6600 6178 6421 6802 7273 7770 8193 8253 5%
cost/performance
tradeoff, with
$1,500
CMOS as an
alternative. We
noticed some
developments $1,000
ongoing on
porous silicon,
which further
improves RF $500
performance and
stability over
temperature.
$-
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
Porous $- $- $- $- $- $0 $0 $1 $2 $3 $3
RFSOI $1,190 $1,248 $1,447 $1,831 $1,770 $1,877 $2,025 $2,182 $2,341 $2,471 $2,487 6%
CMOS $55 $56 $47 $44 $38 $39 $41 $43 $44 $45 $43 2%
18000
As more bands
16000
are combined
and MIMO is
Component volume (Munits)
14000
becoming
democratized, 12000
the use of
switches is 10000
unlikely to
wane. However, 8000
no significant
growth is to be 6000
expected as
switches tend 4000
to be integrated
with LNAs. 2000
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
Porous 0 0 0 0 0 2 8 19 39 65 62
RFSOI 10760 11143 12373 14249 13267 13645 14500 15618 16734 17629 17693 5%
CMOS 1670 1640 1368 1222 1014 1011 1008 1043 1066 1075 1034 0%
$3 500
$-
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
SiGe $ 161 $ 146 $ 192 $ 265 $ 287 $ 320 $ 352 $ 385 $ 410 $ 430 $ 440 7%
RFSOI $1 083 $1 096 $1 230 $1 908 $1 917 $2 155 $2 393 $2 667 $2 958 $3 237 $3 265 9%
CMOS $ 15 $ 12 $ 33 $ 51 $ 56 $ 61 $ 65 $ 67 $ 65 $ 64 $ 63 2%
14000
between SiGe
10000
BiCMOS with its
state-of-the-art
noise figure and 8000
0
CAGR 2022-
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2028
SiGe 3147 3269 3347 3654 3534 3685 3854 4059 4255 4378 4419 4%
RFSOI 3576 3835 4574 6248 6051 6436 6880 7449 8020 8621 8696 6%
CMOS 50 35 105 242 259 271 280 287 288 288 293 2%
• The filter market is the largest in the 2018-2028 filter market forecast by technology ($M)
RFFE segment and the one with the $12,000
most important technology split.
• SAW-type filters are widely used and
mature. For demanding filter $10,000
BAW SMR and FBAR, we expect IPD $12 $17 $52 $93 $120 $137 $143 $219 $238 $269 $285 15%
FBAR $821 $691 $823 $1,293 $1,400 $1,538 $1,568 $1,837 $1,817 $1,784 $1,790 4%
significant penetration of XBAR BAW SMR $591 $568 $700 $1,154 $1,161 $1,242 $1,303 $1,393 $1,434 $1,479 $1,487 4%
technology. ML SAW $32 $35 $74 $179 $234 $303 $395 $484 $574 $665 $686 20%
TC SAW $167 $137 $263 $651 $687 $786 $835 $892 $923 $938 $962 6%
SAW $3,596 $3,625 $3,483 $3,680 $3,517 $3,731 $3,850 $4,048 $4,225 $4,273 $4,226 3%
80000
There were over
70000
50Bunits of filters
Component volume (Mu)
produced in 2021 60000
and 2022, and we
expect 80Bunits 50000
of filters to be
reached by 2027. 40000
The share held by
30000
SAW filters will
erode over time,
20000
though it will
remain high. ML 10000
SAW and XBAR
are expected to 0
CAGR
progress the 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
most as these MLC 4063 4286 5477 6232 5979 6372 6882 7861 8619 9522 9555 8%
technologies are
XBAR 0 0 0 0 0 0 36 169 397 622 791 0%
ramping up.
IPD 101 127 318 556 698 812 871 1389 1548 1789 1891 18%
FBAR 5273 4601 4977 7932 8329 9054 9225 10861 10769 10568 10598 4%
BAW SMR 3586 3427 4017 6553 6648 7152 7724 8476 8978 9400 9525 6%
ML SAW 152 197 547 1425 1869 2479 3283 4050 4925 5778 5971 21%
TC SAW 992 849 1324 2788 2852 3270 3500 3835 4087 4278 4401 7%
SAW 30172 30655 29971 32626 31155 32999 34570 36986 39148 40251 39837 4%
• As for switches, RFSOI will remain the 2018-2028 tuner market value forecast breakdown by technology ($M)
mainstream technology for antenna $2 000
tuners, allowing good-enough
performance at a low cost, while CMOS- $1 800
based tuners will remain an alternative in
the supply chain. $1 600
10000
RF SOI will be the Component volume (Munits)
leading technology
for antenna tuners. It
8000
allows high-power
handling and high
linearity. MEMS is an
even better 6000
alternative due to its
tuning range, but
adoption has so far
4000
been limited by
manufacturing
difficulties. Qorvo’s
expertise has helped 2000
here, and we expect
MEMS to grow over
the next five years. 0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
MEMS 0 0 0 0 1 14 35 58 79 100 100 109%
BST 166 157 124 92 36 10 0 0 0 0 0 -100%
RFSOI 4197 4941 5738 7930 7659 8098 8521 9297 9721 10037 10074 5%
CMOS 205 195 167 132 124 127 133 146 161 171 158 4%
Integrated approach
Advanced
RF CMOS FD SOI RF SOI
4th gen
Next-gen
QTM545
AiP
Custom FEM
Custom AiP
FDSOI is $1,000
expected to
capture a
significant
share of the $800
market over the
forecast period.
Market ($M)
However,
$600
CMOS, likely to
be produced on
RFSOI wafers,
will remain the $400
dominant
technology as
we do not
foresee a move $200
to FDSOI from
Apple.
$-
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
FDSOI $- $- $- $4 $36 $72 $219 $244 $262 $311 $350 46%
CMOS $- $16 $255 $751 $777 $810 $680 $694 $714 $738 $737 -1%
250
CMOS has been Component volume (Munits)
the go-to solution
for MMICs, as
Qualcomm 200
capitalizes on its
decade of WiGig
expertise. 150
However, FDSOI
was introduced by
Samsung in 2022,
shortly followed by 100
Mediatek. We
expect Qualcomm
to switch to FDSOI
for its next- 50
generation AIP by
2024.
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2022-
2028
FDSOI 0.0 0.0 0.0 0.8 7.6 14.5 47.9 53.0 57.8 69.9 80.4 48%
CMOS 0.0 4.4 63.5 178.1 181.7 183.6 166.0 179.5 190.4 202.8 208.7 2%
• GaAs HBT and Bi-HEMT will remain mainstream for sub-7GHz cellular PAs. Both technologies ensure good linearity, high efficiency, low
leakage, and output power requirements for HPUE compliance: PC2/26dBm at the antenna port. MNOs’ demands for even higher output
power – PC1.5/29dBm – will be achieved using two coherent PAs operating at 26dBm. Therefore, there is no need for higher output power
technology. However, related to the dual-stream 2x2 MIMO up-link trend, some OEMs might find any solution that provides much wider
bandwidth than GaAs appealing in order to merge the main bands in the UHB frequency range – N77/N78 and N79 – onto a single PA.
• RF SOI is the mainstream technology for switches and tuners, with RF CMOS as an alternative. Among the emerging technologies, MEMS is
the most promising, especially for antenna tuning. We believe there is a sweet spot for MEMS tuners for high-end antenna tuning cases.
However, manufacturing MEMS tuners at high volumes remains a pain point. PCM switches promise to drastically improve RonxCoff
performance, which has not been sufficient for the technology to penetrate this market so far.
• While SiGe Bi-CMOS is a great fit for high-performance LNAs, RFSOI will take most of the growth due to the attractiveness of its switch
integration capability. In some cases, advanced CMOS remains a competitive alternative.
• High-performance filter technology such as BAW SMR, FBAR, or XBAR will keep gaining momentum as multi-radio coexistence will become
more of a concern. However, legacy SAW technology will maintain a high take-rate from OEMs seeking only a ‘good enough’ solution. On
the other hand, emerging ML SAW technology will continue to gain momentum as it offers thermal compensation along with multi-band
filtering integration in a small form factor. MLC and IPD filters will still be used in the long run due to their cost advantages.
• CMOS MMICs have been the fastest and most cost-effective way to bring 5G mmWave AiP to market. However, the technology has some
drawbacks, such as output power limitation and power consumption linked to efficiency. One piece of evidence of such limitations is the
spectacular disaggregation of commercial AiP done by Apple. Putting the MMIC in a FEM attached to a passive and implementing a larger
antenna has enabled a better radio link budget due to the higher antenna gain. Very recently, we saw FDSOI emerging in this market in 22
and 28nm, and we are expecting this technology to further penetrate the market over the forecast period. Despite the momentum in
FDSOI, we believe that Apple, which is likely to introduce its own mmWave chipset by 2024, will remain conservative and stick to advanced
RF CMOS, though it could be using RFSOI.
800
cellular PAs.
We have 500
excluded
GaN/Si from 400
our forecast
as this
solution 300
would be too
costly at the 200
system level.
100
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
GaAs 6" 460 407 445 640 613 632 660 697 730 758 765 4%
900
switch
600
substrate, with
12” wafers 500
growing the
400
most. Few
silicon wafers 300
are consumed
200
for switches,
and we foresee 100
porous silicon as
0
an option. 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
CAGR
2022-2028
Porous 8" 0 0 0 0 0 0 0 1 2 3 3
RFSOI 12" 2 12 59 102 103 109 115 122 127 131 135 5%
RFSOI 8" 653 646 624 691 620 626 657 699 743 775 774 4%
Si 8" 58 58 48 43 35 35 35 37 38 39 37 1%
800
Since the
introduction of 5G, 700
12” RFSOI became
the main wafer
Wafer starts (kwafers)
500
Wafer starts for
antenna tuners
represent around
Wafer starts (kwafers)
400
450-500 kwafers
per year in the 5G
era. 12” RFSOI is
300
employed for
premium phones,
while significant
growth in silicon 200
wafer starts can
be attributed to
MEMS. GaAs 100
wafer starts are
attributed to BST
tuners.
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
GaAs 6" 6 6 5 3 1 0 0 0 0 0 0 -100%
RFSOI 12" 12 14 67 172 185 200 207 225 225 225 230 4%
RFSOI 8" 369 403 359 339 273 260 267 281 298 309 302 2%
Si 8" 11 11 10 8 7 8 9 10 12 13 13 9%
4500
4000
mmWave is 100
contributing to
growing wafer
starts. FDSOI
will outgrow the 80
Wafer starts (kwafers)
market as three
chipset
providers have
60
adopted or will
adopt this
technology. We
believe Apple 40
will leverage
CMOS, though
the chip could
be made on 20
RFSOI
substrates.
0
CAGR
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2022-2028
FDSOI 0 0 0 0 3 6 20 22 24 30 34 47%
CMOS 0 0 29 81 79 78 63 65 69 73 75 -1%
• The RF front-end market is currently navigating fluctuations following the market recovery from Covid-19 in 2021 and
the market decrease in 2022. Fortunately, many wafer suppliers learned from the recent supply shortage and created
long-term customer agreements.
• Another parameter impacting the wafer start forecast is the move to larger diameters, which is underway and ongoing
for several RF front-end technologies.
• GaAs, RF SOI, silicon, LT and LN are the main substrates needed for RF front-end, driving most of the volume.
• For filters, a new category of substrates is emerging, requiring bonding or layer transfer. Multiple material stacks can be
found on the market, such as LT on LT, LT/LN on silicon, LT/LN on sapphire, and LT/LN on glass. Therefore, in this report,
we group these different engineered substrates in the multi-layer SAW (ML SAW) segment.
• The use of FDSOI substrates is increasing significantly in order to serve MMIC demand for mmWave FEM and AiP
applications.
• According to the Ericsson Mobility report, mobile data traffic continues to grow exponentially and is expected to follow
the same trajectory over the next five years.
• 5G has been developed as a technology capable of absorbing this data traffic as it is the only technology capable of
meeting efficiency requirements in growing the radio network capacity. At some point, 5G simply will not be a question
anymore.
Global mobile network data traffic (EB per month) Global mobile network data traffic and year-on-year growth (EB per month)
• Wireless connectivity will continue to play a vital role in the way people live, and businesses operate.
According to the GSMA, 5G will represent more than 70% of MNOs’ investments in all regions except sub-
Saharan Africa.
• 5G has been adopted quickly in pioneer markets such as Asia Pacific, North America and Europe. A new wave
of 5G rollouts is starting in emerging markets such as India, Indonesia, and Latin America.
• OEMs have adapted their 5G connectivity product offerings to these different markets.
Mobile Network
Operators will
invest massively
in 5G over the
next three
years.
Consequently,
5G’s footprint
will continue to
expand.
• During MWC 2022, HTC showcased VR experiences on its Vive platform. More recently, HTC launched a 5G flagship
model, the Desire 22 Pro, to provide users with a gateway to the Metaverse.
• More generally, OEM efforts to improve the AR/VR experience are keeping pace.
• Recent examples of MNOs’ initiatives for in-venue augmented experiences through 5G:
• Super Bowl LV, Raymond James Stadium in Florida, February 2021: Verizon
enabled fans to compete virtually via the NFL Ticketholder App.
• Super Bowl LVI, SoFi stadium in Los Angeles, February 2022: Verizon
provided fans with a 5G Multiview experience via the NFL Ticketholder
App.
• Fifa World Cup, Qatar 2022: Ericsson and Oreedo delivered seamless,
immersive, high-performance experiences.
• Since 2020, mobile phone app developers have been designing with AR/VR capabilities in mind.
• Android has the biggest AR/VR mobile application ecosystem.
• The following are some of the most popular apps that integrate AR/VR.
Source: https://www.juniperresearch.com/
Core). The 5GC interconnects with an EPC. Source: GSMA 5G Implementation Guidelines July 2019
Note: data correct to January 2022
• In this option, 5G devices must support NR-only technology,
though LTE protocols will be maintained as a fallback.
• The 4G market was built on FR1 in low, mid, and high bands (2.6GHz).
• 5G kicked off the rise in frequency for capacity expansion.
• First 5G wave in FR1 UHB, the sub-6GHz era → N77/N78 bands popular, with N79 drawing less interest.
• Meanwhile, FR2-1, i.,e. mmWave, emerged concurrently.
• Second 5G wave in FR1 UHB, the sub-7GHz era → interworking of NR and Wi-Fi, i.,e. NR-U.
• Further FR2-2 spectrum will be rolled out. Implementation remains to be seen.
• 6G will continue the trend toward spectrum expansion.
• Expansion in FR3, i.e. 7-20GHz, a candidate for urban capacity BUT incumbent services use Satcom (Starlink, OneWeb using Ka-
band).
• Expansion in FR1 in the ultra-low band, a candidate for very large IoT networks. Extends from very large sites to the countryside.
• Making FR4 available in sub-THz frequencies opens up new capabilities for wireless sensing.
6G-related
FR4: > 92GHz
5G-related
FR2-2: 52-71GHz
FR2-1: 24-52GHz
FR3: 7-20GHz
3GHz
5G NR N90,N97 (SUL)
LTE
2GHz
NTN N255, N256
LTE
1.5GHz 5G NTN
5G NR SUL: N81, N82, N83, N89
SUL: Supplementary Uplink 5G NR
LB 5G refarming N71,N12,N28,N20,N5,N8 SDL: Supplementary Downlink
NTN: Non-Terrestrial Network 5G refarming
LTE
0.6GHz
Note: this list is not exhaustive RF Front-End for Mobile 2023 | Report | www.yolegroup.com 114
WORLDWIDE RADIO ALLOCATION FOR 5G
Licensed Shared - unlicensed Considered – under test/auction
N258
N79
• The FCC has already auctioned almost 5GHz of spectrum in the mmWave market:
• Auction 101 2019/01: $700M on 28GHz spectrum
• Auction 102 2019/05: $2B on 24GHz spectrum
• Auction 103 2020/03: $7.5B on 37, 39 and 47GHz spectrums
• More recently, the FCC has moved its focus to the sub-7GHz spectrum, targeting more than 600MHz
for 5G deployment:
• Auction 105 2020/07: $4.5B on 3.55-3.65GHz CBRS PAL: 3.5GHz band use already opened by creating a three-
Recent auctions tiered framework of users: Incumbents, Priority Access (PAL) and General Authorized Access (GAA).
in the C-band
will drive the • Auction 107 2020/12: $81B on 3.7-3.98GHz spectrum: 3.7-4.2GHz has been made available for 5G, making the C-
next wave of 5G band the largest opportunity for the 5G roll-out in the USA. The FCC is accelerating the transition schedule for
deployments. Fixed Satellite Service (FSS) operations to the upper portion of the band (4GHz-4.2GHz).
• Auction 108 2022/08: $428M on 2.5GHz spectrum, dedicated mainly to rural areas.
Source: FCC
• This spectrum is most commonly used around the globe to launch 5G networks. Today, it is considered the most
appropriate spectrum to serve a wide range of use cases cost-effectively.
• GSMA recommends that governments make 2GHz available in the sub-7GHz spectrum in the 2025-2030 time frame
in order to deliver 5G services at a performance level consistent with IMT-2020 (5G) requirements.
• Consequently, most market opportunities will come from this spectrum, explaining the strong interest but high cost
for MNOs. Enhanced mobile broadband (eMBB) and FWA will be the main contributors to value in this area.
Sub-7GHz
• However, the mmWave spectrum is not to be neglected, even though few business opportunities have emerged from
spectrum will
it, as an inflection point around 2028 is to be expected. Indeed, by this timeframe, mmWave will likely be the only way
drive the
to add capacity to meet ever-growing data consumption.
highest values
for 5G in the
next decade,
with eMBB as
the largest
contributor.
mmWave will
come in a
second wave
from 2028
onward.
Source: The socio-economic benefits of mid-band 5G services, GSMA 2022 Source: The socio-economic benefits of mid-band 5G services, GSMA 2022
Note: ‘Mid-band’ is used by GSMA and OEM to refer to the spectrum between 1GHz and 6GHz. It is equivalent to the terminology ‘sub-7GHz’.
• The 6GHz band has been launched, with the standardization of 5G NR band 104 for licensed 5G services in June 2022 as part of 3GPP Release
17. The specification for the full band will be addressed in Release 18 by 2024.
• Nevertheless, countries have started making some decisions on the 6GHz band.
• Full 6GHz (5925-7125 MHz) band made available in the US as unlicensed.
• Europe only allocated a low part of the band as unlicensed.
• China will likely use the full 6GHz band as a cellular licensed band.
Source: ECC Report 302, tentative 6GHz plan, May 2019 RF Front-End for Mobile 2023 | Report | www.yolegroup.com 119
SUB-7GHZ SPECTRUM – 2/2
6GHz band for WiFi6E/WiFi7
• It is important to remember that WiFi offers an offload mechanism to cellular data, carrying most of the traffic. WiFi is designed to coexist
with cellular data and is here to stay.
• The main change from WiFi5 to WiFi6 is the move from OFDM to OFDMA, facilitating more efficient resource sharing. In parallel, signal
modulation order is increasing from 256 QAM to 1024QAM, enabling higher capacity but also posing challenges as better SNRs are required.
This impacts technology as the signal needs to be far more linear.
• Multiuser MIMO is extended on the DL and beamforming is used by the AP.
• Other features improve overall connectivity resiliency and efficiency, such as BSS coloring, or target wake-time.
• Wider 160MHz (previously 80 MHz in WiFi5) channels are becoming available to enable very high-capacity connections. As a comparison,
100MHz instantaneous bandwidth is used in 5G cellular.
• 6GHz band usage for WiFi (WiFi6E) will increase bandwidth capacity.
• WiFi7 will incrementally expand bandwidth and modulation order. WiF7 specification is due for 2024.
• 5G is improving the quality of service delivered to the user by providing larger network capacity, better radio efficiency, and scalable latency.
This is key to securing the long-term viability of current use cases, such as video streaming, while opening the door to new, use cases, such as
cloud gaming. There is growing interest in new use cases at the MNO level. While immersive AR/VR applications are progressing in the
industrial segment, there is still no killer application in the consumer segment. Nevertheless, in the US, operators are pushing for the
adoption of AR and VR to transform the fan experience in stadiums and venues. In China, technology demonstrations from government-led
projects can be seen. Holographic calls have been demonstrated during MWC for the last two years running. These demonstrations show off
5G’s potential and capabilities.
• Besides mobile and consumer applications, Fixed Wireless Access (FWA) has become a strong driver for 5G penetration. FWA offers an
interesting alternative to fiber, especially in rural or suburban areas. For more information on the FWA market, please check our FWA 2023
report.
• MNO strategies for 5G differ between markets, including where they involve spectrum allocation/auctions. MNOs are investing in spectrum
to extend or secure their current assets, thus determining their strategy in the radio network rollout. This is the key driver of the growing
complexity in RF front-end that OEMs must cope with. Besides spectrum expansion, MNOs are also looking to transform their core network
architecture, moving from 5G NSA to 5G SA. However, this aspect does not strongly impact the RF front end, instead having more
implications for the baseband.
• From a regional standpoint, both China and the US rapidly implemented 5G in 2020, though with different strategies.
• China has established a strong rollout policy using sub-7GHz bands: 700MHz, 2.5GHz, 3.5GHz, and 4.8GHz. The next step is the use of the
licensed 6GHz band for cellular 5G, which will likely delay adoption of mmWave in the country.
• In the USA, mmWave was the first 5G platform, explaining the country’s first position in this technology. Aside from this, spectrum refarming
on the low and mid-band has enabled country-wide coverage. The long-awaited release of C-band spectrum is the latest development
receiving attention. As a result, the momentum for mmWave has decreased, though we estimate investments in the technology will
continue.
• Japan and South Korea are very advanced markets, with most of the population covered by the 5G sub-7GHz spectrum. Japan was the first
country where all operators deployed 5G mmWave base stations, whereas Korean MNOs were far behind their objective, with less than 1,000
mmWave BTS at the end of 2021.
• Europe employs low- and ultra-high band spectrum, while interest in mmWave is predicted in some markets such as Finland and Italy.
• India is the last big market to enter 5G, using low and ultra-high band spectrum while eyeing mmWave.
• Finally, 3GPP Release 17 has extended the mmWave spectrum up to 71GHz, providing more available spectrum. This signifies that the
industry will mature to prepare for new use cases. Additionally, new bands have been made available for Non-Terrestrial Networks (NTN).
• Qorvo had great momentum in 2021 with Outer: 2022, $3110M Broadcom
supply for flagship models at all OEMs, Wisol
and stabilized its market share in 2022. Maxscend
25%
• Broadcom has just entered the FEM Others
market, with UHB modules for the iPhone 30%
13, thanks to its flagship filter technology,
21%
and is continuing to grow.
25%
• Other Chinese players are emerging, but
Maxscend remains the domestic leader
for RF modules. It gained 7% market
share in 2022.
• Murata leads the filter market with its 2022 vs. 2021 Filter market share (%)
SAW and MLC technology. Murata
acquired Resonant for its XBAR filter
technology to reinforce its position in the
market.
1%1%
• Qualcomm is maintaining its position in 6%
the discrete filter business, with market
share inherited from RF360. The company 6%
1%3%
is building on this expertise with its ultra- 6%
SAW and newly developed ultra-BAW
filters supporting from 2.7GHz to 7GHz for 6%
cellular and WiFi7. Murata
12% Qualcomm
• Qorvo has a piece of the discrete filter 41%
43%
Qorvo
market, mainly due to its antennaplexing 11%
offering leveraging its BAW SMR Inner: 2021, $2652M Broadcom
technology. Outer: 2022, $2527M Wisol
33%
• Qorvo maintained its leading position in 2022 VS. 2021 TUNER MARKET SHARE (%)
tuner supply to the main OEMs.
Nevertheless, it faced growing
competition and consequently lost 2%
market share in 2022. The company
announced an improved antenna tuning 2% 1%1%
5%
product line, aiming to remain the market
leader and is leveraging its MEMS 6% 1%
3%1%
technology to do so. 5%
• Qualcomm’s share of the PMIC market 2022 VS. 2021 PMIC MARKET SHARE (%)
(ET) is highly correlated with its share
of the baseband market. It is losing
share to Mediatek, which keeps
gaining ground in the mid-tier and
premium segment. The company
launched its seventh generation ET in 10%
2022.
10%
• Mediatek is gaining share as the
11%
company continues to lead in the low- 34%
end and mid-tier segment while its 12%
36%
efforts toward the premium segment Qualcomm
• Qualcomm faced growing competition in 2022 vs. 2021 mmWave components market share (%)
2022 but is still leading the market. It
provides AiP solutions for multiple
mmWave phones. However, Apple does
not use Qualcomm’s off-the-shelf
offering. 1% 3%
> 100 $M
front-end
players are
trying to
> 100 $M > 100 $M > 100 $M
master most of
< 100 $M < 100 $M < 100 $M
the value
chain
internally.
> 100 $M > 100 $M > 100 $M
*
**
PAs
LNAs
* PA driver
** mmWave MMIC
I.H.P SAW
Development
**
* Prototype/Unconfirmed
** For infrastructure
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 140
SILICON FOUNDRY CAPABILITY FOR RF – 1/2
200mm 200 or 300mm 300mm
Node (nm) 350 250 180/150 130/110 95/90 65/55 45/40 28 22 16 14 12
RF CMOS CR013G N28HPC+ 22ULP 16FFC
LNA
RF SOI 0.13SOI
Switch Switch N40SOI
sub-6GHz mmWave 5G, automotive
SiGe
Main foundries for
+ GaN Radio transceiver
RF devices hold a Switch Switch/LNA and PA
large portfolio of RF CMOS 65LP/55LP 40LP/28LP 14LP
technologies. TSMC
has introduced its RF SOI 7RF/7SW 8SW/UC10* UC11/12/13* RF SOI45 22FDX 12FDX
N40SOI technology
and is expected to SiGe 5PAX 7WL 8XP 9HP mmWave
mmWave
start volume + GaN/Si Low power
production this SiGe PA Automotive radar, LiDAR, infrastructure Automotive radar
year.
GlobalFoundries RF CMOS
will increase its
FDSOI capacity RF SOI LNA, switch LNA, switch
with new facility ,
adjacent to ST’s
300 mm existing in RF CMOS TS18 TS13 TPS110 TPS65LP Automotive radar
Crolles, France.
RF SOI SB18HC Switch/LNA
SiGe Switch
SBC18H3 Automotive radar, LiDAR, infrastructure
RF CMOS
0.2µm LNA
RF SOI Switch 0.13µm
Switch
SiGe
* pSemi dedicated process RF Front-End for Mobile 2023 | Report | www.yolegroup.com 141
SILICON FOUNDRY CAPABILITY FOR RF – 2/2
RF CMOS
LNA
On top of RF SOI Switch 0.055µm
0.13µm IPD
increasing their Tuner
capacity, the SiGe
foundries are
expanding their 0.18µm
services, and RF SOI
RF-IPs.
Switch
RF SOI 0.13µm Tuner
Switch
RF SOI 0.18µm 0.13µm Tuner
LNA
Sapphire
The substrate
supply chain for Glass
RF components
is well
consolidated.
Okmetic has LTO/LNO
expanded its
high-resistivity
RFSI ®
LT/LN
substrate
portfolio with
Ultra-Flat
substrates (UF- GaAs
RFSi®) for ML-
SAW filters.
RF SOI
Silicon
* Huntersun changed its name to OnMicro and spun off its filter business under the name Huntersun-MEMS.
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 144
RF FRONT-END COMPANIES IN CHINA
Per company headquarter location
Wuzhou
Acoustic filters
Xiamen/Zhangzhou
Shenzhen/Hong Kong
Chinese fabless
companies have
expanded their
RFFE portfolio
through
modularization OM9902-11 OM9902-11 OM9576-11
and increasing 4G MMBA PA MMBA PA 5G L-PAMiF N77/78
HB/MB/LB, 3G/LTE/NR,HB/MB/LB, 1T2R
complexity. 4HB/5MB/5LB
New modules
could support
advanced 5G
phases 5N and
7, in the UHB
and sub-6GHz.
2018-Q2 2018-Q3 2018-Q4 2019-Q1 2019-Q2 2019-Q3 2019-Q4 2020-Q1 2020-Q2 2020-Q3 2020-Q4 2021-Q1 2021-Q2 2021-Q3 2021-Q4 2022-Q1 2022-Q2
• The RF front-end market reached over $19B in 2021 but flattened in 2022 following demand weakness and inventory
correction.
• In 2021, Qualcomm led the market with its end-to-end approach, followed by Broadcom’s custom PA module offering.
Skyworks and Qorvo have similar company profiles with a broad RF portfolio serving all market segments, although
Skyworks has been more exposed to increased Chinese competition. Murata’s revenue was low, suffering from growing
competition on its filters and restructuring its module portfolio.
• In 2022, traditional players were impacted by the deterioration in the macroeconomic environment. Except for
Broadcom, revenues generated by companies in RFFE-related segments have decreased.
• The Chinese RF front-end ecosystem has been growing, with fabless companies mainly emerging, most taking a share
of their local market. Maxscend, Vanchip, and Smarter Micro are the most relevant examples, though Maxscend recently
invested in upgrading to a fablite business model. There is a long list of companies with great ambitions and access to
financial capital through public offerings on the STAR market. However, not all initiatives will succeed, and we expect
consolidation to happen in the mid-term.
• To sum up, Chinese RF front-end players capture a limited fraction of the market, as OEMs still rely on the leading
players for their premium products, but the largest Chinese companies are on their way to catching up with the leaders.
NR Licensed
LTE/NR
5G operates on
the LTE and NR NR-U
licensed eLWA eLAA
spectrums. The
use of
LWA LAA
unlicensed LTE-U
spectrum is also
5G
very important.
LAA has already
penetrated 4G
handsets and
will be further
used in 5G
handsets. Shared Unlicensed
CBRS, LSA Wi-Fi
MulteFire
Years
20 17 18 19 20 21 22 23 24 25 26 27 28 29 30
R15 R16
Release 17
introduces new
and enhanced
5G devices and
applications,
and extends the
mmWave
spectrum.
System
capabilities
have been
improved with
non-terrestrial
network
enablement
and topology
expansion.
• More MIMO antennas/layers for U: up to eight for single-user MIMO and twelve for
multi-user MIMO. This is dedicated for CPE, FWA, vehicles and industrial devices.
• Frequency-selective precoding in large UL bandwidths.
• Spatial multiplexing and diversity techniques .
• Uplink enhancements: power domain aggregation for CA/DC and enhanced transmit
power efficiency.
• NR multi-band UL carriers: the UE can perform random access on any of these UL
carriers and transmit data over multiple bandwidth parts over multiple bands
According to
without activation/deactivation of secondary cells.
Release 18,
handsets should
operate in NR-DC • NR DC enhancement targeting sub-7GHz and mmWave: Layer 1/2-based intercell
mode. mobility: configuration of multiple cells, dynamic switching, etc.
• NR-DC with selective activation of cell groups via L3 enhancement
• Conditional handover enhancements, including Master Cell Group and Secondary
Cell Group
• Smart repeaters can go a long way towards solving coverage and range challenges for the mmWave spectrum. In addition, they
are synchronized with network gNB equipment, and don’t need fiber backhaul connections.
• Unlike traditional network devices, they dynamically optimize the overall network performance according to changing
environmental conditions and improve point-to-point and line of sight connections by leveraging active phased array antennas.
• Smart repeaters can be used indoors, mounted on utility poles outdoors, or cascaded in a mesh network topology.
Smart
repeaters will
accelerate
the time to
market and
reduce the
cost of
deploying
mmWave
networks.
Sub-7 mmWave
Tx/Rx CP-OFDM
Tx
Waveform Rx
5G sub-7GHz
and 5G
mmWave will Tx DFT-s-OFDM
use different Rx CP-OFDM
(though still Source: Research Gate Source: Research Gate
OFDM-based) Performance Comparison of OFDMA and SCFDMA in LTE Fourier Transform Based Transmission Systems
Systems for Broadband Wireless Communications
waveforms than
LTE and larger 256QAM
Modulation
5 to 100MHz 50 to 400MHz
Sub-carrier spacing
Sub-carrier spacing
Sub-carrier spacing
120kHz
30kHz
5MHz 10MHz 20MHz
Channel
15 kHz
5 to 50MHz 50 to 200MHz
60kHz
15kHz
CA band combination
release 17, the 1000
number of band Inter band 3CC
combinations has
increased by 800
Inter band 2CC
more than four-
fold. This will 600
have a direct Intra band mixed contig/non
impact on the RF contig
front-end in 400 Intra band non contig
mobile handsets.
(b) 200 Intra band contig
(a)
• The sidelink relay can be either a network-to-device relay, where the relay UE
connects a remote UE to the network, or a device-to-device relay, where the
relay UE connects the first remote UE to a second remote UE.
• Release 17 has expanded sidelinks beyond automotive applications, including
device relays that can efficiently extend network coverage. Thus, release 17
Release 17 enables the meeting of key requirements at device level, such as reducing
introduces the latency and power consumption with RedCap devices. Release 17 has also
main features Sidelink Device-to-Device
introduced the NR-L bands to ensure coexistence between adjacent bands in
that will impact
the licensed spectrum.
devices for
sidelink • Release 18 is bringing improved performance, efficiency and further use case
operations, expansion. It supports unlicensed spectrum, multi-beam operation and carrier
mainly through aggregation. The release supports L2/L3 devices for device-to-device relays.
the RedCap
For network-to-device relays, it completes some features needed to cover
concept and NR
mobility scenarios not covered in release 17. This release also introduces
Light bands.
Release 18 advanced positioning.
Sidelink Network-to-Device
expands sidelink
capabilities with
CA and support
for unlicensed
spectrum.
• 4x4 MIMO DL • Band > 2.5GHz • 4 Rx chains (LNA, switch, filter) needed
• 2x2 MIMO UL (N77,N78,N79,B41/B7/B39) • 2 PA modules for these bands (*)
• Two coherent Tx • Band N41,N77,N78,N79 • PA integration in the diversity path (**)
• Diversity Tx
(*) Though a 2x2 MIMO UL is preferred by the carriers, some companies are validating a single Tx link as an adequate solution for low-cost RFFE, e.g. China Mobile.
(**) A diversity Tx link would improve UE transmit gain. However, no commercial release of this feature has been seen so far. We expect Apple to be one of the OEMs to first
implement this feature in its iPhones.
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 166
HIGH-POWER USER EQUIPMENT – POWER CLASS 2
A must for 5G NR TDD bands
RFFE block
> 3dBm > 31dBm
• This approach enables better performance for the transmit gain, thus resulting in a higher signal-to-
noise ratio. However, it increases the bill of materials and poses additional challenges for smartphone
OEMs for integration in phone cases.
This option is • UL MIMO operation can be applied on top of diversity Tx to enable UL and DL data rate
preferred by improvement.
network
operators as it • To reduce the BOM, the second UL path and the diversity Tx can be mutualized. In some cases, for
provides better entry-level phones, i.e., phones below 3000 RMB or $460, this feature is not necessary.
performance and 23dBm
is likely to reduce
the cost of
network RFFE block
deployment. < 31dBm
However, it poses
big challenges to
smartphone PA Switch Filter
Note
OEMs regarding
integration and < 3dBm 23dBm In some cases, HPUE is required for
cost of this configuration, meaning the use
implementation. of two Tx RF chains at 26dBm, as in
BB TRX RFFE block
the Sprint network, for instance.
< 31dBm
PA Switch Filter
• The dual approach is most beneficial from a data-rate point of view. Reports from the field show that 5G
NSA is scaling very rapidly and will probably support over a billion subscribers by the end of this year.
• The majority of 5G SA deployments today are confined to a handful of companies, such as T-Mobile,
Vodafone Germany, the Netherlands, and Korea Telecom. The main MNOs are currently employing a wait-
and-see approach.
5G NSA is
currently • The MNO migration from 5G NSA to SA will undergo several phases. LTE and NR UEs will co-exist in the NSA
scaling very scenario. Local or regional SA UEs are expected in some regions, where the users will switch from NR to LTE
rapidly. if roaming or outside of NR coverage.
Several
countries are
starting with
this approach
before
transitioning
to 5G SA in
the future.
Source: Samsung
Migration roadmap from 4G to 5G
• E-UTRAN New Radio Dual Connectivity (EN-DC) technology enables the use of both
4G and 5G to improve coverage and reliability from a UE perspective but also reaches
higher DL data rates.
• Mobile devices that support EN-DC will have two concurrent radio connections to the
EPC, one via 5G NR and the other via LTE.
• Disparities between 4G and 5G coverage are likely due to higher frequencies often
used in 5G. Source: GSMA
Various companies have demonstrated EN/DC capabilities through dedicated tests. Some illustrative
examples are included here:
1- Ericsson and Qualcomm have demonstrated their Radio System mmWave Streetmacro 6701 base
station, Snapdragon X60 5G Modem-RF system, and third-generation QTM535 mmWave antenna
modules using a form-factor test smartphone. The demonstration combined eight 100MHz mmWave
EN-DC tests
channels in the n257 (28GHz) band, with two 20MHz of 4G LTE CA for a total aggregated BW of 840MHz
have to deliver a 5Gbps data rate for a single user.
demonstrated
increased
speeds and
capacity,
leveraging the
4G core
network.
2- Samsung surpassed the symbolic 5Gbps threshold to reach 5.23Gbps in March 2021, using commercial
devices, a BTS, and a Samsung 20+ phone in their lab. They also combined eight 100MHz 28GHz
mmWave channels with two 20MHz of 4G LTE.
1600
1400
1200
Band combinations
• The number of inter-band combination in 1000
number of band
combination of E-UTRA bands with FR1. The
number of possible combinations increases
combinations. 600
further to 4201 with FR2 and 2053 when
combined with FR1 and FR2. These numbers 400
represent a six-fold increase compared to the
number of configurations in release 16.
200
• With this increased number of combinations,
release 17 can leverage the highest bandwidth 0
2 3 4 5 6
from LTE and 5G networks and increase Inter band EN-DC configurations
aggregated speed. Modems inside number E-UTRA + FR1
503 1526 1360 460 35
• Contrary to EN-DC, which aggregates LTE and 5G to enhance performance, New Radio Dual
Connectivity (NR-DC) is unique to standalone 5G, connecting the master node to the 5G Core
Network (5G CN).
• It can connect to two different NR base stations. For example, one covering sub-7GHz, and another
using the mmWave spectrum. It significantly improves network performance, with higher data rates
NR-DC is (UL and DL) and better coverage without LTE involvement. Combined with SA capabilities such as
expected to ultra-low latency, network slicing, and improved security, NR DC will push the boundaries of 5G
combine FR1
technology.
and FR2
bands,
delivering an
increased
data rate in
addition to SA
capabilities:
low latency
and network
slicing.
5G NR-DC
(*) RAN products used in the demo: AIR 6449 (sub-6GHz) and Streetmacro 6701 (mmWave)
Source: https://www.ericsson.com/en/news/2021/4/ericsson-mediatek-dual-connectivity-5g-standalone-path
• In the case of two contiguous LTE and NR tones transmitting a signal, low Inter-Modulation Distortion
(IMD) is expected, resulting in low desensitization, i.e., noise on the receive band.
• The problem is when two non-contiguous LTE and NR tones transmit a signal. Then, noticeable IMD
Dual appears, which could desensitize one of the receive bands, thus degrading the downlink
connectivity
performance.
will be
imposed by
carriers as per
their spectrum
allocations. RF
front-end
players will
have to cope
with
desensitization
issues in the
design of
modules and
components.
• Line-Of-Sight (LOS) propagation: 28GHz and 39GHz are subject to 21dB and
24dB higher LOS, respectively, than the 2.6GHz band.
• Rain attenuation: According to the FCC’s Office of Engineering and
Technology Bulletin on mmWave Propagation, rain attenuation ranges
from 0.05dB/km to 25dB/km (@28GHz) and 0.08dB/km to 35dB/km
(@39GHz). This translates to up to 2.5dB (@28GHz) and 3.5dB(@39GHz) of
rain attenuation for every 100 meters with a rainfall intensity of 150mm/h.
• Foliage attenuation: According to the FCC bulletin referenced above,
foliage attenuation at a 10-meter depth is estimated to be 17dB on the
28GHz band, which is 11dB higher than 2.6Ghz. For the 39GHz band, this
attenuation is 19dB.
• Building Penetration Loss (BPL): is higher in commercial buildings than Source: GSA, November 2022
residential due to building materials and window isolation techniques.
28GHz is subject to 6dB to 14dB higher BPL over 2.6GHz. 39GHz is subject
to 8dB to 17dB higher BPL.
Source: Qualcomm
Source: Qualcomm
Source: Qualcomm
Source: Qualcomm
V
FR2
Dual PA module
Ka with different
band
combinations
The RF front-
end
UHB
architectures
from the
iPhones 12 to 14
are superficially HB
very similar.
FR1
MB
LB
IPD
FBARFilter FBAR
FBAR Filter FBAR Filter
filter
Filter filter
Filter
BAW Filter
IPD
Filter Filter
Filter
A deeper look Filter
BAW
Filter
Filter
Switch
Switch
inside Switch
modules PA
Switch
LNA
bank
PMIC
highlights the PA
Switch
LNA
bank
PMIC
growing bill of
materials. Switch
Avago AFEM8240
#dies: 14
Qorvo QM 76285
#dies: 11
PA LNA PMIC
bank
SKY 58245
#dies: 5
LNA
> 100GHz InP
PA
FR2-1 RFSOI
MMIC
20-52GHz CMOS FDSOI
FR3
GaN/Si
PA
7-20GHz
BAW SMR/FBAR
XBAR
Filter
ML-SAW
SAW/TC SAW
BST MEMS
Sub-7GHz
Switch Tuner
CMOS/RFSOI
FR1
P-Si?
GaAs CMOS/RFSOI
SiGe BiCMOS
LNA
GaAs CMOS/RFSOI
GaN/Si?
PA
GaAs
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 190
POWER AMPLIFIER CLASS AND TECHNOLOGY
Which type of PA are used for mobile handsets?
• Peak to average ratio (PAPR) describes the signal variation between its peak and average power levels; it is usually used
to evaluate the output envelope variations. As such, it is defined as the ratio between 𝑃𝑃𝑒𝑎𝑘 , the maximum power and
𝑃𝐴𝑣𝑔 , the average power value of the transmitted signal.
𝑃𝐴𝑣𝑔
𝑃𝐴𝑃𝑅 =
𝑃𝑃𝑒𝑎𝑘
• This parameter is of importance in recent OFDM and OFDMA modulations due to the addition of a larger number of
independently modulated sub-carriers.
• In the case of subcarrier waveform constructive addition, peaks at much-higher-than-average levels are formed in the
output envelope.
• Furthermore, the operation at high PAPR forces the PA to operate at a large OBO to support the strict linearity
requirements, which leads to a significant decrease in efficiency. This is less acceptable for higher output RF power levels
that cause more power wastage and consequently necessitate complex and more expensive heat transfer solutions.
• Higher power devices with a large dynamic range, the improvement of clipping and filtering or crest factor reduction
techniques, or OFDM hybrid options can be used to accommodate linearity and efficiency requirements.
Broadband
and/or
reconfigurable
PAs are
required to
fulfill the
current FBW
requirements.
FBW: Fractional Bandwidth
The recent 3GPP releases significantly increased the number of bands and their bandwidth, which led to
increased constraints on PAs. Nowadays, broadband and/or reconfigurable PAs are required to fulfill
these FBW requirements. This opens the way for various approaches, such as load modulation and
Doherty (and pseudo-Doherty) architectures in mobile handsets.
• Doherty PAs have a proven track record of efficiency in infrastructure and for high-power
applications. They may make a comeback in 5G mobile handsets, as both the level of output power
(HPUE) and PAPR constraint increases while requiring better efficiency for lower battery usage.
• For example, recent academic works have described a reconfigurable PA covering the 2.6GHz band
with a 34% FBW and -35.7dBc ACLR, and a 31dBm average output power reaching 42% PAE without
DPD using SOI technology.
Doherty PAs
are
considered
candidates
for high-
efficiency
HPUE.
Source: EUMW2020
• Whether discrete or integrated into a module, the LNA for 5G must accommodate higher frequency
than for LTE.
• Obviously, integrated LNAs provide better loss performance at system level, especially at higher
The design of frequencies such as 3.5GHz.
5G LNAs has • In any case, the LNA noise figure must be improved as frequency increases:
been
anticipated by
the
semiconductor
industry, which G: LNA gain
already has IIP3: LNA linearity
optimized f: Frequency of operation
designs for Wi- F: Noise figure
Fi at 5GHz. We PDC: DC power consumption
believe CMOS
SOI will prevail
as it offers an
• SiGe HBT and SOI CMOS technologies compete in this area. SiGe offers better gain and noise figures
interesting
integration while SOI CMOS offers better linearity. Typical LNA figures of merit for both current technologies in
path. production are in the range of 200 at 5GHz (Wi-Fi).
• However, SOI CMOS scaling enables not only an LNA performance improvement, but also the ability
to integrate switches with LNA onto a single die, resulting in a better form factor.
As a
• Further RF SOI improvements in design and scaling are 800
mainstream
ongoing worldwide in foundries. RF SOI should reach a
technology, 700
RFSOI performance plateau towards 2025. psemi
600 TowerSemi
performance • PCM (Phase Change Material) also offers an interesting
RonCoff (fs)
improvement is technology that would enable further integration of switches 500
ST Microelectronics
ongoing. MEMS directly in the backend. As of today, we have not seen any sign GlobalFoundries
finally hit the 400
of maturity of this technology developed by TowerSemi. PCM Infineon
market and is
expected to
uses manufacturing processes reaching state-of-the-art below- 300 GaAs
emerge in 2023 10fs RonCoff and record cycling endurance. Prior to this MEMS
200
for next- demonstration, PCM switches were only investigated by PCM
generation research institutes such as XLIM or HRL. 100 Magnachip
tuners.
0
2005 2015 2025
Timeline
• 𝐹𝑚𝑎𝑥 is a critical parameter for transistor and technology choice; it defines the frequency of 0dB power
gain. RF applications are usually developed below the transistor 𝐹𝑚𝑎𝑥 /3 frequency, as a rule of thumb.
• 𝐹𝑚𝑎𝑥 is expected to gain importance for future generations of wireless communication beyond sub-
7GHz 5G, where some advanced nodes such as CMOS (e.g. RF/FD/PD SOI) 22 or 28nm nodes can go
as high as 90GHz for RF FEM applications. SiGe HBT technologies are also a crucial contender, as they
provide higher 𝐹𝑚𝑎𝑥 . For example, the IHP 0.13µm SG13G2 BiCMOS process has a 𝐹𝑚𝑎𝑥 of 450GHz.
• A 𝐹𝑚𝑎𝑥 larger than 0.5THz is expected to become a standard requirement for some FEM applications
𝐹𝑚𝑎𝑥 is a in five to ten years. This might go higher with usage of THz in 6G.
critical
technology
parameter
that will play
an important
role beyond-
5G
development.
• Coexistence is a well-known issue for Wi-Fi 2.4GHz and cellular bands, requiring tight filtering.
• Some 3.5GHz bands will already pose coexistence challenges (B42 in Japan, CBRS and C-band in the USA, Wi-Fi 5GHz
and Wi-Fi 6GHz in the USA), leading to similar cases requiring high-performance filtering.
Legend
! Means less than 200MHz between bands
3000
3200
3400
3500
4300
5300
5500
5700
6100
6600
6900
7000
3800
2500
2700
3600
3900
4500
4700
5000
5200
5400
5600
5900
6300
6500
6700
5100
7100
2400
2600
2900
4000
4200
4400
4600
4900
5800
6000
6200
6400
2800
3100
3700
4800
6800
3300
900 N77
by BAW and
FBAR
technologies FBAR XBAW
along with MLC
and IPD. Multi-
layer SAW
technology will 200 Multilayer N41
N4
also have 100 SAW 0
opportunities. N66
75 N25 N7
N3
50 N28 N2 N1 BAW SMR
25 N71
N5
SAW
10 TC SAW
Frequency (GHz) 1 2 3 4 5 6 7
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 201
SURFACE ACOUSTIC WAVE FILTER TECHNOLOGIES
Principles, strengths and weaknesses
Fabless
companies
Filter
foundries
Substrate
providers Manufacturing
equipment Testing
equipment
Filter
IDMs
• The primary impact of 5G is the introduction of new radio bands over a wide range of frequencies.
This comes together with MIMO and CA requirements while keeping previous connectivity on, i.e.
3G/LTE, GNSS, and Wi-Fi/BT. Therefore, the number of antennas will soon increase from between two
to six, to between six to ten, but with increasing space constraints.
• As a result, antennas need to become smaller, so antenna efficiency and bandwidth are dropping.
Antenna tuning mitigates this effect by enabling the antenna to reach the right performance point
for each frequency band.
5G will impact
the number
of antenna
tuners
required in
mobile
phones.
• The goal of antenna tuning is to improve antenna efficiency. With the ever-smaller phones, the
antenna size, as well as its positioning in the phone continues to be subject to new constraints, and
this results in decreased overall efficiency at the antenna system level.
• To improve this efficiency and to allow a single antenna to address multiple bands, antenna tuning
has become a must-have.
• Antenna efficiency is divided into two components: the antenna’s radiation efficiency, which is
strongly linked to match the antenna’s electrical length and the signal wavelength, and mismatch
Two
losses, which happen at the antenna feed when the antenna impedance does not match the RF line
components
impedance.
of antenna
efficiency are Antenna efficiency = radiation efficiency + mismatch losses.
addressed by
two different • To reduce mismatch losses, impedance tuning is possible, and it allows adaptation of the antenna
tuning impedance to the signal frequency and improves transmission efficiency.
approaches. • In order to improve radiation efficiency, aperture tuning is possible. In this case, the antenna’s
‘electrical length’ is adapted to the signal wavelength.
• Advanced phones include aperture tuning as the multitude of frequency bands, as well as the need
for simultaneous transmission for carrier aggregation of different frequency signals make the
frequency matching in the antenna a critical aspect for data transmission.
• Most advanced phones include both approaches to optimize total antenna efficiency.
• The condition for resonance in a monopole antenna is for the element to be an odd multiple of a
quarter-wavelength, λ/4, at least for its electrical length. The electrical length can differ from the
physical length. As an example, the presence of a nearby object creates a parasitic capacitance and
will decrease the velocity of the wave, thus increasing the electrical length of the antenna.
• In practice, it is possible to modify the electrical length of an antenna, either by adding a ‘loading coil’
in series to electrically lengthen the antenna or by adding a capacitor in series to electrically shorten
it.
Aperture
tuning is • Aperture tuning is based on optimizing the electrical length of the antenna. A multiple-throw switch
performed by is linked to the antenna, and can add capacitance in series, an inductance, or shorten the system in
modifying the order to tune the antenna to the right resonant frequency.
antenna’s • Antenna tuning for multiple frequencies is also challenging, as the antenna electric length must
electrical
match two different frequencies. In practice, the lengthening or shortening effect of the added
length.
inductance/capacitance is also correlated to its position on the antenna. Each resonant frequency has
a voltage peak at resonance on the antenna. Placing a tuning element close to the voltage peak will
have a strong effect. On the other hand, placing it close to a zero voltage will reduce its effect.
Therefore, to offer two different antenna lengths depending on the required frequency, a clever
placement of the tuners is necessary in order to place each one of them closest to the desired
frequency peak, and the farthest from the unwanted zero frequency.
• This way, it is possible to optimize the antenna for a multitude of frequencies, and therefore offer a
large bandwidth for its use.
• The main characteristics of aperture tuning elements are linked to their behavior in both the off-state
and the on-state, as well as their linearity and peak power handling.
From Qorvo’s perspective, the main characteristics to look for are as follows:
• Tuner switches:
Antenna tuners • Use switches with low RON and COFF to minimize the losses in the system.
are like a switch
regarding the • Use high-linearity tuner switches to avoid impact on radiated spurious emissions (RSE) and
RonCoff TIS.
performance,
except for the • Switches must be multimode for tuning 2G/3G/4G/5G standard frequency ranges.
fact that it must
• Switches should be capable of handling high RF voltages for broadband antenna applications.
tolerate a
higher voltage • Tuning components:
range, so a
specific design • Use capacitors with a capacitance value greater than 0.5pF to avoid using a component with
is required. high tolerances.
• Avoid using inductors with an inductance value greater than 36nH.
• Mobile smartphones and handsets generally are limited when it comes to battery size and life, and
thus efforts are necessary to take advantage of the available resources.
• Nowadays, 4G/5G PAs face higher challenges than previous generations when it comes to linearity
and efficiency. Firstly, the increasingly complex modulation schemes require higher PAPR; this ratio
has an important impact on PA efficiency. Considering that a PA is most efficient when running at
maximum output power, its efficiency decreases for lower powers. Another factor that can
significantly reduce PA efficiency is its bandwidth, where covering a broader frequency range tends
The number to reduce its efficiency.
of bands and
PAs will • Digital pre-distortion (DPD) can reduce AM-AM distortion by pre-distorting the signal, i.e. introducing
significantly an ‘invert distortion’ at baseband level and before entering the RF transmit path. When this signal is
impact fed into a PA operating in compression, the output signal is correct. This technique allows use of the
battery PA in the compression region and the improvement of its efficiency.
lifetime. • However, this technique relies on powerful signal processing and is far from optimal for all small
handsets. Average power tracking and envelope tracking were introduced to optimize the handset’s
battery life.
• APT determines the appropriate supply voltage to avoid wasting energy when the PA is running
below its maximum output power, while preserving linearity on the whole bandwidth.
• In practice, APT adjusts the PA voltage supply to function as close as possible to the compression
point, controlled by a slow protocol that adapts the voltage supply periodically. It does not change
until the next 3GPP control slot.
Source:Qualcomm
• This solution is suitable for large operating bandwidth at the expense of some efficiency, i.e. some
energy is still dissipated as heat. A typical BW of 100MHz is currently required and power
requirements are expected to increase in the future.
• Envelope tracking is another solution to improve the RF PA efficiency i.e., improve its capability to
convert DC power to RF power, and thus reduce heat dissipation.
Envelope
tracking is
more efficient
than APT but
has BW
limitations.
• The main idea is to deliver the appropriate DC power to the PA to get the maximum output RF power
instantly and continuously. Envelope trackers, or supply modulators in general, dynamically modify
the voltage supply to track the RF signal envelope.
• This optimizes the DC power delivered and avoids unnecessary heat dissipation like APT. Its ability to
rapidly change allows it to be more efficient than APT.
• Nowadays, this is the go-to solution and can be combined with APT in some cases.
The evolution of
power • A large ET BW is important to track signal amplitude
management is
closely related to without distortion. An ET BW two to three times the
100MHz ET
the trends in RF channel bandwidth is used, as a rule of thumb.
components, first
driven by larger
bandwidth (i.e.
CA and 5G • Thus, 60MHz ET BW was sufficient for LTE, while a 100MHz 60MHz ET
bands), better BW is necessary for 5G NR applications. Larger ET BW will
linearity and be needed for future 5G NR releases. This is mainly driven
lower power
consumption.
by carrier aggregation, but also due to the fact that some
40MHz ET
There is now an MNOs secured large spectrum bands, such as China
additional need Mobile (160MHz), while China Unicom and China Telecom
to boost the PA have a combined 200MHz.
performance 20MHz ET
(HPUE).
APT only
• The capability of a supply modulator to handle 2G/3G/4G/5G PAs, or at least several standards
simultaneously, will become a necessity for future designs, opening the door for better circuit
integration.
• High output voltage is also an essential factor as it allows for a higher output power for PC2 and PC1.5
HPUE power levels, and better PA efficiency.
• Recent research has shown that an NR 130MHz ET BW is achievable using the hybrid switching
supply modulator approach, with an output voltage supply reaching 5.8V using a 3V battery.
Recent research • Better overall performance can be attained with higher nominal voltage batteries. However,
has shown that nowadays the conventional Li-ion batteries used for consumer applications cannot go much higher
an NR 130MHz
than 3.8V. Placing batteries in series seems to be too complex a solution for handset applications at
ET BW is
achievable. the moment.
• A coupler can be regarded as a power measurement device. It is a passive circuit that often
comes in the form of an MLC or IPD device.
• It samples a fraction of the RF power passing through it depending on the coupling value
(i.e., 3dB, 10dB etc.) of the device.
• This component is generally used at the output of a PA module and fed to the baseband as a
means to control the transmitted signal power and to correct it for linearity dynamically in
the baseband. This operation is known as DPD (Digital Pre-Distortion) and is frequently used
in mobile phones.
• Dual-directional couplers are currently used in high-end phones to enable the monitoring of
forward and reverse power simultaneously. Monitoring the reverse power mode helps to
dynamically select the best radio link for the receive side.
• Dual-directional couplers are starting to become integrated in silicon-based LNAs/switches
as a high Q-factor becomes available on this technology platform.
• 5G is improving the quality of service delivered to the user by providing larger network capacity, better radio efficiency, and scalable latency.
This is key to securing the long-term viability of current use cases, such as video streaming, while opening the door to new, use cases, such as
cloud gaming. There is growing interest in new use cases at the MNO level. While immersive AR/VR applications are progressing in the
industrial segment, there is still no killer application in the consumer segment. Nevertheless, in the US, operators are pushing for the
adoption of AR and VR to transform the fan experience in stadiums and venues. In China, technology demonstrations from government-led
projects can be seen. Holographic calls have been demonstrated during MWC for the last two years running. These demonstrations show off
5G’s potential and capabilities.
• Besides mobile and consumer applications, Fixed Wireless Access (FWA) has become a strong driver for 5G penetration. FWA offers an
interesting alternative to fiber, especially in rural or suburban areas. For more information on the FWA market, please check our FWA 2023
report.
• MNO strategies for 5G differ between markets, including where they involve spectrum allocation/auctions. MNOs are investing in spectrum
to extend or secure their current assets, thus determining their strategy in the radio network rollout. This is the key driver of the growing
complexity in RF front-end that OEMs must cope with. Besides spectrum expansion, MNOs are also looking to transform their core network
architecture, moving from 5G NSA to 5G SA. However, this aspect does not strongly impact the RF front end, instead having more
implications for the baseband.
• From a regional standpoint, both China and the US rapidly implemented 5G in 2020, though with different strategies.
• China has established a strong rollout policy using sub-7GHz bands: 700MHz, 2.5GHz, 3.5GHz, and 4.8GHz. The next step is the use of the
licensed 6GHz band for cellular 5G, which will likely delay adoption of mmWave in the country.
• In the USA, mmWave was the first 5G platform, explaining the country’s first position in this technology. Aside from this, spectrum refarming
on the low and mid-band has enabled country-wide coverage. The long-awaited release of C-band spectrum is the latest development
receiving attention. As a result, the momentum for mmWave has decreased, though we estimate investments in the technology will
continue.
• Japan and South Korea are very advanced markets, with most of the population covered by the 5G sub-7GHz spectrum. Japan was the first
country where all operators deployed 5G mmWave base stations, whereas Korean MNOs were far behind their objective, with less than 1,000
mmWave BTS at the end of 2021.
• Europe employs low- and ultra-high band spectrum, while interest in mmWave is predicted in some markets such as Finland and Italy.
• India is the last big market to enter 5G, using low and ultra-high band spectrum while eyeing mmWave.
• Finally, 3GPP Release 17 has extended the mmWave spectrum up to 71GHz, providing more available spectrum. This signifies that the
industry will mature to prepare for new use cases. Additionally, new bands have been made available for Non-Terrestrial Networks (NTN).
• The cell phone market recovered in 2021 following the dip caused by the Covid-19 pandemic in 2020. However, pre-
Covid-19 levels have not been reached due to chip supply shortages.
• In 2022, the smartphone industry was seriously impacted following a global macroeconomic downturn: a market decline
with high inflation caused by geopolitical tensions such as the Russia-Ukraine war and tensions between China and
Taiwan. This downturn resulted in consumer hesitancy in purchasing new phones, thus pushing OEMs to enter an
inventory correction phase. The Zero-Covid policy in China further destabilized the smartphone manufacturing industry.
• Despite these challenging conditions, 5G phone production reached parity with 4G phone production in 2022, though
we believe the market penetration of 5G would have been higher without the aforementioned external factors. 5G
resilience can be explained by the fact that OEMs have prioritized their premium segments.
• We expect the mobile market to be flattish in 2023, as we don’t see any short-term improvement in the global economic
situation and, thus, no impetus for recovery.
• In the longer term, growth should come from emerging countries such as India, which might offset the Chinese market
decline over the next few years.
• The RF front-end market made a leap forward in 2021 as an effect of post-Covid-19 recovery and 5G penetration.
• CY2022 ended up flat following the smartphone market decline associated with lower-than-expected 5G penetration.
Consequently, the bill-of-material growth engine has been in low gear.
• During the last two years, 5G has evolved, bringing increasing complexity from, for instance, the need for MIMO UL/DL
support, an increasing number of CA combinations, and more coexistence scenarios to manage. Consequently, the bill of
materials has increased.
• The 4G semiconductor technology portfolio has evolved to accommodate 5G specifications. For instance, GaAs-based
PAs are still used for mobile phones, RFSOI remains the technology of choice for switches, LNAs, and antenna tuners,
and SAW/BAW technologies remain the filter standard. What is noticeable is the increase in the number of variations of
such technology. Take SAW-type filters, for example, for which bonding techniques are used to improve RF performance
while minimizing size.
• New technologies are now emerging, such as the long-awaited MEMS solution for antenna tuning that hit the market in
mid-2022, introduced by Qorvo, and FDSOI, which is becoming mainstream for mmWave transceivers.
• Meanwhile, other technologies are phasing out, such as BST for antenna tuners, while other technologies at the
development stage are trying to position themselves to tackle the market. Some examples are porous silicon, which
would provide improved switch linearity performance to the RF front-end industry, or GaN/Si PAs with state-of-the-art
power-added efficiency and wideband operation capability.
• The RF front-end market is still led by large players, while some competition is emerging from China. However, all
leading players have faced reductions in their revenue from the smartphone market.
• On the competition from China, the growth of these fabless/fablite players is mainly coming from the local market pull
at this stage.
RF Front-End Module
Contact our RF for Fixed Wireless Access
Comparison 2022 – Vol. 1 –
Sales Team 2023 – Focus RF Front End
Focus on Apple
for more
information
Contact our
Consumer > Phone
Sales Team
for more
information
Yole Group, including Yole Intelligence, Yole SystemPlus and PISEO, are pleased to
provide you a glimpse of our accumulated knowledge.
Please note that the entire document and its data is copyrighted © Yole Intelligence
2022.
To share our data with your own network, within your presentations, press releases,
dedicated articles and more, please contact our Public Relations department to make
sure you get up-to-date, licensed materials.
We will be more than happy to provide you our latest results and appropriate formats
of our approved content.
Yole Group
Corporate
Presentation
• Electronic Systems
• Emerging Technologies
CUSTOM
REPORTS MONITORS TRACKS
SERVICES
Insight Insight Insight Insight
› Yearly reports › 4 times per year updated market › Teardowns of phones, smart › Specific and dedicated projects
› Market, technology and strategy data and technology trends in home, wearables and › Strategic, financial, technical,
analysis units, value at wafer level automotive modules and supply chain, market and other
› Reverse costing and reverse › Direct access to the analyst systems semiconductor-related fields
engineering › Bill-of-Materials › Reverse costing and reverse
› Performance analysis Format › Block diagrams engineering
Format SERVICE
› Excel files with data
› PDF files with analyses graphs Format Format
› PDF files with analyses
› PDF files with analyses and key facts
› Excel files with graphics and
data
S
› Web access
› Web access
› PDF and Excel files
› Excel files with graphics and data
› High-resolution photos
› Web access Topics Topics
› Advanced Packaging › Photonics, Imaging & Sensing
Topics Topics › Lighting & Displays
› CIS
› Photonics, Imaging & Sensing › Compound Semiconductor
› Consumer: Smartphones, smart › Power Electronics & Battery
› Lighting & Displays › DRAM & NAND home, wearables › Compound Semiconductors
› Power Electronics & Battery › Micro-controller › Automotive: Infotainment, › Semiconductor Manufacturing
› Compound Semiconductors › Processor ADAS, Telematics and Packaging
› Semiconductor Manufacturing › RF › Computing & Memory
and Packaging › Semiconductor Test Equipment
› Computing & Memory › Semiconductor Subsystems
› Wafer Start
115+ reports per year 11 different monitors 220+ teardowns per year 190 custom projects
quarterly updated Daily updates per year
London Frankfurt
Boise
Nantes
Lyon Seoul Phoenix Boston
Austin
Raleigh
Shanghai Tokyo
Hsinchu
Melaka
25 years in the
120+ annual
semiconductor
conferences
industry
Our unique
position
allows us to
obtain 5,000 players 1250+ teardown
detailed and
accurate
interviews tracks available
information per year
to meet your
needs.
6,800+ 100+
companies’ analysts
news relayed worldwide
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 228
A UNIQUE AND PROVEN METHODOLOGY
BOTTOM-UP, TOP-DOWN AND
MARKETING EXCELLENCE INDUSTRIAL EXPERTISE
AND BEST-IN-CLASS NETWORK
• Top-down
• Market segmentation ➢ End market demand analysis
➢ Per application ➢ Market forecasts at system and
➢ Per technical needs component levels down to wafer and
➢ Per technology adoption and equipment
supply chain’s tendencies • Bottom-up
• Primary research and direct ➢ Ecosystem analysis
interviews with key players ➢ Consolidate industrial players’ revenue at
component, module and system levels
• Industrial experts in all our fields of
investigation
STATE-OF-THE-ART TECHNOLOGY
“Thanks to its unique semiconductor AWARENESS
market intimacy, its understanding of the • Technology analysis
industrial environment and its vision on ➢ Competitive landscape and technology comparison
future technologies adoption, Yole Group ➢ Reverse costing
➢ Reverse engineering
supports its customers at every stage of
their growth” • Technology life cycle
➢ Development cycles
➢ Supply chain adoption
➢ HV manufacturing and evolutions
• Performance testing and analysis
RF Front-End for Mobile 2023 | Report | www.yolegroup.com 229
OUR NETWORK IS THE ENTIRE SUPPLY CHAIN ACROSS 6 MARKETS
6 KEY MARKETS
Academic/ Design & Material & Front-end Back-end OEMs & Consulting &
research engineering equipment manufacturing manufacturing system finance
integrator
NORTH AMERICA
[email protected]
+1 833 338 4999 CUSTOM PROJECT SERVICES
Yole Intelligence General terms
[email protected] | +33 6 27 68 69 33
and conditions
EMEA Yole SystemPlus
[email protected] of sales
[email protected] | +33 2 72 17 89 85
+49 69 9621 7675