SIGNAL AND SYSTEM LEVEL SIMULATIONS ON WIDEBAND
INTERCEPT RECEIVERS
A MASTER’S THESIS
in
Electrical and Electronics Engineering
Atilim University
by
İLTER KARADEDE
JANUARY 2014
SIGNAL AND SYSTEM LEVEL SIMULATIONS ON WIDEBAND
INTERCEPT RECEIVERS
A THESIS SUBMITTED TO
THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF
ATILIM UNIVERSITY
BY
İLTER KARADEDE
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF
MASTER OF SCIENCE
IN
THE DEPARTMENT OF ELECTRICAL ELECTRONICS ENGINEERING
JANUARY 2014
Approval of the Graduate School of Natural and Applied Sciences, Atılım
University.
_____________________
Prof. Dr. İbrahim AKMAN
Director
I certify that this thesis satisfies all the requirements as a thesis for the degree
of Master of Science.
_____________________
Assoc. Prof. Dr. Elif AYDIN
Head of Department
This is to certify that we have read the thesis “Signal and System Level Simulations
on Wideband Intercept Receivers” submitted by “İlter KARADEDE” and that in our
opinion it is fully adequate, in scope and quality, as a thesis for the degree of Master
of Science.
_____________________
Assoc. Prof. Dr. Ali KARA
Supervisor
Examining Committee Members
Assoc. Prof. Dr. Elif AYDIN
_____________________
Assoc. Prof. Dr. Ali KARA
_____________________
Asst. Prof. Dr. Fatma ÇALIŞKAN
_____________________
Asst. Prof. Dr. H. Taha HAYVACI
_____________________
Asst. Prof. Dr. Serdar ÖZYURT
_____________________
Date: 31.1.2014
I declare and guarantee that all data, knowledge and information in this document
has been obtained, processed and presented in accordance with academic rules and
ethical conduct. Based on these rules and conduct, I have fully cited and referenced
all material and results that are not original to this work.
İlter KARADEDE
Signature:
ABSTRACT
SIGNAL AND SYSTEM LEVEL SIMULATIONS ON WIDEBAND
INTERCEPT RECEIVERS
Karadede, İlter
M.S.,Electrical Electronics Engineering Department
Supervisor: Assoc.Prof.Dr. Ali Kara
January 2014, 51 pages
Electronic Warfare (EW) simulations are mostly designed for only receiver frontend or emitter parameter measurements. This thesis presents signal and system level
simulations and emitter parameter measurements on proposed structures. To that end,
a proposed wideband intercept receiver is employed and emitter environment is
designed using commercial simulation tool. Then, parameter measurement part is
employed to measure emitter parameters in a different simulation tool. Finally,
simulation results are discussed for system level simulations for wideband intercept
receivers and emitter parameter measurements.
Keywords: Intercept receiver, Channelized receiver, Electronic Warfare receivers,
System level simulations, Emitter parameter measurement
iii
ÖZ
GENİŞBANT RADYO KESTİRİM ALICILARINA YÖNELİK SİNYAL VE
SİSTEM DÜZEYİNDE BENZETİMLER
Karadede, İlter
Yüksek Lisans, Elektrik Elektronik Mühendisliği Bölümü
Tez Yöneticisi: Doç. Dr. Ali Kara
Ocak 2014, 51 sayfa
Elektronik Harp (EH) simulasyonları çoğunlukla alıcı ön-ucu veya yayıcı
parametre ölçümleri için tasarlanır. Bu çalışmada, önerilen yapılarla sinyal ve sistem
düzeyinde ve yayıcı parametre ölçümleri ve sonuçları sunulmaktadır. Bu amaçla,
ticari bir simulasyon aracı kullanılarak yayıcı çevresi ve önerilen genişbant kestirme
alıcısı gerçekleştirilir. Ardından, başka bir simulasyon aracı kullanılarak
yayıcı
parametrelerinin ölçümü için parametre ölçüm bölümü gerçekleştirilir. Son olarak,
sistem düzeyindeki simulasyonlar için genişbant kestirme alıcısı ve yayıcı parametre
ölçümlerinin sonuçları ele alınır.
Anahtar Kelimeler: Kestirme alıcısı, Çok kanallı alıcı, Elektronik Harp alıcıları,
Sistem düzeyi simulasyonları, Yayıcı parametre ölçümleri
iv
To My Parents
v
ACKNOWLEDGEMENTS
I express sincere appreciation to my supervisor Assoc. Prof. Dr. Ali Kara for his
advice, criticism, guidance and insight throughout the research.
I would like to thank my whole family; for their care on me from the beginning of my
life, their prayers, and their trust on me that I could accomplish this task.
vi
TABLE OF CONTENTS
ABSTRACT ............................................................................................................. iii
OZ .......................................................................................................................... iv
DEDICATION .......................................................................................................... v
ACKNOWLEDGEMENTS ..................................................................................... vi
TABLE OF CONTENTS ........................................................................................ vii
LIST OF TABLES .................................................................................................... ix
LIST OF FIGURES ................................................................................................... x
LIST OF ABBREVIATION .................................................................................... xii
CHAPTER
1. INTRODUCTION ............................................................................................ 1
2. EMITTER ENVIRONMENT GENERATION ................................................ 6
2.1 Emitter Generation ................................................................................... 7
2.2 Noise Effect ............................................................................................... 12
3. SIMULATION ON RECEIVER CONFIGURATION AND RECEIVER
FRONT-END DESIGN .................................................................................... 15
3.1 Frequency Channelization ....................................................................... 17
3.2 ADC and Switch State ............................................................................. 22
4. EMITTER PARAMETER MEASUREMENT PART...................................... 25
4.1 Pulse Envelope ........................................................................................ 26
vii
4.2 Double Threshold ..................................................................................... 27
4.3 Emitter Parameters .................................................................................... 30
4.3.1 Time of Arrival (TOA) and Time of Departure (TOD) .................. 30
4.3.2 Pulse Width (PW) ........................................................................... 33
4.3.3 Radio Frequency (RF) .................................................................... 34
5. SIMULATION RESULTS .............................................................................. 35
5.1 Signal Separation ..................................................................................... 37
5.2 Parameter Measurement Results ............................................................. 40
6. CONCLUSION ................................................................................................ 47
REFERENCES ...................................................................................................... 50
viii
LIST OF TABLES
TABLE
1. The gain, loss, NF and cascaded
NF of a receiver channel .................................................................................... 14
2. 3 emitters radio frequency parameters
due to the changes of NF .................................................................................... 45
3. 10 emitters PW parameters
due to the changes of NF.................................................................................... 46
ix
LIST OF FIGURES
FIGURES
1. LFM emitter signal generator in VSS ................................................................ 8
2. Two LFM pulses generated in VSS ................................................................... 9
3. Spectrum of LFM pulse in Fig. 2 ....................................................................... 10
4. An emitter signal generator in VSS .................................................................... 10
5. Generated emitter signal in VSS ......................................................................... 11
6. Spectrum of pulse in Fig. 5 ................................................................................. 11
7. Cascaded noise figure analyze of a receiver channel .......................................... 13
8. Complete system block diagram ........................................................................ 16
9. Some generated emitter signals at 1-5GHz band .............................................. 17
10. 0.5-2GHz problematic channel belongs to first structure ................................. 18
11. The first and second channels of the second approach ...................................... 19
12. VSS output of the second proposed structure for 1-2GHz channel ................... 20
13. Two channels of second conversion stage ......................................................... 21
14. Dynamic time shared switch .............................................................................. 23
15. Captured pulse by the Receiver ......................................................................... 27
16. A radar pulse with various features ................................................................... 29
17. Flow chart for TOA and TOD indices measurement algorithm ........................ 32
18. Pulse indices graph ............................................................................................ 33
19. Flow chart of simple pulse and LFM pulse frequency measurement
algorithm ........................................................................................................... 35
x
20. Three pulses received with a channel ................................................................ 38
21. One of the separated pulses in three different pulses ......................................... 39
22. FFT of pulse in Fig. 21 ...................................................................................... 40
23. Matlab GUI for parameter measurements .......................................................... 42
24. Average error versus NF .................................................................................... 44
xi
LIST OF ABBREVIATIONS
ADC
-
Analog to Digital Converter
ADS
-
Advanced Design System
CFAR
-
Constant False Alarm Rate
CW
-
Continuous Wave
DR
-
Dynamic Range
EW
-
Electronic Warfare
FT
-
Fourier Transform
FFT
-
Fast Fourier Transform
GUI
-
Graphical User Interface
IIP3
-
Third Order Intercept Point
LFM
-
Linear Frequency Modulated
LO
-
Local Oscillator
MTD
-
Moving Target Detection
NF
-
Noise Figure
PRF
-
Pulse Repetition Frequency
PRI
-
Pulse Repetition Interval
PW
-
Pulsewidth
RF
-
Radio Frequency
SNR
-
Signal to Noise Ratio
TOA
-
Time of Arrival
TOD
-
Time of Departure
xii
CHAPTER 1
INTRODUCTION
Electronic Warfare (EW) receiver design has important trade-offs between some
parameters such as sensitivity, instantaneous bandwidth, dynamic range, number of
simultaneous signals, probability of intercept and dynamic range [1]. If high
probability of intercept is desired, then minimum spurious signals, large
instantaneous bandwidth and high sensitivity can be arranged and this causes
challenging receiver design [1,2]. One of the most important problems for intercept
receivers are no detection of threats and false alarms. The platforms are directly
effected by these two problems. To provide efficient measurements of intercepted
signals and to minimise these two problems are the main design considerations of
EW receivers. Therefore, the receiver design procedure must be based on to provide
elimination of unwanted spurious signals, which are internally generated by the
receiver, high sensitivity, large instantaneous bandwidth and high probability of
intercept [1].
Today, mostly wide open, scanning superheterodyne and parallel channelized
receivers are used commonly [3]. Mostly, radars classifications are based on the type
of waveforms, and they can be pulsed or continuous waveforms. Radar operations
can be rated on its pulse repetition frequency (PRF) as high, medium and low PRF,
and they change due to the operation limits [4]. Radars are used in many operations
such as ground radars, air radars and sea radars. In EW, detection of the enemy
threats firstly depends on intercept receivers and emitter parameter measurements.
For this purpose, measurements of the pulse parameters take the first place of threats
detection [5].
Having a high dynamic range and reasonable wide instantaneous bandwidth are the
advances of digital receiver techniques which make them charming in EW receiver
1
design. Furthermore, generating a suitable counter measure is easy when the
intercepted signals are digitally processed [1].
In EW, extraction of pulse parameters is very important part of an intercept system
[5]. The aim of emitter parameter measurements depends on some important steps.
The most important four steps are signal sorting, signal identification, jamming
information assignment and extraction of radar parameters. Firstly, the deinterleaving of various emitter signals and collecting emitter pulses of individual
radars are included in signal sorting. Secondly, classifications of emitter types are
included in signal identification. Then, assignment of jamming information and
emitter parameter extractions are included in measurement of unknown emitter
parameters. At first, emitter parameters measurement is happened in discrete
components in circuit form. But now, digital signal processing techniques can help
the real time processing [5].
On the other hand, the designers use various simulation tools efficiently. In order
to make a cost efficient, easy and complete intercept receivers, these tools are user
friendly [3]. There are various simulation tools for commercial and educational
purposes on the market. These tools are used in many areas, such as
telecommunication and EW. Advanced Design System (ADS), Simulink and AWR
Visual System Simulator (VSS) are the most commonly used simulation tools. VSS
for high frequency design is more convenient than other simulation tools, because
VSS is made for high frequency designs.
In order to measure emitter parameters VSS can be used again, but these
measurements are made in system level and there could be no understandable and
visual data observed. To overcome this situation, Matlab is used in this thesis. There
are again various software tools in the market, but Matlab is more user friendly than
the other tools in the market. However, the most important reason for using Matlab,
VSS is allowed to use only with Matlab instead of other simulation tools. Matlab is
very helpful with its built in functions, figures and Graphical User Interface etc. In
the reference given in [6], a system, which is used Doppler radar to detect targets, is
designed in VSS to receive Doppler radar signals. In that study, Moving Target
Detection (MTD) is done with the co-operation of VSS and Matlab. Received signals
2
are sent to Matlab for MTD. At last, the results gathered from MTD are shown with
several graphs. Moreover, Constant False Alarm Rate (CFAR) is observed with
Matlab in [6]. In that study, the signals received by the VSS are processed
simultaneously with Matlab. However, in this thesis, the signals received by the VSS
is firstly saved with VSS as a data file and this data file is processed in Matlab.
The wideband channelized intercept receiver design and emitter parameter
measurements efforts contribute to signal level simulations by using AWR design
environment and Matlab co-operations. The aim is to reach a high sensitivity and a
high probability of intercept. For this purpose, parallel channelized receiver structure
was selected and applied according to proposed structure in [2]. According to [2], it
can be necessary to intercept simultaneous signals with a channelized receiver. The
bank of filters with adjacent frequencies can separate signals. Due to this proposal,
the minimum bandwidth of the filter is directly related to the desired minimum
pulsewidth. For instance, 100ns pulsewidth can be detected with a receiver that has
10MHz filter bandwidth. However, this kind of filter bandwidth is not practical in
real [2].
On the other hand, there is another approach, which is given in [2], provides a
proper parallel channelization for wideband intercept receivers. According to the
design, it is decided that the intercept receiver has a bandwidth of 0.5-18GHz and
this bandwidth is divided in to 2GHz channels due to the reference [2]. This
approach not only provides 2GHz channels but also it has some down and up
conversion stages. According to this design, all the bands are directly up or down
converted to 2-4GHz subbands except the 2-4GHz band. In the lights of this
approach, some bands adjacent to 2-4GHz band are not converted directly. These
bands are converted up and down to the 2-4GHz subbands to avoid spur problem
which is generated by the mixer. However, there is another way to convert these
channels. Adjacent to 2-4GHz channels are divided again into two channels and up
or down converted to 2-4GHz band based on this approach which is used in this
thesis by using VSS [2,3].
In order to measure emitter parameters, a signal processing part is performed in
this thesis. There is an effective and attractive proposal given in the reference [5]. A
3
hardware noise gate to capture pulse envelope proposed in [5] is implemented in
Matlab. Emitter pulse parameters such as radio frequency (RF), pulsewidth (PW) and
time of arrival (TOA) are measured with a given algorithm proposed in [5]. In this
study, the multipath fading effect is neglected at the beginning of the system so
system is assumed to work if true emitter signals are detected. If compared with other
system level simulations, the channelized receiver part and emitter parameter
measurement part in signal processing are performed at the same study with
proposed approaches and done in different simulation tools. Emitter signals are
generated in VSS firstly and these emitter signals are received with a wideband
channelized intercept receiver in VSS again, at last these received emitter signals are
processed to measure emitter parameters in Matlab.
In the scope of this thesis, various radar signals such as pulse radar and Linear
Frequency modulated (LFM) pulse radar signals are generated to measure emitter
parameters by using VSS and Matlab. A wideband channelized intercept receiver has
a frequency band of 0.5-18GHz is performed to intercept more than one emitter
signals at the same time. After, signal processing part is employed to measure emitter
parameters. In order to achieve these goals, two different software platforms are used
efficiently. To generate and receive emitter signals, VSS is used in this thesis.
Finally, to measure emitter parameters, Matlab is used.
This thesis presents the simulations and results of the signal and system level
simulations and emitter parameter measurements for intercept receiver. Emitter
environment is designed to provide synthetic emitter signals which are used as the
input signals. Simulations on receiver configurations are described and designed as
proposed in [2]. This proposed approach is performed to receive more than one
emitter signal. After that, emitter parameter measurement part is performed as
proposed in [5], and implementation of this part is done in Matlab to measure emitter
parameters which are radio frequency, pulsewidth and time of arrival. Hereafter, the
results are compared and examined by changing several Noise Figures of the
receiver. The study done in this thesis is concentrated on software implementations
of channelized receiver and signal processing with proposed approaches that are
given in the literatures as [2] and [5], respectively.
4
The rest of the thesis is organised as follows. In the second Chapter, emitter
environment for emitter signal generations is performed and noise effect of the
receiver is discussed. In the third Chapter, receiver configurations are described in
details. In Chapter four, implementations of signal processing part are described in
detail. In the fifth Chapter, the results of the simulations are discussed. Finally,
conclusion is presented in Chapter six.
5
CHAPTER 2
EMITTER ENVIRONMENT GENERATION
In order to create a proper simulation for the channelized intercept receiver, there
must be more than one emitter signal exist and fall in different channels. For this
purpose, synthetic emitter signals are generated as emitter environment using VSS
[3]. As mentioned previously, both synthetically generated pulses and real radar
pulses taken from the literature are used as inputs to the system.
It is important to study with real emitter pulses to test and examine the receiver
part and signal processing part. By this way, reliability of the system is more
convenient. Nonetheless, synthetic radar pulses are added to emitter environment
generation part to increase the number of emitter signals. Two different type of
emitter pulse types are generated in this study. One of them is pulse radar signal and
the other is LFM pulse radar signal. PW, PRI and so TOA’s are selected by the user.
Furthermore, to get a realistic emitter environment a certain amount of noise is added
to the signal generation part. In addition, multipath effect is not added to the emitter
environment part of the system.
A sequence of rectangular pulses is used for a typical emitter signal generations.
According to the operation of civilian and military applications, there are some
parameters of emitters vary, and by this way various types of emitters can be
obtained [7, 8, 9, 10]. Generally, emitter parameters are classified, and there are no
too many number of emitters could be obtained in open sources. There is a very
popular source which gives some emitter parameters in [11], and they are used in this
study effectively [3]. Two kinds of modulated emitter types are used in this study and
they are pulsed radar and LFM pulse radar waveforms as discussed. When modelling
the noise effects, White Gaussian Noise is added to the system. The use of several
groups of blocks in VSS can easily help to generate both pulse radar signals and
LFM pulse radar signals.
6
In the emitter generation part of this thesis, it is desired that generated radar
pulses, which are pulse radar and LFM pulse radar, are received and measured in
VSS and Matlab precisely. Sample emitter generation blocks can be seen in Fig. 1
and Fig. 4 for both pulse radar and LFM pulse radar.
2.1
Emitter generation
As discussed at the beginning of this chapter, it is decided to add both pulse radar
and LFM pulse radar signals to emitter generation part. The main part of an emitter
signal, duration of τ centered at t=0 rectangular pulse function
amplitude of the signal
( ) with the
is as follow,
( )
( )
( ) is as follows,
In order to generate a periodic function,
( )
where the periodic function
(1)
∑
(
)
(2)
( ) has PRI (T) and duration of pulse is . Then,
periodic rectangular pulse function and coherent modulated CW ( ) that is given as
( )
( )
[
( )]
(3)
( )
]
(4)
Then, ( )is represented as,
( )
where
[
( ) is complex envelope of ( ) and
7
( ) can be written as,
( )
( )
(5)
( ) is as follows,
Therefore, inphase and quadrature parts of
( )
( )
[ ( )
( )]
(6)
where
( )
( )
( )
( )
( )
For Linear Frequency Modulated pulse complex radar signal
( )
where
∑
(
( )
(7)
( ) is as follows,
)
(8)
is bandwidth which is subtraction of starting frequency from ending
frequency, and
is sweep duration of LFM signal. To generate emitter signals, VSS
blocks were used effectively. As shown in Fig. 1, LFM pulse radar signals were
produced VSS blocks in this thesis.
Figure 1. LFM emitter signal generator in VSS.
8
Fig. 2 shows two pulses of LFM radar and its important parameters which are
generated in VSS. PRI, PW and bandwidth of LFM pulse radar signal can be seen in
Fig. 2.
Figure 2. Two LFM pulses generated in VSS.
As shown in Fig. 2, an LFM emitter signal is generated with a 32μs PRI and 8μs PW
can be seen in Fig. 2. In Fig. 2,
PRI and
is starting frequency,
is ending frequency, T is
is duration of the pulse. An example of LFM pulse generator is shown in
Fig. 1 in VSS. There were several emitters built to simulate the receiver. Fig. 3
shows the spectrum of generated LFM pulse in Fig. 2.
9
Figure 3. Spectrum of LFM pulse in Fig. 2.
According to Fig. 3, frequency spectrum of LFM pulse is obtained and this can be
supported by the theory with taking Fourier Transform (FT) of time domain signal.
Further information about FT can be found in [12]. Another example of an emitter is
shown in Fig. 4. Some blocks are used for generation of radar pulses in VSS, and this
simulator is very useful to generate emitter with its various parameters. Fig. 4 shows
an example of emitter generation blocks which has a simple pulse radar.
Figure 4. An emitter signal generator in VSS.
10
In Fig. 5, there is an emitter which has a 750MHz, 0.02μs pulsewidth(PW) and 0.2μs
PRI. Furthermore, to generate a simple radar pulse,
( ) is represented as 0 in
Equation 3. By this way simple pulse radar signal generated in VSS is shown in Fig.
5
Figure 5. Generated emitter signal in VSS.
Fig. 6 shows that the spectrum of the generated emitter signal in Fig. 5 and the
parameters which are carrier frequency, PW and PRI can be measured using this
spectrum. The envelope of this signal is a sinc function and this theory can be
supported by taking FT of time domain signal [3].
Figure 6. Spectrum of pulse in Fig. 5.
11
To support the receiver that is described in the next sections, many emitters were
used with both simple pulse radar and LFM pulse radar. The receiver operational
bandwidth was decided to cover 0.5-18GHz like very common commercial receivers.
In order to generate emitter environment, there were several emitters generated with
high sampling rate and it had to satisfy the Nyquist rate which causes a reduction in
simulation speed.
2.2
Noise effect
In order to simulate receiver, noise effect was assumed to the system. In the emitter
generation part, noise was modeled with White Gaussian Noise. 30dB Signal to
Noise Ratio (SNR) was assumed for emitter environment part. Receiver sensitivity is
affected by either internal generated noise of the receiver or video detector
characteristics. By keeping the RF gain high enough in front of the detector, only the
receiver noise level changes the sensitivity. Noise of the resistors can be called as a
noise generator that is in series. At the input of a receiver, thermal noise power can
be as follows
(9)
Where
is the power,
is bandwidth of the receiver,
Boltzmann’s constant (
is the temperature and
is
). Then the noise figure is as follows
(10)
where the receiver output noise is
, the receiver RF gain is
and
is thermal
input noise. For standardization of the equations, the room temperature is specified
by the authorities [2]. The receiver gain is described as
⁄
where
is the input and
(11)
is the output power then combining Equation 10 and 11
12
are as follows
⁄
⁄
Thereafter the output signal to noise ratio
to noise ratio
⁄
⁄
(12)
always smaller than the input signal
and the noise figure is greater than unity. Several devices can be
connected cascaded to the system [2]. In order to calculate the NF of a channel in this
thesis, which can be seen in Fig. 7, Equation 13 is as follows,
(13)
where,
filters NFs.
filters.
and
, which are inverse of the filter losses, are first and second
and
and
first amplifier. Finally,
are gains, which are inverse of losses, of the first and second
are the first and second amplifiers’ NFs and
is the NF of mixer and its gain is
is the gain of
. These NFs, gains
and losses are expressed in power ratios. With the help of these equations, noise
effect of the proposed receiver was applied in VSS. Budget analyze tool of the VSS
was used to determine noise figure of the receiver. Fig. 7 shows the noise figure
analyze of a channel of the receiver.
Figure 7. Cascaded noise figure analyze of a receiver channel.
13
Cascaded NF analyze can be seen in Fig. 7. The gain, loss and NF of the blocks,
which are filters, amplifiers and mixer, are stated in the Table 1 below. At the bottom
of the Table 1, cascaded NF is listed as in order.
Table 1. The gain, loss, NF and cascaded NF of a receiver channel.
Filter 1
Amplifier 1
Fılter 2
Mixer
Amplifier 2
Gain(dB)
-0.5
20
-1
-10
30
NF(dB)
0.5
3
1
10
5
Cascaded
0.5
3.68
3.69
3.92
4.427
NF(dB)
When the Equation 13 is performed to a receiver channel shown in Fig. 7 and the
parameters, which are given in Table 1, are applied to the equation, NF of the
receiver is calculated as follows,
(14)
As shown in Fig. 7, Noise Figure at the output of the channel is nearly 4.5dB.
Noise Figure is changed and tested in the next section.
14
CHAPTER 3
SIMULATION ON RECEIVER CONFIGURATIONS AND RECEIVER FRONTEND DESIGN
Due to their selectivity and high sensitivity, superheterodyne receivers are the most
commonly used receivers for communication. Most of the radar receivers are this
type of receivers. To measure signal information properly, this type of receivers are
mostly used in EW. On the other hand, parallel combinations of narrowband
superheterodyne receivers generate a channelized wideband receiver. The most
proper way to cower wide frequency band is to make a channelized receiver. In order
to cover a 2-18GHz band with a receiver, dividing the whole band into 1GHz parallel
bands with filters is an approach. This can be called a channelized approach;
however, the receiver cannot be called as a channelized receiver. It can only be
called a channelized front end. To make a channelized receiver, channels must have
filters and conversions stages [2]. To collect emitter signals, which are produced in
emitter environment part, a wideband channelized receiver is applied as proposed in
[2]. This wideband channelized receiver design covers 0.5-18GHz. The most
important part of the receiver is frequency channelization operations. In order to
analyze and collect more than one signal, frequency channelization is used [2, 9].
Channelized wideband receiver mainly contains large number of filters for every
channel. Emitter signals are separated from each other by receiver channels [2]. To
analyze different emitter signals simultaneously, frequency channelization is
necessary. The wideband channelized receiver block diagram is shown in Fig. 8. In
Fig. 8, the system starts with an emitter environment part to generate emitter signals
for simulation. In the first conversion stage, 0.5-18GHz frequency band is down or
up converted to 2GHz subbands. First conversion stage has 11 channels and every
channel has filters, a mixer, a local oscillator and amplifiers. However, in the third
channel, which is 2-4GHz channel, there is no mixer and local oscillator, because it
has already desired band. Therefore, N is the number of channel corresponds to 11
channels which can be seen in Fig. 8. After that, a switch is employed. The
15
operational band of the switch is 2-4GHz, and the duration of the every switched
channel is given maximum PRI of the emitter environment part. The second
conversion stage is employed after the switch. In this conversion part, 2-4GHz
channels are down converted to 200MHz subbands. In the second conversion, there
are 10 channels and M corresponds to channel number of the second conversion
stage which can be seen in Fig. 8. Then a 16 bit ADC is employed to convert analog
data to digital. At last, signal processing is employed to measure emitter parameters.
According to implemented receiver as shown in Fig. 8, a sample emitter signal which
has a 750MHz pulse is expected to receive with the first channel, which corresponds
to 0.5-1GHz channel, and it is up converted to 2750MHz. After first conversion, this
emitter signal is switched and second conversion begins. After that, the emitter
signal, which is received and converted by the first conversion, is received by the
fourth channel, which corresponds to 2.6-2.8GHz channel, and this signal is down
converted to 0-200MHz channel. At the end of the second conversion, the emitter
signal frequency is now 50MHz. Next, this signal is digitized and sent to signal
processing part.
1
≈
Emitter
Signals
Emitter
Environment
1
.
.
.
.
.
.
.
≈.
LO
N
1st Conversion
≈
.
.
.
.
.
.
.
.
1
.
.
.
.
.
.
.
≈
N
A
D
C
10001
LO
M
Switch
2nd Conversion
Figure 8. Complete system block diagram.
16
Parameter
Measurement
3.1
Frequency channelization
The desired receiver band is set to 0.5-18GHz and it is decided that the whole band
is firstly separated to 2GHz sub-bands [2]. According to proposal of [2], two receiver
structures can be implemented to cover 0.5-18GHz. For the first proposed structure,
the whole band is directly up or down converted to 2-4GHz channels. Then, the first
channel is set to 0.5-2GHz, the second is set to 2-4GHz and the last channel is set to
16-18GHz. All input bands are converted to 2-4GHz bands for matching the input
frequency. This structure has a very critical disadvantage. The channels adjacent to
2-4GHz channel are affected by spurious signals [2]. As shown in Fig. 9, some
generated emitter signals can be seen between 1-5GHz bands. These emitter signals
are some of the generated signals which are marked on the Fig. 9.
Figure 9. Some generated emitter signals at 1-5GHz band.
For instance, it was mentioned before, 0.5-2GHz channel is a problematic channel
and when the first structure was implemented, it was expected that 1200MHz and
1500MHz emitter signals were up converted to 3200MHz and 3500MHz
respectively. However, it was seen that many spurious signals were been formed,
furthermore it can be seen that these spur signals’ powers are very close to expected
17
signals in Fig. 10. The 0.5-2GHz problematic channel output can be seen with its
unwanted spur signals in Fig. 10.
Figure 10. 0.5-2GHz problematic channel belongs to first structure.
Up converted actual emitter signals 3200MHz and 3500MHz are marked in Fig. 10,
but it is seen that there are 3175MHz, 3225MHz, 3425MHz and 3525MHz unwanted
signals in Fig. 10, which are generated by mixers, can be seen at the same time.
Moreover, some of these signals are stronger than the actual signals.
On the other hand, due to this structures spurious problem, another approach is
implemented. For the second approach, instead of dividing the whole band to 24GHz channels, channelization process starts with 0.5-1GHz and goes with 1-2GHz
and follows with 2-4GHz channel. After 2-4GHz channel, the receiver channels
continue with 4-5GHz and follows with 5-6GHz. After the 5-6GHz channel, the
channelization continues with 6-8GHz channel and the last one is 16-18GHz. To
summarize this case, unlike the first structure the channels adjacent to 2-4GHz
channel, which are 0.5-2GHz and 4-6GHz channels, are divided in to 0.5-1GHz, 12GHz, 4-5GHz and 5-6GHz channels with respectively [2]. According to [2], 0.51GHz and 4-5GHz bands should be converted to 2-3GHz subband. 0.5-1GHz
subband should be converted 2.5-3GHz channel and 4-5GHz subband should be
18
converted to 2-3GHz. For 1-2GHz and 5-6GHz subbands, they should be converted
to 3-4GHz channels. Hereafter, channelization continues with 2GHz subbands to the
end of 16-18GHz subband [2]. This means that the first conversion contains 11
channels. Fig. 11 shows the first and second channels in the second approach. The
other channels have same types like in 1-2GHz channel but related blocks and its
values can be different. Each channel of this approach operates as a narrowband
superheterodyne receiver and their parallel combinations construct a wideband
channelized receiver [3].
Figure 11. The first and second channels of the second approach.
According to this approach, which is performed in this thesis, it is expected that
1200MHz and 1500MHz emitter signals are received by 1-2GHz channel. Moreover,
as stated in [2], unlike the first structure there should be no unwanted spurious
signals at this band which can be seen in Fig. 12.
19
Figure 12. VSS output of the second proposed structure for 1-2GHz channel.
It can be easily seen that 1200MHz and 1500MHz emitter signals are converted to
3200MHz and 3500 MHz respectively in Fig. 12. Furthermore, there are not any
spurious signals close to actual emitter signals or stronger than the actual ones.
Before the second channelization operation, a dynamic switch was employed and
this is discussed in next sections. After the first channelization, these subbands are
again separated to 200MHz subbands. In order to reduce presence of more than one
emitter signals at the same channel, second channelization is performed. Emitter
parameter measurements could be easier by this way; because, less emitter falls in a
same channel. Fig. 13 shows two channels of the second channelization stage. The
other channels have same types but their filters and local oscillators (LO) values can
be different.
20
Figure 13. Two channels of second conversion stage.
In this stage, it operates like the first conversion stage, but its operational bands
and channels bands are different. In Fig. 13, two channels from the second
conversion can be seen, and the switched channels from the first conversion are
separated by 200MHz channels. The channelization starts with 2.0-2.2GHz channels,
continues with 2.2-2.4GHz channel and lasts with 3.8-4GHz channel. This means
that the second conversion contains 10 channels.
Therefore, to keep signal quality, design considerations were implemented to this
stage such as, NF and SNR etc. Finally, tests showed that second stage was
implemented successfully.
After the second stage, it was seen that if emitter numbers are increased, there
could be more than one emitter signal fall at the same channel. To avoid this
problem, another solution was employed in the signal processing part. In the signal
processing part, which is described in next sections, presented algorithm can separate
signals.
21
3.2
ADC and Switch state
In order to convert analog signal to digital one, 16 bit ADC is used in this study.
One of the most important parameters of the receiver is dynamic range (DR). In this
case, ADC selection is an important part too. For a digitized signal the resolution of a
converter is limited by SNR. For this case, it is expected that DR of the ADC must be
bigger than the DR of the receiver; however, if the DR of the ADC is too bigger than
the receiver, DR will effect badly, and it is not desired. Receiver dynamic range can
be calculated with Equation 15 which is given in [2],
[
(
)
(
)]
(15)
Where IIP3 is the third order intercept point of the receiver, NF is the Noise Figure of
the receiver and BW is the receiver bandwidth. When the equation 15 is applied to
the receiver configurations, which are 40dB
, 4dB NF and 2GHz BW, DR is
calculated as follows,
(16)
Under the light of these situations, 16 bit ADC is used in this system [13]. To
calculate DR of an ADC is given in [2] as follows
(
where
)
(17)
is the number of quantization level. Number of quantization level and
number of bits is related to each other, such as 16 bit ADC has 16 quantization
levels. According to 1 bit change in ADC causes 6dB change in DR. Therefore, it is
important to adjust ADC not to affect DR of the system badly. SNR and DR of the
ADC are directly proportional to each other. 16 bit ADC has a maximum SNR of
96.32dB. Therefore, it is suitable to use 16 bit ADC if the receiver has a dynamic
range under 90dB. The ADC input signal sampling rate should satisfy the Nyquist
criterion. The bandwidth of the ADC is 200MHz at the second conversion stage, so
sampling rate must be greater than twice or more of the bandwidth. At the same time,
22
when the sampling rate of the ADC increases, the digitally processing speed
decreases. Therefore, processing time can be effected badly [13]. For further
information about ADC can be found in [13].
Figure 14. Dynamic time shared switch.
To avoid loss of frequency information, a different time shared switch state had
been used in this study. Generally, the switches are used with activity detection
stages. For instance, activity detectors are positioned between at the end of the
channels and inputs of the switch. Activity detectors detect all the channels if a signal
is present, that channel is switched and all the other channels are delayed. In this
activity detection, the switch input is activated by the leading edge of pulses detected
with a threshold detector, by this way only active channel signal, which has pulse or
pulses, passes to the second conversion stage and the other channels outputs are
delayed till the trailing edge of the pulse is ended. This process continues with the
same procedure for other channels [14]. For further information about activity
detection and switch can be found in [14]. However, VSS starts to produce signals
from the lowest frequency to highest frequency. In this case, this activity detection
cannot be applicable. Because, the switch keeps open the channels with respectively.
For example, if there are 500MHz, 1500MHz, 2500MHz and 5000MHz emitter
signals, activity detector is firstly activated by 500MHz signal, then this signal passes
through to first input of the switch. After that, the same procedure like for the first
23
500MHz signal, continues with 1500MHz for the second input, 2500MHz for the
third input and 5000MHz for the fourth input of the switch, respectively. Therefore,
this type of activity detector and switch are useless in this case. Instead of this type,
dynamic switch was used to switch all channels in this thesis. For this situation,
another way was followed. Firstly, to keep open switched channel, maximum PRI
(T) of the emitter environment is used to determine switched channel duration.
According to this situation, all channels are kept open with the same duration
respectively, but channels are delayed with multiplication of channel number and
maximum T. To keep channels in respectively, delay is used as can be seen in Fig.
14. The switching state stars with the first channel and goes second and then ends
with Nth channel respectively, and multiplication of PRI and the channel number
gives delay time of each channel. N is the input number of the switch and this
number must be same as with the channel number of the first conversion stage.
Switched input channel numbers are stored with a look up table not to loose channel
information. This channel information will be helped to find frequency information
in next sections.
24
CHAPTER 4
EMITTER PARAMETER MEASUREMENTS PART
In this part of the study, it is decided that emitter parameters such as PW, TOA,
TOD and RF are measured which are parameters of pulsed radar and LFM pulsed
radar emitters. In order to measure emitter parameters, a direct and efficient approach
is performed as described in literature [5]. In early years, emitter parameter
measurements were employed with circuits in the receiver. However, nowadays
emitter parameter measurements are performed with software platforms. Fast speed
computer based processor can help to measure emitter parameters efficiently.
In this thesis, a simple and efficient algorithm is performed as given in [5]. Before
implementing this approach, averaging filter is used to obtain envelope of the signal.
After that, double threshold levels are performed to get the envelope of the signal.
When the previous signal is lower than the upper threshold and present signal is
higher than the upper threshold leading edge is detected by this way. If previous
signal is upper then the lower threshold and present signal is lower than the lower
threshold trailing edge is detected by this way. The leading edge and trailing edge
locations give TOA and TOD, respectively [5]. According to [5], TOA and TOD
locations are used to separate pulses in the same channel. In addition to this,
increasing the range between two threshold levels can give better measurements, but
the receiver sensitivity is affected badly by this way. To avoid this, 3dB difference is
generally used [5].
In order to measure radio frequency of the emitter signals, FFT is performed as
mentioned in [5]. Firstly, FFT is employed to the signal, after that maximum peak of
the sampled FFT is found and another threshold level is used to determine the radio
frequency of the emitter signals. If just one peak exists on the upper side of the
threshold, maximum peak is measured as simple pulse radio frequency and if more
than one peak exists on the upper side of the threshold level, the first and second
25
maximum peaks are used to determine LFM pulse radio frequency, which is
described in this Chapter.
4.1
Pulse envelope
In order to have constant amplitude (flat), pulse must be perfect. Because of
temperature effect and non-linear characteristics of devices, the amplitude of the
pulses is not flat. Frequency changes do not affect the pulse envelope and from pulse
to pulse it has a specific issue. Further information about pulse envelope can be
found in [15], [16], [17], [18] and [19].
In this study, pulse envelope y(n) is calculated by using averaging filter of the
amplitude of A(n) of input pulse [15]
( )
∑
(
)
(18)
where the filter length is N. This filter is implemented in the signal processing part of
this study. The reason of using this filter is to reduce the noise of the signal. In Fig.
12 an average filtered pulse can be seen.
For the calculation of the pulsewidth, window based power calculation is done to
designate %50 of the pulse [15]. On the pulse,
( ) in average power
( ) is as
follows [15]
( )
| (
∑
)|
(19)
where the window initial sample is n and the window size is N. When the average
power calculation of the pulse is done, start and stop points of the pulse can be
calculated by double thresholding.
26
Figure 15. Captured pulse by the Receiver.
When average power is obtained to pulse seen on Fig. 15, then the pulse envelope is
formed like in Fig. 16. After this stage, parameters’ measurement of the emitter
signals begins.
4.2
Double threshold
In order to measure emitter parameters, double threshold noise gate is used in this
study. In this study, as mentioned in section 2.2, signals can be affected by noise. In
Matlab environment to detect the pulse, double threshold noise gate was applied in
the signal processing part as introduced in [5]. The receiver may be triggered
multiple times if the input signal is very close to the threshold level. In this case,
receiver detects incorrect information.
On the other hand, to avoid this problem double threshold is applied. According to
this case, the signal must exceed the upper threshold level for triggering to obtain
27
data, and drop under the second threshold to state end of the pulse and stop achieving
data [5].
Fig. 15 shows the double threshold on averaged power on pulse. The range is
arbitrary chosen between two thresholds. If the range between thresholds is too large,
receiver sensitivity can be affected and if the range is too short, data acquisition can
be affected in emitter parameter measurements’ part. In general, the range between
two thresholds is selected 3dB not to effect receiver sensitivity in a bad way [5].
Rise time and fall time locations can give important clues about the type of the
radar. These rise and fall regimes can be generated intentionally or unintentionally on
pulse shaping, but bandwidth of the pulse is limited by pulse shaping. To increase
main lobe of the amplitude and to decrease side lobes pulse shaping can be applied
[15]. On the rising edge, the rise time can be represented by the time difference
between %10 and %90 points of average amplitude [16]. Furthermore, fall time has
the same response but in trailing edge. The rise, fall time and threshold values can be
seen on Fig. 16.
28
Figure 16. A radar pulse with various features.
29
Emitter parameters
4.3
In this stage, pulse radar and LFM pulse radar parameters are measured which are
captured and received by the receiver and measured in parameter measurement part.
Describing the pulsed radars can be performed with carrier frequency, which
depends on radar operation, PW, modulation and pulse repetition frequency (PRF)
[2]. In this study, simple pulsed and LFM pulse waveform are used to generate
emitter environment. Hence, radio frequency, PW and TOA parameters are measured
in emitter parameter measurement part.
4.3.1
Time of arrival (TOA) and time of departure (TOD)
In the proposed approach of [5], time reference of all pulses can be obtained by
TOA. Some emitters generate pulses with staggered PRI, in other words PRI varies
within specific periods. However, with stable PRI kind of emitters, practical
parameter is TOA. In this thesis, in order to provide time reference for measuring
received pulse parameters, TOA is used. Also, TOA is used to separate pulses if
more than one pulse exists at the same channel. By measuring each sample’s
amplitude levels of the TOA and TOD with thresholds, TOA and TOD information
indices can be found in Matlab. Positive edge detection occurs when the previous
sample is lower than the upper threshold and the present signal is higher than the
upper threshold level. TOA index is found by this way. Algorithm states this index as
a TOA index. On the other hand, negative edge detection occurs when previous
signal amplitude is higher than the lower threshold and the present is lower than the
lower threshold level. This time the algorithm states that indices as a TOD indices
[5]. These TOA and TOD indices are used to find number of pulses. Moreover, if
there is more than one pulse, these pulses are found by using TOA and TOD indices.
In addition to this, every TOA and TOD indices indicate one pulse and by this way
signal separation for PW measurements of the emitters can be done which is
described in section 4.4.3.
A flow chart starts from the IF signal, which is obtained at the end of the receiver,
and lasts with TOA and TOD measurements for driven algorithm is presented in Fig.
30
17. According to this algorithm flow chart, IF signal is digitized with an ADC, after
that envelope of the whole channel is obtained with averaging filter and double
threshold approach is applied. After applying threshold levels, TOA and TOD
indices are found which is described in this section.
31
IF Signal
2.5
2
1.5
Amplitude
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
1
2
3
Time(s)
4
5
6
x 10
ADC
-5
2.5
Amplitude
2
Envelope
1.5
1
0.5
0
1
3
Time(s)
4
5
6
x 10
-5
3
2.5
2
Amplitude
Treshold 1
Treshold2
2
1.5
1
0.5
0
0
1
2
3
4
Time(s)
5
6
7
x 10
-5
Pulse Presence (Pulse Indices
location and Number of Pulse
Determination)
TOA and TOD
measurement for
all pulses
Figure 17. Flow chart for TOA and TOD indices measurement algorithm.
32
4.3.2
Pulse Width (PW)
In order to measure PW of the emitter signals, a simple method is used in this
thesis. A high pass filter was used to find PW in early years. Because when the pulse
passes through a high pass filter, leading edge is turned into positive spike and
trailing edge is turned into negative spike. A great accuracy was achieved for PW
measurements by using a counter, because positive spike location is used to start
count and the negative spike location is used to stop count. Today, complete shapes
of the pulses can be achieved with fast ADCs, because PW can be sampled with a
high rate [5]. Therefore, according to proposed approach in [5], calculated TOA and
TOD indices are very helpful to indicate PW with simple Equation 20 given in [5]
[
where pulse width indices length is
]
(20)
. TOA and TOD indices can be seen in Fig.
18.
Figure 18. Pulse indices graph.
33
4.3.3
Radio frequency (RF)
Radio frequency is one of the most important parameter for emitter signals.
Sampled data frame is converted from time domain to frequency domain in order to
measure emitter frequency. For this measurement Fast Fourier Transform (FFT) is
performed in this thesis, which can be found in [12], to measure radio frequency of
the emitters. In this study, RF is calculated by a simple equation which is given in
literature [5]. According to [5], the location of the maximum FFT output sample
(
) is an approach to find frequency of the emitter signal. In order to use this
approach for determining the emitter frequency, given data frame length ( ), local
oscillator frequency (
) and sampling frequency ( ) are used. Therefore the
equation is as follows,
(21)
In this study, there are two kinds of emitter signals generated. These are simple pulse
and LFM pulse signals. In order to measure simple pulse radar signal frequency, it is
sufficient to use Equation 21. However, Equation 21 is not sufficient for LFM pulses.
To avoid this, a simple threshold algorithm is used as an approach of this thesis like
in Section 4.3. After FFT is applied to the pulse sample, a threshold is applied under
the maximum peak of the FFT output sample. This threshold can help to determine if
the signal is simple pulse or LFM pulse. According to this threshold, maximum peak
of the FFT sample is found firstly and this threshold is located under 3dB of the
maximum peak. After that, if one peak exists between maximum peak and threshold
level, this peak is described as maximum peak location of the FFT sample, then this
location (
) is used in Equation 21 to measure emitter signal frequency.
Furthermore, if there is more than one peak between the maximum peak of the FFT
sample and threshold, the second maximum peak location is found. By this way, the
range between these two maximum peak locations gives LFM signal frequency
bandwidth.
The
algorithm
flow
chart
34
can
be
seen
in
Fig.
19.
IF Signal
ADC
Pulse Separation
Pulse Presence
(Pulse Indices
location and
Number of Pulse
Determination from
Section 4.4.1)
FFT
Max. Peak
Power
Treshold
Peak
Detections
and Decision
Simple Pulse
Frequency
LFM Pulse
Frequency
Figure 19. Flow chart of simple pulse and LFM pulse frequency measurement algorithm.
35
CHAPTER 5
SIMULATION RESULTS
RF front-end design and digital signal processing procedures were employed
successfully on AWR-VSS environment and Matlab with stated proposals.
Simulation was performed with several emitters. These emitters were used both pulse
signal and LFM pulse signal.
In the literature, most of the proposed approaches are concentrated on theoretical
aspects of simulations for emitter parameter measurements. Furthermore, less
importance is given on performing receiver and parameter measurement stages at the
same study. They are mostly concentrated on receiver stage or parameter stages.
Some of these studies can be found in literatures [1], [5], [20] and [21].
In the scope of this thesis, emitter generation is employed in emitter environment
part to simulate the channelized receiver and parameter measurement parts. Emitter
environment part is implemented in VSS with both simple pulse and LFM pulse
radar signals. Secondly, to receive generated emitter signals from the emitter
environment part, a wideband channelized intercept receiver is performed with first
and second conversion stages. This receiver has a 0.5-18GHz operational bandwidth
as can be found commercial ones in open market. After receiving stage, these
received emitters’ signal parameters are measured in Matlab. For the wideband
channelized receiver, a proposed structure is implemented as described in [2]. For the
parameter measurement part, a proposed method is implemented as described in [5].
Finally, a Graphical User Interface (GUI) is designed to demonstrate measured
emitter parameters.
36
5.1
Signal separation
To measure desired emitter parameters, some algorithms, which are defined in the
parameter measurement part of the study, were used in this study. For time domain
analysis, same method, which was described in sections 4.4.1 and 4.4.2, was
employed because both pulsed radar and LFM pulse radar signals can be analyzed
similarly in time domain. In order to measure the frequency of the pulses, FFT is
used which is described in section 4.4.3.
At the beginning of this study, it is assumed that only one signal will fall on one
channel in receiver, however; when number of emitters was increased to get test data,
it was seen that there could be more than one signal falls on a channel. To avoid this
problem, signal processing algorithm was improved to separate all emitter signals at
one channel. In the processing part, different emitter signals separated firstly with an
approach described in section 4.4.1, than the parameter measurement is employed as
mentioned in section 4.4.1 and 4.4.2. By this way, emitter separation can be done
efficiently.
37
Figure 20. Three pulses received with a channel.
38
In Fig. 20, there are three pulses received at the same channel. It is desired to
separate these three different pulses. As seen on Fig 20, an LFM and basic pulsed
emitter signals are received on the same channel of a receiver, and the written
algorithm checks the envelope of the whole channels and then determines the
number of the pulses, thereafter the algorithm starts to analyse pulse parameters
which are described in section 4.4.1 with flow chart in Fig. 17. Fig. 21 shows an
emitter signal which was received and separated from the other signals which can be
seen in Fig. 20.
Figure 21. One of the separated pulses in three different pulses.
In Fig. 20 there are 3 different pulses can be seen and these three different pulses are
separated to measure their parameters. After separation is done by the approach
which is described in section 4.4.1, parameter measurement part is performed as
described in section 4.4.2. Furthermore, for frequency measurement, the location of
pulses indices, which are obtained from section 4.4.1, are used to determine FFT
sample location as mentioned in section 4.4.3.
39
5.2
Parameters’ measurement results
To measure emitter parameters such as pulsewidth (PW), time of arrival (TOA)
and frequency, approaches and algorithms, which are described in section 4.4.1,
4.4.2 and 4.4.3, were employed in Matlab. In time domain analyses, section 4.4.2 is
used to process pulse radar and LFM pulse radar signal as mentioned before, but in
frequency domain different approach is used which is described in section 4.4.3 and
the driven algorithm can be seen as an algorithm flow chart in Fig. 19. A received
and measured emitter signal can be seen in Fig. 22. This emitter signal in Fig. 22
belongs to a channel which is shown in Fig. 20.
-10
X: 5.753e+07
Y: -12.32
-20
Power(dBm)
-30
-40
-50
-60
-70
2
4
6
8
Frequency(Hz)
10
12
14
x 10
7
Figure 22. FFT of pulse in Fig. 21.
Exact frequency information of the signal in Fig. 21 and 22 is 57MHz, but it was
measured 57.53MHz due to noise effect of the receiver, which is described in section
2.2, in RF front-end.
A sample emitter parameter measurement is shown in GUI in Fig. 23. It can be
40
seen that some emitter parameters are measured and monitored. These parameters are
PW, TOA and radio frequency measurements. As written “Emitter 1” in the Fig. 23,
indicates the first pulse as in Fig. 15. “Emitter 2” is measured simultaneously, which
is shown in Fig. 21 and 22 particularly, and “Emitter 2” is shown with GUI in Fig.
23. PW parameters, which are shown in Fig. 23, can be measured with a very small
error, which is expected, when the actual received emitter signals and measured
parameters are compared in this section. To measure radio frequency of the of simple
pulse and LFM pulse signal an algorithm is employed which is described in section
4.4.3 and driven algorithm of radio frequency measurement flow chart is shown in
Fig. 19. Moreover, these emitter signals radio frequencies can be shown with a GUI,
which is created in Matlab to show measured parameters, in Fig. 23.
41
Figure 23. Matlab GUI for parameter measurements.
42
Finally, all channels are analyzed and then emitter parameters are monitored like in
Fig. 23. After parameter measurement part, channel information taken from the
switch state is used to determine actual emitter signals. According to “look up table”
in switch state channel information is taken and added to measured frequency and it
gives actual radio frequency of the emitter.
In order to test the receiver performance, Noise Figure (NF) of the receiver, which
is described and realized in section 2.2, was set to 8, 10, 12 and 14dB respectively. In
open literature, there are a lot of commercial wideband channelized receivers. Some
of these receivers and their specifications can be found in [22, 23]. The receivers’
NF’s are varied between 8dB and 21dB in given literature [22, 23]. In order to
achieve accurate results in parameter measurement, 8dB NF is performed in this
thesis. Furthermore, to observe noise effects on emitter’s parameter measurement,
10dB, 12dB and 14dB NF is performed. On the other hand, there are different
channels in receiver and some of their channel specifications, which is described in
section 3.1, are different, so NF of the channels can be different. But they are varied
around the desired NF of the receiver in very small differences. Therefore, overall
receiver NF is described as desired NF.
Several emitters are added to test PW, TOA, TOD and radio frequency parameters
of emitter signals. Test results of measured emitter parameters with different NFs can
be seen in Fig. 24. TOD parameter is not added to Fig. 24, because PW is measured
with using time difference between TOA and TOD, so TOD will have the same
curve with TOA parameter.
43
Av. Error vs. NF
10
9
Average Error in %
8
7
6
5
PW Av. Error
4
TOA Av. Error
3
Freq. Av. Error
2
1
0
6
8
10
12
14
NF in dB
Figure 24. Average error versus NF.
It can be desirable to report the PW in a logarithmic scale, where the PW varies
with a wide range. This means that, fine time resolution is used to measure short
pulses and coarse time resolution is used to measure long pulses [2]. In addition to
this, according literature given in [2], a pulse, which has a pulsewidth of hundred
nanoseconds, can be measured with an accuracy of 50ns resolution; however a pulse,
which has tens of microsecond pulsewidth, can be measured with an accuracy of 1µs
resolution. In addition to this, a study which is described in [20], 50ns resolution is
achieved for 1µs PW. On the other hand, in this thesis, PW of the “Emitter 9” was
measured 0,7966µs (actual is 0,8µs) and another PW of the “Emitter 10” was
measured 15,011µs (actual is 15µs) which can be seen in Table 3. With the help of
these measurements, it is understood that with the same time resolution, long pulses
can be measured more accurate than short pulses.
According to the Fig. 24, when NF of the receiver increases, average error of the
PW, TOA and frequency of the emitters increase in logarithmic scale. Frequency
error measurement is not increased like PW and TOA because frequency
measurement is employed in frequency domain. Under the lights of the Fig. 24, when
NF is high (above 10dB), the accuracy of the measurement decreases after quickly
44
after 10dB. Table 3 shows that 10 emitters’ PW parameters measured in different
NF’s. NF of the receiver is changed and measured parameters are observed to test the
system. According to [2], frequency accuracy can be 1MHz and it can even be
10MHz. Moreover, the literature [2] states that 1MHz resolution is high frequency
accuracy. Under the light of this situation, high frequency accuracy can be achieved
under 12dB NF of the receiver.
Table 2. 3 Emitters’ radio frequency parameters due to different NF values
Actual Emitter
Frequency in MHz
Emitter 1
Emitter 5
Emitter 10
750
5000
16555
750,01
5000,021
16555
750,11
4999,54
16555,67
751,26
5501,12
16555,81
753,34
4996,32
16556,62
Measured
Frequency in MHz
for 8dB NF
Measured
Frequency in MHz
for 10dB NF
Measured
Frequency in MHz
for 12dB NF
Measured
Frequency in MHz
for 14dB NF
45
Table 3. 10 Emitters’ PW parameters due to different NF values.
Emitter
1
Emitter
2
Emitter
3
Emitter
4
Emitter
5
Emitter
6
Emitter
7
Emitter
8
Emitter
9
Emitter
10
Actual Emitter PW
in µs
0,5
0,5
0,5
0,5
5
5
0,7
10
0,8
15
Measured PW in µs for
8dB NF
0,4979
0,4977
0,4969
0,4974
4,9834
4,981
0,6934
9,973
0,7966
15,011
Measured PW in µs for
10dB NF
0,4961
0,4949
0,4954
0,4957
4,9484
4,973
0,6831
9,862
0,7836
15,03
Measured PW in µs for
12dB NF
0,481
0,4821
0,4813
0,4763
4,8565
4,85
0,6614
9,714
0,734
15,27
Measured PW in µs for
14dB NF
0,441
0.4453
0,4445
0,45
4,6465
4,621
0,63
9,623
0,689
15,61
46
CHAPTER 6
CONCLUSIONS
In this thesis, a wideband channelized intercept receiver and emitter parameter
measurements are performed to receive and measure generated emitter parameters. A
proposed structure is implemented to cover 0.5-18GHz band which is given in [2].
To cover 0.5-18GHz band and to receive and measure more than one emitter signal,
wideband channelized receiver structure is performed in this thesis. AWR Visual
System Simulator is used to generate emitter pulses and this simulator is used again
to implement 0.5-18GHz band channelized receiver. Finally, in order to measure
emitter parameters, Matlab is used.
The literature studies are generally focused on hardware design and
implementations or software implementation on simulations tools. Moreover, these
studies in literature are concentrated on theoretical approaches. In this study, both
front-end and digital signal processing applications are implemented at the same time
with different simulation tools effectively.
A co-operation between AWR Visual System Simulator and Matlab is done in this
thesis. In order to simulate implemented receiver and emitter’s parameter
measurement parts, emitter environment part is performed to generate emitter
signals. Simple pulse radar and LFM pulse radar signals are generated to use them as
emitters. Secondly, receiver front-end, which is proposed in [2], is implemented in
VSS. At last, the outputs of the receiver are simultaneously analyzed in the signal
processing part of this study in Matlab.
The combination of VSS and Matlab and its stages start with first channelization
part and goes with switch state, then continues with second channelization part,
finally ends with digital signal processing part. This is a new study of implementing
two different simulation tools for wideband channelized intercept receiver and
emitter parameter measurements. In addition this, due to limitations of VSS, activity
47
detection cannot be employed. Therefore a dynamic switch is performed to avoid this
problem
System level design for digital wideband channelized receiver and emitter
parameter measurement for Electronic Warfare (EW) and Electronic Intelligence
(ELINT) is presented in this study.
For ESM operations, channelized receivers are employed due to their superior
performances on capability to handle simultaneous signals, fine frequency
resolutions and good dynamic ranges [2]. In the scope of this thesis, good capability
to handle simultaneous signals is achieved with more than one emitter signal can be
measured at the same channel. In literature [2], maximum 4 emitter signals can be
measured at same channel is acceptable. Moreover, 1MHz frequency resolution and a
dynamic range between 65 and 95dB are acceptable [2]. These goals are achieved
with 1MHz frequency resolution and 82dB dynamic range.
In order to cover the desired 0.5-18GHz band, the whole band is divided into
2GHz subbands and all bands are down or up converted to 2-4GHz band. However,
this structure has a disadvantage. Adjacent to 2-4GHz subbands have unwanted spur
problems [2]. To overcome this situation a different method is proposed in [2] and
implemented in this thesis efficiently. According to second approach in [2], the
channels adjacent to 2-4GHz channel are divided into two channels again. Thus, the
channels adjacent 2-4GHz channels are performed as 0.5-1GHz, 1-2GHz, 4-5GHz
and 5-6GHz. In addition to this, 0.5-1GHz and 4-5GHz channels are converted to 23GHz band, then 1-2GHz and 5-6GHz channels are converted to 3-4GHz band. By
this way, spur problem is eliminated. The wideband channelized receiver tests, which
are presented in Chapter 2, show that second approach is sufficient to avoid spur
signals adjacent to 2-4GHz channel. Emitters’ parameter measurement is done when
the proposed approach in [2] is implemented with using AWR Visual System
Simulator.
Another proposed approach, which is described in [5], is employed to measure
emitter parameters. By using a proposed approach in [5], pulse width, time of arrival
and radio frequency of the emitter signals are found. In Chapter 4, it is described that
48
TOA and TOD parameters are measured by using averaging filter and double
threshold. Averaging filter is used to obtain pulse envelope. In addition to this,
proposed approach in [5] is not used pulse envelope to measure parameters, but in
this thesis pulse envelope is used before double threshold. Moreover, PW
measurement is done with using TOA and TOD parameters. Radio frequency
measurement of simple pulse and LFM pulse are done with using FFT as described
in Chapter 4.
In this thesis, two approaches given in [2] and [5] are combined to receive and
measure LFM and simple pulse radar parameters which are PW, TOA and RF. A
proposed second approach in Chapter 2 is performed to receive emitter signals and
averaging filter is used to obtain pulse envelope, after that another proposed
approach in Chapter 4 is used to measure emitter parameters. In addition to this,
second approach in [2], which proposes 0.5-18GHz receiver structure, is performed
in VSS and another approach in [5] is performed in Matlab to measure emitter
parameters. To combine these approaches and measure emitter parameters these two
simulations tools, which are VSS and Matlab, are combined.
The measurements and tables, which are presented in Chapter 5, show that the
emitter parameters PW, TOA and radio frequency are measured with a great
accuracy. To increase the accuracy of the measurement, the threshold levels can be
adjusted to lover levels for time domain measurements. This adjustment can give
better measurements. Moreover, double threshold approach presented in Chapter 4 is
a very efficient way to measure emitter parameters. According to the results and
error calculations with comparison, proposed approach to cover 0.5-18GHz band
given in [2] and parameter measurement part works and implemented successfully.
49
REFERENCES
[1]
C. Pandolfi, M. Bartocci, G. Gabrielli, P. E. Longhi, A. Megna, and B.
Orobello, “Compact Wideband Downconverter Module for Electronic
Warfare Applications”, 6th European Radar Conference, pp. 355-357
Rome, September – October. 2009.
[2]
J. B. Y. Tsui (1986), Microwave Receivers with Electronic Warfare
Applications, 1st ed., John Wiley& Sons.
[3]
A. Kara, İlter Karadede, “Signal and System Level Simulations for
Wideband Intercept Receivers: Eliminating Spur Signals”, 9th International
Conference on Electronics, Computer and Computation, Ankara, pp.215219, 2012.
[4]
Mahafza BR, Elsherbeni AZ (2004). Matlab Simulations for Radar Systems
Design. Chapman & Hall/ CRC Press Company, USA.
[5]
B. M. Albaker, N. A. Rahim, “Detection and parameters interception of a
radar pulse signal based on interrupt driven algorithm”, Scientific Research
and Essays, vol. 6, pp.1380-1387, March. 2011
[6]
http://awrcorp.com/download/faq/english/examples
[7]
D. L. Adamy, (2009), Tactical Battlefield Communications Electronic
Warfare, Artech House.
[8]
M. I. Skolnik, (2001), Introduction to Radar Systems, 3rd ed., McGrawHill.
[9]
D. D. Vaccaro, (1993), Electronic Warfare Receiving Systems, Artech
House.
[10]
S. D. Berger, “DRFM Linear Range Gate Stealer Spectrum”, IEEE
Transactions on Aerospace and Electronic Systems, vol. 39, pp.725-735,
April. 2003.
[11]
CEA Tech. Pty Ltd, Jane’s Radar and Electronic Warfare Systems, 2002.
[12]
S. J. Orfanidis (2010), Introduction to Signal Processing, Prentice Hall.
[13]
J. B. Tsui (2004), Digital Techniques for Wideband Receivers, SciTech
Publishing.
[14]
D. E. Allen, “Channelised receiver A viable solution for EW and ESM
system”, IEE PROC, Vol. 129, Pt. F, No. 3, June. 1982.
50
[15]
C. Erdem, A. Kara, “Towards real time implementation of specific emitter
identification on pulse waveforms:optimizations on a low cost commercial
platform”, October. 2012.
[16]
J. Eilevstjohn, J. E. Odegard , E. Malnes , “Radar Pulse Classification Using
Compressed Intrapulse Feature Vectors”, IEEE DSP Workshop, 2000.
[17]
S. UR Rehman, K. Sowerby, C. Coghill, “RF Fingerprint Extraction from
the
Energy
Envelope
of
an
Instantaneous
Transient
Signal”,
Communications Theory Workshop (AusCTW), 2012.
[18]
O. Ureten and N. Serinken, “Detection of radio transmitter turn-on
transients”, Electronics Letters, vol. 35, no. 23, pp. 1996–1997, 1999.
[19]
A. Kawalec and R Owczarek, “Specific emitter identification using
intrapulse data”, EURAD, First European Radar Conference, pp, 249-252,
2004.
[20]
E. Carpentieri and S. Cuomo, “An adaptive threshold algorithm for
detection of pulse radar signals”, IEEE Radar Conference, 2008.
[21]
H. Jiang, W. Guan, L. Ai, “Specific Radar Emitter Identification Based on a
Digital Channelized Receiver”, 5th International Congress on Image and
Signal Processing (CISP), pp. 1855-1860, 2012.
[22]
http://www.mw-elisra.com/
[23]
http://www.teledynedefence.co.uk/
51
Download

SIGNAL AND SYSTEM LEVEL SIMULATIONS ON WIDEBAND