5G Development

with MATLAB
5G Development with MATLAB

1. Technology and Design

2. New Architectures and Algorithms

3. Accelerating Prototypes and Field Trials

4. System Verification and Testing

1. Technology and Design
What is 5G?

5G (5th generation wireless systems) is the next major phase of mobile

telecommunications standards. The scope of 5G will ultimately range
from mobile broadband services to next-generation automobiles and
connected devices.

The initial 5G New Radio (NR) specification was completed in June

2018 and published in the 3GPP Release 15 specification. Now, a
variety of industry players, including network equipment vendors,
network operators, semiconductor vendors, and device manufacturers,
are developing new products that implement the new standard.

If you are already familiar with 5G, feel free to skip ahead to sections
2–4, which discuss strategies for doing 5G design and development
with MATLAB®: new algorithm design (Section 2), accelerating
prototyping and field trials (Section 3), and system verification
(Section 4).

5G Development with MATLAB 4

What’s driving 5G?

Two major trends are behind the race to 5G: the explosive growth
in demand for wireless broadband that can carry video and other
content-rich services, and the Internet of Things (IoT), where large
numbers of smart devices communicate over the Internet. To achieve
these objectives, 5G will provide extreme broadband speed, ultralow
latency, and ultrareliable web connectivity.

5G networks and devices will require substantially different

architectures, radio access technology, and physical layer algorithms.
Dense networks of small cells will complement macro base stations,
operating at millimeter wave technologies and employing massive
MIMO antenna arrays. And the processing components within network
equipment and user devices will become more integrated
and adaptive.

Innovations like hybrid beamforming are stretching the old ways of

developing wireless systems. These highly integrated technologies
require a corresponding integration of engineering domain expertise
and tools.

5G Development with MATLAB 5

5G Terminology and Expected Applications

By providing higher bandwidth capacity than current 4G–supporting eMBB—Enhanced Mobile Broadband
broadband, 5G will enable a higher density of mobile broadband
• For high-capacity and ultrafast mobile communications for phones
users and support ultrareliable device-to-device and massive
and infrastructure, virtual and augmented reality, 3D and ultra-HD
machine-type communications.
video, and haptic feedback

URLLC—Ultrareliable and Low Latency

Key 5G Parameters
• For vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I)
communications, autonomous driving
Latency in the air link <1 ms
mMTC—Massive Machine-Type Communications
Latency end-to-end
<10 ms • For consumer and industrial IoT, Industry 4.0 mission-critical
(device to core)
machine-to-machine (MC-M2M)
Connection density 100x vs. current 4G LTE

Area capacity density 1 (Tbit/s)/km2

System spectral efficiency 10 (bit/s)/Hz/cell

Peak throughput
10 Gbit/s
(downlink) per connection

Energy efficiency >90% improvement over LTE

5G Development with MATLAB 6

5G Standard: Timeline

The 3GPP standardization group defines the wireless 5G standard, Current 5G development focuses on enabling technologies such
with help from many participants and contributors around the as flexible baseband and RF technologies, hybrid beamforming,
globe. Release 15 of the 3GPP standard, finalized in June 2018, and massive MIMO systems; rapid prototyping and field trials; and
introduces the 5G standard. Physical layer algorithms, millimeter wave verification of compliance with the new standard specifications.
technology, and massive MIMO architectures are expected to be
significantly different from 4G LTE technologies.

FIRST RELEASE OF 5G SECOND RELEASE OF 5G

SPECIFICATION: 06/2018 SPECIFICATION: 12/2019

3GPP STANDARDIZATION TIMELINE

5G RESEARCH, PROTOTYPE, AND TRIAL 5G STANDARD 5G PRODUCT 5G DEPLOYMENT

RELEASE 10 RELEASE 11 RELEASE 12 RELEASE 13 RELEASE 14 RELEASE 15 RELEASE 16

LTE ADVANCED (PHASE A) LTE ADVANCED (PHASE B) LTE ADVANCED (PHASE C)

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021

5G Development with MATLAB 7

A New Physical Layer for 5G

The 5G physical layer will depart from 4G LTE in a number of ways, Sections 2–4 of this ebook discuss 5G design and development
in order to improve spectral efficiency and data rates. One distinctive strategies in detail: new algorithm design (Section 2), fast prototyping
feature is a significant jump in the number of active antennas and (Section 3), and system verification and field trials (Section 4).
antenna arrays, and the related issues of beamforming and millimeter
wave RF signal processing. New modulation and coding schemes,
power and low-noise amplifier designs, and channel models all need
to be developed.

DATA SIGNALS

5G Development with MATLAB 8

Millimeter Wave Higher-Frequency Operation

Higher data rates (multi-Gbps) drive the need for greater bandwidth High frequencies will provide larger bandwidth availability and
systems, and the available bandwidth in the spectrum up through smaller antenna dimensions for a fixed gain, or higher gain for a given
7.125 GHz is not sufficient to satisfy these requirements. (For reference, antenna size. However, this increases modem complexity in baseband
current cellular operation is below 3 GHz.) This has moved the target and RF designs. To study the performance, we also need an accurate
operating frequency bands up into the millimeter wave (mmWave) channel model for the new frequencies in 5G.
range for the next generation of wireless communication systems.
For example, equipment developers have performed 5G NR trials
with mobile network operators to demonstrate the viability of higher
frequency transmissions. Those trials operate in the midband spectrum
from 3.3 GHz to 5.0 GHz, as well as the mmWave spectrum at 28
GHz and 39 GHz, showcasing the unified 3GPP-based 5G NR design
across diverse spectrum bands.

PEAK RATE PEAK RATE

1 Gbps 50 Gbps

18 28 38 60 GHz

FREQUENCY BAND

LEGACY BANDS NEW BANDS

5G Development with MATLAB 9

Massive MIMO: More Antennas

Another key technology for achieving greater spectral efficiency is

massive MIMO. Massive MIMO, sometimes referred to as large-scale
MIMO, is a form of multiuser MIMO in which the number of antennas
at the base station is much larger than the number of devices per
signaling resource. The large number of base station antennas relative
to user devices results in a channel response that is quasi-orthogonal
and has the potential to yield huge gains in spectral efficiency.

Designers face a challenge when scaling the number of antennas to

hundreds:

• The simulation speed of traditional antenna design tools is slow for

large antenna arrays.

• It is difficult to simulate the antenna coupling.

• Hybrid beamforming is needed to optimize the number of the RF

chains.
A massive MIMO antenna array for a Huawei 5G field trial.
See Section 2 for details on design architectures and algorithms.

5G Development with MATLAB 10

Fast Prototyping of 5G Systems
with Hardware Testbeds
Engineers working on 5G designs have realized the value of rapid
design iterations and of placing proof-of-concept prototypes in
field trials quickly. Hardware testbeds employed as flexible and
reconfigurable design platforms have proven to be dependable and
efficient for the rapid design and verification of new concepts, as
well as for their deployment in precommercial field trials. The tools
and workflows that interface to testbeds must support rapid design
iterations and rapid deployment of new algorithms or design changes.

A fast prototyping workflow is discussed in Section 3.

5G Development with MATLAB 11

Learn More

Ready for a deeper dive? Explore these resources to learn more about 5G
technology development and design.
Watch
Design 5G Wireless Technologies with MATLAB and Simulink Chalk Talk (32:00)
Understanding and Modeling the 5G NR Physical Layer (74:01)
5G NR Physical Layer Modeling in MATLAB (25:21)
5G Explained (12 Videos)

Explore
5G Wireless Technology Development

Read
Advancing the 5G Wireless Standard at Convida Wireless: An Insider Look
2. New Architectures and Algorithms
New Design Architectures and Algorithms for 5G

The leap in 5G broadband speeds will be enabled by massive MIMO

communication in the millimeter wave (mmWave) frequency range and
by new radio algorithms that achieve more efficient use of spectrum.
New design architectures and algorithms will affect every aspect of
5G systems, from antennas to RF electronics to baseband algorithms. RF Transceivers
The performance of these subsystems is so tightly coupled that they Nonlinearity, noise, and memory

must be designed and evaluated together.

Waveforms ANTENNA Antennas

5G, LTE, WLAN, custom Array elements, configuration
TRANSMITTER

Digital CHANNEL
Baseband DAC PA
Front End

DIGITAL PHY RF FRONT END

Digital
Baseband ADC LNA
Front End

RECEIVER

Measurements
EVM, BER, ACLR

Channel
Algorithms Noise, interference, fading
Mixed-signal
Coding, synchronization, beamforming, DPD Discrete-time and continuous-time

5G Development with MATLAB 14

New 5G Physical Layer Algorithms for Greater
Flexibility and Spectral Efficiency
The 5G wireless communication standard will provide significantly
higher mobile broadband throughput with its enhanced mobile
broadband (eMBB) mode. Several techniques and features have been
standardized for 3GPP Release 15.

Among the key elements of 5G NR in 3GPP Release 15 are:

• Shorter slot durations, corresponding to increased subcarrier

spacing, for increased signal bandwidth and shorter latency

• New channel coding methods such as LDPC for data and polar
codes for control information, for more efficient error correction and
improved data rates

• Spatial channel models for operation at current (<7.125 GHz) and

mmWave (>24 GHz) frequencies
Using the 5G Waveform Generator to generate a 100 MHz fixed reference
These elements have the potential to improve bandwidth, latency, channel (FRC) with 30 kHz spacing according to TS 38.101-1.
and system efficiency, but they can add complexity and delay to your
design. New tools available for use in MATLAB and Simulink let you
quickly characterize the new features of 5G NR and make critical
design tradeoff questions at an early stage.

5G Development with MATLAB 15

New 5G Physical Layer Algorithms for Greater
Flexibility and Spectral Efficiency continued
To help engineers developing 5G standard-compliant products and
researchers exploring future 5G technologies, Mathworks has introduced
5G Toolbox. As the 5G standard evolves, 5G Toolbox will strive to
evolve its functionality in lock step to provide algorithms and apps for
analysis, design, simulation and testing of 5G communications systems.

5G Toolbox provides:
• Uplink and downlink 5G NR waveform generation including
predefined (test models, FRCs) and custom waveforms based on
5G NR Mixed numerology, CP-OFDM, and SC-FDMA

• 5G uplink and downlink physical channels and signals such as

PDSCH/PUSCH, PDCCH/PUCCH, synchronization burst, DMRS
(demodulation reference signals), CSI-RS, SRS, and PT-RS

• Channel models as specified in TR 38.901 including tapped delay

line (TDL) and clustered delay line (CDL)

• Customizable link-level simulation reference design, enabling you to Results of block error rate simulations using polar coding.
measure throughput of a downlink (PDSCH) or an uplink (PUSCH) 5G
link over 2D or 3D channel models

• Synchronization procedures including cell search and MIB decoding

in the context of PBCH beam sweeping

• Standard measurements including EVM and ACLR

5G Toolbox functions are implemented as open, editable MATLAB code,

so you can easily understand and customize algorithm implementations.

5G Development with MATLAB 16

New 5G Physical Layer Algorithms for Greater
Flexibility and Spectral Efficiency continued
Using 5G Toolbox, you can measure the impact of different algorithms and design choices on system performance by simulating end-to-end system
performance along with realistic 5G propagation channels.

CHANNEL MODEL: CP-OFDM CHANNEL PDSCH DL-SCH

DL-SCH PDSCH PRECODING CP-OFDM TIMING SYNCH
CDL OR TDL DEMOD ESTIMATION DECODING DECODING

HARQ

5G Resource Grid 3GPP 38.901 Channel Model Throughput Measurement

Using a 5G downlink reference model, you can insert custom algorithms to simulate and
optimize end-to-end link performance and validate standard compliance.

5G Development with MATLAB 17

New 5G Physical Layer Algorithms for Greater
Flexibility and Spectral Efficiency continued
You can investigate the impact of filtering and the RF front-end on the waveform and explore tradeoffs for ACLR and EVM.

Measuring the impact of filtering on an NR-TM waveform.

5G Development with MATLAB 18

Massive MIMO Arrays for High-Gain Beamforming

5G mmWave designs require massive MIMO antenna arrays with Having many antenna elements in a small area makes it practical to
hundreds of antenna elements on base stations (eNodeB). Because the achieve a high beamforming gain. The highly directional beams help
area of an antenna array is reduced in proportion to the wavelength, offset the increased path loss at the higher operating frequencies,
an array for mmWave frequencies can be up to 100x smaller than an because the beams steer power in a specific direction.
array for microwave frequencies.

UEs in a group using beams with same elevation angle (left) and a hybrid beamforming array architecture (right).

5G Development with MATLAB 19

Behavioral Simulation for Massive MIMO

Typical array designs include parameters such as array geometry, design from system architecture and signal processing algorithms.
element spacing, the lattice structure of the elements, element tapering, MIMO simulation times are also typically 10x longer than 3G and 4G
and the effects of mutual coupling. By adjusting the design parameters, simulations.
you can achieve tapering of the rows and columns of the array to
Behavioral-level simulation of the antenna array system can address
reduce side lobe levels.
these challenges. Simulating at the behavioral level reduces the
Achieving an optimal design thus requires combined models of simulation time. This enables engineers to experiment with different
the antenna arrays and beamforming algorithms to simulate their array architectures and algorithms, simulate the performance of the
interaction and impact on system performance. This puts a strain on array and associated algorithms, and iteratively adjust parameters to
current 3G and 4G design tools, which typically separate antenna mitigate the effect of antenna coupling.

Massive MIMO antenna array design, which requires simulating the interactions between antenna, RF, and digital subsystems.

5G Development with MATLAB 20

Optimizing Tradeoffs Between Antenna Gain
and Channel Capacity
As a part of the design process, the larger antenna gains achieved with narrower beams must be balanced with the fact that MIMO systems are
based on scattering environments that also depend on broader beam patterns to maximize channel capacity.

This is one of the key tradeoffs to assess, and it can be done iteratively with behavioral simulation.

The figures below show how you can develop and visualize behavioral simulations with Phased Array System Toolbox™ and Antenna Toolbox™
for MIMO array design tasks.

Configure the Initial Array Experiment Interactively with Design Parameters

Use antenna element models, such as omnidirectional or cosine Vary the array size, array geometry, element spacing, and tapering.
elements, and rapidly move to more accurate analysis using patterns Visualize the resulting geometry, 2D and 3D directivity, and the grating
computed with electromagnetic (EM) tools or measured in the lab. lobe diagrams.
Visualize the array performance characteristics, such as the radiation
This example shows a beam pattern and grating lobe diagram for
pattern shown below.
66 GHz 64x64 element design, designed with Antenna Toolbox.

8x1 ULA subarray and corresponding radiation pattern. Beam pattern and grating lobe diagram for 66 GHz 64x64 element design.

5G Development with MATLAB 21

Designing Antenna Array Elements

Detailed design of antenna elements, with appropriate radiation

patterns, can then be added to improve the fidelity of an antenna
array model.

The figure at right illustrates an element pattern generated using a full

wave EM solver in Antenna Toolbox. The toolbox uses the method
of moments (MoM) algorithm to compute port properties such as
impedance, surface properties such as current and charge distribution,
and field properties such as the near-field and far-field radiation
pattern.

You can use Antenna Toolbox to visualize antenna geometry and

analysis results in 2D and 3D. You can also integrate antennas and
array models into wireless systems and use impedance analysis
to design matching networks. The toolbox also provides radiation
patterns for simulating beamforming algorithms.

The Antenna Designer app, which provides interactive tools for

antenna selection, design, and analysis.

5G Development with MATLAB 22

AI-Driven Optimization for 5G Antenna
and Array Design
Antenna optimization is critical to satisfy 5G requirements. Many variables Antenna Toolbox uses the surrogate model assisted differential evolution
affect the antenna performance, such as dimensions, amplitude tapers, for antenna synthesis (SADEA) optimization technique. It is an AI-driven
and phase shifts for each element of the array. Manually covering a huge design method, based on machine learning and evolutionary computation
design space to find the best configuration can take a lot of time and techniques. It performs global optimization by creating a surrogate model.
experience. The technique is specifically designed for electromagnetic modeling, in
which each optimization run can be time-consuming.
To automate this process, you can use optimization methods based on
artificial intelligence. The Antenna Designer and Antenna Array Designer
apps provide predefined objective functions and built-in constraints, so you
don’t need to be an optimization expert to set up the problem.

Automating antenna optimization using the Antenna Array Designer app in Antenna Toolbox.

5G Development with MATLAB 23

Hybrid Beamforming

While smaller wavelengths enable massive MIMO implementation Hybrid beamforming is a technique to partition beamforming between
within small form factors, signal path and propagation challenges the digital and RF domains to reduce the cost associated with the
associated with mmWave frequencies also increase. To achieve number of RF signal chains. Hybrid beamforming combines multiple
better beamforming control and flexibility, it would be ideal to have array elements into subarray modules, with one T/R module dedicated
independent weighting control over each antenna array element, with to a subarray in the array.
a transmit/receive (T/R) module dedicated to each element. But this is
generally not practical due to cost, space, and power limitations.

DAC RF RF ADC

BASEBAND RF RF BASEBAND
PRECODING ARRAY ARRAY COMBINING

DAC RF RF ADC

Hybrid beamforming architecture.

5G Development with MATLAB 24

Hybrid Beamforming continued

A key challenge in hybrid beamforming design is to meet

the required performance parameters while meeting the
RF Subarrays
implementation cost constraints. Simulink® enables unified,
multidomain modeling and simulation of the RF domain and
digital domain components. Circuit envelope simulation in RF
Blockset™ ensures fast simulation of the hybrid system.
Digital Beamformers
In Simulink, you can evaluate the number of elements and
precoding parameter values to ensure system-level performance
is met across a range of steering angles. The digital and RF
beamforming weights can be calculated in MATLAB® and
incorporated into the Simulink model.

The figure shows a section of a multidomain model containing

digital beamforming weights used to shape the signals feeding
the RF subarrays, where phase shifts are applied. The resulting Array Pattern
hybrid weights produce the desired array pattern.

Hybrid beamforming design in MATLAB.

For a detailed discussion, see the white paper Hybrid Beamforming for Massive MIMO Phased Array Systems. The white paper uses a 64x64 element, 66 GHz millimeter wave example and demonstrates antenna array
modeling and partitioning of beamforming between the digital and RF domains.

5G Development with MATLAB 25

 Modeling and Linearizing
Power Amplifiers
The linearity of power amplifiers (PAs) is the critical specification of power amplifiers are nonlinear and are affected by finite memory, the
every transmitter. Backing off power amplifiers to operate in the highly characterization of power amplifiers strongly depends on the type of
linear region is simply not a viable commercial solution, especially at signal used to drive it. For example, the behavior of the PA depends on
higher frequencies and for larger bandwidths. For this reason, digital the bandwidth of the driving signal, on its spectral occupation, and on
predistortion (DPD) techniques are applied to increase the efficiency its peak-to-average-power ratio (PAPR).
of the transmitter and at the same time limit spectral regrowth and
Because of this complexity, DPD algorithms are often developed in the
interchannel interference.
lab, using rapid prototyping platforms that enable you to test algorithms
Developing a quality DPD algorithm is challenging, because it together with the actual PA. While this approach is useful to validate
requires a deep understanding of the effects introduced by the power and fine-tune the algorithms, it is harder to apply when the actual PA is
amplifier and adjacent subsystems, such as the antenna. Because not yet available, or to explore the algorithmic DPD design space.

Example of a memory polynomial model of a Power Amplifier identified with MATLAB

using measurement data, and simulated in RF Blockset using Circuit Envelope.
Conceptual representation of amplifier linearization using digital predistortion.

5G Development with MATLAB 26

 Modeling and Linearizing
Power Amplifiers continued
For these reasons, a model-based approach is recommended to Once the model is identified, it can be used within a system simulation
explore and develop DPD algorithms before lab prototyping and environment together with realistic (and standard-compliant) baseband
testing. However, this is only possible when a good quality model of signals, models for the low-power RF transmitter and observer receiver,
a power amplifier is available. RF Blockset provides models of power the antenna termination expressed with S-parameters, and different
amplifiers at different levels of abstraction, including generalized types of adaptive DPD algorithms. With this approach, designers
memory polynomial models that are a convenient derivation of Volterra can innovate more rapidly and validate new ideas while taking into
series. The behavioral models are identified using power amplifier account dispersive and nonlinear effects that are otherwise hardly
input/output characteristics, coming from circuit-level simulation done reproducible and understood in the lab.
with IC design tools or from actual measurements. Designers can
leverage the provided identification routing or use their own procedure
to compute the series coefficients. In a matter of seconds, users can
experiment using different polynomial orders and memory depth.

Closed-loop transceiver model with power amplifier and adaptive DPD algorithm. The
lab-validated AD9371 models include real-life effects. The Volterra series model of the
power amplifier includes non-linearity and memory effects. The loop simulation includes
low and high-power effects, timing, and frequency selectivity over the signal bandwidth.
The adaptive DPD algorithm improves the device linearity within the signal bandwidth. Plots showing spectrum analysis (bottom right) and received constellation (top left).

5G Development with MATLAB 27

Learn More

Ready for a deeper dive? Explore these resources to learn more about new
architectures and algorithms for 5G systems.
Watch
What Is 5G Toolbox? (2:06)

Spatial Multiplexing and Hybrid Beamforming for 5G Wireless Communications (28:35)

Modeling RF Power Amplifiers and Increasing Wireless Transmitter Linearity with DPD Using MATLAB (25:11)

Explore
5G Wireless Technology Development

Beamforming

RF Systems

Channel Modeling

Massive MIMO
3. Accelerating Prototypes and Field Trials
Accelerating Prototypes and 5G Field Trials

To evaluate the performance of new 5G algorithms and architectures,

engineers need to develop proof-of-concept prototypes and prepare
Algorithms
the new designs for field trials. Typically, the prototypes are built using
FPGA hardware with embedded processors for some portions of the and
designs. These platforms are commonly known as hardware testbeds, Models
and they enable rapid prototyping and testing of new technologies
and design changes in the field.

It can be difficult for a typical R&D team to implement FPGA-

based radio prototypes and testbeds without outside assistance.
R&D engineers have strong signal processing and communications
algorithm development expertise, but relatively little experience with
hardware implementation. This experience gap is compounded by a
tool and workflow gap. While R&D engineers typically use a high-level
language like MATLAB, hardware engineers use their own design tools
and hardware description languages (HDLs).

Leading companies instead have turned to Model-Based Design with

MATLAB and Simulink to bridge these gaps and enable R&D engineers
to quickly move new 5G algorithms and design changes to their FPGA-
based test platforms.

Rapid prototyping of new algorithms on FPGA-based hardware for 5G field trials.

(Image courtesy of Huawei)

5G Development with MATLAB 30

Rapid Prototyping with
Model-Based Design
Many wireless engineers rely on MATLAB to develop and simulate With Model-Based Design, 5G R&D teams can:
physical layer communication algorithms. Model-Based Design with
• Explore algorithms and develop a standard-compliant reference and
MATLAB and Simulink enables wireless engineers to use a common
test bench in MATLAB
model for both simulation and implementation on the FPGA or SoC
(system-on-chip) prototyping platform. They no longer need to maintain • Develop a hardware-accurate model in Simulink using fixed-point
different representations of the same design, and they can use MATLAB algorithm blocks
code as a reference and test bench to verify the hardware design.
• Partition the model into subsystems to target the FPGA fabric
Using Model-Based Design, the R&D and system engineers can and processor on the prototype
develop and debug hardware prototypes of their algorithms without
hardware specialists. They can quickly update prototypes in response • Automatically generate target-independent or target-optimized
to changes in requirements. Instead of handing the hardware team an HDL and C code
abstract algorithm to interpret and implement, they can hand off an • Automate prototyping on SDRs and other FPGA or SoC hardware
efficient, fully verified VHDL® or Verilog® implementation.
• Integrate and test the generated code into the full radio
platform design

5G Development with MATLAB 31

Rapid Prototyping with
Model-Based Design continued
This process produces fully verified algorithm implementations, running ALGORITHM DESIGN (MATLAB)
C and HDL code automatically generated from a Simulink model.
Algorithm Golden
Available support packages for commercially available software- Exploration Reference
defined radio (SDR) frameworks further automate integration into a
real-time radio application that transmits and receives signals. Test Bench

To deploy the verified algorithms onto a custom hardware testbed, you

can integrate the generated code with other custom interfaces and IMPLEMENTATION
external IP cores using standard FPGA development tools. DESIGN (SIMULINK)

When changes are required, you can quickly modify the algorithm Architecture VERIFICATION
model, verify it in simulation, regenerate the code (typically within
Timing Fixed-point
minutes), integrate the code in the FPGA development environment,
and synthesize it for FPGA implementation. The generated code is well
structured, readable, and functionally accurate.
HARDWARE PROTOTYPE
(HDL CODER ™ , EMBEDDED CODER ®)

C Code HDL

Processor FPGA

SDR Platform

Using Model-Based Design for rapid prototyping of 5G algorithms.

5G Development with MATLAB 32

Case Study: Developing a 5G NR
Cell Search Design
This example shows how to deploy a 5G NR cell search algorithm on The steps in transforming the algorithm into HDL code for FPGA
FPGA hardware. The algorithm is partitioned between hardware and prototyping or implementation are as follows:
software.The SS Block Detector will be deployed to hardware, while the
• Create the behavioral golden reference in MATLAB
Search Controller will be deployed to software.
• Model the hardware and software architecture in MATLAB
The software controls the hardware, which runs in two modes. The first
mode searches for signal synchronization blocks (SSBs), returning a • Model the fixed-point streaming hardware in Simulink, connected
list of detected Primary Synchronization Signals (PSSs). The software to MATLAB
determines the strongest cell ID, passing it back into the hardware in
demodulation mode, where it OFDM demodulates the SSB resource grid • Generate HDL code from the Simulink model
to detect its Secondary Synchronization Signal (SSS).

Hardware-software partitioning of the 5G NR HDL Cell Search algorithm. The SS Block Detector
will be deployed to hardware, while the Search Controller will be deployed to software.

5G Development with MATLAB 33

 Case Study: Developing a 5G NR
Cell Search Design continued
Deployment to hardware requires algorithms to be adapted to work on a continuous stream of data.

Managing the stream of data to balance resource usage with latency and throughput is often a big challenge in hardware deployment. Converting
to fixed-point data helps make hardware more efficient but reduces precision, which is another tradeoff to balance.

Simulink helps you visualize these tradeoffs and can simulate the hardware implementation using the same MATLAB tests and visualization. This
approach enables you to verify the hardware model against the MATLAB reference models.

This reference application successfully detects cell IDs both in simulation and over-the-air running on a Xilinx® Zynq® SDR.

Plots comparing the Simulink implementation vs. the MATLAB algorithm for PSS detection, SSS detection, and the grid for the decoded signal.

To learn more, download an example of building an NR HDL Cell Search for implementation with HDL Coder™, and use 5G Toolbox to verify the
HDL implementation model.

5G Development with MATLAB 34

HDL Implementation on FPGA and ASIC

Model-Based Design doesn’t stop at prototyping. Wireless engineers The model produces hardware-independent HDL code that can be
are successfully using MATLAB and Simulink with automatic HDL used on any FPGA or ASIC. Optimizations for other architectures can
generation to produce algorithm implementations that meet the be performed by modifying the model, verifying the results in Simulink,
performance, size, and power requirements of production FPGA and and regenerating the HDL code. HDL Coder integrates with SoC and
ASIC designs. The iterative workflow enables rapid development and FPGA design to provide target-optimized implementations. This can
verification of highly efficient hardware implementations of algorithms accelerate the development of SoC and FPGA designs, enabling teams
for multirate filtering, PAPR suppression, digital predistortion, and to complete this work in days or weeks rather than in months.
baseband processing.

SPEC

TYPICAL
SYSTEM/ HARDWARE HAND RTL VERIFICATION
DEVELOPMENT ALGORITHM RTL
ARCHITECTURE CODING
PROCESS
RTL SYNTHESIS

DEVELOPMENT TIME

MODEL

RTL VERIFICATION
MODEL-BASED SYSTEM/ HARDWARE
ALGORITHM ARCHITECTURE RTL
DESIGN GENERATE RTL SYNTHESIS
COLLABORATE
ITERATE
VERIFY

Using Model-Based Design to accelerate algorithm implementation on FGPAs and ASICs.

5G Development with MATLAB 35

Who uses this workflow for 5G prototyping?

Ericsson: Deploying 5G Hardware Testbeds

To complete and deploy a full proof-of-concept on a hardware
testbed, a development group at Ericsson has integrated algorithms
developed with MATLAB and Simulink with other system components
in a standard FPGA development environment. Specifically, they
integrated HDL code automatically generated from Simulink using
HDL Coder with manually coded custom interfaces and IP cores from
FPGA vendors for components such as soft processors, DMA, and
physical interfaces. When changes are required, the engineers make
the change in Simulink, regenerate the code (typically within minutes),
integrate the code in the FPGA development environment, and
synthesize it for FPGA implementation. Ericsson engineers have noted
that the generated code is well structured,
readable, and functionally accurate.

For details on how Ericsson is using hardware testbeds for fast

hardware prototyping and automatic HDL code generation with
MATLAB and Simulink, watch the video Radio Testbed Design Using
HDL Coder (22:44).
“When a new design is required because a new design idea
emerges, using HDL Coder, we were able to demonstrate that the
change was acceptable and have a working demonstration in less
than one week.”

— Tomas Andersson, Ericsson

5G Development with MATLAB 36

Learn More

Ready for a deeper dive? Explore these resources to learn more about
accelerating 5G prototypes and field trials.
Watch
5G NR HDL Cell Search Reference Application (5:27)

Read
Nokia: Model-Based Design in Nokia 5G (Conference Proceedings)
QUALCOMM: Connecting Systems and HDL World – Rapid RTL Generation (Conference Proceedings)
Deploying 5G NR Wireless Communications on FPGAs: A Complete MATLAB and Simulink Workflow
(White Paper)
Taking Wireless Infrastructure Systems Design from 3G to 5G and Beyond (White Paper)

Explore
HDL Code Generation and Verification
4. System Verification and Testing
5G System Verification and Testing

After completing baseband, RF, and antenna array designs, it is

essential to verify whether the system designs yield satisfactory results
in the lab and in the field under real-world conditions. To do so,
engineers need to work with live 5G signals and waveforms using
interfaces to test and measurement equipment.

5G Development with MATLAB 39

5G Live-Signal Generation and Reception

By connecting to software-defined radio (SDR) and RF instrument During field testing, it is important to be able to quickly modify test
hardware, you can use MATLAB® and Simulink® to perform over the-air parameters and test scripts to accommodate different test scenarios.
tests to validate your 5G designs in simulation, in the lab, or in the Using MATLAB provides the flexibility to customize tests and to
field under real-world conditions. The test benches, signal generators, diagnose and debug subtle issues that are difficult to address in the
scopes, and measurements used at the simulation stage can be reused more constrained software environments and encrypted waveforms
for hardware testing. This approach eliminates the need to recreate that test and measurement instruments typically provide.
tests in a different software environment and reduces test development
MATLAB and Simulink support a range of available SDR hardware,
time and errors.
as well as RF signal generators and spectrum analyzers from RF
You can capture live 5G or LTE signals for analysis and comparison to instrument vendors such as Keysight, Rohde & Schwarz, National
baseband simulation results. Algorithm designers can use the captured Instruments, and Anritsu. SDR support packages are available for Xilinx
signals to test their algorithms, and the RF team can use this setup to Zynq and FPGA Radios; USRP® N, X, and E Series Radios; PlutoSDR;
verify their RF design. Download an example that demonstrates live LTE and RTL-SDR. From these options, you can choose the hardware that’s
signal generation and capture capability. most appropriate for your requirements and budget.

TRANSMITTER

SUPPORTED HARDWARE

Digital DAC PA
Baseband
Front End

DIGITAL PHY RF FRONT END

Software-Defined Radio

Digital
Baseband ADC LNA
Front End

RF Signal Generator Spectrum Analyzer

RECEIVER

MATLAB and Simulink support for over-the-air (OTA) testing with a range of available SDR and RF instrument hardware,
simplifying comparison of OTA tests to simulation results.

5G Development with MATLAB 40

5G Live Signals via Software-Defined Radio

A software-defined radio (SDR) is a wireless device consisting of a

configurable RF front end with an FPGA or programmable SoC to perform
digital functions. Commercially available SDR hardware can operate at
different frequencies to transmit and receive standard-compliant and custom
signals to test baseband algorithms and RF front end designs.

You can use MATLAB and Simulink with SDR hardware as a cost-effective,
real-time platform for a range of wireless engineering tasks, including:

• Over-the-air lab and field testing with live RF signals

• Signal capture for posttest analysis

• Rapid verification with custom radio functions

A typical SDR device with RF front end and FPGA board.

BASEBAND RF SIGNAL RF SIGNAL BASEBAND

WAVEFORM GENERATION CAPTURE WAVEFORM
GENERATION HW Support HW Support ANALYSIS
package package

Over-the-air

+ +
LTE System Toolbox™ SDR Platform SDR Platform LTE System Toolbox™

MATLAB and LTE System Toolbox with supported SDRs. This provides a flexible, cost-effective environment for live signal generation and capture.

5G Development with MATLAB 41

BASEBAND RF SIGNAL RF SIGNAL BASEBAND
5G Live Signals via Software-Defined Radio
continued
With an interactive app or a few lines of MATLAB code, you can adapt SDR hardware to generate baseband signals that span a wide range of
possibilities—a candidate 5G signal, a 4G LTE standard-compliant signal, or a fully customized signal.

Generation (left) and analysis (right) of a 5G NR test model (NR-TM) waveform.

5G Development with MATLAB 42

Generation and Capture of Live 5G Signals
Using RF
BASEBAND
WAVEFORM
Instruments
RF SIGNAL
GENERATION
RF SIGNAL
CAPTURE
BASEBAND
WAVEFORM
GENERATION HW Support HW Support ANALYSIS
package package
The second method for working with live signals uses commercially To transmit signals, you can generate the 5G baseband signal,
available test instruments. These instruments employ finely calibrated download it to a signal generator, configure and control the
RF characteristics, making them suitable for precise lab, field, or parameters in the baseband signal, and upconvert the IQ waveform to
Over-the-air
production tests. However, 5G experiments and field trials require RF and play back over the air. On the receiver side, you can retrieve
greater software flexibility
+ and analysis capability than is typically IQ data into a PC for analysis, perform visualization and
+ analysis,
found onLTEsuch instruments.
System These limitations can
Toolbox™ SDRbe overcome by using
Platform programmatically
SDR configure
Platform acquisition parameters, and
LTE System down-convert
Toolbox™
MATLAB and Instrument Control Toolbox™, which supports interfaces the RF signal to baseband in the hardware.
and standards such as VISA, TCP/IP, SPI, USB, and GPIB; equipment
from different manufacturers; and instrument types, such as
signal generators and vector signal analyzers.

BASEBAND RF SIGNAL RF SIGNAL BASEBAND

WAVEFORM GENERATION CAPTURE WAVEFORM
GENERATION Instrument Control Instrument Control
ANALYSIS
Toolbox Toolbox

Over-the-air

+ +
LTE System Toolbox™ Signal Generator Spectrum Analyzer LTE System Toolbox™

Over-the-air testing with MATLAB and test and measurement instruments.

5G Development with MATLAB 43

Analysis of Massive Field Trial Data Sets

Network equipment vendors and wireless operators have been

conducting public field trials to demonstrate their 5G technology and
characterize its operation in a real-world setting.

In 5G field trials, many parameters are measured and monitored

dynamically, such as reference signal received power (RSRP) and
reference signal received quality (RSRQ). System performance is
recorded by capturing signals and sweeping across a large range of
parameters.

After taking the measurements, designers can extract useful and

actionable findings from the data. Compared with 4G systems, 5G field
tests capture massive amounts of data due to high data rates and the
need to capture and analyze massive MIMO radiation patterns.

The field trial system requires a test and analysis system that supports
scalable data capture, data processing, analysis, and sharing of results.

Huawei field trial from Tokyo.

5G Development with MATLAB 44

Test Data Capture and Analysis

Test engineers performing field tests often use commercially available information such as antenna pattern files. It is helpful to have a
test instruments. However, 5G field trial analysis requirements go far common environment or tool that can access various kinds of data
beyond simple measurements. from different equipment vendors.

Field trial analysis software must be able to import data directly from For very large data sets, teams can use computing clusters or cloud
test instruments or from stored data in a variety of formats. The data storage that scales from a single workstation to compute clusters or
represents captured signals, name and time stamps, and configuration process big data sets.

Using MATLAB to capture, process, and analyze field test data stored locally or in the cloud.

5G Development with MATLAB 45

Data Processing and Visualization

Test engineers may want to store raw captured data or show the results
to their management, partners, or customers after analysis.

For example, a team might want to show cell handover points on

a map, decide if signal-to-interference-plus-noise ratio (SINR) is
satisfactory, and how the RSRP varies. In performing this task, it can be
necessary to quickly generate standard waveforms and repeat the test
process many times. Cell ID Change Points

Analysis of the signal in conformity with the standard and handover point and cell ID number.

5G Development with MATLAB 46

Prediction of Signal Propagation and Coverage

You can use prediction tools as a lower-cost alternative to field testing physical environment and computes corresponding values such as path
for some stages of the design and evaluation process. These tools loss, angle of departure, and angle of arrival.
use propagation models and a description of the 3D environment
A prediction tool must be able to import and use both buildings and
to predict multipath reflections, losses, and signal coverage on a
terrain data to generate realistic predictions. Other propagation
geographic map.
impairments must also be included such as loss due to weather or
Ray tracing techniques are preferred for urban 5G use cases where foliage. Download an example that demonstrates urban channel link
analysis of multipath is important. A ray tracing propagation model analysis and visualization using ray tracing.
finds the possible propagation paths given a description of the

Visualization of propagation paths between points (left) and cellular signal coverage (right) on street maps.

5G Development with MATLAB 47

Visualization of 5G Field Test Results on Maps

Visualization is critical in field testing. Test engineers need to If properly implemented, this visualization architecture enables
superimpose captured signals as well as performance and parameter engineers to comprehensively describe system performance in real-
data on a representation of a geographic map. world scenarios and demonstrate results to inform network planning
decisions.
For some 5G applications such as vehicle-to-vehicle communication
and base station coverage analysis, engineers need to situate
propagation data on a map including visualization of 3D buildings
and terrain.

Visualization of data throughput capacity (left) and drive test measurements (right) on street maps.

5G Development with MATLAB 48

Analysis and Results Sharing

For large-scale postprocessing of data and central SHARED

management, system architects can deploy their 5G field DIRECTORY

testing software as applications on enterprise systems. These

applications can be hosted on a dedicated server or cloud,
so the system architects can remotely check the test data and
centrally manage system updates. Using MATLAB application WEB
APPLICATION
deployment tools, the development team can leverage their
MATLAB PRODUCTION SERVER
previous work, and everyone on the team (system architects WEB SERVER

and developers, field engineers, and management team) can Visualization

remotely view and validate performance and analysis results. DESKTOP

APPLICATION
You can implement enterprise applications with MATLAB
Decoding
Production Server™, which lets you incorporate custom
algorithms, tests, and analytics into web, database, and Pre-Process APPLICATION SERVER
BATCH
APPLICATION
production enterprise applications running on dedicated
servers or a cloud. See an example of this deployment process.
This 5G field test workflow has been used by cellular service
DATABASE SERVER
providers and leading suppliers and is ideally suited for the
A B C
demands of 5G field trials.

IT architecture of a 5G field trial.

5G Development with MATLAB 49

 Learn More

Ready for a deeper dive? Explore these resources to learn more about 5G
system verification and testing.
Watch
How to Generate a 5G Waveform for System Verilog Verification Using 5G Toolbox

What Is Antenna Toolbox?

What Is Instrumentation Control Toolbox?

Learn
Verifying LTE Designs Using Live Signals and Test and Measurement Equipment

Explore
Verify Wireless Signal Designs with RF Test Equipment

Testing 5G NR Devices with Standard Waveforms

© 2020 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc. See mathworks.com/trademarks for a list of additional trademarks. 4/20
Other product or brand names may be trademarks or registered trademarks of their respective holders.