A P P L IC AT ION NO T E

Electronic Warfare Signal

Generation:
Technologies and Methods
Introduction
Productive and efficient engineering of electronic warfare (EW) systems requires the generation
of test signals that accurately and repeatably represent the EW environment. Simulation of
multi-emitter environments, in particular, is vital to ensure realistic and representative testing.

Currently, these multi-emitter environments are simulated with large, complex, custom systems
that are employed in the system qualification and verification stage, and not widely available to
EW design engineers as R&D test equipment. Designers working on optimization and pre-
qualification are therefore at a disadvantage compared to wireless engineers performing similar
tasks. Engineers often learn of the nature and magnitude of performance problems later in the
design phase, leading to delays, design rework, and solutions that are not well-optimized.

This application note will summarize the available technological approaches for EW signal and
environment simulation, and the latest progress in flexible, high-fidelity solutions. For example,
recent innovations in digital-to-analog converters (DACs) have brought direct digital synthesis
(DDS) signal generation into the realm of practicality for EW applications through advances
in both bandwidth and signal quality. DDS solutions and other innovations in agile frequency
and power control will be discussed in the context of improving design-phase EW engineering
productivity.

Find us at www.keysight.com Page 2

 Realism and Fidelity in Multi-Emitter Environments
Validation and verification of EW systems is heavily dependent on testing with realistic signal
environments. EW test realism increases as high-fidelity emitters are added to create density.
In addition to emitter fidelity and density, platform motion, emitter scan patterns, receiver
antenna models, direction of arrival, and multipath and atmospheric models enhance the
ability to test EW systems under realistic conditions. EW systems are now designed to identify
emitters using precise direction finding and pulse parameterization in dense environments of 8
to 10 million pulses per second.

The cost of test is as important as test realism, as the relationship between cost and test
fidelity is exponential. As test equipment becomes more cost effective and capable, more EW
testing can be performed on the ground—in a lab or chamber—rather than in flight. Even though
flight testing can add test capability, it does so at great cost and is typically done later in the
program lifecycle, adding risk and further cost to the program through missed deadlines if the
system under test (SUT) fails. It is far better to test early in a lab environment with as much
realism as possible where tests can be easily repeated to iteratively identify and fix problems.

Challenges of Simulating Multi-Emitter Environments

The modern spectral environment contains thousands of emitters—radios, wireless devices,
and tens to hundreds of radar threats—producing millions of radar pulses per second amidst
background signals and noise. A general overview of the threat frequency spectrum is shown in
Figure 1.
Pulse density (log)

Acquisition, GCI

Fire control
Early warning

VHF UHF L S C X Ku

A B C D E F G H I J

Figure 1. A general representation of the threat density vs. frequency band in a typical operational environment.
The full RF/microwave environment would be a combination of the threat and commercial wireless environments.

Find us at www.keysight.com Page 3

Simulating this environment is a major challenge, especially in the design phase, when design
flexibility and productivity are at their greatest. The situation is very different from the typical
wireless design task, where a single signal generator can produce the required signal, perhaps
augmented by a second signal generator to add interference or noise.

In EW design the multiplicity and density of the environment—and often the bandwidth—make
it impractical to use a single source or a small number of sources to simulate a single emitter
or a small number of emitters. Cost, space, and complexity considerations rule out these
approaches.

The only practical solution is to simulate many emitters with a single source, and to employ
multiple sources—each typically simulating many emitters—when required to produce the
needed signal density or to simulate specific phenomena such as angle-of-arrival (AoA).

The ability to simulate multiple emitters at multiple frequencies depends on the pulse repetition
frequency, duty cycle and number of emitters, and ability of the source to switch between
frequency, amplitude, and modulation quickly.

A limiting factor in the use of a single signal generator to simulate multiple emitters is pulse
collisions. Figures 2 and 3 show the number of pulse collisions expected for the cases of low
and high pulse repetition frequency (PRF).

Low PRF emitter density vs. pulse collision percentage

100
3000 emitters
90
Pulse collision perccentage

1000 emitters
80
70
512 emitters
60
50
40 256 emitters
30
20 128 emitters

10
36 emitters
0
0 1 1 2 2 3 3 4
Millions of pulses per second

Figure 2. As the number of emitters grows, the number of pulse collisions grows even when all emitters use low PRF.

Find us at www.keysight.com Page 4

 High PRF emitter density vs. pulse collision percentage
100
4 emitters
90
3 emitters
Pulse collision percentage

80
70
60 2 emitters

50
40
30
20
10
1 emitter
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Millions of pulses per second

Figure 3. The percentage of pulse collisions climbs very quickly as high-PRF emitters are added to a simulation.

A source’s agility is a factor in its ability to simulate multiple emitters. Source frequency and
amplitude settling time (whichever is greater) is the transition time between playing one pulse
descriptor word (PDW) and the next.

Total pulse density for a single source is limited by the sum of the transition time and the width
of the transmitted pulses, a lockout period parameter as shown in Figure 4. It is obviously
desirable that the lockout period be as short as possible, and therefore that the source settling
times be as brief as possible.

Pulse width
Frequency switching time

Amplitude switching time

Transition time

Lockout period

PDW PDW Time

sent playback

Figure 4. The ability to simulate multiple emitters depends not only on emitter parameters like PRF and pulse width,
but also the frequency and amplitude switching speeds and setting times of the signal source used to synthesize
the emitters. If the source is switching, it cannot play a pulse. If it is playing a pulse, it cannot switch. The source is
unavailable to simulate a different threat during the lockout period.

To simulate high pulse density and the possibility of some overlapping pulses, it is often
necessary to combine multiple sources. As more sources are added to the test configuration,
pulse density should scale easily and seamlessly, ultimately reaching the desired tradeoff of
satisfactory simulation realism and cost.

Find us at www.keysight.com Page 5

Technology Improvements Simplify System Integration and
Reduce Cost
Simulating more threats to create more pulse density ultimately requires more parallel
simulation channels, even if the simulation channel can switch frequency and amplitude quickly.
This is because pulses begin to collide in the time domain as the number of emitters, their PRFs,
and their duty cycles grow larger 1. Pulses that overlap in the time domain must be played out of
parallel generators or selectively dropped based on a PDW priority scheme. Unfortunately, the
increased realism of a higher-density environment comes at a substantially higher system cost,
as shown in Figure 5.
Cost

Value

Legacy simulation
technology Modern
simulation technology

Fidelity
Figure 5. Simulation fidelity and cost increase exponentially. System integrators and evaluators must decide the level
of cost vs. fidelity that ensures system performance. New simulation technologies enable more simulation realism and
fidelity at lower cost.

In the past, simulations have generally been created with a separate component for each
emulation function, such as signal generation, modulation/pulsing, attenuation or amplification,
and phase shift. The same PDW would be sent to each functional component to provide output
on a pulse-to-pulse basis. For instance, a synthesizer would generate the output frequency,
while a separate modulator would create pulsed modulation and/or AM/FM/PM modulation.
Amplifiers and attenuators would adjust the signal to the desired output power level. An
example of this system topology is shown in Figure 6.

PDWs Control parameters

Pulse RF Amplifier/ EW simulation

generator generator attenuator output

Synchronization

Figure 6. In the traditional approach, PDW control parameters are sent in parallel to multiple functional
elements, on a pulse-to-pulse basis, to generate and modify the desired signal. This approach results in a 1.  Philip Kazserman, “Frequency of pulse
complex system, demanding precise synchronization. coincidence given n radars of different
pulse widths and PRFs,” IEEE Trans.
Aerospace and Electronic Systems, vol.
AES-6. p. 657-662, September 1970.

Find us at www.keysight.com Page 6

Since multiple functional components are required to produce each output channel, time
synchronization is a significant configuration and operational challenge. A wide variety of
settling times and latencies must be fully characterized to optimize pulse density by minimizing
lockout periods.

This approach can be scaled directly to create multiple coordinated channels, as shown in
Figure 7. However systems configured in this way require a large footprint, occupying more rack
space, and cost escalates quickly.

Figure 7. A signal generation approach using separate functional elements can be scaled up in a straightforward
manner to increase pulse density and generate a more realistic environment. Unfortunately cost and space
requirements scale up rapidly as well. 1

The controller shown in Figure 7 would route PDWs to channels based on emitter parameters
such as frequency, amplitude, and pulse repetition frequency and also the availability of each
channel to implement the PDW. Since a channel cannot execute the parameters of two different
PDWs at the same time, one could be shunted to a backup channel or dropped according to its
priority.

Ultimately, EW receivers must be able to handle 8-10 million pulses per second where most of
the pulse density occurs at X-band. EW receivers must be able to handle pulses arriving at the
same time at different frequencies from different angles. Creating pulses that are coincident
with one another in the time domain should be a goal of simulation to increase simulation
realism.

Though Figure 7 describes a very capable system, the level of integration the system elements
is rather low. Recent developments in analog and digital signal generation technologies
are enabling a higher degree of integration, and solutions which are more cost- and space-
efficient, as described in the section, “Increasing Integration in EW Test Solutions.” There are
several methods of controlling simulations depending on test objectives.

1.  Reproduced by permission from

David Adamy, EW 101: A First Course in
Electronic Warfare, Norwood, MA: Artech
House, Inc., 2001. © 2001 by Artech
House, Inc.

Find us at www.keysight.com Page 7

Control of Hardware-in-the-Loop Testing
Depending on the integration of simulation elements and the simulation length, scenarios can
be played from list memory or streamed over a digital interface such as low-voltage differential
signaling (LVDS). List mode plays PDWs from list memory for shorter scenario lengths with some
ability to trigger between lists for an adaptive (closed-loop) simulation in response to the SUT.

For example, there is often a need to switch between one simulated threat mode to another in
response to identification and jamming by the SUT. For long scenario lengths with fast control
over scenario changes, PDWs can be streamed over the LVDS to the signal generation system
operating in an agile controller mode. In this case, simulation software generates batches of PDWs
according to simulation kinematic granularity and streams them ahead of their desired play time.

In either method of control, the goals are to stress the SUT with increasing pulse density,
depending on the number of simulation channels available and the parameters of the threats to
be simulated. As pulse density increases, PDWs can be dropped according to a priority scheme as
they increasingly collide in the time domain and there are insufficient signal generation channels
to play them.

Creating AoA
In addition to creating emitters with the desired fidelity and density, it is also important to
match the geometry and kinematics of EW scenarios since the AoA of a radar threat to the
EW system changes slowly compared to other parameters such as center frequency and pulse
repetition frequency.

EW systems measure AoA and estimate distance using amplitude comparison, differential
Doppler, interferometry (phase difference), and time difference of arrival (TDoA). Precise AoA
measurements enable precise localization of radar threats. New stand-off jamming systems
use active electronically-scanned arrays capable of precise beam forming to minimize loss of
jamming power due to beam spreading towards a threat. Moreoever, EW receivers with better
AoA capability reduce the need for pulse de-interleaving and sorting. Consequently, AoA is an
increasingly-important test requirement.

Techniques for creating AoA

In the past, AoA was created with a combination of signal sources and analog phase shifters,
attenuators, and gain blocks in the cable path to the SUT. Analog elements in the cable path
took up a lot of space, had limited resolution, and were expensive.
As an alternative, and depending on their architecture, sources can be linked together to
create phase coherent output, allowing for finer control over creating phase fronts to the
SUT. Similarly, amplitude control at the source can be used to create appropriate amplitude
differences at SUT receive channels.

The ability to control AoA to meet modern test requirements depends on the architecture of
the source. At a minimum, it should be possible to lock the local oscillators (LOs) of multiple
sources together so that they all share the same phase. Often, calibration is required to finely
align phase and timing between sources.

Creating small, accurate, and repeatable differences in phase or frequency between channels is
the next challenge. Sources based on DDS architecture allow AoA to be controlled digitally in a
numerically-controlled oscillator. Phase alignment in a DDS source is then a matter of sharing
reference clocks. Calibrations to provide accuracy and repeatability can be uploaded to a table
to be applied in real-time.

Find us at www.keysight.com Page 8

Overview of Source Technologies for EW Test
The characteristics and tradeoffs of EW signal generation systems are largely determined
by the core synthesizer and oscillator technologies used. To provide insight into the most
significant choices for EW engineers, this section will summarize the three principal
technologies currently available:

–– Direct analog synthesis (DAS)

–– Phase-locked loop or indirect analog synthesis (PLL, frequently fractional-N)
–– Direct digital synthesis (DDS)

General source requirements

Signal sources used to test EW systems must be broadband. Traditionally, a frequency range
of 0.5 to 18 GHz was required. Frequency requirements have expanded dramatically in recent
years, now beginning near DC and extending as high as 40 GHz. This allows systems to
simulate early warning, fire control, and missile-seeker radars from a single output channel.

In addition to wide frequency coverage, sources for EW test must have fast frequency and
amplitude switching speeds to simulate different radars operating in different modes in
different frequency bands.

PLLs and fractional-N synthesis

Indirect synthesis
Most general-purpose sources today are PLL-based, where a broadband oscillator such as a
voltage- controlled or YIG-tuned oscillator is locked to a stable reference in a phase-locked
loop (PLL). The PLL improves signal quality by reducing phase noise and spurious signals in
the output. To provide a combination of wide frequency range and fine frequency resolution,
PLL-based sources have been configured with a combination of sum and step loops or a single
loop with fine fractional division capability. These fractional-N PLLs offer excellent signal
quality and fine frequency resolution in a cost-effective single-loop configuration, making them
a good choice for general purpose signal sources.

Unfortunately, the required control loop filtering in PLLs results in a significant settling or loop-
response time. This limits the ability of the synthesizer to switch frequency quickly. Due to their
comparatively large transition time, these sources are limited in their ability to simulate multiple
radar threats out of a single channel, even if they have the necessary broadband frequency
coverage and frequency resolution. They also lack phase-repeatable switching capability.

Find us at www.keysight.com Page 9

Direct analog synthesis
A direct analog synthesizer typically contains several stable frequency references multiplied
or divided from the same crystal oscillator reference. These frequency references (and their
harmonics) can be switched in and out of the signal path and multiplied, divided, added, and
subtracted to provide fine frequency resolution quickly. The frequencies of these references are
chosen to reduce the amount of multiplication stages required such that phase noise increases
only moderately as frequency is increased. Division to lower frequencies reduces the phase
noise.

Since the switches and arithmetic operators used in the DAS approach operate very quickly
and do not need loop filtering, these synthesizers have very high frequency agility. They have
therefore been a common architecture for EW test solutions.

However, DAS technology has several drawbacks. First, numerous stages are required to
achieve the desired frequency resolution. Switching parallel and series multiplication, division,
and mixing stages requires more hardware than PLLs and generally reduces reliability. Second,
circuit noise from each stage is cascaded, and phase noise is multiplied through the stages.
Finally, each stage adds components which increase size, weight, and cost.

On the positive side for EW applications, DAS has the potential for limited phase-repeatable
frequency switching. However, though all frequencies are usually derived from the same
reference, divider ambiguities generally preclude full phase-coherent switching.

DDS now suitable for EW applications

The DDS approach, based on DAC circuits, is a natural fit for the needs of EW signal simulation.
However, until recently, DACs were not available with the required combination of fast sample
rates and high purity.

Fast sample rates are needed to produce outputs with very wide bandwidth, so that a minimum
of multiplying stages can be used to produce the desired output frequencies. The use of either
a large number of multiplying stages or a DAC of insufficient purity would limit the effective
spurious-free dynamic range (SFDR) of the EW synthesizer.
In concept, a DDS is one of the simplest types of signal generators. In a frequency-tunable
DDS, data from a numerically-controlled oscillator is converted to analog form by a DAC and
lowpass filtered to remove image frequencies and harmonics. A block diagram of the major
elements of a DDS is shown in Figure 8.

Frequency Numerically Reconstruction

Analog
control controlled DAC lowpass
output
register oscillator filter

Reference oscillator Fclk

Figure 8. Principal functional blocks of a direct digital synthesizer.

Find us at www.keysight.com Page 10

The numerically-controlled oscillator itself consists of two elements: a phase accumulator (PA)
and a phase-to-amplitude converter (PAC) as shown in Figure 9. In modern DDS designs, these
are often implemented using field programmable gate arrays (FPGAs) or dedicated integrated
circuits.

Phase Phase-to-
∆Φ Phase Φ(t)
Frequency increment amplitude To DAC
accumulator
calculator converter

Figure 9. Functional block diagram of a numerically-controlled oscillator.

In frequency synthesis, a frequency control word—a delta phase—is sent to the phase
accumulator along with the digital reference clock. For each clock cycle, this delta phase
is added in the phase accumulator with high precision. The phase value generated by the
accumulator is then converted to a sinusoidal amplitude in the phase-to-amplitude converter.
The digital sine wave is then sent to the DAC and output at a frequency given by the DDS tuning
equation, where N is the number of bits in the frequency control word 1:

∆phase
fout = fclk
2N

This equation demonstrates that higher output frequencies are achieved by greater DAC clock
frequencies while resolution is controlled by the number of bits in the frequency control word
and phase accumulators. The numerically controlled oscillator behaves as a divider to the
reference clock to provide frequencies with high resolution according to the bit depths of the
phase register and frequency control word. Note that transitions to new frequencies happen in
one clock cycle.

1.  David Buchanan, “Choosing DACs for

direct digital synthesis,”Analog Devices
Application Note 237, Available:(http://
application-notes.digchip.com/013/13-
14876.pdf)

Find us at www.keysight.com Page 11

Advantages of DDS
The new Keysight UXG agile signal generator uses DDS technology made possible by a
Keysight-proprietary DAC to generate multi-emitter simulations. DDS has several advantages
over other synthesis technologies for EW applications:

–– Digital control of extremely fine frequency and phase tuning increments within a single
clock cycle. In the Keysight UXG, frequency resolution is one Hertz and phase resolution
is sub-degree. Fractional-N techniques can provide micro-Hertz resolution, but frequency
changes are much slower due to PLL filtering. DAS techniques provide rapid frequency
switching, but at a cost in frequency resolution.
–– Fast frequency hopping with phase continuity and phase repeatability to simulate multiple
pulse-Doppler radars at different frequencies while maintaining their original phase.
This combination of phase control and hopping speed is unique to the Keysight UXG.
DAS techniques offer hop speed and frequency/phase repeatability only under limited
conditions.
–– Modulation is created in the digital domain, providing numerical precision and
repeatability.

There are other advantages to using DDS that are of interest to the EW engineer. Many DDSs
employ a digital modulator for amplitude, frequency, and phase modulation for creation of
digitally-modulated signals in the numerically-controlled oscillator. Linear frequency modulated
(LFM) chirps and Barker codes can also be directly synthesized using the numerically-
controlled oscillator. Chirp bandwidth depends on the bandwidth of the bandpass filters after
each multiplication stage and whether the signal is crossing a band.

Microwave source architecture using DDS

Modern EW applications require frequency coverage to 40 GHz, along with high agility and high
purity. Digital signal processing technologies for numeric signal creation have been adequate
for some time, but wideband DAC performance has been inadequate for these applications.
Available DACs with very wide bandwidth and high clock rate have not been sufficiently pure,
while DACs with good signal purity and high bit depth have been limited to lower frequency
clocks and narrower bandwidth.

Recent DAC innovations from Keysight provide an example of a DAC and DDS suitable for EW
test applications. The DAC has been designed for RF applications, with a combination of high
bit depth and excellent purity, including spurious-free dynamic range and phase noise. The high
sample rate of the DAC supports a wide bandwidth DDS that allows microwave frequencies to
be synthesized with a low number of multiplication stages. Limiting multiplication stages limits
the phase noise and spurious signals present in microwave output.

EW testing also requires precise signal amplitudes, over a wide range of power levels. These
power levels must be switched as fast as frequencies are changed, without signal distortion
from attenuator settling. As with the DAC, these demands have led Keysight to develop a new
series of FET switches to implement a solid-state attenuator with high agility, low distortion,
and an amplitude range of 120 dB. The agile amplitude range of the attenuator is 80 dB
anywhere in the 0 dBm to -120 dBm output range.

The architecture of a true DDS based, agile microwave signal generator utilizing developments
in DAC and FET switching technology is shown in Figure 10.

Find us at www.keysight.com Page 12

 Lowpass
filter bands

Freq
Amplifier
doublers
Numerically Digital to Electronic & Analog out
controlled analog x2n mechanical 0.01-40 GHz
oscillator converter attenuators

Frequency Pulse Amplitude

Phase Pulse time
LFM Pulse width
Pulse parameter list & external digital PDW interface
Figure 10. High-level block diagram of a DDS-based agile signal generator, covering 0.01 to 40 GHz.

Signal generation begins with a DDS, optimized for very low spurious output, since spurs
increase for each doubling stage. A sequence of doubler circuits is then used as needed to
create signals up to 40 GHz. Each multiplication stage employs bandpass filters to remove
unwanted signals from the multipliers.

The FET-based agile attenuator is then used to produce the desired output levels. This
attenuator provides very fast settling, matched to frequency switching speed, so that the
source can implement open loop power control with high accuracy and no loss in switching
time.

Find us at www.keysight.com Page 13

Increasing Integration in EW Test Solutions
A general trend in EW simulation solutions is to absorb more simulation elements into the RF/
microwave signal source. For example, the Keysight UXG agile signal generator combines the
intra-pulse modulation, pulse modulation, and amplification/attenuation stages into the fast
frequency synthesizer.

By implementing a high level of functional integration, a DDS-based agile source, and

a matching agile attenuator, the UXG can meet important functional and performance
requirements for EW test:

–– Fast frequency, amplitude, and phase switching for fast transitions between multiple
emitters
–– High dynamic range to match the dynamic range of the modern EW receiver
–– A wide, accurate, agile amplitude range to simulate multiple threats with accurate power
levels and switch amplitude as quickly as frequency
–– Low noise floor to test receiver sensitivity as channels are combined
–– Pulse modulation with a high on/off ratio and fast settling with low distortion
–– Intrapulse modulation capability for pulse compression such as Barker codes and linear
frequency modulation
–– Scalable to multi-channel and multi-port threat simulation to increase pulse density and
realism easily
–– Wide frequency range from near DC to 40 GHz to keep pace with modern threat simulation
requirements
–– BCD frequency control interface for backward compatibility with legacy sources previously
used as LOs
–– LVDS interface to allow high-rate PDW streaming—EW simulation sources need a fast,
full-featured interface for streaming complete PDWs at a high rate rather than frequency-
only control

In systems with a traditional, distributed architecture (as described in Figure 7) the

synchronization of an agile LO with functions such as pulse modulation, frequency/phase
modulation, and amplitude control is a considerable challenge. In an integrated EW test
solution such as the UXG, this synchronization is automatic, provided by the test equipment
itself. By simplifying hardware and system complexity, this integrated approach promises to
improve both performance and reliability.

Find us at www.keysight.com Page 14

Conclusion
A variety of technologies have been used to generate the signals needed for effective EW
simulation. Each of these technologies has brought a different combination of benefits
and challenges. The highest fidelity solutions have provided very realistic simulations of
the EW environment but their use has been limited by their complexity and expense.

Recent innovations in core hardware such as DACs and FPGAs have enabled new
solutions with the hardware simplicity and reliability of traditional test equipment. These
solutions will provide dramatic improvements in solution cost and size, bringing high-
fidelity EW environment simulation to a much earlier phase in the design process. Using
realistic EW environment simulation at the optimization and pre-verification stages of
design will improve performance, speed the design process, and reduce overall costs.

Learn more at: www.keysight.com

For more information on Keysight Technologies’ products, applications or services,
please contact your local Keysight office. The complete list is available at:
www.keysight.com/find/contactus

Find us at www.keysight.com Page 15

This information is subject to change without notice. © Keysight Technologies, 2014-2019, Published in USA , January 18, 2019, 5992-0094EN