eBook

Innovations in Phased Array

Antennas for 5G
March 2019

S P O N S O R E D B Y
Table of Contents

3

Introduction
Gary Lerude
Microwave Journal, Technical Editor

4

Strong Defense Outlook Offers Continued Growth for
RF Technologies
Asif Anwar
Strategy Analytics, Newton, Mass.

8

Spatial Multiplexing for 5G Wireless Communications
Honglei Chen and Rick Gentile
MathWorks, Natick, Mass.

13 Phased Array Antennas & The Roadmap to 5G Wireless

MACOM

15

Tile Arrays Accelerate the Evolution to
Next-Generation Radar
Doug Carlson
MACOM, Lowell, Mass.

19 Breaking Through the Cost Barrier for Phased Arrays

Doug Carlson
MACOM, Lowell, Mass.

22

Software and Hardware Near-Field Transformations
for 5G OTA Testing
Benoît Derat, Corbett Rowell, Adam Tankielun and Sebastian Schmitz
Rohde & Schwarz, Munich, Germany

26

Sub-6 GHz mMIMO Base Stations Meet 5G’s Size and
Weight Challenges
Walter Honcharenko
MACOM, Lowell, Mass.

2
Introduction

Innovations in Phased Array Antennas for 5G

In just a few years, 5G has morphed from a distant wireless concept being debated by arcane 3GPP
committees to a consumer brand, as fixed wireless access and mobile services begin to roll out globally. 5G
has been described as revolution, not evolution, requiring the end-to-end network be transformed to support
the use cases envisioned for 5G:
• Enhanced mobile broadband (eMBB), with date rates above 10 Gbps.
• Ultra-reliable and low latency communications (uRLLC), with availability of 99.9999 percent and air
interface latencies under 1 ms.
• Massive machine-type communications (mMTC), supporting at least 1 million low data rate devices
within a square kilometer.
Achieving 5G’s performance goals requires many enablers, including additional spectrum, optimized
radio waveforms, a greater density of base stations and antennas with spatial diversity, the latter to cover
more users from each cell site. For the sub-6 GHz spectrum, the antenna architecture is called massive MIMO
(mMIMO), referring to the relative number of transmit and receive channels. Where LTE-Advanced uses 8 x 8
MIMO, mMIMO jumps to 64 or higher. Base stations operating at millimeter wave frequencies have the added
challenge of path loss, requiring similarly large arrays — 64, 128 or 256 elements — steering multiple beams
to achieve the coverage and capacity.
Regardless of frequency, these arrays borrow from the RF and beamforming technology of active phased
arrays developed for defense applications, now the standard architecture for military radar and many
electronic warfare and military communications systems.
This eBook explores innovations in phased array systems and their application to the 5G radio access
network (RAN), beginning with an article by Strategy Analytics, discussing how the military market will
continue to feed technology for commercial systems.
An article written by MathWorks introduces the theory of spatial multiplexing, laying the foundation for
beamforming arrays.
Several articles contributed by MACOM discuss the development of phased array antennas for 5G.
A significant challenge porting military technology to commercial applications is cost, as commercial
markets have a very different price-performance balance than the military. MACOM describes a unique
tile array, initially developed for weather and air traffic control radar, now being applied to 5G. The design
uses commercial RF and digital components, assembled with high volume surface-mount manufacturing
processes, to meet 5G base station cost targets. Applying the concept to sub-6 GHz 5G, MACOM outlines the
design of a mMIMO base station integrating the antenna, RF and baseband circuitry in a compact housing
compatible with small cell mounting constraints.
While designing a phased array antenna is challenging, over-the-air (OTA) conformance testing is equally
difficult. To address this, an article contributed by Rohde & Schwarz covers the options for OTA testing at
sub-6 GHz and millimeter wave frequencies, including far field, compact antenna test ranges and software/
hardware near-field transformations.

Gary Lerude, Microwave Journal Technical Editor

3
Strong Defense Outlook Offers
Continued Growth for RF
Technologies
Asif Anwar
Strategy Analytics, Newton, Mass.

The defense sector has been a technology incubator for RF technologies for decades. This
article explores the latest market forecasts from Strategy Analytics and how spending on military
systems will dictate demand for RF technologies such as GaN.

S
trategy Analytics forecasts global defense
spending will continue to grow, driven by force
modernization requirements and political intent
in the U.S., Western European and other advanced na-
tions. A need to counter both resurging conventional
threats from near-peers, coupled with ongoing asym-
metric wars against non-state or pseudo-state actors,
will drive military equipment and capability and support
procurement opportunities for the defense industry
supply chain to a forecast of $827 billion, part of the
projected $2.58 trillion global defense budget in 2027
(see Figure 1).
The U.S. will remain the largest defense market in
the world. Strategy Analytics’ model forecasts U.S.
1,000
900
5-Year SIPRI Figure 10-Year Strategic Analytics Forecast
800
700
$ Billion
defense spending will grow to $866.6 billion in 2026.
The emphasis by the Trump administration to renew
U.S. leadership across the world was reflected in the
fiscal year (FY) 2018 budget and has been further ce-
mented by the most recent FY 2019 budget request:
$686 billion for the U.S. DoD with a number of major
warfighting investments across airborne, naval, ground
and space platforms and systems (see Figure 2). This
will maintain momentum behind the U.S. defense in-
dustry, sustaining growth for major suppliers like Boe-
ing, General Dynamics, Lockheed Martin and Raytheon,
as well as the enabling technology supply chain epito-
mized by companies such as Microsemi, Qorvo, Tele-
dyne and Wolfspeed.
Globally, an emphasis on improved
capabilities at the system level will
drive demand for military radar, mili-
tary communications and electronic
warfare (EW), and capabilities will con-
tinue to provide opportunities for en-

600
500 abling technologies such as GaN.
400
300 RADAR
200 Active electronically scanned array
100 (AESA) architectures enabled using
0
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 GaN and GaAs semiconductor tech-
APAC CALA MEA Europe North America nologies will continue to underpin
radar systems and serve to catalyze
s Fig. 1 Global defense spending outlook, by region. military radar market growth through
WWW.MWJOURNAL.COM/ARTICLES/30993
4
 F-35 Joint Strike Fighters 10.7
Virginia Class Submarines 7.4
DDG-51 Arleigh Burke Destroyers 6.0
Columbia Class Submarine 3.7
KC-46 Tanker Replacements 3.0
M-1 Abrams Tank Modifications 2.7
B-21 Long Range Strike Bomber 2.3
P-8A Aircraft 2.2
Ground Based Midcourse Defense 2.1
Evolved Expendable Launch Vehicles 2.0
Joint Light Tactical Vehicles 2.0
F/A-18s 2.0
CVN-78 Class Aircraft Carrier 1.8
AEGIS Ballistic Missile Defense (SM-3) 1.7
CH-53K King Stallion 1.6
Global Positioning System 1.5
Littoral Combat Ship 1.3
AH-64E Attack Helicopters 1.3
Patriot Advanced Capability (PAC-3) Missile Segment Enhancements 1.1
THAAD Ballistic Missile Defense 1.1
Space Based Infrared System 0.8
Armored Multi-Purpose Vehicles 0.8
Joint Air-To-Surface Standoff Missile-Extended Range 0.6
Ground Based Strategic Deterrent 0.3
Amphibious Combat Vehicles 0.3
0 2 4 6 8 10 12
$ Billion

s Fig. 2 Major U.S. defense program investments in FY19.

2027. Strategy Analytics Advanced Defense Systems

100
service’s forecasts for the global military radar mar- 90
ket, covering expenditure as well as system shipments 80
across the land, air, sea and space domains, projects the 70
Share (%)

global military radar market will grow at a compound 60

average annual growth rate (CAAGR) of 4.6 percent
over the 2017 to 2027 timeframe to be worth over
$21.5 billion.
Suppliers of military radar systems are increasingly
implementing AESA architectures at the core of their
product offerings, highlighting advances in perfor-
mance as well as lifecycle and total cost of ownership
advantages over traditional radar designs. An AESA
radar comprises a large number of transmit-receive
modules that feed and collect multiple signals via an
antenna array. Potential advantages of an AESA archi-
tecture include high beam steering agility, very low ra-
dar signature when illuminated and extremely low side
lobes. Being able to digitally control transmit-receive
module gain allows for refined power management,
which is vital for reduced or low probability of intercept
(RPI, LPI) operation. Beam steering agility also facilitates
reduced or low probability of intercept scan patterns.
From an operational perspective, the implementation
of AESA architectures in military radar systems enables
improvements in system reliability and reductions in to-
tal cost of ownership. Use on fast jet platforms is often
cited as a good example where these metrics are best
exemplified with mean time between failures being im-
proved significantly from the 300 hours typically cited
for conventional radar systems. Coupled with “grace-
ful degradation,” where a radar system continues being
functional even as individual transmit-receive modules
fail, means that aircraft can stay operational with repairs
5
50
40
30
20
10
0
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
APAC CALA MEA Europe North America
s Fig. 3 Forecast share of the military radar market, by
region.
being performed alongside the regular maintenance
schedule of the platform.
The core enabling semiconductor technologies for
AESA architectures have typically been GaAs-based,
but as GaN technology has matured, the defense in-
dustry has looked to GaN as the new core enabler for
AESA-based military radar. GaN offers the advantages
of increased power, efficiency and robustness to im-
prove the performance of land-, air- and sea-based
military radar systems. This does not mean that GaAs
technology will no longer be used, any more than it
would be unwise to suggest that there will no longer be
demand for vacuum tube-based RF transmitters. How-
ever, GaN does arguably offer the added flexibility of
being able to displace GaAs, other RF semiconductors
and vacuum tube technologies as the RF transmitter
source in radar systems. With these advantages, GaN is
becoming a key enabling technology and military radar
demand for GaN is forecast to grow at a CAAGR of 19
percent through 2027.

25
20
15
$ Billion
10
5 planning and countermeasures activity that can either
0
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
s Fig. 4 Global EW market forecast.
40
35
30
25
$ Billion
20
15
10
5 Military communications operate under an umbrella
0 of heterogeneous networks that enable the provision
2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
Radio SATCOM Datalink Network
s Fig. 5 Global military communications market outlook, by
segment.
Strategy Analytics also predicts:
• N
 orth America will continue to represent the largest
regional end market, but the fastest growth will come
from demand in the Asia-Pacific region (see Figure 3).
• Airborne radar will represent the largest market, both
in dollars and total shipments.
• Early warning, surveillance and fire control radars will
account for around 76 percent of the global military
radar market.
• L-, S- and C-Band will represent the largest market,
followed by radars operating at X-Band, which reflect
the primary frequencies used by surveillance, early
warning and fire control radars.
• The total number of radar shipments is forecast to
grow at a CAAGR of 4.6 percent through 2027 to
reach 1,607 units. Fire control radar and early warn-
ing and surveillance radar shipments will account for
48 percent of 2027 military radar shipments.
• The associated market for semiconductors and other
components will grow from $2 billion in 2017 to reach
$5 billion in 2027.
EW
Operational requirements to establish freedom of ac-
tion in contested and congested environments, as well
as the ability to counter modern agile radar and com-
munications will drive opportunities for the EW market.
There is a renewed push to upgrade conventional EW
capabilities that support anti-access/anti-denial (AA/
AD) strategies. This will be coupled with the ongoing
requirement to combat asymmetric threat scenarios.
Future systems will employ wideband solid-state semi-
conductors to enable artificial intelligence (AI)-based
machine learning algorithms to provide cognitive analy-
sis of the threat environment. EW will play an important
role in tackling the increasing complexity that comes
with operating in a spectrally constrained environment.
Companies providing systems and enabling technol-
ogies will need to focus on solutions that employ wide-
band materials, such as GaN, and AESA architectures
to enable machine learning-based cognitive analysis,
augment or circumvent the traditional threat library.
Strategy Analytics forecasts the global EW market will
grow to $20 billion by 2027 (see Figure 4). The associ-
ated market for semiconductors and other components
for RF-based EW systems will grow at a CAAGR of 8.4
percent through 2027. Future EW program will increas-
ingly use GaN, making this semiconductor technology
a staple ingredient in EW systems. This will be coupled
with requirements for direct and faster digital synthesis
of RF signals across the full frequency spectrum.
COMMUNICATIONS
of interoperable voice, video and data services across
Other
a global environment, segmented according to security
policies, transmission requirements and the individual
needs of the end user. In terms of the networked bat-
tlespace, this can be summarized as:
• Upper level networking, consisting of infrastructure
and networking components.
• Mid-level networking providing high capacity back-
haul.
• Support to the tactical edge for end-users and sensors.
Similarly, 5G serves as an aggregator technology that
encompasses a range of network types and technolo-
gies to serve traditional voice, video and data require-
ments to the end user, as well as enabling capabilities
for connectivity across devices, including vehicles, ma-
chines, sensors and devices.
Phased arrays, beamforming, mmWave frequencies,
SATCOM, GaN, duplex communications and shared
spectrum access are among the crossover technologies
that will become common across both commercial and
military communications.
Communicating voice, data and video simultane-
ously and securely over wider and higher bandwidths
in an increasingly complex spectrum environment will
underpin the trends in military communications system
design and supporting components, including software-
defined architectures, solid-state technologies such as
GaN, radio-satellite communications and integration
with wireless networks.
Strategy Analytics forecasts spending on global mili-
tary communications systems and services will grow to
over $36.7 billion in 2026, a compound annual growth
rate of 3.5 percent (see Figure 5).
RF GaN GROWTH
Demand from military radar, EW and communications
applications will provide the primary drivers for GaN
market adoption, and this will be coupled with ongoing
demand from the rollout of commercial wireless infra-
6

300
200
$ Million
100
0 opportunity will cross the $1 billion barrier by 2022,
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
s Fig. 6 Historic defense system demand for RF GaN.
structure. RF GaN market growth continued to acceler-
ate in 2017, with revenues growing at over 38 percent
year-on-year.
The military radar segment will remain the largest
user of GaN devices for the defense sector. Substan-
tial production activity in AESA radars for land-based
and naval systems, in particular, is driving increasing
demand for RF GaN, as many systems in development
move to production. RF GaN demand from the military
sector grew by 72 percent year-on-year in 2017, and
this will grow at a CAAGR of 22 percent through 2022
(see Figure 6).
As highlighted earlier, operational requirements to
operate in contested and congested environments, as
well as the ability to counter modern agile radar and
Catch up on the latest industry news with the bi-weekly
video update Frequency Matters from Microwave Journal
@ www.microwavejournal.com/frequencymatters
Sponsored By
7
communications, will drive opportunities for RF GaN
from the EW market. Communicating voice, data and
video simultaneously and securely over wider and
higher bandwidths in an increasingly complex spectrum
environment will underpin trends in military communi-
cations system design and associated component de-
mand, also increasingly favoring RF GaN.
Strategy Analytics forecasts that the total RF GaN
with defense sector demand slightly greater than
commercial revenue at that time.
CONCLUSIONS
As countries look to maintain a mix of both conven-
tional and leading-edge capabilities, to counter both
symmetric and asymmetric threats, global defense
spending is forecast to reach $2.6 trillion by 2027. The
emphasis on improving capabilities at the system level
will drive demand for military radar, EW and communi-
cations systems and provide growth opportunities for
enabling technologies such as GaN.
Companies will need to be able to “scale, integrate,
incorporate and disrupt” to differentiate themselves
in addressing the challenges and opportunities in the
defense sector, as well as exploiting adjacent mar-
ket growth opportunities benefiting from these same
capabilities.n
Spatial Multiplexing for 5G
Wireless Communications
Honglei Chen and Rick Gentile
MathWorks, Natick, Mass.

I
ncreasing demand for higher data rates and chan-
nel capacity is driving the need to use the RF spec-
trum more efficiently. As a result, 5G wireless sys-
tems will use mmWave frequency bands to take
advantage of the increased bandwidth. The higher op-
erating frequencies enable large-scale antenna arrays,
which can be used to mitigate severe propagation loss
in the mmWave band. Large arrays can also be used to
implement a MIMO system in which unique signals can
be transmitted from different antenna elements in the
array. MIMO systems enable spatial multiplexing tech-
niques that can be used to improve data throughput.
The core idea of spatial multiplexing is to create
multiple subchannels in scatterer-rich environments so
that multiple data streams can be transmitted and re-
covered independently. This is achieved by applying a
set of precoding and combining weights derived from
the channel matrix. This concept is discussed first with
an all-digital solution based on precoding in a MIMO-
OFDM system that uses a spatial channel model. This
channel model incorporates array pattern information to
improve model fidelity.
Because 5G systems require large antenna arrays, ap-
plying digital weights on each antenna element is not
always practical due to cost and space limitations. Hy-
brid beamforming techniques can be applied in a mixed
RF and digital beamforming system to alleviate these
restrictions. In a hybrid beamforming system, both the
precoding weights and the combining weights are com-
binations of baseband digital weights and RF band ana-
log weights. On the transmit side, the baseband digital
terer environment can send multiple data streams
simultaneously across the channel. For example, the
channel matrix of a 4 × 4 MIMO channel becomes
full rank because of the scatterers. This means that
it is possible to send as many as four data streams at
once. The goal of spatial multiplexing is less about
increasing the signal-to-noise ratio (SNR) and more
about increasing the information throughput.
The idea of spatial multiplexing is to separate the
channel matrix into multiple modes so that the data
stream sent from different elements in the transmit array
can be independently recovered from the received sig-
nal. To achieve this, the data stream is precoded before
the transmission and then combined after the reception.
In the MATLAB models that follow, the precoding and
combining weights can be computed from the channel
matrix by:
[wp, wc] = diagbfweights(mimompchan);
The information received by each receive array el-
ement is simply a scaled version of the transmit array
element, which results in multiple orthogonal sub-
channels within the original channel. The first sub-
100
SISO LOS
MIMO Multipath
10–1
MIMO Multipath Stream 1
MIMO Multipath Stream 2
10–2
BER

weights modulate the incoming data streams to form

input signals at each RF chain, and the analog weights 10–3
then translate the signal at each RF chain to the signal
radiated at each antenna element. The process is re-
10–4
versed on the receive side. This article includes an ex-
ample with hybrid weights and compares the achieved
spectral efficiency with the all-digital case. 10–5
–10 –8 –6 –4 –2 0 2 4 6 8 10
SPATIAL MULTIPLEXING Eb/N0 (dB)

The idea behind spatial multiplexing is that a

MIMO system in a multipath channel with a rich scat- s Fig. 1 BER of MIMO multipath streams compared to a single
line-of-sight link.
WWW.MWJOURNAL.COM/ARTICLES/30837
8
 Data Bits

Preamble MIMO OFDM Encoder, Precoding OFDM Tx Beam MIMO Rx Beam

Tx Channel Rx Mapper Tx Steering Channel Steering

From
MIMO Channel Sounding Recovered Demapper, MIMO OFDM
Estimate Data Bits Decoder Equalize Rx

V For MIMO Channel

SVD
Feedback Estimate

(a) (b)

s Fig. 2 Channel sounding (a) and data transmission/reception (b) flow.

Elevation Angle 0° Azimuth Angle 0°

90 90
120 60 120 60
0
150 30 150 30
Directivity (dBi)

Directivity (dBi)
–50
Z 180 0 180 0

Y
X
–150 –30 –150 –30

–120 –60 –120 –60

(a) (b) –90 (c) –90

s Fig. 3 Transmitter array geometry (a) with azimuth (b) and elevation (c) beam patterns.

channel corresponds to the dominant transmit and
receive directions, so there is no loss in the diversity
gain. In addition, it is possible to use other subchan-
nels to carry information, as shown by the bit error
rate (BER) curves in Figure 1, including the first two
subchannels. The gain of the second data stream is
not as high as the first stream, since it uses a less
dominant subchannel. Still, the overall information
throughput is improved. This concept will be applied
to a MIMO-OFDM system.
MIMO-OFDM SYSTEMS
MIMO-OFDM systems are common in wireless sys-
tems due to their robustness with frequency-selective
channels and high data rates. With antenna arrays that
implement spatial multiplexing, efficient techniques to
realize the transmissions are necessary. As an example,
consider an asymmetric MIMO-OFDM single-user sys-
tem in which the maximum number of antenna elements
on the transmit and receive ends are 1024 and 32, re-
spectively, and up to 16 independent data streams.
For clarity, a single link (one base station communicat-
ing with one mobile user) is modeled, but this structure
could be extended for more complex configurations.
The link between the transmitter and receiver relies
on channel sounding to generate the channel informa-
tion needed for transmit beamforming. For a spatially
multiplexed system, the availability of channel informa-
tion at the transmitter allows for precoding to be applied
to maximize the signal energy in the direction and chan-
nel of interest. Assuming a slowly varying channel, this
process is facilitated by first sounding the channel with a
reference transmission. The receiver then estimates the
channel and feeds this information back to the transmit-
ter, as shown in Figure 2. In this system, a preamble
signal is first sent over all transmitting elements. It is sub-
sequently processed at the receiver after accounting for
the channel effects. The receiver performs pre-amplifi-
cation, OFDM demodulation, frequency domain chan-
nel estimation and calculation of the feedback weights
based on channel diagonalization using singular value
decomposition per data subcarrier. The specific system
configuration is as follows:
% Single-user system with multiple streams
prm.numSTS = 16; % Number of independent data
streams, 4/8/16/32/64
prm.numTx = 32; % Number of transmit antennas
prm.numRx = 16; % Number of receive antennas
prm.bitsPerSubCarrier = 6; % 2: QPSK, 4: 16-QAM, 6: 64-
QAM,
8: 256-QAM
prm.numDataSymbols = 10; % Number of OFDM data symbols
A rectangular array at the transmitter is used, based
on the desired number of data streams and transmit
antennas. Figure 3 shows the array geometry of the
transmitter and the beam patterns in azimuth and el-
evation. For simplicity, assume the steering angle is
known with respect to the mobile location. In actual
systems, the steering angle would be obtained from

9
an angle-of-arrival estimation at the receiver, as a part
of the channel sounding or initial beam tracking pro-
cedures.
Multiple options exist for modeling spatial MIMO
channels, typically selected based on the level of fidelity
that is needed for the analysis. For example, 5G channel
models and the WINNER II channel models are spatially
defined MIMO channels where the array geometry and
location information can be specified. For this discus-
sion, consider scattering-based channels with a single-
bounce path through 100 scatterers placed randomly
within a circle between the transmitter and receiver. The
channel model allows path loss modeling and both line-
of-sight (LOS) and non-LOS propagation conditions. For
this analysis, a non-LOS propagation and isotropic an-
tenna element pattern with linear geometry are config-
ured. The same channel is used for both sounding and
data transmission. The data transmission has a longer
duration controlled by the number of data symbols.
Z a 16-element square array with four RF chains. Each
Y that each antenna is connected to four phase shifters.
X
(a)
Elevation Angle 0°
90
120
0
150
Directivity (dBi)
180
–150
–120
–90
(b)
s Fig. 4 Receiver array geometry (a) with azimuth beam
pattern (b).
The receive antenna array, shown in Figure 4, pass-
es the propagated signal to the receiver to recover the
original information embedded in the signal. Similar to
the transmitter, the receiver used in a MIMO-OFDM sys-
tem contains many stages, including OFDM demodula-
tor, MIMO equalizer, QAM demodulator and channel
decoder. For this MIMO system, the displayed receive
constellation of the equalized symbols offers a qualitative
assessment of the reception. The actual BER offers the
quantitative figure by comparing the actual transmitted
bits with the received decoded bits (see Figure 5).
Parameters can be modified to vary the number of
data streams, transmit/receive antenna elements, base
station or array locations and geometry, channel models
and their configurations to study the parameters’ indi-
vidual or combined effects on the system. This frame-
work can be used for further analysis but, so far, all of the
precoding and combining were based on an all-digital
system. As the antenna array size increases, an all-digital
system may not be feasible, which leads to the use of
hybrid beamforming.
HYBRID BEAMFORMING
In a hybrid beamforming system, both the precoding
weights and the combining weights are combinations of
baseband digital weights and RF band analog weights.
This type of architecture is shown in Figure 6.
For this discussion, assume a set of larger arrays:
The transmitter consists of a 64-element square ar-
ray with four RF chains, and the receiver is based on
antenna is connected to all RF chains, which means
This type of array can be modeled by partitioning the
1.0
60
0.5
Quadrature Amplitude
30
0
–0.5
0
–1.0
–30
–1.0 –0.5 0 0.5 1.0
In-Phase Amplitude
–60
RMS EVM (%) 1.5 Peak EVM (dB) –9.8
Peak EVM (%) 32.2 Avg MER (dB) 36.2
Avg EVM (dB) –36.2
s Fig. 5 Constellation diagram for the MIMO OFDM system.
10
 ..
..
..
DAC RF
..
.. Baseband . .
NS . Digital
Precoding NTRF
RF Analog
Precoding
DAC RF
..
..
..
Phase Shifter
s Fig. 6 Hybrid beamforming architecture.
array aperture into four completely connected subar-
rays. To maximize the spectral efficiency in the system,
each RF chain can be used to send an independent
data stream. In this case, the system supports up to
four streams.
A scattering environment with six scattering clusters
randomly distributed in space is used to define the chan-
nel. Within each cluster, there are eight closely located
scatterers, for a total of 48 scatterers. The path gain for
each scatterer is obtained from a complex circular sym-
metric Gaussian distribution.
As described earlier, in a spatial multiplexing system
with all-digital beamforming, the signal is modulated
by a set of precoding weights, propagated through the
channel and recovered by a set of combining weights.
Mathematically, this process can be described by Y =
(X*F*H+N)*W where
• Ns is the number of data streams.
• Nt is the number of transmit elements.
• Nr is the number of receive elements.
• Nrf is the number of RF channels.
• X is an Ns-column matrix whose columns are data
streams.
• F is an Ns × Nt matrix representing the precoding
weights.
• H is the channel representation.
• W is an Nr × Ns matrix representing the combining
weights.
• N is an Nr-column matrix whose columns are the receiver
noise at each element.
• Y is an Ns-column matrix whose columns are recovered
data streams.
Since the goal of the system is to achieve better spectral
efficiency, obtaining the precoding and combining weights
can be considered an optimization problem in which the
optimal precoding and combining weights make the prod-
uct of F*H*W a diagonal matrix, so each data stream can
be recovered independently.
In a hybrid beamforming system, the signal flow is
similar. Both the precoding weights and the combining
weights are combinations of baseband digital weights
and RF band analog weights. The baseband digital
weights convert the incoming data streams to input sig-
..
.
..
.
RF ADC
.. .. ..
. . Baseband
..
NT NR
RF Analog
Combining NRRF
Digital
Combining .
.. RF ADC
.
..
.
nals at each RF chain, and the analog weights then con-
vert the signal at each RF chain to the signal radiated or
collected at each antenna element. Note that the analog
weights can only contain phase shifts. Mathematically, it
can be written as F = Fbb*Frf and W = Wbb*Wrf, where
• Fbb is an Ns × NtRF matrix.
• Frf is an NtRF × Nt matrix.
• Wbb is an NrRF × Ns matrix.
• Wrf is an Nr × NrRF matrix.
Since both Frf and Wrf can only be used to modify
the signal phase, there are extra constraints on the op-
timization process to identify the optimal precoding and
combining weights. Ideally, the resulting combination of
Fbb*Frf and Wrf*Wbb are close approximations of F and
W that are obtained without those constraints. Unfortu-
nately, optimizing all four matrix variables simultaneously
is quite difficult. Many algorithms exist that arrive at sub-
optimal weights with a reasonable computational load.
This example uses an approach that decouples the op-
timizations for the precoding and combining weights. It
first uses the orthogonal matching pursuit algorithm to
derive the precoding weights. The hybrid weights can be
computed as
[Fbb, Frf] = helperHybridPrecodingWeights(H, NtRF, Ns
, At);
Once the precoding weights are computed, the re-
sult is used to obtain the corresponding combining
weights. Assuming the channel is known, the uncon-
strained optimal precoding weights can be obtained by
diagonalizing the channel matrix and extracting the first
NtRF dominating modes.
The transmit beam pattern for both cases is shown
in Figure 7. The response patterns show that even in
a multipath environment, the number of dominant di-
rections is limited. The beam pattern using the hybrid
weights is similar to the beam pattern obtained using
the optimal weights, especially for the dominant beams.
This result means that the data streams can be success-
fully transmitted through those beams using hybrid
weights.
One of the system-level performance metrics of a 5G
system is the spectral efficiency. Figure 8 compares the
spectral efficiency achieved using the optimal weights
11
 with the proposed hybrid beamforming weights. The
Z Z simulation assumes one or two data streams. The transmit
Az 0 Az 0
El 90 El 90 antenna array is assumed to be at a base station, with a
focused beam coverage of 60 degrees in azimuth and 20
El El degrees in elevation. The signal can arrive at the receive
Y Y array from any direction. The resulting spectral efficiency
Az 90 X Az Az 90 curve is obtained from 50 Monte Carlo trials for each
X Az El 0 Az 0 El 0
Az 0 SNR. Figure 8 shows that the spectral efficiency improves
El 0 El 0
(a) (b) significantly by increasing the number of data streams. In
addition, hybrid beamforming can perform close to what
s Fig. 7 Transmit patterns for all-digital (a) and hybrid (b) optimal weights can offer and uses less hardware.
cases.
SUMMARY
18
Ns = 1 Optimal
MIMO systems enable spatial multiplexing techniques
Spectral Efficiency (Bits/s/Hz)

16 Ns = 2 Optimal that can be used to improve data throughput. This im-

14 Ns = 1 Hybrid provement is achieved by applying a set of precoding
Ns = 2 Hybrid
12 and combining weights derived from the channel matrix.
10 Because 5G systems require large antenna arrays, apply-
8 ing digital weights on each antenna element is not always
6 practical. Hybrid beamforming can be applied in a mixed
4 RF and digital beamforming system to lower system
2 costs. System modeling techniques can be used to ex-
0 plore system tradeoffs before any hardware is built, which
–40 –35 –30 –25 –20 –15 –10 –5 0 can prevent costly errors and project delays.n
SNR (dB)
s Fig. 8 Spectral efficiency of all-digital and hybrid systems.

12
Phased Array Antennas & The
Roadmap to 5G Wireless
MACOM
A
s the market and media hype around 5G contin-
ues to grow, there is a tacit acknowledgement
that we have miles to go before 5G becomes a
reality. Initial industry standards for 5G aren’t expected
to be ratified until Summer 2018 at the earliest, and
there are many regulatory issues and a myriad of tech-
nology challenges still to be resolved before 5G is ready
for mainstream commercialization.
Yet, despite these daunting challenges, the promised
benefits of 5G are so profound that one can’t help but
get excited about it. Improved mobile phone connec-
tivity is just the tip of the iceberg when one considers
5G’s implications for transportation, industrial and enter-
tainment applications, among many others. In a recent
whitepaper from industry research firm IHS Markit, 5G is
heralded as the catalyst that will vault mobile technology
into the realm of general-purpose technologies (GPTs)
that drastically alter society, à la the printing press, the
steam engine and electricity.
s Fig. 1 An overview of the markets and applications impacted by 5G (image above
courtesy of Nokia)
HTTPS://WWW.MACOM.COM/BLOG
13
MIMO VS MASSIVE MIMO
Achieving the promise of 5G will require, among oth-
er things, major innovations in the way basestations are
architected. Today we rely on multiple input, multiple
output – or MIMO – antennae configurations to multiply
the capacity of antennae links for wireless basestations.
These antennas provide the ability to concentrate signal
strength into smaller areas of space, boosting overall
efficiency and throughput by guiding the signal to the
precise location it’s needed. By adding additional anten-
nas, this beamforming capability is improved.
But whereas conventional basestations may house
between two and eight antennas, 5G basestations will
need anywhere from 64 to hundreds of antennas ar-
rayed in a “massive MIMO” configuration to provide the
requisite data rates. This phased array antennae design
comprises an active electronically scanned array (AESA),
which enables signals to be electronically steered with
much greater beamforming precision than MIMO can
support today.
HIGH-PERFORMANCE, LOW
COST ACTIVE ANTENNAS
When it comes to the architecture
and assembly of massive MIMO 5G
systems, we see many parallels with
the new generation of Multifunction
Phased Array Radar (MPAR) active an-
tennae systems targeted for use for
military and civil air traffic control, and
weather system tracking applications.
And while you might not typically as-
sociate this class of radar system with
cost-sensitive commercial applications
like 5G, you might be surprised to
learn that MPAR technology leverages
design and manufacturing efficiencies
that dramatically reduce the cost of
the end system.

s Fig. 2 First generation MPAR systems leverage an array of
Scalable Planar Array (SPAR) Tiles.
First generation MPAR systems feature an array of
Scalable Planar Array (SPAR™) Tiles in a flat panel con-
figuration comprised of hundreds to thousands of active
antennas. SPAR Tile technology, developed in a collab-
oration between MACOM and MIT Lincoln Laboratory,
embodies a new cost-conscious approach to phased
14
array radar system development, leveraging highly-in-
tegrated antenna sub-systems, and volume scale com-
mercial packaging and manufacturing techniques.
Tile-based AESAs create the foundation for a new
generation of high-performance, agile radar systems
that can be built quickly and flexibly tailored and scaled
for deployment across a range of applications – at 5X
less cost than conventional slat array architectures. Con-
tinued innovations in phased-array-based technologies
like MPAR will help to unlock the full promise of 5G, al-
lowing basestation OEMs to simplify design and manu-
facturing processes, and get to market faster with 5G
technology.
To learn more about the benefits of MPAR technol-
ogy for 5G, check out this article in Microwave Product
Digest. For a deep-dive read on the MPAR technology
architecture, head over to Microwave Journal.
Disclaimer
All financial guidance projections referenced in this post were made
as of the publication date or another historical date noted herein,
and any references to such projections herein are not intended
to reaffirm them as of any later date. MACOM undertakes no
obligation to update any forward-looking statement or projection at
any future date. This post may include information and projections
derived from third-party sources concerning addressable market
size and growth rates and similar general economic or industry
data. MACOM has not independently verified any information
and projections from third party sources incorporated herein. This
post may also contain market statistics and industry data that are
subject to uncertainty and are not necessarily reflective of market
conditions. Although MACOM believes that these statistics and
data are reasonable, they have been derived from third party
sources and have not been independently verified by MACOM.
Tile Arrays Accelerate the
Evolution to
Next-Generation Radar
Doug Carlson
MACOM, Lowell, Mass.

D
efense and civil radar infrastructure has his- veillance. With AESA
torically evolved over lengthy cycles, fueled technology, meteo-
by massive investments of time, innovation rologists can better
and capital. Defense applications have been predict and assess
chiefly responsible for driving this development, yield- severe storms and
ing significant radar performance improvements, often save lives. AESAs will
with concomitant increases in total system cost. These also provide the sen-
initiatives have catalyzed the transition from convention- sor capability that will
al, mechanically-steered radar architectures to active bring drones and un-
electronically scanned array (AESA)—or active antenna manned vehicles into
array— that enable huge gains in temporal and spatial the mainstream. This
precision, leveraging advanced multi-beam capabilities. will transform society,
AESAs’ multi-function capability is similarly compelling, fundamentally alter-
affording newfound agility to consolidate disparate ra- ing transportation
dar systems onto a single platform (see Figure 1). and commerce.
The performance characteristics of AESAs position this For AESAs to suc- s Fig. 1 The active phased array
technology as the clear successor to legacy defense ra- cessfully transition architecture enables the beam to be
dar systems, and AESA implementations are already well from defense to civil electronically steered and multiple
underway in this domain. Deployable across ground, sea and commercial ap- radarssystem.
to be combined in a single
and air, AESA technology dramatically strengthens the plications, however,
sensor mesh network and improves situational aware- there are technical and economic hurdles to overcome.
ness throughout the modern battlefield. Propagated to Continued reliance on conventional RF components
civil radar infrastructure, AESAs have the potential to pro- and cumbersome assembly techniques, among other
foundly affect the safety and security of citizens. A single factors, has blocked the path to mainstream adoption.
network of multifunction AESA radars can improve air To gain a perspective on where this development activ-
traffic control, providing direct economic benefit to the ity is headed, it’s helpful to understand its origins and
country and bolstering homeland defense capabilities, roadmap to date.
while simultaneously dramatically improving weather sur-
WWW.MWJOURNAL.COM/ARTICLES/27652
15
R&D ROADMAP
Next-generation active antenna technology has its roots
in development programs dating back to the 1960s. With
the maturing of GaAs monolithic microwave integrated
circuit (MMIC) technology in the 1980s, AESA develop-
ment accelerated. The U.S. Defense Advanced Research
Projects Agency (DARPA) worked in partnership with
MACOM and other technology companies on the Mi-
crowave/Millimeterwave Monolithic Integrated Circuits
(MIMIC) program and then the Microwave Analog Front
End Technology (MAFET) program. These initiatives
helped advanced compound semiconductor technolo-
gies make the leap from laboratory research to commer-
cially manufactured RF devices, yielding the industry’s
first multi-watt MMICs. Further development of these
semiconductor and packaging technologies led to the
first mass-produced RF modules and components to use
mainstream printed circuit board (PCB) technology and
surface mount assembly techniques. This activity was fol-
lowed by DARPA programs aimed at developing com-
pound semiconductors for higher output power, higher
s Fig. 2 Slat array.

efficiency and higher frequency operation. GaN technol-

ogy became a major focus of development and invest-
ment, culminating with high frequencies (to 500 GHz),
yield, uniformity and reliability.
Initiated in 2014, DARPA’s Arrays at Commercial Tim-
escale (ACT) program is designed to streamline develop-
ment and manufacturing cycles for next-generation radar,
electronic warfare (EW) and communications systems by
leveraging best practices established in the commercial
domain. It aims to achieve a digitally-interconnected
phased array building block, from which larger systems
can be built without a full redesign for each new appli-
cation. This approach to radar system implementation is
anticipated to shorten time-to-market and reduce costs,
both necessary for AESAs to achieve mainstream adop-
tion for civil and commercial applications.

SLAT VS. TILE

The cost and mainstream viability of AESAs correlate
to the cost of the electronic components that comprise
them and the way they are architected in the array. From
the transmit/receive (T/R) module to the RF boards and
cabling, there are many considerations to weigh that im-
pact final system cost.
Approximately half the cost of the phased array por-
tion of the radar is attributable to the T/R module, the
cost of which is determined by the types of MMICs,
housing and substrates used in the module. Convention-
al T/R modules for radar applications employ ceramic-
based materials for the substrates and are manufactured
using chip-and-wire assembly processes, during which
MMIC die and other ICs require additional touch labor.
These add considerable expense compared to plas-
tic encapsulated MMICs used in commercial systems.
The multi-layer RF boards and cables employed in the
phased array design account for almost all the rest of
the cost (assembly and test processes and the structure
represent approximately 10 percent of the total cost). As
with the T/R modules, the underlying architecture of the
array design can add considerable cost.
This is exemplified via the conventional approach to
constructing large phased arrays, commonly known as
16
s Fig. 3 Tile array.
the slat array architecture, which has a series of slats ar-
ranged perpendicularly to the face of the array (see Fig-
ure 2). The slat approach has a couple of advantages: it
provides a large surface area on which the T/R modules
and supporting components can be attached. Also, the
thermal load from the high power amplifiers can be dis-
tributed across a large volume, factoring in the depth of
the slat and the aperture area. The downside is that a
large number of RF boards and copious cabling are re-
quired to channel the RF, DC and control signals on and
off the slats. This adds considerable cost to the design.
An alternative approach—the tile array—overcomes
these disadvantages through a more streamlined archi-
tecture, where the array is constructed of layers that are
oriented parallel to the face (see Figure 3). Antenna ele-
ments and RF beamformers are integrated in a single,
multilayer RF board, with the T/R modules mounted
on the back of the board. This approach significantly
reduces the area of the RF boards and dramatically re-
duces the number of cables and connectors. Costs can
be further reduced through the use of T/R modules de-
signed to leverage volume-scale commercial packaging
and manufacturing techniques. With this approach, the
T/R module MMICs are assembled in industry standard,
quad flat, no-lead (QFN) packages that are directly sol-
dered onto an inexpensive PCB that is soldered to the
back of the tile. Simple metal pads lining the edge of
the PCB serve as the RF and DC interconnects between
the T/R module and the back of the tile.
Comparing the relative transmit power per unit area
versus the cost per unit area for the slat and tile array
architectures, a greater than 5x cost reduction can be
achieved with the tile array at both high power and low
power outputs.1
MULTIFUNCTION RADAR
Working together, MACOM and the Massachusetts
Institute of Technology (MIT) Lincoln Laboratory have
optimized the tile array architecture, demonstrating the
cost efficiencies that can be achieved using commer-
s Fig. 4 SPAR™ Tile prototype developed by MACOM and
MIT Lincoln Laboratory.
Aperture PCB Assembly DC/Logic Tx Beamformer
Interconnects Driver
Panel
Structure
Standoffs
Backplane
PCB
T/R
Aperture Modules
PCB
s Fig. 5 SPAR Tile construction.
cial manufacturing processes. Sponsored by the Fed-
eral Aviation Administration (FAA) and National Oceanic
and Atmospheric Administration (NOAA), the MACOM/
Lincoln Laboratory Multifunction Phased Array Radar
(MPAR) is designed to be the next-generation civil ra-
dar, integrating eight separate legacy radar functions
into a single multifunction platform. The first-generation
MPAR (see Figure 4) uses an array of Scalable Planar
Array (SPAR) Tiles to detect and track weather systems,
aircraft and airborne objects.
While most legacy radar systems mechanically ro-
tate and tilt the radar dish to scan different sectors of
the airspace, the SPAR™ Tile-enabled radar comprises
hundreds to thousands of T/R elements in a stationary
flat panel array that electronically scans the airspace. As
shown in Figure 5, the MPAR SPAR Tile consists of a front
aperture printed circuit board (APCB) containing the radi-
ating elements, transmit and receive beamforming net-
works, power and logic distribution. The T/R modules are
surface-mounted to the back of the APCB using standard
industry manufacturing processes. A second PCB, the
backplane, contains the DC power supplies, general pur-
pose processors and high level logic. It is combined with
the APCB to form the SPAR Tile. The two PCBs are con-
nected with low cost, high performance interconnects,
and a simple mechanical structure holds the two PCBs
together and forms the structure for the assembly of the
full array.
The first deployed MPAR prototype is being validated
for weather observation by the National Severe Storms
Laboratory (NSSL) in Oklahoma. It also provides a base-
line platform for the development of backend architec-
ture and AESA data-driven weather modeling algorithms.
When fully implemented, this system will help increase
forecast accuracy for severe weather events such as tor-
nadoes, enabling earlier major storm warning and other
benefits. For the FAA, MPAR systems can dramatically
improve air traffic awareness and homeland defense sur-
veillance.
MACOM has been maturing the SPAR Tile technol-
ogy for almost a decade. From concept demonstration
to the deployment of the first prototype system in a re-
alistic end-use environment, the Technology Readiness
Level (TRL) and Manufacturing Readiness Level (MRL) of
the SPAR Tiles have advanced. Based upon the success
of the first prototype system, MACOM has begun the
first phase of scaled manufacturing of SPAR Tiles, with
over 90 tiles for the FAA/NOAA Advanced Technology
Demonstrator (ATD) be-
ing delivered for a full
system demonstration.
This has allowed, for the
first time, meaningful sta-
tistics to be gathered on
the performance distribu-
tion from the end-to-end
manufacturing process.
Over 6,000 T/R modules
have been built and test-
ed, yielding data such as
shown in Figure 6.

17
 opportunity to develop a new and more cost-effective
approach to radar system production, an approach
400
combining an innovative RF system architecture with
commercial manufacturing processes.
300 Tile-based AESAs create the foundation for such a new
Count

generation of high performance, agile radar systems that

200 can be built quickly and cost effectively and flexibly tai-
lored and scaled for deployment across defense, civil and
100 commercial applications. The design and assembly tech-
niques used for the tile array MPAR address both commu-
0 nications and sensing applications, enabling active anten-
2.80 3.15 3.50 3.85 4.20 4.55 4.90 na capability at a cost point that makes this technology
Noise Figure (dB) viable for a wide range of commercial use cases: internet
in the sky, 5G, sense-and-avoid for airborne drones and
s Fig. 6 Noise figure distribution of the ATD T/R modules. The
radar for autonomous vehicles.■
histogram approximates a normal distribution with a mean noise
figure of 3.7 dB.
Reference
1. Jeffrey S. Herd and M. David Conway, “The Evolution to Modern
TILE-BASED AESA OPPORTUNITIES Phased Array Architectures,” Proceedings of the IEEE, 2016, Vol.
With legacy civil radar infrastructure for air traffic 104, Issue 3, pp. 519–529.
control and weather surveillance approaching the end
of life and government-mandated spending reductions
impacting key defense programs, there is a unique

Catch up on the latest industry news with the bi-weekly

video update Frequency Matters from Microwave Journal
@ www.microwavejournal.com/frequencymatters

Sponsored By

18
Breaking Through the Cost
Barrier for Phased Arrays
Doug Carlson
MACOM, Lowell, Mass.

F
or the U.S. military and contractor ecosystem
that services it, intensifying security threats at
home and abroad are being met, in part, with
increased investments in phased array radar sys-
tems. Phased arrays promise enhanced threat detection
and mitigation and faster, more agile communication
across the modern battlefield. But, where legacy ra-
dar systems were afforded massive government fund-
ing and ample time for development and deployment,
next-generation phased array radar technology does
not have nearly the same luxury.
The growing proliferation of unmanned aircraft sys-
tems (UAS)—drones—spells a perilous new front in
modern warfare and homeland security, whether they
are state funded or off-the-shelf recreational varieties,
particularly as swarming techniques come to the fore.
To answer this threat, a more agile phased array radar
platform is required, one that is easily and quickly manu-
facturable, deployable, field replaceable and adaptable
to changing needs. Much like the drones, these radars
must be nimble and affordable.
Drones are not the only threats driving this devel-
opment effort, of course. Continued advancements
in traditional warcraft and weaponry require a simi-
lar evolution toward more flexible and precise radar
platforms. In all cases, the strong need to invest in
scalable radar technology is counterbalanced by an
equally strong need to reduce strain on the defense
budget.
advent of commercial 5G wireless infrastructure is fuel-
ing considerable innovation in massive MIMO antenna
designs and software-defined beamforming. In the
government domain, the Multifunction Phased Array
Radar (MPAR) program, sponsored by the U.S. Federal
Aviation Administration (FAA) and National Oceanic
and Atmospheric Administration (NOAA), successfully
spearheaded the consolidation of multiple independent
legacy radar functions into a more
cost-effective, unified, multifunc-
tion platform for weather sur-
veillance and air traffic control
(ATC).
The end goals of these
commercial and government
initiatives are similar: they
promote the use of afford-
able, scalable and highly inte-
grated phased array antenna

INTERESTING BEDFELLOWS
The development of next-generation phased
array radars has followed two parallel paths,
which are bending toward convergence. The
WWW.MWJOURNAL.COM/ARTICLES/31325
19
technologies to enable faster, more accurate and more
sensitive transmit and receive (T/R) capabilities. This, in
turn, allows wireless operators to increase subscriber
coverage and affords civil radar with expanded fields of
view to encompass aircraft, airborne objects and weath-
er systems.
Although it might seem unlikely that commercial
and government defense entities can reconcile their
respective requirements for ultra-low-cost and battle-
field ruggedness, the gap between them is, perhaps
surprisingly, not that far apart. The prevailing next-
generation phased array architecture targeted for 5G
networks and civil ATC has achieved considerable
cost reduction by using commercial-scale manufactur-
ing processes and significant reductions in component
count to improve ruggedness and reliability in harsh
operating environments. This integrated and affordable
architecture—the tile-based planar array—is poised for
mainstream adoption in commercial 5G and civil ATC
applications alike and is already deployed in the field,
with defense trials set to commence.
INCREASED INTEGRATION
To better understand how tile-based planar phased
array radars map to the specific end needs of defense
applications, it is helpful to discuss the
underlying architecture. The salient ar-
chitectural goal of the planar array is to
allow for a flat design where layers are
oriented parallel to the face of the ar-
ray (see Figure 1a)—not perpendicu-
lar as with legacy slat/brick-based con-
figurations (see Figure 1b). The key
reasons why the planar architecture is
more cost-effective than the slat/brick
architecture are because it is:
• Easier to manufacture.
• Significantly reduces cabling and
connectors.
By moving to a planar configura-
tion, the array can be built like a PCB, (a)
data architecture, enabled by RF SOCs with integrated
RF, analog and digital circuitry. RF SOCs preclude the
need for cumbersome analog down-conversion. The ca-
pability to sample signals up to 56 GSPS enables direct
RF sampling at very high frequencies, with the option to
down-sample, which eliminates the need for a superhet
receiver and discrete data converters tied to very spe-
cific frequency plans. This architecture also eliminates
the exciter technology needed for superhet frequency
conversion.
At the board level, this architecture yields a smaller
system footprint, with digital flexibility and increased
I/O across a very wide frequency range, streamlining
the data pipeline and providing a scalable pathway for
increasing the number of RF channels. Ultimately, RF
SOCs eliminate a significant amount of legacy electron-
ics and the associated power consumption from the ar-
ray, playing a key role in enabling the software-defined
coherent beamforming essential for efficient and pre-
cise radar tracking.
From the array architecture to the RF SOC layer, the
higher integration achievable with tile-based planar ar-
rays translates into significant reductions in component
and connector count, reducing cost while simultaneous-
ly boosting system reliability and ruggedness. The ease
(b)

with surface-mount attachment of

components to the back of the ap- s Fig. 1 Tile-based planar phased array (a) vs. slate or brick configuration (b).
erture board. The RF content —T/R
modules comprised of highly integrat- DC/Logic Backplane Panel
Tx
Interconnects PCB Structure
ed MMICs—is mounted to the back T/R Driver
Aperture Modules
of the board, with an additional layer PCB Standoffs
housing the DC voltage regulation
and capacitor components (see Fig-
ure 2), which enables a single voltage
to be fanned out within the system.
This more elegant approach to chan-
neling the RF, DC and control signals
eliminates copious cabling and the as-
sociated cost, weight and layout com-
plexities, while simultaneously provid-
ing a compact airflow design.
Further cost and weight reduction
can be achieved within the planar ar- Air Flow
ray by eliminating the conventional su-
per heterodyne (superhet) receiver in
favor of a direct sampling, high speed s Fig. 2 Construction of the tile-based planar array.

20
s Fig. 3 MPAR 10-tile array deployed at the National Severe Storms Laboratory in Oklahoma.

of manufacturing with the “PCB-like” planar architecture

ensures radar tiles can be produced quickly and cost-
effectively, leveraging commercial, high volume best
practices.

FIELD PROVEN
The viability and scalability of the planar array radar
architecture is being demonstrated through initial de-
ployments, successfully underway. The first planar array-
based radar system deployed in conjunction with the
MPAR program is in use today by the National Severe
Storms Laboratory (NSSL) in Oklahoma (see Figure 3). It
features a 10-tile planar architecture that is radiating and
actively supporting NOAA weather tracking. A 76-tile
array is being installed at the NSSL to support program
software development, and another large-scale array is
being tested at Hanscom Air Force Base.
The MPAR program has now been integrated into the
Spectrum Efficient National Surveillance Radar (SENSR)
program, a U.S. government-sponsored, cross-agency
planning and coordination initiative aimed at two goals:
consolidating the national radar infrastructure into a sin-
gle multifunction platform and freeing up valuable radio
spectrum for reallocation to 5G services. The SENSR
program brings together the FAA, DoD and the De-
partment of Homeland Security (DHS) in a joint effort to
radically streamline civil radar infrastructure (see Figure
4) and minimize the spectrum constraints on future 5G
wireless roll-outs.
In addition to servicing traditional civil applications
like airport terminal and weather surveillance, the SENSR
program is addressing coastal and border surveillance,
assessing airborne threats to homeland security—focus-
ing on drones and missiles. Leveraging its earlier invest-
ment in the successful MPAR program, the government
agencies spearheading SENSR may reasonably see
planar arrays as a strong candidate for the underlying
phased array radar platform.
s Fig. 4 The SENSR program aims to address multiple U.S.
radar needs with a single phased array radar. Source: Raytheon.
In parallel, tile-based radar technology is set for use
by SRC, a not-for-profit research and development com-
pany, to underpin its advanced SkyChaser multi-mission
radar system, which is being optimized for military ap-
plications. The SkyChaser radar is designed to detect
and track close-in and long-range targets while on the
move in vehicle-mounted configurations, if desired. This
impressive capability distinguishes SkyChaser as an ex-
tremely versatile means of conducting air surveillance,
counter-fire target acquisition, short-range air defense
and drone sensing and avoidance.
SUMMARY
The higher levels of integration, affordability and
manufacturing efficiency inherent to the planar array
radar architecture, from the system to the component,
positions it uniquely to help address the evolving mili-
tary and homeland defense needs. Early deployments
have demonstrated scalability and flexibility, and paral-
lel advancements in commercial 5G infrastructure will
drive the cost structures needed to adopt planar phased
array radar for mass deployment for civil radar and the
modern battlefield.n

21
Software and Hardware
Near-Field Transformations
for 5G OTA Testing
Benoît Derat, Corbett Rowell, Adam Tankielun and Sebastian Schmitz
Rohde & Schwarz, Munich, Germany

Increased capacity in 5G mobile communications requires rolling out massive MIMO base
stations with network and mobile terminals at both sub-6 GHz and mmWave frequencies.
Dynamic beamforming and the absence of RF test ports on the devices being tested make
over-the-air (OTA) measurement pivotal to 5G deployment. Fortunately, OTA testing solutions
employing software and hardware near-field transformations are meeting the challenges.

5
G new radio (NR) communication systems will
increase the capacity of mobile radio networks
using frequency bands in the sub-6 GHz region,
called frequency range 1 (FR1) by 3GPP, and the
mmWave range (FR2). New technological approaches
selected by the industry and 3GPP promise greater
bandwidth at lower operational expense.
In FR1, the main innovation effort is focused on the
base station, with the enabling of massive MIMO tech-
niques.1 4G systems use single-user MIMO, where the
user equipment (UE) calculates the inverse channel
matrix to extract separate data streams. 5G multi-user
MIMO (MU-MIMO) shifts the complexity from UEs to the
base station by using a pre-coding matrix. Here, each
data stream is received independently by separate re-
ceivers. Beamforming with antenna arrays of 64 to 512
elements reduces interference to adjacent users using
MU-MIMO. In addition to facilitating the adoption of
MU-MIMO to increase capacity, beamforming has other
advantages. Its lower energy consumption brings a re-
duction in overall network operating costs by targeting
individual UEs with their assigned signals.
Communication systems in the FR2 range use large
available bandwidths at frequencies around 28 and 39
GHz. The impact is more than 60 dB path loss at 1 m
distance and large electromagnetic field absorption in
WWW.MWJOURNAL.COM/ARTICLES/30835
22
nearby objects. As with FR1 systems, the solution is to
employ antenna arrays and beam steering, improving
the gain on both the mobile device and base station
sides of the network.
Whether for FR1 or FR2, 5G deployment relies on
the performance of highly integrated solutions com-
bining modem, RF front-end and antenna. The chal-
lenge is to define new methods and setups for perfor-
mance evaluation, as RF test ports tend to disappear
and beam steering technologies require system-level
testing. In this context, both antenna and transceiver
performance criteria must be measured OTA: effective
isotropic radiated power (EIRP), total radiated power
(TRP), effective isotropic sensitivity (EIS), total isotro-
pic sensitivity (TIS), error vector magnitude (EVM), ad-
jacent channel leakage ratio (ACLR) and spectrum
emission mask (SEM). Assessing these OTA raises the
critical question of the required measurement dis-
tance. Antenna characteristics are usually measured
in the far field. Using direct far-field probing and ap-
plying the Fraunhofer distance criterion (R = 2D2/λ), a
75 cm massive MIMO device under test (DUT) radiating
at 2.4 GHz should be evaluated in a chamber with at
least 9 m range length. Even a 15 cm smartphone trans-
mitting at 43.5 GHz needs a 6.5 m testing distance. This
distance is required to create a region encompassing

s Fig. 1 Spherical measurement system (ATS 1000), capable of
near-field software transformation, measuring a 28 GHz array.
the DUT where the impinging field is as uniform as pos-
sible and approaches a plane wave with phase deviation
below 22.5 degrees, known as the quiet zone.
Research shows that actual far-field behavior in the
peak directivity region can start much closer than the
Fraunhofer distance.2 These results proved, for exam-
ple, that the far-field EIRP or EIS of a 15 cm DUT radiat-
ing at 24 GHz can be assessed at a distance as short as
1.14 m. Distance reduction of about 70 percent comes
at the price of increased longitudinal taper error, caused
by the deviation of the apparent phase center from the
center of the measurement coordinate system. Also,
sidelobe levels cannot be evaluated accurately at short-
er distances.3 While direct far-field measurements at
shorter distances are not convenient for all applications,
there is an incentive to do so when conditions of appli-
cation are verified. This is because large OTA anechoic
chambers have high costs of ownership and limited dy-
namic range. Typical applications may be in the “white
box” case, where the antenna location within the device
and its aperture size are known.
NEAR-FIELD TO FAR-FIELD
Direct far-field measurements under “white box” as-
sumptions may be inappropriate when the radiation ap-
erture is larger than the quiet zone, the antenna cannot
be precisely identified within the DUT or multiple an-
tennas transmit simultaneously, e.g., from two extreme
edges of a DUT which does not fit within the quiet zone.
The “black box” scenario must then be considered,
where the radiating currents can flow anywhere within
the DUT. A first efficient approach to treat such cases in
23
a compact environment is to employ software near-field
to far-field transformations (NF-FF), for which the quiet
zone size question becomes irrelevant. Mathematical
implementations of NF-FF may vary, but the concept
is generally the same: at least two polarization compo-
nents of the electromagnetic field (E, H or a mixture of
the two) are measured in magnitude and phase over a
surface encompassing the DUT. The measured data is
processed using functions to propagate the fields to-
ward larger distances and extract far-field radiation
components. From the Huygens principle, the knowl-
edge of two phasors is enough to reconstruct exactly
all six field components outside the surface. Alternative
transformation methods use spherical wave expansion,
plane wave expansion or integral equation resolution,
with techniques to improve computational efficiency or
accuracy by taking parameters such as spatial sampling
rate, scanning area or truncation into account.
Figure 1 shows a commercial system capable of
both direct far-field and near-field measurements with
spherical scanning around the DUT using a conical cut
positioner. On this system, the DUT is positioned on a
turntable rotating in azimuth, while a dual-polarized Viv-
aldi antenna is mounted at the tip of a boom rotating
in elevation. An RF test port available at the DUT con-
nects one port of a vector network analyzer (VNA); the
measurement antenna ports connect to two other termi-
nals of the VNA, enabling near-field assessment through
measurements of complex S-parameters.
Near-field measurement methods often rely on un-
derlying assumptions about passive or RF-fed antenna
testing:
• The antenna feed port is accessible with a signal fed
to the antenna that is used as a phase reference.
• The RF signal is a continuous wave signal.
• Reciprocity applies so that transmit (Tx) and receive
(Rx) patterns at the same frequency are identical.
There are workarounds available in Tx cases where
such assumptions do not apply. For example, techniques
can address the case of a DUT transmitting a modulated
signal with no access to the antenna feed port. Hard-
ware and processing implementations to retrieve the
propagation phase vary, for example using interfero-
metric techniques or multi-port phase coherent receiv-
ers4 with the addition of a dedicated phase reference
antenna. For systems like those in Figure 1, this antenna
is typically attached to the azimuth turntable. Alternative
approaches include phaseless methods when the phase
information is retrieved from magnitude measurements.
However, the Rx mode is more complex. First, the
reciprocity assumption does not apply to mobile phone
and base station devices, as the Rx RF component chain
is, in general, different from the Tx RF chain. For a DUT
with no test port, the power available at the Rx input
of the RF front-end generated by an impinging wave
coming from the probe antenna (here used as the trans-
mitter) cannot be straightforwardly predicted in the near
field. In other words, it is not possible to isolate the in-
trinsic receiving properties of the DUT in the far field
from near-field coupling effects resulting from the test
setup. There is also no access to a phase reference, so
the NF-FF software transformation becomes inapplica-

Planar Wavefront
Spherical Wavefront
s Fig. 2 Compact antenna test range with a roll-edge reflector
collimating a spherical wavefront into a planar wavefront.
ble. Therefore, EIRP can be evaluated accurately in the
near-field using NF-FF software but not EIS.
TRANSCEIVER PERFORMANCE MEASUREMENTS
Another key question is the OTA evaluation of radio
transceiver performance, such as EVM, ACLR or SEM.
Software NF-FF approaches are designed for process-
ing periodic portions of the RF signal (the carriers) that
determine propagation. However, this part of the signal
is of no interest to assess these performance parame-
ters, so the challenge is to extract information from the
carrier modulation.
The first difficulty is that these quantities depend
strongly on the signal-to-noise ratio (SNR) at the receiver
(a spectrum analyzer in the Tx mode or the DUT in the
Rx mode). This can be overcome by first assessing the
complete 3D Tx or Rx pattern to determine the peak
direction. Demodulation and EVM or other measure-
ments can then be conducted at this specific location.
The question remains whether the obtained values are
reliable and reflect the results obtained in the far field.
In the case of a single transceiver, the near-field EVM
must be the same as the far-field EVM so long as the
SNR is above a certain threshold dependent on modu-
lation scheme, e.g., better than 20 dB. For multiple in-
dependent transceivers operating simultaneously, the
near-field EVM may not be straightforwardly related to
the far-field EVM because of positional dependence of
the noise figure in the near field.
HARDWARE NEAR-FIELD TRANSFORMATIONS
Alternative testing methods enable OTA assessment
in the near field without a software transformation, rath-
er a hardware-based one. The idea is to physically cre-
ate far-field conditions in a specified quiet zone region
within a short range. This is known as “indirect far field.”
A compact antenna test range (CATR) uses a mirror to
transform a spherical wave into a planar wave and vice-
versa. Using Fermat’s principle of least time, a planar
wave can be focused on a single point using a parabolic
mirror. If a measurement antenna is placed at this focal
point, using the reciprocity principle, a plane wave can
be generated as the parabolic mirror reflects a certain
planar component of the incoming spherical wave from
the measurement (or feed) antenna into the quiet zone
where the DUT is placed (see Figure 2).
The error inside a CATR system comes from two
main sources: the mirror geometry—the edge treat-
ment and smoothness of the surface which limit the fre-
24
quency range—and the feed antenna characteristics. If
the reflector is built with a simple parabolic section, the
sharp edges cause diffraction, which significantly con-
taminates the quiet zone by producing ripples as large
as 2 dB. Techniques to mitigate this phenomenon in-
clude serrations and rolled edges to scatter the energy
away from the quiet zone. The size and shape of the
serrated/rolled edges determine the lowest operating
frequency, where the surface roughness determines the
upper frequency. Feed antenna pattern characteristics
have a direct impact on the size of the quiet zone, as the
mirror essentially projects the radiation pattern of the
feed antenna onto the quiet zone. The reflector size with
serrated/rolled edges is generally at least 2× the DUT/
quiet zone size, where a reflector with sharp edges is 3
to 4× the size of the quiet zone. The optimum reflec-
tor to DUT separation distance is (5/3)× the focal length
of the reflector. An optimal focal length can be derived
from the manufacturing shape tolerances with a range
of roughly 0.3 to 1 for the ratio of focal length to para-
bolic diameter.
Since the quiet zone size is dependent on the reflec-
tor characteristics instead of range length, it is much
easier to create a large quiet zone inside small enclo-
sures. Figure 3 shows the measured magnitude quiet
zone size of 27 cm at 28 GHz of a CATR, similar to the
one shown in Figure 2, with a 42 cm × 42 cm reflector.
This CATR setup fits within a chamber as small as 2 m
× 1.5 m × 0.85 m. A direct far-field measurement sys-
tem featuring the same quiet zone size would require a
14.5 m range.
Such technologies are of great interest for testing
UEs or base stations operating in 5G NR FR2, promising
a significant decrease in the size of test environments.
In addition, CATR has the same capabilities as a far-field
30
27
24
21 –1
–2
18
15
12
9
–3
6
3
–4
–5 –6
0
0 3 6 9 12 15 18 21 24 27 30
50 cm Range Length Spherical System
1.5 m Range Length Spherical System
42 cm x 42 cm Reflector CATR 2 dB
s Fig. 328 GHz, 2 dB amplitude taper quiet zones for a roll-
edge compact range system vs. direct far-field systems.

s Fig. 4 R&S PWC200 showing the PWC antenna array and
calibration array mounted on a great-circle cut positioner.
s Fig. 5EVM of a single 20 MHz carrier measured using the
R&S PWC200.
system, i.e., instantaneous and direct measurements of
RF transceiver performance in both Tx and Rx. As the
path loss of such a system only occurs between the lim-
ited region where waves propagate between the feed
and the reflector, the dynamic range of a CATR system is
better than a direct far-field approach. Using Figure 3 as
an example, the CATR system has a focal length of 0.7
m compared to the equivalent far-field range length of
14 m, resulting in a path loss difference of 26 dB.
PLANE WAVE SYNTHESIS
A CATR reflector is typically built using a solid piece
of aluminum to maintain the strict surface geometry
requirements. The 5G FR2 DUT size requirements al-
low for compact and rather light reflectors (20 to
40 kg). In the 5G FR1 range, reflector weight significantly
increases, up to hundreds of kilograms for base station
DUTs. The cost, fabrication time and handling of large
heavy mirrors becomes prohibitive. A lightweight and
cost-effective alternative is to use an “electronic ver-
sion” of the CATR mirror.3 By combining the radiation
of multiple antennas assembled in a phased array and
fed with pre-determined signal magnitude and phase, a
plane wave within a defined quiet zone is created. A ver-
sion of this near-field focusing technique was used for
several years for the measurement of large phased array
radars at MIT Lincoln Labs and has been proposed as an
OTA measurement baseline for base stations by 3GPP.5
25
Figure 4 shows the plane wave converting (PWC)
system reported at the 2018 European Conference on
Antennas and Propagation.3 It comprises an array of 156
wideband Vivaldi antennas with a beamforming network
of phase shifters and attenuators, located at the rear.
This PWC array is 1.8 m wide and creates a spherical
quiet zone of 1 m diameter at a distance as short as 1.5
m in the frequency range from 2.3 to 3.8 GHz. In the
setup of Figure 4, the DUT (here the calibration antenna)
is mounted on a combined axis positioner, enabling full
spherical measurement. The calibration antenna is used
for evaluating the appropriate compensations of the in-
dividual RF channels of the PWC array, as well as deter-
mining the path loss of the entire test system. The PWC
system is reciprocal and has only one RF input/output
which can either be connected to a signal generator, a
spectrum analyzer or a VNA, enabling measurement of
devices with or without RF test ports.
Figure 5 shows the single-carrier EVM measured with
the PWC for an OFDM signal with five 20 MHz carri-
ers in the range of 2.35 to 2.45 GHz, using a Rohde &
Schwarz vector signal generator. The output power is
5 dBm and fed into a 60 cm × 60 cm patch array DUT.
The demodulation is carried out by a Rohde & Schwarz
vector signal analyzer connected to the PWC, where
the measurement span is 30.72 MHz. The EVM is as low
as 0.41 percent, roughly corresponding to the internal
EVM of the measurement instruments. EVM results were
below 0.5 percent for the other four carriers, showing
that the PWC adds negligible EVM to the measurement
setup.
SUMMARY
Near-field techniques employing software transforma-
tions are suitable for evaluation of EIRP and TRP quantities.
When Rx or demodulation is involved with a DUT using
multiple non-identical RF transceivers, methods utilizing
hardware field transformations such as CATR and PWC
overcome the limitations of software NF-FF. They also pro-
vide compact and reliable alternatives to direct far-field
measurements, making them well-suited for 3GPP RF con-
formance testing of UEs and base stations.n
References
1. C. L. I, C. Rowell, S. Han, Z. Xu, G. Li and Z. Pan, “Toward Green
and Soft: A 5G Perspective,” IEEE Communications Magazine,
Vol. 52, No. 2, February 2014, pp. 66–73.
2. B. Derat, “5G Antenna Characterization in the FF,” IEEE EMC &
APEMC 2018, Singapore, May 2018.
3. C. Rowell and A. Tankielun, “Plane Wave Converter for 5G Mas-
sive MIMO Base Station Measurements,” 12th EuCAP2018, Lon-
don, U.K., April 2018.
4. Derat et al., “A Novel Technology for Fast and Accurate Specific
Absorption Rate Measurement (SAR),” iWAT, Karslruhe, Germany,
March 2013.
5. Rohde & Schwarz, “2D Compact Range for Testing of AAS Base
Stations,” TSG-RAN WG4 #87, R4-1806605, May 2018.
Sub-6 GHz mMIMO Base Stations
Meet 5G’s Size and Weight
Challenges
Walter Honcharenko
MACOM, Lowell, Mass.

Base station deployment and site acquisition constraints require smaller, lighter 5G massive
MIMO (mMIMO) radios and antennas. Improved signal processing, high efficiency devices and
integration from discrete lineups to front-end modules (FEM) make it possible.

T
he RF and microwave industry has made con-
siderable progress toward enabling commer-
cial sub-6 GHz 5G wireless infrastructure, with
mmWave fixed wireless trials progressing in
parallel. The early excitement surrounding 5G has given
way to a defined set of industry standards, and compo-
nent and system manufacturers have variously aligned
on practical, scalable 5G base station architectures that
deliver the promise of faster data throughput and ex-
pansive capacity to serve subscriber, IoT and other ap-
plications.
The evolution from 4G to 5G—and the anticipated
100× capacity improvement required by our ever-
increasing demand for data—requires a fundamental
change in cellular communication RF system architec-
ture and design. With so many users, devices, automo-
biles, smart meters, low-power wide-area devices and
other machine-to-machine communication, the capacity
of 4G cellular systems employing fixed sector antenna
systems will fall short. At the highest level of communi-
cation theory, it is well known that maximizing through-
put over a wireless channel requires maximizing signal-
to-noise ratios (SNR) or signal-to-interference+noise
ratios (SINR). High density cellular networks are typically
interference limited, not noise limited, forcing the RF
WWW.MWJOURNAL.COM/ARTICLES/31754
26
architecture to evolve into a system where interference
is managed. Here is where mMIMO systems come into
the picture. mMIMO, with many more transceivers and
antenna elements than are used in 4G systems, uses
beamforming signal processing to direct RF energy to
users and reduce interference by dynamically steering
antenna beams away from interfering sources, in both
azimuth and elevation. By steering RF energy to users
and away from interference, SINR, throughput and over-
all system capacity improve.
mMIMO CHALLENGES
With 5G antenna array and mMIMO technology com-
ing to fruition, wireless network operators will face de-
ployment challenges as they make the transition from
4G LTE to 5G base stations, an incremental evolution
that will see both technologies comingled for what
is likely to be an extended period. Occupying similar
physical footprints, 4G LTE and 5G base stations will,
wherever possible, populate existing co-located cell
towers and rooftop installations, configured as they are
today to minimize interference and coverage gaps.
As 5G base stations proliferate across existing sites,
available installation space will shrink dramatically,
space that is already at a premium from continued 4G

s Fig. 1 Typical cell tower installation, with two tiers of radios
and antennas.
LTE base station deployments in some regions. Indeed,
many cell towers have already been pushed to the brink
of their hosting capacities, evidenced by the increasing-
ly and chaotically-cluttered towers dotting today’s metro
environments. Figure 1 shows a typical tower installa-
tion with two tiers of antennas, radios, RF cables and
power feed lines, which represent approximately 250 kg
weight on each sector. Wind loading, ice loading and
moment arms are key factors as base stations multiply
on a tower, with concern for base station resilience and
service continuity in poor weather conditions.
These challenges must be met with smaller, denser
sub-6 GHz 5G base station designs. In parallel, base
station weight and volume remain critical consider-
ations for system designers, given the nontrivial labor
and equipment costs imposed on wireless operators for
installation and subsequent maintenance. Where oper-
ating costs were calculated based on just the aperture
size of the antenna, tower operators have largely moved
to a pricing model where charges are calculated using
base station weight, aperture area and volume. Initial
installation costs are also determined by the location,
weight and type of installation: tower or rooftop, one or
two people, crane, etc. The initial 4G systems were split
27
into a radio head and an antenna, where the radio was
often on the ground and the passive antenna mounted
on the tower. In other installations, both radio and an-
tenna were located on the tower, with commensurate
costs. 5G mMIMO antennas, by definition, will position
the active electronics on the tower, immediately behind
the antenna, in a single integrated unit.
Of course, base station size and weight have always
and will always be central concerns for RF component
providers, base station designers and operators. A
looming shortage of tower and rooftop real estate will
only exacerbate these problems. On the path to real-
izing commercial-scale mmWave 5G connectivity, site
acquisition will become infinitely more difficult, given
the 100 m spacing between base stations that frequen-
cy and physics require for uniform coverage. mmWave
base station equipment installed on lamp posts, street
signs, bus stop shelters and other structures will need to
be far lighter and less obtrusive than anything that has
come before.
Site acquisition challenges will be compounded by
concerns over the effective isotropic radiated power
(EIRP). While 4G LTE and sub-6 GHz 5G base stations
may exhibit comparable EIRP levels when accounting
for beamforming gain, increasingly higher frequencies
will require higher RF power to compensate for build-
ing penetration losses and boost the EIRP to achieve
comparable indoor coverage. Diffraction losses, aper-
ture efficiencies and path loss all suffer as a function of
frequency (i.e., 6 to 12 dB/octave), while penetration
losses increase dramatically at higher frequencies due
to the skin depth and conductivity of coated glass, con-
ductive (moist) masonry, brick faces and other materials.
Health and safety requirements dictate that EIRP
emission limits (1 mW/cm2) and exclusion zones remain
within acceptable levels in the transition from 4G LTE to
5G, so raising EIRP levels will naturally introduce some
placement challenges. These will be compounded with
the implementation of mMIMO beamforming tech-
niques if theoretical maximum power is used. Where
conventional antennas point horizontally, beam steered
antenna arrays will radiate in many directions, even
down into pedestrian walkways. This health and safety
concern will introduce additional constraints for acquir-
ing urban 5G base station sites, intensifying the pressure
to design smaller, lower power base stations that can be
flexibly deployed while preserving safety.
REDUCING SIZE AND WEIGHT
When it comes to optimizing sub-6 GHz base station
size and weight, several design considerations must be
considered, from the component to the system, with
power consumption, efficiency and thermal dissipation
the most important.
Antenna aperture size is wholly dependent on the
number of onboard antenna elements, which depends
on the desired network capacity and expected interfer-
ence. Whether the array has 64, 128 or 192 elements,
the physical dimension is determined by the physics of
the array, scanning angle requirements, grating lobe
performance and beam widths. The volume and depth
of the base station, determined by the underlying elec-
tronics and heat sinking, can absolutely be optimized;
here, we see plenty of room for improvement.
A key system size consideration frequently over-
looked with 5G mMIMO is the dramatic increase in sig-
nal processing hardware required, compared to typical
LTE systems. The mMIMO system may have 192 antenna
elements connected to 64 transmit/receive (TRx) FEMs
with 16 transceiver RFICs and four digital front-ends
(DFE), a 16× increase in digital signal processing com-
pared to the four transceivers in a typical LTE 4T MIMO
system (see Figure 2). Add a 5× increase in bandwidth
when moving from 20 to 100 MHz, for example, and the
signal processing multiplier is staggering. The stackup
in Figure 3 illustrates the functions in a typical mMIMO
integrated antenna and radio. The top panel contains
4T Antenna mMIMO Antenna
Integrated Assembly
the antenna elements, and the layers below the antenna
contain the RF and digital circuitry. While the TRx FEM,
RFIC and DFE layers are shown as separate boards, in
practice the three functions will be combined into one
or two densely packed boards to minimize interconnec-
tions.
Perhaps even more jarring than the additional hard-
ware in a mMIMO system is the attendant impact on
power consumption and heat dissipation. Previously,
power amplifier (PA) power consumption was the most
important factor when designing base station heat sinks
and power supplies. Now, the power consumption of
the signal processing electronics is approaching, and in
some cases, eclipsing that of the onboard PAs.
The significant increase in signal processing hard-
ware can be offset to an extent by optimizing the sig-
nal and waveform conditioning algorithms applied
to the transmitted signals. Legacy signal condition-
ing algorithms such as crest factor reduction and digi-
tal predistortion (DPD) were primarily developed for
macro base stations with very high-power PAs, a more
complicated and taxing processing workload than is
required for the smaller, lower power PAs populating
mMIMO antennas. These algorithms easily consumed
75 percent of the available signal processing resourc-
es in the DFE processors, whether custom ASIC/SOCs
or FPGAs. By streamlining these algorithms for 5G

mMIMO architectures and redistributing the functional-

ity across several logic blocks, a minimal set of optimized
algorithms will improve signal processing efficiency, re-
ducing overall power consumption.
To illustrate the 16× multiplier of the digital signal
processing and transceivers in a mMIMO system, a func-
tional block diagram is shown in Figure 4. This architec-
A/D

A/D

A/D

A/D
D/A

D/A

D/A

D/A

ture is typical of all mMIMO designs, with some differ-

DFE
ences in the partitioning of logic (e.g., 8- or 16-chan-
Signal Conditioning
nel DFEs) or discrete component instead of integrated
CFR/DPD/Tx/Rx FEMs. This example shows, from left to right, 64 RF and
FEM transceiver paths divided among 16 transceiver RFICs
RFIC
driving four DFEs. The DFEs process the digital data
4T LTE RRH mMIMO Radio from the 64 channels and are connected to the beam-
forming processor and baseband interface processor.
CPRI CPRI
The emergence of RF SOCs with direct sampling ana-
Baseband Baseband log-to-digital converters (ADC) and digital-to-analog
(a) (b) converters (DAC) capable of ~60 GSPS help to shrink
the size and weight of 5G antennas by reducing the
s Fig. 2 Architecture of a conventional 4T LTE remote radio steps required for analog up- and down-conversion in
head (a) vs. a mMIMO radio containing 192 antenna elements conventional transceiver architectures. This reduces the
and 64 TRx FEMs (b).
overall component count and cost by eliminating mix-
ers, converters and local oscillators.

FEM BENEFITS
The emergence of GaN on Si PAs provides wide-
band performance and superior power density and ef-
ficiency compared to LDMOS devices, meeting the ex-
acting thermal specifications while preserving valuable
PCB space for the tightly-clustered mMIMO antenna
arrays. Space-saving multifunction MMICs and multi-
Baseband chip modules (MCM) are supplanting discrete ICs and
single-function devices, enabling integrated RFICs for
5G base stations. FEMs are benefiting from a similarly
s Fig. 3 Notional mMIMO radio stackup. streamlined design approach using integrated assem-
28
 FEM 4 Channel RFIC/Transceiver 16 Channel DFE Function
Pulse
D/A P Q DPD CFR Shaping IFFT

Q P Pulse
A/D Shaping FFT

FEM
Pulse
D/A P Q DPD CFR Shaping IFFT
Pulse
A/D Q P Shaping FFT

FEM 4 Channel RFIC/Transceiver Beamforming/

Channel
Pulse
D/A P Q DPD CFR Shaping IFFT Estimation/ to Baseband
Coefficient
Q P Pulse Management
A/D Shaping FFT

FEM
Pulse
D/A P Q DPD CFR Shaping IFFT

Q P Pulse
A/D Shaping FFT

64 Total FEMs 16 x of 4 4 x of 16
in 64 TRx Architecture Channel RFIC Channel DFE

s Fig. 4 Typical mMIMO radio functional block diagram.

Radome
To Filter
and TX In AI or Mg
Antenna Housing

PA Control

RX Out

50 Ω

s Fig. 5 Simplified FEM block diagram.

blies incorporating PAs, T/R switches, matching circuits,

low noise amplifiers, digital step attenuators, controllers
Thermal
and DPD couplers packaged in compact packages (see Vias
Figure 5). With drain efficiencies approaching 60 per-
cent and optimized integration of the Tx and Rx com- Thermal
ponents, as well as DPD feedback paths, using FEMs in Pads

mMIMO radios and TRx boards has many benefits: Bonded

• Reuse portions of the transceiver board layout.
• Optimize device-to-heat sink thermal management.
• Optimize power levels, feedback loops, VSWR and
control circuitry.
• Manage isolation and noise within the FEM.
• Enable dynamic power saving modes.
• Improve final yields compared to discrete designs,
since the integrated FEM is fully tested.
Using the FEM design approach, redesigning a mMIMO
radio for a different number of antenna elements, frequen-
cy band or power level is simplified, as FEMs are “plug and
Fins
s Fig. 6 Integrated mMIMO radio.
play” modules, with standardized interfaces, control logic
and RF levels part of the design methodology.
THERMAL DESIGN
In all mMIMO designs, where the antenna and elec-
tronics are contained in one enclosure (see Figure 6), the
majority of the product engineering focus is managing
thermal performance. The engineering efforts for sig-

29
 Radome
AI or Mg
Housing
Thermal
Vias
Thermal
Pads
Bonded
Fins
s Fig. 7 Integrated mMIMO radio cross section showing layers and heat conduction paths.
nal processing, RF design, digital design, board layout,
power design are indeed complex, but ultimately the
mechanical/thermal/design and product environmental
requirements will determine the volume and weight.
Conventional 4G radio heads are built with the radio in-
side the heat sink, fins surrounding the entire package.
With a mMIMO design, the antenna and its radome are
very poor conductors of heat, limiting thermal dissipa-
tion to the rear of the mMIMO radio.
By leveraging advanced packaging techniques for
the MMICs and MCMs within the FEM, additional cool-
ing and space-saving benefits can be achieved. Figure
7 illustrates a simplified mMIMO design, not including
the power supplies and fiber interfaces. The case hous-
ing has extruded fins bonded into the housing to save
casting weight and increase thermal efficiency. The TRx
board integrates the FEMs and RFICs, with the FEMs
conducting heat down through thermal vias, while the
heat from the RFICs conducts out through the lid. This
allows heat to be dissipated in multiple directions, rath-
er than unidirectionally from the FEMs and RFICs. Heat
can be removed through the top lid and the bottom of
the package through the ground vias and baseplate,
30
Thermal Dissipation
distributing heat more efficiently and enabling a cooler
device in a smaller footprint. Alternatively, the FEM can
channel heat through both the thermal vias and the lid,
to dissipate as much heat as possible.
SUMMARY
The impending proliferation of 5G base stations op-
erating sub-6 GHz, later at mmWave, will undoubtedly
strain tower and rooftop deployment flexibility and site
acquisition options for the foreseeable future. By alleviat-
ing the signal processing and conversion workload and
exploiting higher levels of integration, from discrete com-
ponents to FEMs, significant reductions in base station
size and weight can be achieved.
The expected roadmap for FEMs, SOCs and full
single module solutions, from optical in to RF out, is a
natural progression of technology. Integration of the op-
tical interfaces, with direct sampling RF Tx and Rx and
the required signal conditioning, will define a true SOC.
These evolving capabilities will enable 5G mMIMO base
stations to become ubiquitous, fitting comfortably in the
contours of our metro and suburban landscapes.n
RF Matters Here …
MACOM is committed to delivering industry-leading diode design
and application-specific solutions
Ka-Band 40 W Terminated Switch for our customers worldwide.
Ideal for mmW Radar, SATCOM &
5G Infrastructure
5G Infrastructure
Insertion Loss & Isolation vs. Frequency
0.0 0
-0.2
Part #: MASW-011094
-4
Join the Next Generation
MACOM’s 65-year legacy of innovation
Insertion Loss S21 (dB)

-0.4 -8
Isolation S21 (dB)

-0.6 -12
is driving the industry’s broadest
-0.8 -16
-1.0 -20 portfolio of MMICs, diodes and
-1.2 -24 transistors for the entire RF signal
-1.4 -28 chain. These trusted high-performance
-1.6 -32
-1.8 -36
RF devices enable your most critical
-2.0 -40 applications including Radar, SATCOM,
20 22 24 26 28 30 32 34 36 38 40 T&M, ISM and current 4G LTE to next-
Frequency (GHz)
gen 5G connectivity.
Ka-Band Solution
Ka-Band SATCOM SATCOM Solution With our state-of-the-art technology
IF GAIN POWER FRONT END and high-performance products,
BLOCK MIXER PRE-AMPLIFIER DRIVER AMPLIFIER SPDT
IF, L-BAND
we’re helping customers achieve
leading bandwidth, power, packaging
TX 29 - 30 GHz
and reliability.
LO DRIVER xN VCO
AMPLIFIERS

RX 18 - 22 GHz

IF GAIN MIXER GAIN BLOCK LNA

BLOCK

Learn more at:

www.macom.com/diodes