SIMULATED ASSESSMENT OF INTERFERENCE EFFECTS IN DIRECT

SEQUENCE SPREAD SPECTRUM (DSSS) QPSK RECEIVER

THESIS

Luis S. Rojas, Captain, Chilean Air Force

AFIT-ENG-14-M-64

DEPARTMENT OF THE AIR FORCE

AIR UNIVERSITY

AIR FORCE INSTITUTE OF TECHNOLOGY

Wright-Patterson Air Force Base, Ohio

Distribution Statement A: Approved for Public Release; Distribution Unlimited

The views expressed in this thesis are those of the author and do not reflect the official
policy or position of the United States Air Force, the Department of Defense, or the
United States Government, Chilean Air Force, Chilean Ministry of Defense or Chilean
Government.

This material is declared a work of the U.S. Government and is not subject to copyright
protection in the United States.
AFIT-ENG-14-M-64

SIMULATED ASSESSMENT OF INTERFERENCE EFFECTS IN DIRECT

SEQUENCE SPREAD SPECTRUM (DSSS) QPSK RECEIVER

THESIS

Presented to the Faculty

Department of Electrical and Computer Engineering

Graduate School of Engineering and Management

Air Force Institute of Technology

Air University

Air Education and Training Command

in Partial Fulfillment of the Requirements for the

Degree of Master of Science in Electrical Engineering

Luis S. Rojas, E.E.

Captain, Chilean Air Force

March 2014

Distribution Statement A: Approved for Public Release; Distribution Unlimited

AFIT-ENG-14-M-64

SIMULATED ASSESSMENT OF INTERFERENCE EFFECTS IN DIRECT

SEQUENCE SPREAD SPECTRUM (DSSS) QPSK RECEIVER

Luis S. Rojas, E.E.

Captain, Chilean Air Force

Approved:

//signed// 11 March 2014

Dr. Richard K. Martin, PhD (Chairman) Date

//signed// 11 March 2014

Dr. Robert F. Mills, PhD (Member) Date

//signed// 11 March 2014

Dr. Michael A. Temple, PhD (Member) Date
AFIT-ENG-14-M-64
Abstract

This research developed and validated a generic simulation for a direct sequence
spread spectrum (DSSS), using differential phase shift keying (DPSK) and phase shift
keying (PSK) modulations, providing the flexibility for assessing intentional interference
effect using DSSS quadrature phase shift keying receiver (QPSK) with matched filtering
as a reference. The evaluation compares a comprehensive pool of jamming waveforms at

pass-band that include continuous wave (CW) interference, broad-band jamming, partial-
band interference and pulsed interference. The methodology for jamming assessment
included comparing the bit error rate (BER) versus required jamming to signal ratio (JSR)

for different interferers using the Monte Carlo approach. This thesis also analyzes the

effect of varying the jammer bandwidth for broad-band jammers including broad-band

noise (BBN), frequency hopping interference (FHI), comb-spectrum interference (CSI),

multi-tone jamming (MTJ), random frequency modulated interference (RFMI) and linear

frequency modulated interference (LFMI). Also, the effect of changing the duty cycle for

pulsed CW waveforms is compared with the worst case pulsed jamming equation. After

the evaluation of different interferers, the research concludes that pulsed binary phase shift

keying (BPSK) jamming is the most effective technique, whereas the CW tone jamming
and CW BPSK interference result are least effective. It is also concluded that by finding an
optimum bandwidth, FHI and BBN improves the required JSR by approximately 2.1 dB,
RFMI and LFMI interference by 0.9 and 1.5 dB respectively. Alternately, MTJ and CSI

improves their effectiveness in 4.1 dB and 3.6 dB respectively, matching the performance
of the pulsed BPSK jammer.

iv
Dedicated to my beloved wife and my funny little daughter.

v
 Acknowledgments

I would like to express my gratitude to my advisor Dr. Richard K. Martin for the
useful comments, remarks and engagement through the learning process here in AFIT.
Furthermore, I would like to thank the Chilean Air Force for giving me the opportunity
and for trusting in me to follow this master program. My sincere thanks also to the staff
of the IMSO for its support and effort for my integration and adaptation since I arrived to

AFIT. Also, I would like to thank my friend Nick Rutherford for his support and good time
shared during this master program. Finally, I would like to thanks to my classmates Fawwaz
Alsubaie, Matthew Crosser, Ethan Hennessey and Richard Rademacher, for interchanging

experiences, knowledge and time regardless of language gap.

Luis S. Rojas

vi
 Table of Contents

Page
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Assumptions and Resources . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

II. Direct Sequence Spread Spectrum Theory . . . . . . . . . . . . . . . . . . . . . 5

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Spread Spectrum Communications . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Direct Sequence Spread Spectrum System Model . . . . . . . . . . . . . . 6
2.4 Pseudonoise (PN) Sequences . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Interference in Direct Sequence Spread Spectrum Systems. . . . . . . . . . 13
2.6 Interference Techniques Classification . . . . . . . . . . . . . . . . . . . . 14
2.7 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

III. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Evaluation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Direct Sequence Spread Spectrum Model . . . . . . . . . . . . . . . . . . 24
3.3.1 Phase Shift Keying Modulation . . . . . . . . . . . . . . . . . . . 28
3.3.2 Differential Phase Shift Keying Modulation . . . . . . . . . . . . . 29

vii
 Page
3.4 Jamming Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4.1 Broad-band Interference . . . . . . . . . . . . . . . . . . . . . . . 33
3.4.1.1 Broad-band Noise . . . . . . . . . . . . . . . . . . . . . 33
3.4.1.2 Random Frequency Modulated Interference (RFMI) . . . 34
3.4.1.3 Linear Frequency Modulated Interference (LFMI) . . . . 37
3.4.1.4 Frequency Hopped Interference (FHI) . . . . . . . . . . 38
3.4.1.5 Comb-Spectrum Interference (CSI) . . . . . . . . . . . . 40
3.4.1.6 Multi-Tone Jamming (MTJ) . . . . . . . . . . . . . . . . 40
3.4.2 Narrow-Band Interference (NBI) . . . . . . . . . . . . . . . . . . . 41
3.4.2.1 Tone Jammer . . . . . . . . . . . . . . . . . . . . . . . . 42
3.4.2.2 Binay Phase Shift Keying (BPSK) Interference . . . . . . 43
3.4.2.3 Partial-Band Noise . . . . . . . . . . . . . . . . . . . . . 44
3.4.3 Pulse Jamming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5 Jamming Performance Evaluation . . . . . . . . . . . . . . . . . . . . . . 48
3.5.1 Receiver Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5.2 Receiver Performance with Bandpass Filters . . . . . . . . . . . . 49
3.5.3 Simulation to Determine the Jamming to Signal Ratio (JSR) . . . . 53

IV. Results and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.1 Simulation of Broad-band Jammers . . . . . . . . . . . . . . . . . . . . . 55

4.2 Simulation of Continuous Wave and Narrow-band Noise Jammers . . . . . 56
4.3 Simulation of Pulsed Jammers . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Jammer Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.4.1 Effects of Jamming under Bandwidth and Duty Cycle Variation . . 62
4.4.2 Narrow-band and Pulsed Jamming Comparison . . . . . . . . . . . 65
4.4.3 Optimum Bandwidth for Pulsed Broad-band Jammers . . . . . . . 67
4.5 Analysis for Optimum Jammers . . . . . . . . . . . . . . . . . . . . . . . 69

V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.1 Research Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.2 Results Found . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3 Recommendations for Future Research. . . . . . . . . . . . . . . . . . . . 76

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

viii
 List of Figures

Figure Page
2.1 DSSS BPSK modulator [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 DSSS BPSK demodulator [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Linear feedback shift register [5]. . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Basic DSSS with interference signal [5]. . . . . . . . . . . . . . . . . . . . . . 13

3.1 DSSS Transmitter Model [24]. . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.2 The square root raised cosine filter frequency response. . . . . . . . . . . . . . 26
3.3 DSSS Receiver Model [13], [25]. . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.4 Normalized frequency response of the RF filter (left) and despreading filter

(right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.5 Time domain response (left) for 1000, T s =1 ms, symbols and normalized PSD
(right) for BBN using a bandwidth W J =201.5 KHz. . . . . . . . . . . . . . . . 34

3.6 BBN spectrograms showing 100, T s =1 ms, symbols and W J =201.5 KHz (left)

and W J =62 KHz (right) bandwidths. . . . . . . . . . . . . . . . . . . . . . . . 35

3.7 The RFMI time domain response (left) and normalized PSD (right) for two,

T s =1 ms, symbols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.8 The RFMI spectrogram showing three, T s =1 ms, symbols. . . . . . . . . . . . 36

3.9 Time domain response (left) for one, T s =1 ms, symbol and normalized PSD

(right) for CW LFMI for three symbols. . . . . . . . . . . . . . . . . . . . . . 37

3.10 The LFMI spectrogram showing three, T s =1 ms, symbols. . . . . . . . . . . . 38

3.11 Time domain (left) and normalized PSD (right) for the FHI showing five, T s =1
ms, symbols and bandwidth W j =62 KHz. . . . . . . . . . . . . . . . . . . . . 39

3.12 The FHI spectrogram for 10, T s =1 ms, symbols and bandwidth W j =200 KHz. . 39

ix
Figure Page
3.13 The time domain CSI response (left) for one, T s =1 ms, symbol and normalized
PSD (right) response for five frequencies, 1000 symbols and ∆ f =0.5 Hz. . . . . 40

3.14 The CSI spectrogram for 50, T s =1 ms, symbols and bandwidth W j ≈ 62 KHz. . 41
3.15 Time domain MTJ response (left) for one, T s =1 ms, symbol interval and
normalized PSD for 10 random phase tones and 1000 symbols. . . . . . . . . . 42
3.16 The MTJ spectrogram for 100, T s = 1 ms, symbols and bandwidth W J ≈62 KHz. 42
3.17 Time domain response showing one, T s =1 ms, symbol phase transition (left)

and normalized PSD (right) for tone jammer for 1000, T s =1ms, symbols. . . . 43
3.18 The tone jammer spectrogram for 100, T s =1 ms, symbols. . . . . . . . . . . . 44
3.19 Time domain response for 1000, T s =1 ms, symbols (left) and normalized PSD

(right) for NBN using a bandwidth W J ≈8 KHz. . . . . . . . . . . . . . . . . . 45

3.20 NBN spectrograms showing 100, T s =1 ms, symbols for W J ≈8 KHz. . . . . . . 45

3.21 Time domain response for 1000, T s =1 ms, symbols and normalized PSD for
pulsed BPSK. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

3.22 Pulse BPSK spectrograms showing 5, T s =1 ms, symbols. . . . . . . . . . . . . 47

3.23 BER performance for QPSK in AWGN channel. . . . . . . . . . . . . . . . . . 50

3.24 BER performance for DQPSK and DBPSK in AWGN channel. . . . . . . . . . 51

3.25 Simulated BER performance for DSSS QPSK receiver in AWGN channel
illustrating the effect of band-pass filtering. . . . . . . . . . . . . . . . . . . . 52
3.26 Comparison of input S NRS im and S NROut of the band-pass RF filter (S NRRF )
and despreading filter (S NRDS ). . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1 Broad-band jammer BER performance versus JS RRF for Eb /No =7 dB. . . . . . 56
4.2 CW and NBN BER performance versus JS RRF for Eb /No =7 dB. . . . . . . . . 57
4.3 Pulsed jamming BER performance versus JS RRF for Eb /No =7 dB and ρ
varying according to Equation (4.3). . . . . . . . . . . . . . . . . . . . . . . . 59

x
Figure Page
4.4 BER versus for JS RRF comparing all jammer signals. . . . . . . . . . . . . . . 61

4.5 BER performance versus bandwidth variation for broad-band jammers for
JSR= 5 dB and Eb /No =7 dB, where W J ≈0 represents single tone jamming
frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.6 BER performance effect for duty cycle variation ρ ∈ [10−3 , 1] for pulsed jamming. 64
4.7 BER performance with optimum bandwidth for broad-band jammers for an
Eb /No =7 dB. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.8 Comparison of BER performance for pulsed BPSK and pulsed tone for
changing the duty cycle and Eb /No =7 dB. . . . . . . . . . . . . . . . . . . . . 68
4.9 BER performance comparison for optimal bandwidth pulsed jammers and non-

optimal bandwidth pulsed jammers. . . . . . . . . . . . . . . . . . . . . . . . 69

4.10 Pulsed BPSK, MTJ and CSI, the most effective jammers for Eb /No =7 dB. . . . 71
4.11 BER performance for the optimized jammers at JS RRF =3.1 dB. . . . . . . . . 72

xi
 List of Tables

Table Page
3.1 Signal parameters considered for the simulated receiver. . . . . . . . . . . . . . 49
3.2 PSK and DPSK BER performance comparison. . . . . . . . . . . . . . . . . . 50
4.1 JS RRF in dB required for broad-band jammers to degrade bit error rate (BER)
by one order-of-magnitude. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.2 JS RRF in dB required for CW, Narrow-band noise and pulsed jammers to
degrade BER by one order-of-magnitude. . . . . . . . . . . . . . . . . . . . . 61
4.3 The JS RRF in dB for broad-band jammers. . . . . . . . . . . . . . . . . . . . . 66

xii
 List of Acronyms

Acronym Definition
AWGN additive white Gaussian noise
BBN broad-band noise
BER bit error rate
BPSK binary phase shift keying

CDMA code division multiple access

CSI comb spectrum interference
CW continuous wave

DPSK differential phase shift keying

DBPSK differential binary phase shift keying

DQPSK differential quadrature phase shift keying

DRFM digital radio frequency memory

DSSS direct sequence spread spectrum

FCC Federal Communication Commission

FHI frequency hopped interference
GPS Global Positioning System
ISI inter-symbol interference

LFSR linear feedback shift register

LFMI linear frequency modulated interference

LPE low probability of exploitation

LPI low probability of intercept

JSR jamming to signal ratio

MTJ multi-tone jamming

NBN narrow-band noise

xiii
Acronym Definition
PSD power spectral density

PSK phase shift keying

PN pseudonoise
PBN partial-band noise
QPSK quadrature phase shift keying
RFMI random frequency modulated interference
RF radio frequency

SNR signal to noise ratio

SS spread spectrum

xiv
 SIMULATED ASSESSMENT OF INTERFERENCE EFFECTS IN DIRECT
SEQUENCE SPREAD SPECTRUM (DSSS) QPSK RECEIVER

I. Introduction

1.1 Background

The spread spectrum (SS) technology has been crucial for enabling the coexistence

of wireless devices in military and civil applications. According to [1] the origin of SS
communication was a natural result of the battle for electronic supremacy after the Second
World War.

The first public patent on SS was granted in 1942 and it came from Hedy Lamarr,
the Hollywood movie actress, and George Antheil, an avant-garde music composer. Hedy

Lamar got the idea from her previous husband who worked on wireless torpedo guidance
and its vulnerability to jamming could be avoided by sending messages over multiple

radio frequencies in a random pattern. This idea along with the music knowledge of

Antheil yielded a solution to provide synchronization based on the 88 piano frequencies,

consisting of two rolls perforated with the same pattern where every hole represent a

different frequency and a mechanical device to keep the stability in the rotation frequency
[2].
The concept of spreading information to avoid interference and increase range
resolution was a familiar concept for radar engineers at the end of the Second World
War. The SS concept was known and developed during the 1950s and helped by its
implementation by the development of information theory contributions made by Claude E.
Shannon who in 1947 published a paper revealing that a channel capacity can be maximized
by spreading the signal. Shannon showed that the channel capacity was increased by

1
sending a set of noise-like waveforms and distinguishing them at the receiver via minimum
distance criterion of the received signal and a stored waveform copy.
The correlation concept was first published in 1959 by a German scientist, F. H. Lange.

It was possible because of the effort of Shannon and Norbert Wiener with his work in
filter theory to reduce the noise presence in a signal by comparison with an estimated
noiseless signal. However, since the 1960s most of the SS development occurred for

military equipment. An important SS milestone was the publication of Spread Spectrum

Systems by Robert Dixon in 1976 as the first comprehensive book with unclassified and

commercial applications. During the 1980s another important milestone occurred with the
first authorization for civil use of SS in 1985 by U.S. Federal Communication Commission
(FCC) that marked a start point for the development of commercial spread spectrum devices

in use today [3].

Nowadays, SS techniques are used broadly from military and civilian prospective

with examples such as Wi-Fi, Bluetooth, and Zigbee. Special attention is focused on
wireless sensor networks, cellular telephony, wireless tactical military communications due

its mobility and flexibility, Global Positioning System (GPS), ranging system and data
link systems. One spread spectrum technique with low probability of intercept (LPI),
low probability of exploitation (LPE) and good response to unintended and intended
interference is direct sequence spread spectrum (DSSS) since the energy of the transmitting
signal is distributed across a bandwidth much wider than the message signal itself.
This research motivation consists of simulating a communication receiver that
provides flexibility to assess different interference techniques typically studied for DSSS
under the variations of jamming parameters. This evaluation gives a good approximation
and baseline to evaluate complex communication jamming scenarios at low cost.

2
1.2 Problem Statement

The interference in communication systems and assessing its effects can be decisive
for evaluating existing systems, predicting the quality of data transmission and achieving
reliable communication of digital information. A practical way to evaluate the robustness
of wireless communication systems is by doing on site testing but it can be costly in
regards to availability of the service and resources required. Currently there is a variety of
theoretical background on DSSS describing the effect of interference, however few papers
in the open literature address simulation of a comprehensive pool of interference techniques
or they study particular techniques in isolation. This research focused on developing a
DSSS receiver and evaluating its performance in the presence of continuous wave (CW)

interference, noise interference and pulse interference with flexibility to vary the jamming

signal strength, bandwidth and transmission duty cycle. This simulation can provide a

baseline to assess and discriminate the effectiveness of different interfering techniques in

a DSSS quadrature phase shift keying (QPSK) communication system at low cost. The
simulation is adaptable for adding more complexity and is a good tool for future research.

1.3 Assumptions and Resources

This research considers the interference evaluation of a DSSS receiver for a single

user neglecting the environment effects and the angle-of-arrival of the interfering signals.

This means that received signals are assumed to arrive in the antenna bore-sight where

the desired signal has a constant power simulating a cooperative transmitter with fixed
distance to the receiver. Also the interference assessment assumes a coherent receiver with
ideal carrier demodulation that neglects phase errors and mixer losses. The despreading
mixer considers a Gold sequence that is a perfectly synchronized with the received desired
signal while neglecting delays for multi-path using an ideal additive white Gaussian noise
(AWGN) channel. The phase shift keying (PSK) DSSS receiver utilized for assessing the

3
simulated jamming techniques was implemented in MATLAB® and the Communication
System Toolbox™, version 2013b.

1.4 Thesis Organization

The thesis has been organized in five chapters. Chapter II provides the basic concept
of DSSS and typical implementations, the characteristics of pseudonoise (PN) sequences

used in DSSS, the concept and classification of jamming techniques, the properties of
interference reduction in DSSS and previous work in DSSS jamming. Chapter III presents
the methodology and the receiver model implemented along with the description of the
different jamming models simulated. Chapter IV provides evaluation results for different

jamming techniques and the effects of varying parameters such bandwidth and duty cycle

to optimize the jamming response. Chapter V presents the conclusions, summarizes the

thesis and provides future areas of research.

4
 II. Direct Sequence Spread Spectrum Theory

2.1 Introduction

This chapter provides general concepts of spread spectrum theory, the DSSS receiver
model, the characteristic of PN sequences implemented in DSSS, a general description of
DSSS interference rejection capability along with typical jamming classification schemes.
Finally, some references and related DSSS interference work is provided.

2.2 Spread Spectrum Communications

In a communication system the modulated waveform occupies a bandwidth that is

dependent on the modulation order (bits per symbol) and the modulation technique. In a SS

system the transmission bandwidth is much higher than the minimum bandwidth required
to send information. The spreading bandwidth is accomplished by a spreading signal with

noise-like characteristics that are independent of the data intended to transmit. The signal

recovery or despreading is achieved by correlating the incoming signal with a synchronized

replica of the spreading signal used to spread the information.

Spread spectrum can be classified according to the following modulation formats:

• Direct sequence (DSSS): a form of phase-shift keying modulation.

• Frequency hopping (FHSS): a narrow-band frequency-shift keyed signal is hopped

over a wide band using pseudo-random carrier frequency selection.

• Time hopping (THSS): similar to frequency hopping but the PN sequence selects a
transmission time (slot) within consecutive time frames (a low duty cycle or burst).

• Hybrid: that combines any of the three main types.

This research focused on DSSS modulation implemented as two stages of modulation

type: 1) the incoming data sequence is used to modulate a wide-band code, transforming

5
the narrow-band data sequences into a noise-like signal, 2) a second modulation using a
selected phase shift keying technique.

2.3 Direct Sequence Spread Spectrum System Model

Direct sequence spread spectrum modulation can be defined as a means of transmis-

sion such as the data sequence is spread by using a code that is independent of the data
sequence. The concept of spreading implies a bandwidth expansion far beyond that it is re-
quired to transmit the digital data. For instance a requirement to transmit information using
a data Rd = 200 Kbits/s occupying a spectrum bandwidth W ss = 200 MHz has a spreading
factor of 103 as the ratio between the spreading bandwidth and the signal bandwidth.

The unique characteristic and purpose of DSSS modulation is that provides interfer-

ence suppression, energy density reduction, ranging or time delay measurement [4]. The

interference suppression can be a combination of the presence of users with the intention
of disrupt the communication (jammers or interferer) or users that independently share

a common channel without an external synchronization (multiple access communication).

Also multi-path is considered as self interference that is mitigated by spread spectrum tech-

niques, where delayed versions of the signal arrive to the receiver using alternate paths.

The energy density reduction of spread spectrum provides low probability of

interception and low probability of exploitation. Those characteristics are important to

design a communication system that meets regulations of signal strength, to minimize

detectability and to obtain privacy.

With respect to range delay measurement spread spectrum provides low error in
successive pulse time delay measurement, since the error is inversely proportional to the
spread spectrum signal bandwidth.
A simplified baseline model of DSSS system is depicted in Figure 2.1 [5]. The
information signal x(t) represents an antipodal pulse stream with values ±1, with a given
√
data rate which is modulated by multiplying it with carrier signal P cos ωo t. The resultant

6
 Figure 2.1: DSSS BPSK modulator [5] .

√
product is a binary phase shift keying (BPSK) signal S x (t) = x(t) P cos ωo t. The BPSK
signal is multiplied by an spreading sequence c(t) with a much higher data rate called

chip rate. The effect is a bandwidth expansion given by the convolution of S x (t) and c(t)

in frequency domain. Thus if the signal S x (t) is narrow-band, then the resulting product
S x (t)c(t) is effectively spread a bandwidth approximately equal to the spreading signal.

At the receiver, as is shown in Figure 2.2 [5], the original signal is recovered ideally

by a synchronized replica of the spreading signal. The parameter T̂ d is a delay estimate

of the propagation time from the transmitter to receiver. The signal r(t) is considered

without interference with constant system gain A and a random phase ϕ in the range from
(0, 2π). For spread signal c(t) = ±1, then the product c(t − T d )c(t − T̂ d ) = 1 for optimum
synchronization with T d = T̂ d . For a synchronized signal, the correlator output is the
despread modulated signal (considering a random phase and delay T d .) Subsequently the
signal is filtered in order to remove high frequency components and finally demodulated
using a conventional demodulator. Any unwanted signal will be spread by the same
bandwidth. The advantage in terms of interference rejection is given by the fact that the

7
 Figure 2.2: DSSS BPSK demodulator [5].

incoming signal is multiplied just one time in the receiver whereas the transmitted signal is

multiplied two times in order to recover a good estimate of the original signal x(t).

2.4 Pseudonoise (PN) Sequences

There are two main mechanisms for spreading the signal: transmitted reference (TR)

and stored reference (SR).

The first considers two channels, one for transmitting the data and other for the

spreading waveform which is randomly generated. The main advantage of TR is that it

achieves synchronization easily. However this method has some disadvantages such as the

code is available for any unintended users, the performance degrades at lower signal to
noise ratio (SNR) and it requires a greater bandwidth and power to transmit.
The SR method requires a single channel to generate a pseudo-random spreading
signal which is generated independently by transmitter and receiver. The disadvantage
of this technique is that synchronization is more complex to achieve. However depending
on the code it cannot easily be predicted or exploited by an unintended receiver.

8
 A PN sequence is defined as a deterministic periodic sequence because it is known by
the transmitter and receiver. PN sequence has the main property that statistically is similar
to white noise. The main properties of PN sequences are the following:

• Balance property. The number of binary ones differs from the number of binary zeros
by at most one digit.

• Run property. A run is defined as a sequence of a single type of binary digits. Among
the run of zeros and ones it is desirable that one half of the runs of each type are of
length 1, about one fourth of the length 2, one eight of length 3, and so on.

• Correlation property. If a period of the sequence is compared term by term with

a cyclic shift of itself, the number of agreements differs from the number of
disagreements by no more than one count.

The normalized autocorrelation for a PN coded waveform x(t) with period T can be
defined as:
( )∫ T
1 1 2
R x (τ) = x(t)x(t + τ), dt for −∞ < τ < ∞, (2.1)
K T − T2
∫ T2
1
K= x(t)2 , dt where K is the energy of the signal. (2.2)
T − T2

For a PN waveform of unit chip duration and period p chips, the normalized

autocorrelation function can be expressed as

( )
1  agreements − disagreements 
R x (τ) =   . (2.3)
p length o f sequence
Typically a PN sequence can be generated using a linear feedback shift register (LFSR)
whose output is defined by the number of register stages. A maximal length sequence has
a period given by p = 2n−1 , where the each sequence is repeated every p clock pulses.
Figure 2.3 shows a linear shift register [5] example of four stages X1, . . . , X4. A

sequence is controlled by a clock pulses (not shown). At each clock pulse the content of

9
 Figure 2.3: Linear feedback shift register [5].

the register is shifted by one stage to the right. Also in each clock the stage X3 is modulo
2 added to stage X4 and fed back to the stage X1.

There are two classes of PN sequences; aperiodic and periodic. An aperiodic sequence

does not repeat itself in a periodic way whereas the latter is a sequence that repeats itself

exactly with a specific period.

An aperiodic sequence can be described analytically by a sequence of N plus or n

minus ones as follows:

a1 , a2 , . . . , an , ai = ±1. (2.4)

The aurocorrelation of a PN sequence is defined by:

∑
N−k
C(k) = an an+k k = 0, 1, . . . , N − 1. (2.5)
n=1

As an example of autocorrelation in aperiodic sequence, consider a sequence of four

digits of plus or minus ones from a1 through a4 . The autocorrelation for C(1) is obtained
by shifting the sequence by one symbol:

a1 a2 a3 a4
a1 a2 a3 a4

C(1) = a1 a2 + a2 a3 + a3 a4

10
 An ideal aperiodic sequence would have an autocorrelation function given by:






 k=0
N,
C(k) = 
 (2.6)




0 or ± 1, k , 0

Such sequences are called Barker sequences and only exist for a few values of N.
Specifically they have been found for N=1, 2,3,4,5,7,11 and 13. This kind of sequences is
too short as a spreading function and normally is used for synchronization purposes [6].
In a spread spectrum communication system it is important to have sequences where
the autocorrelation function is large at zero lag because the synchronization can be
accomplished. On the other hand, at non-zero lags it is desirable that the autocorrelation
be low in order to avoid false synchronization. Besides, the cross-correlation between the

sequences used by two communication systems should be low even at zero lag in order to

avoid false correlation between two systems.

A periodic sequence consists of an infinite sequence of plus or minus ones divided

into blocks of length N, where each particular block is the same. A periodic sequence can

be represented as follows [6]:

. . . , aN−1 , aN , a1 , a2 , . . . , aN , a1 , . . .

In every period the number of plus ones differs from the number of minus ones by

exactly one. Hence N is odd number. Thus

N+ + N− = N, (2.7)

|N+ − N− | = 1.

In every period half of the runs of the same sign have length 1, one fourth have length
2, one eight have the length 3, and so forth. Also the number of positive runs equals the
number of negative runs. The autocorrelation of a periodic sequence is two valued. That

11
is, it can be described by:




∑
N 

 k = 0, N, 2N, ...
N,
C(k) = an an+k =
 (2.8)



n=1 
−1, otherwise

where

an+N = an . (2.9)

In this research the simulation will use a periodic code broadly implemented in code
division multiple access (CDMA) and GPS systems which is a Gold code. This code is

generated by a modulo-2 operation between two different preferred m-sequences. The

preferred m-sequence operation consists of choosing a reference m-sequence with a shifted

version or vice-versa. Using two sequences with equal length N, the resultant Gold

sequence is N length as well. For a period N = 2n − 1, there are N possible circular

shift. Then it is possible to obtain N sequences from two preferred m-sequences. The Gold

sequence generated is not an m-sequence with two correlation values instead it has three

low correlation values. The autocorrelation r xx and cross-correlation r xy function for this

Gold sequence can be represented by [7]:







 for τ = 0
1,
r xx (τ) = 
 { −t(n) } (2.10)




 N , −1 , t(n)−2 for τ , 0
N N

and
{ }
−t(n) −1 t(n) − 2
r xy (τ) ∈ , , . (2.11)
N N N

where






n+1

1 + 2
 for n odd
2

t(n) = 
 (2.12)



 n+2
1 + 2 2 for n even

12
 Figure 2.4: Basic DSSS with interference signal [5].

The Gold peak correlation value is t(n)/N and from the above equation less correlation
values occur when n is odd. Where n is the stage number or polynomial degree of the two

preferred m-sequences.

2.5 Interference in Direct Sequence Spread Spectrum Systems.

The main advantage of spread-spectrum waveforms is their ability to reject interfer-

ence. The source of interference can be produced by authorized users transmitting simulta-

neously, or also by a hostile transmitter with the intention of jamming a determined channel

[8]. In Figure 2.4 is represented a basic DSSS to illustrate the interference rejection capa-
bility: the energy in signal x(t) is spread across a bandwidth given by the multiplication

with a PN sequence c(t) and then in the receiver it is multiplied by c(t) again. On the other
hand any non-spread interference signal will be multiplied just one time by the spreading
replica c(t) when received.

Jamming a communication system implies an intentional and deliberate transmission

or retransmission of amplitude, frequency, phase, or otherwise modulated pulsed, CW, or

noise-like signals for the purpose of interfering with a receiver [9].

Spread spectrum techniques consider many orthogonal signal coordinates coexisting
in a link where only a portion of that signal is present at a given time. The total space of

13
a signal of bandwidth W and duration T is given by 2WT [10]. The intended interference
presents a finite power and an ambiguity about the signal coordinates and parameters. Then
there are two main possibilities. To jam all space of possible signals present with the power

distributed across the bandwidth or to interfere some signal coordinates with different levels
of power. The processing gain is an important parameter to measure how the signal spread
is resilience to interference. The general expression is given by:

W ss Rc
Gp = = , (2.13)
Wmin Rd
where W ss is the spreading signal bandwidth for a chip rate Rc and Wmin is the minimum

required bandwidth of the signal to transmit at data rate Rd .

2.6 Interference Techniques Classification

The main waveforms for generating interference can be divided into the following

categories:

• Noise jamming. This waveform consists of injecting interference signal to the

receiver with the goal of covering the desired signal by a band limited white Gaussian

noise of high power. The main advantage of noise jamming is that it does not require
more detailed information about the communication beyond its spread bandwidth

[8]. In this technique the jamming carrier signal is modulated with random noise.
Depending on the bandwidth available noise jamming technique includes [11]:

– Broad-band noise. The intended interference energy is distributed through the

entire receiver bandwidth. The effect of this technique is to reduce the channel
capacity by affecting the SNR at the receiver.

– Narrow-band noise. The interference energy is distributed over a single channel

bandwidth or fraction of that channel bandwidth.

14
 – Partial-band noise. The interference energy is transmitted across multiple
channels not necessarily contiguous.

Noise jamming also is related to the communication system channel capacity

(C), that is the maximum rate at which the information can be transmitted. It
considers a band-limited channel with white noise and with signal power constraints
represented by:

( )
S
C = B log2 1 + (2.14)
No B
where:

C =Channel capacity in bits/seconds.

B =Channel bandwidth in Hz. (2.15)

S =Average received signal power in the channel bandwidth in Watts.

No B =Average noise power in Watts, noise spectral density and bandwidth product.
S
=Signal to Noise Ratio.
No B

In terms of interference if the average noise power increases by adding intended

interference, the channel capacity is affected by SNR reduction in the presence of

noise.

• Tone jamming. In this technique one or more carrier frequencies (tones) are
transmitted wisely in order to interfere one or more channel simultaneously.

Depending on the tones transmitted the technique is called single tone or multi-
tone jamming (MTJ). Single tone jamming consists of transmitting an unmodulated
carrier with an average power J p within the spreading bandwidth. In general
tone jamming can be effective in DSSS systems if jammer power overcomes the

processing gain and if its frequency is centered in the spreading bandwidth. The
phase difference between the jammer and target signal has an effect in the power

15
 processed at the receiver, then at more phase difference more power is required to
overcome the processing gain. In frequency hopping multi-tone jamming (MTJ)
jamming is applicable, however it requires high synchronization and coherency in

phase between the jammer signal and target signal, since the energy is distributed in
multiple frequencies.

• Pulse jamming. Pulse jamming is similar to the concept of partial-band noise. The
transmission is performed over multiple channels by exploiting the concept of duty
cycle with the intent of affecting the target signal a fraction of the time the jammer is
on. The interferer transmits a pulsed bandlimited white Gaussian noise signal with a
power spectral density (PSD) that covers the spread spectrum system bandwidth. The
interference duty factor can be denoted by ρc representing the ratio of time when the

jammer is on relative to the total interval (on and off time). The average interference

power J p with a bandwidth B can be expressed as:

J p = BJo ρc , (2.16)

where Jo is the constant jamming PSD in Watts/Hz.

• Repeater Jammers. This jamming technique consists of a transceiver that senses and

estimates the spread spectrum signal parameters and then amplifies and retransmits

the signal with high power. This jamming technique tries to deal with the main

strength of spread spectrum which is the generation of a high processing gain on the
receiver such that an interferer with no spreading sequence knowledge or spreading

sequence estimation requires a high level of power, depending on the spreading

sequence length, to induce errors and affect the receiver performance.
From the jammer signal perspective, noise jamming techniques require more
power to overcome DSSS the processing gain. If the jammer is able to sense the

incoming DSSS signal and replicate it while keeping certain correlation properties,
then it is expected to require less power for a given effectiveness.

16
 Under the repeater jammer are jamming techniques that try to disrupt portions
of the digital signal required to deny communication. The goal is attack the receiver
during the acquisition time of new signals or users. For DSSS signals, this interval

consists of detecting the magnitude of a tolerance margin out of phase of the signal.
This is a decision circuit that accepts synchronization of a received signal after
detecting a certain energy level after cross-correlating to despread the signal. The
time of tolerance for synchronization is on the order of ±T c .

2.7 Related Work

The main sources of open literature that treat interference or jamming in DSSS,

presents analytical expressions for noise interference considered as AWGN that increases
the receiver noise floor and the receiver performance. For a particular kind of modulation

the jamming symbol error probability is derived from their respective symbol error rate or

bit error rate expressions. In general this type of interference is described as a Gaussian

process and represents the baseline jammer performance. The effectiveness is primarily
a function of theDSSS processing gain. However, particular implementations can have

different results and 1) there is no particular jamming technique that affects all spread

spectrum systems equally and 2) there is no a single spread spectrum that performs best

again all jamming waveforms [12]. The analytical results for partial-band noise, single-
tone and pulse jamming can be found in [13], [7], [11] and [12]. The theoretical and
mathematical analysis for uncoded and coded BPSK DSSS are covered in [13] and [12],

including block and convolutional coding.

Other approaches in interference analysis are described as a denial of service in
wireless computer sensor network and as radio frequency (RF) layer interference analysis.
Authors [14] explore the concept of distributing k interferer nodes to put N nodes out of
service. In sensor networks some strategies consist of identifying the jamming area and
mapping the network traffic using alternate routes. Managing the power and prioritizing

17
the traffic are other strategies mentioned to cope with jamming. Authors [15] generalize
jamming classes in sensor networks as active jamming and intermittent jamming. The
first is based on keeping the channel busy most of the time with a goal to saturate or

disrupt communication. The latter considers a trade-off between energy efficiency and
interferer effectiveness but requires more knowledge of the network protocols. A general
classification for RF intended interference considers the following categories: broad-

band noise, partial-band noise, continuous wave jammers, pulse jammers and multi-tone
jammers [12], [16].

Other work that describes physical RF interference is [17] that covers the implemen-
tation of a real-time reactive jamming (sense and then interfere) on software-defined radio
and evaluation of their performance at physical layer on simulated IEEE 802.15.4 in terms

of packet reception ratio. The three techniques analyzed include noise jamming (always is

present in wireless communication then is the primary source of RF jamming to consider),

single tone jamming and modulated jamming. The simulation results in that single-tone
(continuous jamming) is the most effective technique in terms of the effectiveness and re-

quired jamming gain. The modulated jamming consists of generating the same modulation

of the target signal with the idea of breaking synchronization by imitating the preamble

and header of transmissions, however to produce similar effects to single-tone jamming

significant more jamming power is required. Noise jamming technique also requires more
interference power but significantly less than that required for modulated jamming.

The linear frequency modulated interference (LFMI) interference and comb spectrum
interference (CSI) are mentioned in [18], [19] as a critical interference source to
DSSS, describing mechanism to suppress this interference based on time-frequency
representation. Other work that considers simulation of RF jamming techniques was found
in [20] where broad-band noise, partial-band noise, multi-tone and frequency, follower

jamming are considered on a network users the 802.11p protocol in AWGN and vehicular

18
channels. This study showed that in orthogonal frequency-division multiplexing (OFDM)
signaling under an AWGN channel, partial-band noise and multi-tone jamming (knowing
the pilot frequency locations) have more significant effects. Under vehicular channel the

study showed that partial and broad-band noise have more significant effects in terms of
frame error rate for a given jamming to signal ratio.
Repeater jamming applied to DSSS systems previous works is addressed in [21].

This paper discusses uncorrelated jamming techniques and their respective probability
of bit error performance, then it simulates a repeater jamming based on digital radio

frequency memory (DRFM) technology to generate correlative jamming technique. DSSS

techniques spread the energy of the baseband signal over a wide bandwidth, then they make
it difficult to sense the electromagnetic spectrum and then provide an intended interference

or jamming. The author shows that non-correlative jamming techniques such as narrow-

band noise (NBN), partial-band noise (PBN) and tone jamming applied to DSSS require

high power levels to overcome the integration gain due to PN characteristic of coding
process. Consequently, the energy which is not synchronized with the PN is spread

in the decorrelation process at the receiver. Besides, DSSS receiver could implement
adaptive interference mitigation, notch filters and prediction filter. This paper discuses

the interference on DSSS acquisition code and obtains the probabilities of bit error for non-
fading and Rayleigh fading environment. This research concludes after comparing noise
jamming techniques and a correlative jamming technique that the latter is more effective.
Other research related to repeater jamming is described in [22]. This work includes a
design and simulation of a jammer technique generated by using a compressive receiver
model and adaptive signal extraction. The main task of the jammer is to capture the
DSSS transmitted signal that has been corrupted with AWGN. The author proposes a
model of compressive receiver that performs a continuous and fast scan over the DSSS

frequency band. The output of the compressive filter has the energy at certain times that

19
correspond to the frequency of the input signal. The signal received is passed through the
autocorrelator and then the Levinson-Durbin algorithm is applied. The filter response to a
pulse train is dynamically changing (which is an estimate of the shape being transmitted by

the DSSS transmitter) and is to be transmitted by the jammer. Then it compares the modem
performance without jamming and the modem performance in the presence of jamming.

20
 III. Methodology

This chapter describes the approach and methodology for evaluating different
interference techniques in generic DSSS PSK and differential phase shift keying (DPSK)
receivers. The chapter includes a description of the evaluation parameters, the PSK/DPSK
transmitter and receiver developed and jammer models used for the assessment of different
techniques applicable to DSSS.

3.1 Approach

This thesis evaluates the performance of a DSSS QPSK receiver using Gold sequence
in terms of BER with different jamming waveforms present. First the research develops

and validates synchronized DSSS PSK/DPSK receivers in an AWGN channel. Then

this work simulates different jamming waveforms in a DSSS QPSK receiver determining

the jamming to signal ratio (JSR) to achieve a frame of reference or jamming margin.
Consequently, comparison and analysis are performed including the parameters variations

yielding the most effective jamming.

3.2 Evaluation Techniques

The present research methodology evaluates the receiver BER under interference

conditions using Monte Carlo performance evaluation. This method is a numeric

computation of the BER as the ratio of the number of bits transmitted with error over
the total transmitted bits. The errors are produced by the AWGN introduced in the channel
which correspond to generating N independent Gaussian random variables with zero mean
and variance σ2 added to an observation vector (information symbols). The errors are
counted at every symbol interval by comparing the received symbols and sent symbols and
consequently symbols are mapped to bits to compute the BER. This process is repeated K

times until a required number of errors are obtained. For interference evaluation the same

21
methodology is applied in order to estimate the BER when the intended interference signal
is added to the channel for a given interference power level.
The probability evaluation consists of:

1. Count the number of times that estimated received bits are different from the

transmitted bits. This is the condition for estimating the BER.

2. Estimate the probability as the ratio of the number of times that the condition is
satisfied over the number of trials or number of bits required.

The Monte Carlo simulation provides an estimated probability and the number of
realizations affect the result [23]. From the simulation perspective it is important to define
a tolerance margin or error. The absolute error accounts for the difference between the true

probability or ideal BER (P) and the estimated or simulated BER (P̂).

 
P − P̂
ϵ= . (3.1)
P
Once a desired absolute error is chosen a confidence interval is required to determine

a 100(1 − α)% of the time the error will be present. Then the number of iterations K should
satisfy:

[ ]2
Q−1 (α/2) (1 − P)
K≥ . (3.2)
ϵ2P
To evaluate system performance, simulated results are compared with analytical
expression of receiver performance in terms of BER and the information is presented in
plots of BER vs JSR and Eb /No .
One of the most important parameters, based on how efficiently in terms of energy a
system transmits the information, is the energy per bit to noise power spectral density ratio
Eb /No because it accounts for the noise in the channel. The Eb /No and energy per symbol
to noise power spectral density ratio E s /No relation is given by

22
 Es Eb Eb
= log2 M = k , (3.3)
No No No
where log2 M represent the number of k bits per symbol given modulation index M =
2k . The relation between SNR and E s /No is given by the noise bandwidth and signal
bandwidth utilization and the importance of this metric is that it allows evaluation of system
performance at specific points in the receiver. The mathematical expression for E s /No :

Es S S
= · Ts = · F s · T s, (3.4)
N0 N/Bn 2N

where:

S is the average signal power in Watts.

N is the average noise power in Watts.

Bn is the noise bandwidth equivalent to F s /2 for positive frequencies in Hz.

Fs is the sampling frequency in Hz.

Ts is the symbol duration in seconds.

The parameter used to evaluate the interference effects on the receiver is the JSR. It
is also called the jamming margin of the spread spectrum system, which is the largest

JSR considered to satisfy specific BER performance. In the present thesis, this scalar
is computed with the goal of determining the JSR required to affect the system BER

performance one order-of-magnitude. Therefore the receiver performance at a specific

SNR point is considered and then a range of JSR is simulated to determine the JSR value
where the BER is degraded by one order-of-magnitude relative to no interference being
present. A mathematical expression for JSR and its relation with E s /Jo can be obtained
from:

23
 S
Es = S · Ts = , (3.5)
Rs
where R s is the symbol rate in symbols/s. The jamming power spectral density (Watts/Hz)
across spreading bandwidth W ss can be expressed as:

J
Jo = . (3.6)
W ss

From Equation (3.5) and Equation (3.6)

Es S /R s W ss /R s Gp
= = = , (3.7)
Jo J/W ss J/S J/S

and

W ss Rc T s
Gp = = = = Nc . (3.8)
Wmin R s T c

G p is the bandwidth expansion factor, or processing gain, of the DSSS receiver and is
equivalent to the number of chips Nc per symbol duration due to T s = Nc · T c .

3.3 Direct Sequence Spread Spectrum Model

The transmitter model for simulation is presented in Figure 3.1 [24], where the digital

baseband pulse modulation and pulse shaping filters, were adopted from Communication

System Toolbox™, MATLAB® functions, version 2013b. A message source generates a

stream of random bits that are grouped according to a modulation index M as k bits per
symbol where k = log2 (M). According to the phase modulator, the symbol representation

has M constellation points. Next, the baseband symbol representation is up-sampled and
passed through a square root raised cosine filter to reduce the inter-symbol interference
(ISI) and adapt the signal to the communication channel. The cascade connection of the
up-sampler and the low pass filter is called the interpolator. In this implementation, the
up-sampling factor is 403 samples per symbol and the low pass filter (pulse-shaping filer)
has gain unity and a filter order in symbols corresponding to 8 symbols. Other design

24
 Figure 3.1: DSSS Transmitter Model [24].

parameters include the roll-off factor as a measure of bandwidth occupied over the Nyquist
bandwidth 1/2T s where T s is the symbol duration. Figure 3.2 shows the low-pass root raise

cosine filter response implemented using a roll of factor of β=0.2. For a symbol duration

T s =1 ms it implies an excess bandwidth of ∆ f =100 Hz given by β/(2T s ).

Then the baseband signal is modulated by taking the real part of the product of

complex carrier signal and complex baseband symbols. Finally the passband signal is

spread by using an antipodal Gold coded waveform c(t) of 31 chips. The transmitted spread

signal is S (t) .

On the receiver side, as is shown in Figure 3.3 [13], [25], the received signal is
represented by:
r(t) = s(t) + n(t) + J(t), (3.9)

where n(t) is a random process with zero mean and variance σ2 to represent the AWGN
channel. J(t) is the interfering signal. The first step is to filter the received signal r(t) in
order to eliminate unwanted components out of the spreading bandwidth. The RF filter
designed in the first step is a band-pass Butterworth filter of order 16, defined for WRF =62
KHz that corresponds to the spreading bandwidth. However, to minimize phase distortion,
the received signal is filtered using a zero-phase filter that doubles the Butterworth filter
order. The same process is used for the despreading filter but for a bandwidth WDS =2 KHz.

25
 0

−10

Magnitude (dB)
−20

−30

−40

−50

−1 −0.5 0 0.5 1
Frequency (kHz)

Figure 3.2: The square root raised cosine filter frequency response.

Figure 3.4 presents the frequency response for the RF and despreading filters. The bandpass

WRF and WDS filtering implementation ares used in this thesis not only to allow improving

the SNR by attenuating noise and unwanted signal components, but also to analyze the
bandpass DSSS interference reduction on the receiver prior and after despreading the

received signal [13].

As a second step, the filtered signal is despread by using a known sequence c(t)

and filtered according to the signal bandwidth. As a third step the despread filtered
signal is down-converted to a baseband and passed through the received matched filter
[24], consequently the resultant signal is down-sampled to obtain the received baseband
symbols representation. The received matched filter presents the same design specification
as the transmit pulse shaping filter, however the cascade of the down-sampler and low-
pass filtering the signal is called decimation. In this case the decimation is performed by

26
 Figure 3.3: DSSS Receiver Model [13], [25].

0 0

−5 −5
|H(f)| (dB)

|H(f)| (dB)
2

−10 −10

−15 −15

−20 −20
50 70 90 110 130 150 98 99 100 101 102
Frequency (KHz) Frequency (KHz)

Figure 3.4: Normalized frequency response of the RF filter (left) and despreading filter
(right).

a factor of 403 samples per symbol to recover the original symbol transmitted. Finally the
bit estimation b̂ process is performed by mapping from symbols to bits and then comparing
the estimated bits with transmitted bits in order to compute BER.

27
 3.3.1 Phase Shift Keying Modulation.

For M-ary digital phase modulation the modulated signal can be represented by the
product of a pulse shape g(t) and the carrier signal:

[ ]
S m (t) = Re g(t)e( j2πθm ) e( j2π fc t) , 1 6 m 6 M, 06t6T
[ ]
= g(t) cos 2π fc t + θm

, = g(t) cos(θm ) cos(2π fc t) − g(t) sin(θm ) sin(2π fc t). (3.10)

where

2π(m − 1)
θm = ; m ∈ 1, 2, . . . , M (3.11)
M

The vector representation in terms of orthogonal signal basis is given by [26]:

[√ √ ]
Sm = E s cos(θm ) E s sin(θm ) (3.12)

where E s is the symbol energy. Since the signal waveforms have equal energy, the optimum
detector for AWGN channel is given by the correlation of the received signal vector r and

the vector representation of reference signals. This operation represents the projection of r

vector in the direction of reference signals Sm .

C (r, Sm ) = r · Sm ; m = 1, 2, . . . , M. (3.13)

For the binary case M=2, from Equation (3.12), s1 (t) and s2 (t) are antipodal signals
with equal energy and the bit error probability expression is given by [26]:
 √( )
 
b 
Pb = Q 
2E  .
No 
(3.14)

For M = 4, from the receiver’s perspective the effect is like having two binary phase-
modulation signals in quadrature and it implies that there is no interference to each other.
Thus the Equation (3.14) is also applicable for QPSK.

28
 For M > 4 the analytical expression for the symbol error probability using Gray code
assignment is given by the following expression [26]:

√ ( π )
 2E s 
P s ≈ 2Q  sin , (3.15)
No M
√ ( π )
 2kEb 
≈ 2Q  sin .
No M
As implementing in this research the Gray code assignment produces a constellation
scheme where from consecutive symbol representation the distance between each other is
one bit. Under this code assignment the relation between bit error probability (Pb ) and
symbol error probability (P s ) with k bits per symbol yields:
1
Pb ≈ Ps. (3.16)
k
3.3.2 Differential Phase Shift Keying Modulation.

For DPSK modulation the received phase symbol at a given symbol interval is

compared to the phase of the received symbol at previous signaling interval. The

information transmitted is conveyed in θk where every symbol m to transmit defines:

2π(m − 1)
∆θk = ; m ∈ 1, 2, . . . , M. (3.17)
M
The modulator differentially phase encodes the transmitted symbols from a set of M

symbols. Consequently for transmitting ∆θk at kth transmission interval the transmitter
computes θk = θk−1 + ∆θk modulo 2π and then modulate θk on the carrier [27]. For the first

symbol it can be assumed θk−1 = 0.

To demodulate a differentially encoded phase signal the received signal is projected

onto basis functions cos(2π fc t) and sin(2π fc t) over the interval T s . At the kth signaling
interval, the demodulator output is [26]:
[√ √ ]
rk = E s cos(θk − ϕ) + nk1 E s sin(θk − ϕ) + nk2 , (3.18)
√
rk = E s exp( jθk − ϕ) + nk . (3.19)

29
where θk is the phase angle of the transmitted signal at the kth signaling interval, ϕ is the
carrier phase and nk = nk1 + nk2 is the noise vector. Similarly, the received signal vector at
previous interval yields:

√
rk−1 = E s exp( jθk−1 − ϕ) + nk−1 . (3.20)

The projection of rk onto rk−1 for the complex received signal representation yields:

∗
√ √
rk rk−1 = E s exp( jθk − θk−1 ) + E s exp( jθk − ϕ)n∗k−1 + E s exp( jθk−1 − ϕ)nk + nk n∗k−1 . (3.21)

The previous expression in absence of noise can be considered as the phase difference
∗
θk − θk−1 . Therefore, the mean value of rk rk−1 is independent of the carrier phase. Assuming
that phase difference θk −θk−1 is zero, the exponential factors exp( jθk−1 −ϕ) and exp( jθk −ϕ)
can be absorbed into the Gaussian noise component without changing their statistical
∗
properties and rk rk−1 can be expressed as [26]:

∗
√
rk rk−1 = Es + E s (nk + n∗k ) + nk n∗k−1 . (3.22)

√
For high SNR the term nk n∗k−1 is small than E s (nk + n∗k ) and it can be neglected.
√
Normalizing Equation (3.22) by E s , the decisions components are [26]:

√
x= E s + Re(nk + n∗k−1 ). (3.23)

y = Im(nk + n∗k−1 ). (3.24)

The variables x and y are uncorrelated Gaussian random variables with identical variances
σ2n = No . The received phase is:
(y)
θr = arctan . (3.25)
x
The phase decision can be made by comparing the correct received phase with previous
phase θr−1 . For an AWGN channel the probability of bit error for binary DPSK is given by:

1 Eb
Pb = e− No . (3.26)
2

30
The Pb for binary DPSK comparatively yields poorer performance than binary PSK, with
approximately less than 3 dB for a higher SNR required for a given BER [26].
When M > 2 the DPSK symbol error probability with k bits per symbol in an AWGN

channel and for a large E s /No tends to [5]:

√ ( )
 2E s π 
P s ≈ 2Q  sin √  , (3.27)
No 2M
√ ( )
 2kEb π 
≈ 2Q  sin √  .
No 2M

According to [28] the bit error probability Pb for M > 2 with k bits per symbol can be
approximated by:

 M/2 
1 ∑ 
Pb =  (wi )Ai  (3.28)
k i=1
where wi = wi + w M−i , w M/2 = w M/2 , wi is the Hamming distance of bits assigned to
symbol i and:

[ ] [ ]
(2i + 1)π (2i − 1)π
Ai = F −F , (3.29)
M M

∫ −π/2 [ ]
sin ψ exp −kEb /No (1 − cos ψ cos t)
F(ψ) = − dt. (3.30)
4π π/2 1 − cos ψ cos t
Particularly for M=4, using Gray code assignment, the BER from Equation (3.28) and
Equation (3.29) can be obtained evaluating the integral presented in Equation (3.30):

][ [π]
5π
Pb = F −F . (3.31)
4 4

3.4 Jamming Models

This research considers the following jamming categories: simulated noise interfer-
ence, continuous wave jammers, pulse jammers and multi-tone jammers. The main as-

31
sumption for DSSS is that is essentially not frequency agile system, therefore the most
applicable strategies are noise, tone and pulsed interference schemes. The latter interfer-
ence approach searches for generating spectral components within the RF filter that can

approximately cover the RF bandwidth or equivalently transmit high power intermittent

signals over portion of the RF bandwidth. The receiver model corrupted with noise and
interference can be represented by:

r(t) = S (t) + J(t) + n(t), (3.32)

where J(t) denotes the interference signal and n(t) is a zero mean AWGN.
Noise interference can be modeled as a broad-band interferer that tries to cover the
entire channel bandwidth or as a narrow-band (partial band) jamming by filtering an
assumed AWGN signal for the required bandwidth and controlling the average power to

achieve a required jamming margin. The resultant signal after the filtering process is a

colored version of Gaussian noise due to spectral changes in the original noise signal.

Similarly tone jamming techniques consider a function to control the power for a required
jamming margin considering a tone either with a random phase or tone with random

frequency within the RF bandwidth. For the tone jamming the phase variation generates a
PSD approximately centered at the RF filter depending on the knowledge of the receiver

frequency. For the random frequency interference it is desirable to get a PSD distributed

over the RF bandwidth.

The simulated interference waveforms are categorized as a continuous jammer when

the jamming signal is present during the symbol interval and for all symbols generated
in the message transmitted interval. On the other hand pulse jamming is an intermittent
interference signal with a given pulse duration and pulse repetition interval to determine a
duty cycle as a ratio between on transmission and pulse repetition interval. In this research
all pulsed waveforms generated were derived from continuous waves as a product of the
waveform and a pulse train with a determined duty cycle. The MTJ is considered as a

32
finite number of frequencies within the signal duration either as tones using the same
waveform and transmitted at different frequencies or waveforms that generate different
frequencies describing some spectrum pattern as a frequency hopped interference (FHI) or

comb spectrum interference (CSI).

3.4.1 Broad-band Interference.

The broadband interference for DSSS can be any waveform that occupies a bandwidth

equal to or greater than the spreading bandwidth. In this research the spreading bandwidth
W ss is defined as a function of the chip rate Rc in chips per seconds. For a DSSS passband

signal this bandwidth can be expressed as:

1
W ss = 2Rc = 2 , (3.33)
Tc
and the symbol interval T s = Nc T c , where Nc is the number of chips per symbol and T c is

the chip interval in seconds. Then the spreading bandwidth measured in Herz in terms of

the symbol rate RS , yields :

Nc
W ss = 2 = 2Nc R s . (3.34)
Ts
Therefore within the category of broadband interference can be considered multi-tone
interference, noise interference or any random modulated waveform that exceeds W ss .

In this research the waveforms considered include: broad-band noise (BBN), random

frequency modulated interference (RFMI), LFMI, FHI, CSI and MTJ.

3.4.1.1 Broad-band Noise.

In BBN the spectral components are affected equally and similarly for different

frequencies. This interference technique is the simplest to generate because it only requires
knowledge of the spreading bandwidth. This can be simulated as an AWGN with average
jamming power J p over the simulated receiver bandwidth (F s /2) or using the spreading
bandwidth W ss :

Jo
J p = 2W ss = W ss Jo , (3.35)
2

33
 4 0

2 −5

|DFT| (dB)
Amplitude

2
0 −10

−2 −15

−4 −20
0 200 400 600 800 1000 0 50 100 150 200
Time(ms) Frequency (KHz)

Figure 3.5: Time domain response (left) for 1000, T s =1 ms, symbols and normalized PSD
(right) for BBN using a bandwidth W J =201.5 KHz.

where Jo is the jamming PSD in units of Watts per Hz.

Figure 3.5 shows the time domain response of noise for 1000, T s =1 ms, symbols (left

subplot) and the normalized PSD response (right subplot) using a bandwidth W J =201.5
KHz that corresponds to the simulated bandwidth F s /2. Figure 3.6 show the time-
frequency representation of BBN jammers using two different bandwidths W J =F s /2 and

a filtered noise (colored noise) using the spreading bandwidth W J = W ss implemented with

a Butterworth filter of order 8 to illustrate the frequency distribution differences across

time.

3.4.1.2 Random Frequency Modulated Interference (RFMI).

This jamming technique generates random frequencies from the center carrier
frequency and within the W ss bandwidth, controlled with the frequency deviation factor
∆ f . The frequency randomness is generated by a random instantaneous phase of ±π over
T s . A RFMI can be represented by [29]:

√
J(t) = 2J p cos(2π fc t + θ(t) + ϕ), (3.36)

34
Figure 3.6: BBN spectrograms showing 100, T s =1 ms, symbols and W J =201.5 KHz (left)
and W J =62 KHz (right) bandwidths.

with
∫ Ts
θ(t) = ϕ + ∆ f f (α)dα, (3.37)
0

where

ϕ is the initial phase.

θ(t) is a random process characterizing the signal’s instantaneous phase.

f (α) is the instantaneous random frequency modulated signal in Hz.

∆f is the frequency deviation from the central frequency in Hz.

Ts is the symbol duration in seconds.

As an example of RFMI Figure 3.7 shows the time domain signal (the left subplot)

and the normalized PSD for two symbols (right subplot) with T s =1 ms. The RFMI

waveform generation was adopted from the MATLAB® simulation introduced by Temple
[30]. Figure 3.8presents the corresponding time-frequency plot to illustrate how the random
frequency is varying with respect to the time. Random frequency modulated signals are
used in wide-band radar for intra-pulse modulation due to its properties of low side-lobe

35
 1.5
0
1

|DFT| (dB)
Amplitude
0.5 −10

2
0
−20
−0.5

−1 −30
−1.5
0 0.5 1 1.5 2 0 50 100 150 200
Time(ms) Frequency (KHz)

Figure 3.7: The RFMI time domain response (left) and normalized PSD (right) for two,
T s =1 ms, symbols.

Figure 3.8: The RFMI spectrogram showing three, T s =1 ms, symbols.

autocorrelation, good range resolution and interference suppression. The RFMI can achieve
lower side-lobes than conventional intra-pulse modulation in radar used to increase the
range resolution [31].

36
 0.1
10

0.05 0

|DFT| (dB)
Amplitude

2
0 −10

−20
−0.05

−30
−0.1
0 0.5 1 50 100 150
Time(ms) Frequency (KHz)

Figure 3.9: Time domain response (left) for one, T s =1 ms, symbol and normalized PSD
(right) for CW LFMI for three symbols.

3.4.1.3 Linear Frequency Modulated Interference (LFMI).

A LFMI signal is broadly used in radar and spread spectrum communications also

called chirp spread spectrum signaling. The LFMI signal a waveform whose frequency
varies linearly within the signal duration to generate a high bandwidth maintaining the pulse

duration. The resultant signal can achieve a time-bandwidth product much greater than the

non-modulated pulsed signal where this factor is not greater than 2, due to the passband

bandwidth for non-modulated pulsed signal is typically defined as 2/τ. Mathematically

LFMI can be modeled as having an average power J p and initial phase ϕ:

√ { }
J(t) = 2J p exp j2π fc t + jπµo t2 + jϕ , (3.38)

where µ0 = B/τ is the is linear frequency slope factor from the initial frequency fc and
B in Hz is the frequency deviation over signal duration τ in seconds. The units of µ0 are
S −2 [32]. Figure 3.9 shows the time domain (left subplot) and the normalized PSD (right
subplot) for a generated LFMI signal considering three symbols with T s =1 ms. Figure 3.10
illustrate a LFMI plot of frequency versus time to observe the linear frequency variation.

37
 200

Frequency (KHz)
150

100

50

0
0.5 1 1.5 2 2.5
Time(ms)

Figure 3.10: The LFMI spectrogram showing three, T s =1 ms, symbols.

3.4.1.4 Frequency Hopped Interference (FHI).

Another model of simulated interference to evaluate the DSSS receiver performance is

presented by [18]. The FHI consist of a signal with power J p as a product of a rectangular
window shifted by a time hopping interval T H and different frequency tones fk chosen

randomly within a jamming bandwidth W j and initial phase ϕo . The mathematical model

is:

√ ∑
N
[ ]
J(t) = 2J p ⊓T H (t − kT H ) × exp j2π fk (t − kT H ) + jϕo , (3.39)
k=1
where







 |t| < TH
1, 2
⊓T H =
 (3.40)




0, |t| > TH
2

The simulated FHI is shown in Figure 3.11 for five T s = 1 ms symbols using a T H

with the same duration as the symbol duration, including time domain (left subplot) and
the normalized PSD (right subplot) responses, showing five frequencies in a bandwidth

38
 0.1

0.05 0

|DFT| (dB)
Amplitude

−10

2
0

−0.05 −20

−0.1 −30
0 1 2 3 4 5 50 100 150
Time(ms) Frequency (KHz)

Figure 3.11: Time domain (left) and normalized PSD (right) for the FHI showing five, T s =1
ms, symbols and bandwidth W j =62 KHz.

200
Frequency (KHz)

150

100

50

0
2 4 6 8
Time(ms)

Figure 3.12: The FHI spectrogram for 10, T s =1 ms, symbols and bandwidth W j =200 KHz.

W j =62 KHz. The time-frequency plot in Figure 3.12 shows a FHI for 10, T s = 1 ms,
symbols, using a T H=0.4 ms, resulting in 26 frequencies random uniformly distributed in
a bandwidth W J =200 KHz.

39
 3.4.1.5 Comb-Spectrum Interference (CSI).

The CSI model consists of generating a series of narrow-band signals modulated by a

series of tones distributed over the W ss bandwidth [18].

√ ∑
N
J(t) = 2Pk exp { j2π fk t + 0.01∆ f sin(2π∆ f )} (3.41)
k=1

J(t) is generated by a group of frequency modulated signals, where fk is the central

frequency for each component with a frequency deviation of ±∆ f in Hz from the center
frequency such that ∆ f ≪ fk . Figure 3.13 shows the temporal CSI response (left) for one

T s =1 ms symbol and the normalized PSD for 1000, T s =1 ms, symbols, five frequencies and
a frequency deviation ∆ f =0.5 Hz. Figure 3.14 presents the corresponding time-frequency
plot to illustrate that the comb-like spectrum across the time, showing five frequencies for

a total of 50, T s =1 ms, symbols and ∆ f =0.5 Hz.

0.04
0

0.02
|DFT| (dB)
Amplitude

−10
2

0
−20

−0.02
−30

−0.04
0 0.5 1 60 80 100 120 140
Time(ms) Frequency (KHz)

Figure 3.13: The time domain CSI response (left) for one, T s =1 ms, symbol and normalized
PSD (right) response for five frequencies, 1000 symbols and ∆ f =0.5 Hz.

3.4.1.6 Multi-Tone Jamming (MTJ).

This jammer can be describes as a summation of several tones each of frequency fk and
random phase which is uniformly distributed in the interval [0, 2π] and average power J p .

40
Figure 3.14: The CSI spectrogram for 50, T s =1 ms, symbols and bandwidth W j ≈ 62 KHz.

Under this technique, also called multiple CW tone interference, the total received jamming

power is divided in Nt different random phase CW tones. The tones are usually distributed
over the spreading bandwidth. The MTJ can be modeled by the following expression:

√
∑
Nt
Jp
2 cos(2π fk t + ϕ) where ϕ ∼ U [0, 2π] . (3.42)
k=1
Nt
In Figure 3.15 can be observed the time domain representation of a multi-tone signal
for one, T s =1 ms, symbol interval (left subplot) and the normalized PSD (right subplot)

for 1000 symbols using 10 random phase tones. Figure 3.16 presents the corresponding

time-frequency plot to illustrate that MTJ generates multiple tones with constant frequency

as the time varies.

3.4.2 Narrow-Band Interference (NBI).

In this research NBI is considered a waveform that occupies passband bandwidth that
is much less than the RF bandwidth of the DSSS. Within this category are considered 1)

CW interference signals that occupy a smaller band-pass bandwidth compared to the W ss

bandwidth and 2) partial-band jammers.

41
 0.06

0.04 0

|DFT| (dB)
Amplitude
0.02
−10

2
0

−0.02 −20

−0.04
−30
−0.06
0 0.5 1 60 80 100 120 140
Time(ms) Frequency (KHz)

Figure 3.15: Time domain MTJ response (left) for one, T s =1 ms, symbol interval and
normalized PSD for 10 random phase tones and 1000 symbols.

Figure 3.16: The MTJ spectrogram for 100, T s = 1 ms, symbols and bandwidth W J ≈62
KHz.

3.4.2.1 Tone Jammer.

The tone jammer considered in this research assumes that carrier frequency is known
and the phase is a random variable uniformly distributed over the interval [0, 2π]. The phase
varies in a symbol by symbol basis and J(t) has an average power J p . The mathematical

42
representation is:
√
J(t) = 2J p cos(2π fc t + ϕ) where ϕ ∼ U [0, 2π] . (3.43)

In Figure 3.17 is presented the tone jammer time domain response (left subplot) illustrating
a random symbol transition from the first to the second symbol and normalized PSD

(right subplot) response for tone jammer with random phase for 1000 T s =1 ms symbols.
Figure 3.18 presents the corresponding time-frequency plot to illustrate the tone jammer
with constant frequency across time variation.

1 10

0.5 0
|DFT| (dB)
Amplitude

−10
2

0
−20
−0.5
−30

−1
0.96 0.98 1 1.02 1.04 80 90 100 110 120
Time(ms) Frequency (KHz)

Figure 3.17: Time domain response showing one, T s =1 ms, symbol phase transition (left)

and normalized PSD (right) for tone jammer for 1000, T s =1ms, symbols.

3.4.2.2 Binay Phase Shift Keying (BPSK) Interference.

The BPSK interference model consists of a source of random binary data that is
mapped according to:

θm = π(m − 1); m ∈ 1, 2. (3.44)

Then the jamming signal yields:

√
J= 2J p cos (2π fc t + θm ) . (3.45)

43
 Figure 3.18: The tone jammer spectrogram for 100, T s =1 ms, symbols.

From the above equation θm takes values either 0 or π. This is another waveform that

is similar to the target signal with no knowledge of the spreading sequence and it also can
be used to generate a pulse jamming interference.

3.4.2.3 Partial-Band Noise.

As it was explained noise interference is a function of the signal bandwidth. However,

the noise power can be distributed over a desired bandwidth instead of the total spreading

bandwidth. In this research the partial-band noise or narrow-band noise is simulated using

filtered (colored) AWGN noise signal. The average jamming power J p can be expressed as
a function of ρn that represents a fraction of the spreading bandwidth W ss :
WJ
ρn = ≤ 1, (3.46)
W ss

where W J is the jamming bandwidth. The jammer PSD S j can be represented by [11]:
Jp J p W ss
Sj= = · ,
W J W ss W J
Jo
Sj= . (3.47)
ρn
Jo is equivalent to the noise power spectral density as if the jammer power were spread over
W ss .

44
 4 0

2 −5

|DFT| (dB)
Amplitude

2
0 −10

−2 −15

−4 −20
0 200 400 600 800 1000 0 50 100 150 200
Time(ms) Frequency (KHz)

Figure 3.19: Time domain response for 1000, T s =1 ms, symbols (left) and normalized PSD
(right) for NBN using a bandwidth W J ≈8 KHz.

Figure 3.20: NBN spectrograms showing 100, T s =1 ms, symbols for W J ≈8 KHz.

Figure 3.19 shows the time domain response for1000, T s =1 ms, symbols (left subplot)
and the normalized PSD for NBN using a W J ≈8 Khz (right subplot). Figure 3.20 illustrate
the time-frequency plot for NBN for 100, T s =1 ms, symbols and bandwidth W J ≈8 KHz.

45
 3.4.3 Pulse Jamming.

Pulse jamming is a kind of interference that occurs sporadically for short durations,
periodic or aperiodic [7]. At baseband the pulse burst model can be represented by:

∑
M−1
X p (t) = ⊓(t − mT p ) (3.48)
m=0

where

X p (t) is the baseband train of pulses.

M is the number of pulses in the burst.

Tp is the pulse repetition interval in seconds.

⊓(t) is the rectangular function with a duration of τ seconds.

and



(t) 

 τ
1, if |t| ≤ 2 ,

⊓ (t) = rect =
 (3.49)
τ 



0, if |t| > 2τ .

Any interfering pulsed waveform can be simulated as a product of the pulse train in

Equation (3.48) and a continuous waveform f (t) as follows:

√
Jp
J(t) = 2 X p (t) · f (t), (3.50)
ρ

where J p is the average jamming power and ρ is the duty cycle.

In Figure 3.21 are shown the pulsed BPSK time domain response for one T s=1 ms symbol
(right subplot) with pulse duration τ=100 usec and a duty cycle ρ=0.3 and the pulsed BPSK

46
 3

2
0

|DFT| (dB)
Amplitude
1
−10

2
0

−1
−20
−2

−3 −30
0 0.2 0.4 0.6 0.8 1 60 80 100 120 140
Time(ms) Frequency (KHz)

Figure 3.21: Time domain response for 1000, T s =1 ms, symbols and normalized PSD for
pulsed BPSK.

200
Frequency (KHz)

150

100

50

0
1 2 3 4
Time(ms)

Figure 3.22: Pulse BPSK spectrograms showing 5, T s =1 ms, symbols.

normalized PSD response for 1000, T s=1 ms, symbols (right subplot). Figure 3.22 shows
the pulsed BPSK time-frequency response illustrating the how the spectrum is varying
across time, considering 5, T s =1 ms, symbols.

47
3.5 Jamming Performance Evaluation

The process involved in the interference evaluation consists of the receiver validation,
receiver performance with implemented filters and JSR computation requirements for the
different interference techniques described in Section 3.4.

3.5.1 Receiver Validation.

First the receiver performance is compared with the theoretical expression under an

AWGN channel with ideal conditions. It means that the RF and despreading filters are
not considered and therefore the received signal PSD is not altered by filter coloration. It is

important to contrast results with the analytical expressions in an AWGN channel presented
in Section 3.3. The spread spectrum model presented in Figure 3.3 can be used either for
PSK or DPSK modulation that are the most representative signaling schemes in a DSSS

communication system.
The parameters considered as a baseline for BER receiver simulations are presented

in Table 3.1. In the present research the simulation considered a total of 1000 symbols for

Eb /No ∈ [0, 8] dB in 0.5 dB increments. For each Eb /No level the symbols are generated

until 500 bit errors are found by comparing the symbols received with respect to the
effective symbols transmitted. The average simulation error is less than 30 percent for

a 95 percent confident interval as a trade-off between error and number of bits required,

according to the criteria described in Section 3.2.

In Figure 3.23 is presented the performance of a QPSK DSSS receiver implemented

with a sequence Nc =31 chips that follows the Gold code properties as explained in
Section 2.4. The signal power considered is fixed to1 Watt while varying the noise power
to obtain the required Eb /No levels. It is observed that simulated QPSK performance
is the same as BPSK but efficiently allows doubling of the data rate. The results
presented validates the simulated performance compared with analytical expression since

48
 Table 3.1: Signal parameters considered for the simulated receiver.

DSSS Modulation
QPSK DBPSK DQPSK
Carrier Frequency (KHz) 1 1 1

Symbol Rate R s (Sym/s) 1000 1000 1000

Data Rate Rd (KBits/s) 2 1 2

Code Rate Rc (KBits/s) 62 31 62

Spread Sequence(Chips) Gold(31) Gold(31) Gold(31)

RF Bandwidth WRF (KHz) 62 62 62

Despreading Filter Bandwidth WDS (KHz) 2 1 2

the simulated results closely approach the analytical curves. The blue curve was computed
using Equation (3.14).

In Figure 3.24 is presented the performance of DPSK for a modulation index M=2 and

M=4. It is observed that simulated results approach closely to the analytical results. For

Eb /No ratios greater than 4 dB in order to obtain similar performance differential quadrature
phase shift keying (DQPSK) requires an Eb /No of approximately 1 dB more with respect
to differential binary phase shift keying (DBPSK) performance. Comparing QPSK with
binary DPSK, the first modulation type performs 1.5 dB better. For QPSK compared with

DQPSK, it is observed than the latter is approximately 2.5 dB poorer. In Table 3.2 are
presented simulated BER for PSK and DPSK for Eb /No ratios from 5 to 8 dB showing the
approximate differences discussed previously.

3.5.2 Receiver Performance with Bandpass Filters.

Once the receiver has been validated under ideal conditions, a second step considered
is the receiver performance simulated using the bandpass RF and the bandpass despreading

49
 −1
10
Analytical QPSK/BPSK
Simulated QPSK

−2
10
BER

−3
10

−4
10
0 1 2 3 4 5 6 7 8
E /N (dB)
b o

Figure 3.23: BER performance for QPSK in AWGN channel.

Table 3.2: PSK and DPSK BER performance comparison.

BER ×10−3
QPSK 6.3 4.4 2.4 1.5 0.9 0.5 0.2

DBPSK 20.8 16 9.7 6.1 3.5 2.1 1

DQPSK 31.7 25 18.1 12.4 9.3 6.2 4

Eb /No (dB) 5 5.5 6 6.5 7 7.5 8

filters to simulate a real receiver implementation where the signal is filtered previous to
down-conversion in order to remove the unwanted spectral components and to improve
the SNR and also to estimate the interference reduction after the despreading mixer. In

50
 0
10
Analytical DQPSK
Analytical DBPSK
Simulated DQPSK
−1
Simulated DBPSK
10
BER

−2
10

−3
10

−4
10
0 1 2 3 4 5 6 7 8
Eb/No (dB)

Figure 3.24: BER performance for DQPSK and DBPSK in AWGN channel.

this research and for the purpose of interference analysis the modulation scheme that will

be considered is QPSK because it presents the best performance in PSK modulation and
bandwidth efficiency in terms of data rate, that is, for equal symbol rate it doubles the data
rate compared with BPSK with the same BER performance. Figure 3.25 shows results
of an ideal DSSS QPSK receiver along with the performance using filters as BER versus

Eb /No . From the figure it is possible to observe that the filtered signal is degraded such that
approximately each BER point of the receiver requires 0.5 dB higher Eb /No to obtain the
ideal receiver’s performance neglecting the filter in an AWGN ideal condition that performs
closely to the theoretical curve as was observed in the receiver validation.

The performance degradation is the result of RF and despreading filtering generating

an approximately constant error for the Eb /No levels simulated. However there is an

51
 −1
10
Filtered DSSS/QPSK
Ideal DSSS/QPSK

−2
10
BER

−3
10

−4
10
0 1 2 3 4 5 6 7 8
E /N (dB)
b o

Figure 3.25: Simulated BER performance for DSSS QPSK receiver in AWGN channel

illustrating the effect of band-pass filtering.

improvement in the SNR after each filter where the noise is filtered and attenuated in

greater proportion compared with the signal. In Figure 3.26 are presented the linear relation
of S NRRF prior despreading the received filtered signal and S NRDS after despreading and
filtering the signal (WS im ≈ 201.5 KHz > WRF ≈ 62 KHz > WDS ≈ 2 KHz). The plot shows
the relation in respect to the input S NRS im that numerically represents an average increment

of approximately 4.9 dB after WRF filtering and 14.87 dB from WRF to WDS filtering. The
latter result is tightly close to the theoretical processing gain using a Gold sequence of 31
chips length, i.e., 10 log10 (31) ≈ 14.91 dB.

52
 10
SNR
RF
SNRDS
5

0
SNROut (dB)

−5

−10

−15

−20
−20 −19 −18 −17 −16 −15 −14 −13 −12
SNRSim (dB)

Figure 3.26: Comparison of input S NRS im and S NROut of the band-pass RF filter (S NRRF )

and despreading filter (S NRDS ).

3.5.3 Simulation to Determine the Jamming to Signal Ratio (JSR).

The JSR is a parameter that depends on jamming waveform and it is measured at

the receiver. This research develop a simulation of the receiver to contrast the effect of

different interference signals described in Section 3.4. It implies to vary the jamming signal
increasingly in order to overcome the processing gain. For this purpose it is necessary select
as a baseline the receiver performance in terms of BER at a specific Eb /No ratio in dB. It
simulates a receiver static in respect to the jammer, also the simulated scenario assumes
maximum antenna directivity in the jammer direction, neglecting the antenna pattern and
assuming that the desired signal is received with constant power. From the DSSS receiver

perspective it allows to observe a specific BER performance for a given energy per bit

53
and noise power spectral density that remain constant while the interference average power
varies with respect to the average signal power.
Under the assumption considered previously the performance comparison is based on

the jamming margin required to degrade the receiver performance chosen by one order
of magnitude. From that point it has been considered that the interference overcomes the
processing gain of the DSSS receiver. Chapter IV provides results using the jammer models

presented in Section 3.4 in plots of BER versus JSR after the RF filter in order to evaluate
the performance of a QPSK for different interference waveforms. Finally all jammers are

compared in respect to the JSR and processing gain required to degrade the DSSS BER by
one order-of-magnitude relative to a BER of Eb /No =7 dB with no interference present. The
variation of the jamming bandwidth is explored for the broadband jammers and the duty

cycle for the pulsed jammers with the goal of determining their optimum responses. Finally

the most effective jammers are evaluated on the receiver for the JSR found by changing the

input Eb /No .

54
 IV. Results and Analysis

This chapter presents the results and analysis of different jammers’ performance in
a simulated DSSS QPSK receiver. The parameters considered in the simulation follow
Table 3.1. The sampling frequency for all simulated jammer was 403 KHz to avoid
aliasing based on passband spreading bandwidth W J =62 KHz and using an oversampling
factor of 1.5. The simulation was performed considering a DSSS QPSK receiver due

to its performance and bandwidth efficiency compared with binary PSK or DPSK. The
conditions for the simulation assume that the receiver is perfectly synchronized with
the carrier frequency ignoring mixing losses or errors due to phase variation. Also the

despreading process considers a perfectly synchronized sequence. The results presented

were obtained by generating a message of 1000 symbols length for a fixed Eb /No =7

dB that corresponds to BER=1.4 × 10−3 . The input JSR ∈ [−15, 15] dB in increments
of 2 dB with the goal of determining a BER degradation of one order-of-magnitude.

The Eb /No point chosen and the JSR range is a trade-off between simulation time and

interference assessment accuracy, however the constraints considered allow to compare

between different jamming techniques.

4.1 Simulation of Broad-band Jammers

Figure 4.1 shows the performance of broad-band jammers that include BBN with a

bandwidth greater than the spreading bandwidth W ss , FHI, CSI, MTJ and RFMI with a
bandwidth that covers exactly W ss and centered at carrier frequency. The horizontal axis
represents the JSR at the output of the RF filter (JS RRF ) denoting that for BBN using
the simulation bandwidth F s /2 the jamming power is decreased considerably because the
filter eliminates the spectral component out of the band W ss . The red line represent the

BER=1.4 × 10−2 at which the receiver performance is degraded in one order-of-magnitude

55
 0
10
BBN (F /2)
s
BBN (Wss)
FHI
CSI
−1 MTJ
10 RFMI
BER (dB)

−2
10

−3
10
−20 −15 −10 −5 0 5 10 15
JSR (dB)
RF

Figure 4.1: Broad-band jammer BER performance versus JS RRF for Eb /No =7 dB.

from the baseline at Eb /No = 7 dB. In general all broad-band jammers behave similarly

for the JSR range simulated with the exception of RFMI that produces slightly more
degradation and also requires slightly less JS RRF signal to pass the BER reference point
(red line). The BBN (F s /2) results in the least effect due to its PSD being distributed
uniformly over the simulated channel bandwidth and consequently its average power is
attenuated more by the RF filtering.

4.2 Simulation of Continuous Wave and Narrow-band Noise Jammers

The Figure 4.2 present the JSR curves for CW interferers and for narrow-band
jammers. The CW jammers include LFMI, tone jammer and BPSK. Narrow-band jammers

include filtered noise for two different jammer bandwidths (W J ), including W J =7.75

56
 0
10
LFMI
BPSK
Tone
NBN (Wss/8)
NBN (Wss/16)
−1
10
BER (dB)

−2
10

−3
10
−15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.2: CW and NBN BER performance versus JS RRF for Eb /No =7 dB.

KHz (W ss /8) and W J =3.875 KHz (W ss /16). Results denote that for tone and BPSK the

performance is similar requiring more JS RRF to reach the jamming margin, i.e. the

point where the curve passes the red line. LFMI interference presents a considerably
better jamming performance than BPSK and tone jammers occupying the spreading
bandwidth W ss . On the other hand colored noise perform better than CW interference

occupying fraction of the spreading bandwidth with slight performance improvement when
the bandwidth is 1/8 of W ss compared to 1/16 of W ss . The colored noise interference
using a narrow bandwidth requires approximately 1.63 dB less JS RRF than LFMI and
approximately 8.3 dB and 8.4 dB less JS RRF for BPSK and tone jamming respectively.

57
4.3 Simulation of Pulsed Jammers

For pulse interference the average jamming power spectral density can be expressed
as the ratio between the jammer PSD and the duty cycle ρ as Jo /ρ and the equivalent noise
PSD at the receiver is Noe [7]:

Noe =No + Jo /ρ, (4.1)

and the symbol error probability P s for pulse jammer is

(√ ) (√ )
P s  ρQ 2E s No + Jo /ρ +(1 − ρ)Q 2E s No , 0 ≤ ρ ≤ 1. (4.2)

When ρ is treated as a continuous variable over [0, 1] and Jo ≫ No , the value that
maximizes P s is:



 ( E )−1



s
, Es
> 0.7
0.7 Jo Jo
ρo  
 (4.3)




1,
Es
≤ 0.7
Jo

The worst-case pulse interference from the receiver perspective is more effective than

continuous interference if E s /Jo > 0.7. Substituting ρ = ρo into Equation (4.2), P s when
Jo ≫ No is:




 ( E )−1


0.083 s
, Es
> 0.7
 Jo Jo
Ps  
 √ ( 2E ) (4.4)




Q s
, Es
≤ 0.7
Jo Jo

According to the preceding expression the bit error probability performance is a function
of pulse duty cycle ρ and the optimum ρ decreases as E s /Jo ratio increases. It is the
worst jamming case scenario that maximize the jamming effectiveness. In Figure 4.3 are

presented simulated results for pulsed jamming waveform as BER versus JSR after the RF
filter. Plots are considered for a variable duty cycle as a function of the JSR and for a

58
 0
10
Pulsed Tone
Pulsed BBN
Pulsed BPSK
Pulsed LFMI
Pulsed RFMI
−1
10
BER (dB)

−2
10

−3
10
−15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.3: Pulsed jamming BER performance versus JS RRF for Eb /No =7 dB and ρ varying

according to Equation (4.3).

fixed pulse-width of 0.1 milliseconds, that is equivalent to a bandwidth W J ≈20 KHz. This

computation was concluded after plotting JSR for different duty cycles and due to analysis
of theoretical P s in Equation (4.4) described in [13], [7], [11] and that show the worst
jamming scenario is obtained when the duty cycle decreases as the JSR increases. Clearly

results show that pulse BPSK presents the best performance with a required JS RRF ≈3.05
dB. Pulse tone requires approximately 0.74 dB more of JS RRF to reach the jamming
margin, whereas the pulse LFMI requires 3.38 dB, pulsed RFMI requires 2.25 dB and
pulsed BBN requires 4.28 dB of additional JS RRF to produce the same degradation, being

the least effective of the pulsed jammers simulated.

59
Table 4.1: JS RRF in dB required for broad-band jammers to degrade BER by one order-of-
magnitude.

Jammer BBN(F s /2) BBN(W ss ) FHI CSI MTJ RFMI LFMI

JS RRF 7.18 7.03 7.05 6.74 7.12 6.05 6.67

JS RDS -8.22 -8.04 -8.26 -8.57 -8.17 -7.98 -8.62

Imp. fact. -15.40 -15.08 -15.32 -15.31 -15.29 -14.04 -15.29

4.4 Jammer Comparison

The JSR performance curves have been grouped all together in order to observe the

general behavior once the signal has been received after the RF filter. The main objective of

this measurement is to compare the effects of simulated partial-band interference, narrow-

band interference, broad-band interference, CW and pulsed jammers. In Figure 4.4 is

presented the BER performance of a total of 15 simulated jammers. It is possible to confirm

that CW interference, single tone frequency, has the poorest effectiveness denoting slight
difference between BPSK and single tone jamming. The BPSK jammer performs slightly
better than tone jamming for different symbols generated randomly at the carrier frequency.

The tone jammer is centered at the carrier frequency but the phase varies randomly in
symbol by symbol basis within the interval between 0 and 2π. Also from this plot it is

possible to conclude that pulsed BPSK presents the best performance followed closely by
pulsed tone jamming. The narrow-band noise jammers require approximately 2 dB more

of JS RRF than pulsed BPSK. The broad-band jammer’s performance is similar to pulsed
LFMI and pulsed noise waveforms.

The summary of the numerical JS RRF , expressed in dB, required to degrade the
receiver performance to a BER=1.4 × 10−2 for 15 of the jammers simulated are shown

60
 0
10
BBN (F /2)
s
BBN (Wss)
FHI
CSI
MTJ
−1
10 RFMI
LFM
BPSK
BER (dB)

Tone
NBN (Wss/8)
Pulsed Tone
−2 Pulsed BBN
10
Pulsed BPSK
Pulsed LFMI
Pulsed RFMI

−3
10
−20 −15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.4: BER versus for JS RRF comparing all jammer signals.

Table 4.2: JS RRF in dB required for CW, Narrow-band noise and pulsed jammers to
degrade BER by one order-of-magnitude.

Jammer BPSK Tone NBN P. BPSK P. Tone P. BBN P. LFMI P. RFMI

JS RRF 13.30 13.48 5.04 3.05 3.80 7.33 6.43 5.30

JS RDS -2.39 -2.66 -7.72 -9.61 -8.68 -7.83 -8.58 -8.33

Imp. fact. -15.69 -16.15 -12.77 -12.67 -12.49 -15.16 -15.02 -13.91

61
in Table 4.1 and Table 4.2 respectively. The NBN jammer correspond to the colored noise
with a simulated bandwidth of 1/8 of the spreading bandwidth (W ss ).
In presence of jamming, the processing gain also can be expressed as an improvement

in the JSR at the despreading filter output (JS RDS ) compared to the JSR at after the
RF filter. From data presented in Table 4.1 broad-band noise jamming present less JSR
improvement compared to broad-band jammers, that is, the jamming power decreases in
more proportion in comparison to the signal power. From Table 4.2 and consistent with
Figure 4.4, tone and CW BPSK jamming present the least JSR improvement factor agreeing

with its performance. Similarly pulsed BBN and pulse LFMI present improvement factors
close to the improvement achieved for broad-band jammers with the exception of pulsed
RFMI that performs approximately 1 dB better.

4.4.1 Effects of Jamming under Bandwidth and Duty Cycle Variation.

Considering the comparison of all simulated jammers and the close results of pulsed
jamming and narrow-band jamming, this section analyzes the effect of varying the

bandwidth for the broad-band interferers and the duty cycle for pulsed jammers in order

to find the optimal responses for those jammer with better results in term of effectiveness

in the simulated DSSS QPSK receiver. From the analysis of broad-band and narrow-band
jamming it was observed from Figure 4.1 and Figure 4.2 that changing the bandwidth of

noise jamming varies the effectiveness in terms of JS RRF required and BER degradation. In
order to analyze the effect of bandwidth variation for broad-band interference a simulation

for the baseline Eb /No =7 dB and a JSR=5 dB were run. Figure 4.5 shows BER versus
jammer bandwidth for [0, 62] KHz where W ss =62 KHz and W J ≈0 represents the single
tone jamming frequency. This simulation denotes that approximately from 10 to 40 KHz of
bandwidth, the average response is optimal with a maximum BER degradation depending
on the jamming waveform. The CSI and MTJ were simulated for a total of 100 frequencies
in the spreading bandwidth but the number of frequencies vary according to the bandwidth.

62
 −1
10
FHI
CSI
MTJ
RFMI
Noise
LFMI
BER

−2
10

−3
10
0 10 20 30 40 50 62
Jammer Bandwidth W (KHz)
J

Figure 4.5: BER performance versus bandwidth variation for broad-band jammers for

JSR= 5 dB and Eb /No =7 dB, where W J ≈0 represents single tone jamming frequency.

FHI differs in respect to the total of frequencies generated. For a given time hopping less

than the symbol interval and considering 1000 symbols a total of 2600 frequencies were
generated from random permutations considering the spreading bandwidth centered at the
carrier frequency. The Figure 4.5 shows that CSI and MTJ generate more degradation with

a BER=2.3 × 10−2 using W J ≈22.1 KHz and W J ≈17.71 KHz of bandwidth respectively. The
minimum BER performance was obtained for a W J ≈48.7 KHz. These were CSI and MTJ
cases, due to the tones separation varies with the bandwidth for a fixed number of tones
Nt =100, resulting in a jammer PSD with frequency components distributed farther from

the jamming central frequency f j that are significantly spread after convolving with the
spreading waveform and consequently filtered out the signal bandwidth WDS =2 KHz.

63
 −1
10
Pulsed Tone
Pulsed BPSK
Pulsed Noise
Pulsed LFMI
Pulsed RFMI
BER

−2
10

−3
10
0 0.2 0.4 0.6 0.8 1
Duty Cycle (ρ)

Figure 4.6: BER performance effect for duty cycle variation ρ ∈ [10−3 , 1] for pulsed

jamming.

Figure 4.6 shows BER performance for different pulse jammers when the duty cycle

ρ ∈ [10−3 , 1] that confirms that pulsed BPSK is the most effective pulsed jamming
technique between other pulsed waveforms simulated. The simulation was run for the

JSR required to degrade the performance to a BER=1.4 × 10−2 and for an Eb /No = 7
dB. Pulsed tone and pulsed BPSK jammers show the best performance with a duty cycle
ρ=0.21 using a JSR=5 dB. Pulsed noise simulation was run for a JSR=13 dB with the
maximum BER corresponding to a duty cycle ρ=0.07, pulsed LFMI and pulsed RFMI were
simulated with a JSR of 7 dB with maximum BER at ρ=0.36 and ρ=0.64 respectively. The

JS R requirement and maximum ρ value was the result of convolving the pulsed windows
and broad-band waveforms in the frequency domain. The duty cycle variation ρ for a

64
pulse duration τ=100 useg implies that pulse repetition period T p varies inversely. For
pulsed BPSK and pulsed tone jammers with ρ=0.36 the simulattion showed that the BER

performance changed abruptly independent of the number of required bit for that particular
performance. In this case, the convolution of pulsed window is performed with a narrow-
band CW signal with W J ≈ 2 KHz. Consequently, as ρ increases the T P decreases but the
spectral separation of frequency components for multiples of 1/T p increase, resulting in
an abrupt jammer energy reduction after despread and filter the signal in the bandwidth
WDS =2 KHz. However, the duty cycle relevant for pulsed jammer analysis considered was

ρ=0.21.

4.4.2 Narrow-band and Pulsed Jamming Comparison.

According to Figure 4.4 pulsed tone, pulsed BPSK jammers and the narrow-band noise
jammers were the most effective. However the effect of varying the bandwidth also denoted

that for certain values of bandwidth the BER versus JS RRF curves perform better requiring
less JS RRF to degrade the receiver’s performance in comparison to the broad-band jammers

previously simulated. Based on BER versus bandwidth the jamming’s BER performance

for each jammer at the optimum bandwidth are presented in Figure 4.7. The results show

that both MTJ and CSI perform similar requiring the minimum JS RRF to degrade the BER
performance in one order-of-magnitude with respect to the other jammer presented. MTJ

requires approximately a JS RRF ≈3.03 dB to degrade the BER to a Pb =1.4 × 10−2 , followed
by CSI that requires a JS RRF ≈3.09 dB. In contrast, FHI, LFMI, NBN and RFMI require

at least a JS RRF ≈4.9 dB to degrade the BER to the same level. Also it is observed that
for JS RRF >10 dB MTJ and CSI achieve less degradation compared with the rest of the
jammers simulated.

In Table 4.3 are summarized the JS RRF , JS RDS and the improvement factor for broad-
band jammers simulated with an optimum bandwidth at the BER=1.4 × 10−2 . The results

denote that both MTJ and CSI jammers present the best performance. Also from Table 4.3

65
 Table 4.3: The JS RRF in dB for broad-band jammers.

Jammer FHI MTJ RFMI NBN CSI LFM

JS RRF 4.92 3.03 5.12 4.92 3.09 5.08

JS RDS -7.58 -9.29 -7.50 -7.73 -9.57 -7.42

WJ 13.3 17.71 17.71 8.86 22.1 17.71

Imp. fact. -12.50 -12.33 -12.63 -12.66 -12.66 -12.51

it is observed that the effect of finding an optimum bandwidth results in greater JSR
improvement and consequently more effectiveness in favor of narrow-band jamming using

an optimum bandwidth in comparison to the improvement factors presented in Table 4.1

for broad-band jammers. The improvement factors presented in Table 4.3 are close to the

improvement factors for pulsed BPSK and pulsed tone jammers consistent with the required
JS RRF .

Also the effect of duty cycle variations are presented in Figure 4.8 for the two

most effective pulsed jammers presented in Section 4.4.1, pulsed BPSK and pulsed tone

jammers, curves in blue and green respectively. These results were simulated with a duty

cycle variation according to Equation (4.4), that is, every JSR point has a duty cycle that is
in inverse proportion to the E s /Jo ratio. On the other hand, curves in red and gray represent

the performance when the optimum value of duty cycle is chosen from Figure 4.6 that
corresponds to ρ=0.21 for pulsed BPSK and pulsed tone respectively. This result confirms
that Equation (4.4) corresponds to the worst case pulsed jamming scenario showing that

both pulsed BPSK and pulse tone require less JSR to degrade the BER to a Pb =1.4×10−2 in
comparison to the simulation with an optimum duty cycle ρ=0.21, although the differences
are on the order of hundredths of a decibel. Also it is observed that over a JS RRF =10 dB,
pulsed tone jammer with a duty cycle ρ=0.21, achieves slightly more BER degradation than
pulsed BPSK.

66
 0
10
FHI
CSI
MTJ
RFMI
NBN
−1 LFM
10
BER (dB)

−2
10

−3
10
−15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.7: BER performance with optimum bandwidth for broad-band jammers for an

Eb /No =7 dB.

4.4.3 Optimum Bandwidth for Pulsed Broad-band Jammers.

Another important point derived from bandwidth’s and duty cycle’s analysis is
the BER performance for pulse jamming implemented using a pulsed window and a

broad-band waveforms (waveforms with a bandwidth equal or greater than the spreading
bandwidth W ss ) compared to pulse jamming using waveform with optimum bandwidth.
Previously pulsed BBN, pulsed RFMI and pulsed LFMI jammers required JS RRF values
of approximately 2.2 dB to 4.3 dB more than the pulsed BPSK jammer. Based on this
results and the optimized bandwidth analyzed in Section 4.4.1. Figure 4.9 show simulations
for pulsed jammers with optimized bandwidth along with pulsed broad-band jammers
with non-optimal bandwidth. The pulsed window applied on broad-band waveforms

67
 0
10
Pulsed BPSK
Pulsed BPSK ρ=0.21
Pulsed Tone
Pulsed Tone ρ=0.21

−1
10
BER (dB)

−2
10

−3
10
−15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.8: Comparison of BER performance for pulsed BPSK and pulsed tone for

changing the duty cycle and Eb /No =7 dB.

did not improve the JS RRF as it did when it was applied on CW BPSK and CW tone,

where the JS RRF was between 9 dB to 10 dB approximately. However the combination

of the optimum bandwidth for broad-band waveforms and pulsed window improve the
jamming effectiveness. The pulsed broad-band jammers, using the optimum bandwidth
W J ≈17.71 KHz found previously, resulted in a required JS RRF ≈3.92 dB for pulsed RFMI,

JS RRF ≈3.98 dB for pulsed LFMI and JS RRF ≈3.81 dB for pulsed noise with an optimum
bandwidth W J ≈8.86 KHz, denoting a JS RRF improvement of approximately 2.44 dB, 1.36
dB and 3.51 dB respectively in respect to the pulsed jammers generated from broad-band
waveform presented in Table 4.2. After using the optimum bandwidth for pulsed broad-

band jammers, the results approach to the pulsed BPSK performance with approximately 1

68
 0
10
Optimal Pulsed RFMI
Optimal Pulsed LFMI
Optimal Pulsed Noise
Pulsed RFMI
Pulsed LFMI
−1 Pulsed Noise
10
BER (dB)

−2
10

−3
10
−20 −15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.9: BER performance comparison for optimal bandwidth pulsed jammers and non-

optimal bandwidth pulsed jammers.

dB of difference in the required JS RRF . Also the simulations show that pulsed broad-band

jammers with optimum bandwidth produce more degradation above JS RRF =13 dB.

4.5 Analysis for Optimum Jammers

Considering the different jamming techniques compared and the effect of bandwidth
and duty cycle variation, it is possible to conclude that MTJ, pulsed BPSK and CSI present
the optimum performance in terms of requiring a JS RRF range from 3.03-3.09 to degrade
the DSSS QPSK receiver. The Figure 4.10 shows the optimum jammers denoting that MTJ

requires JS RRF ≈3.03 dB followed by pulsed BPSK that requires JS RRF =3.05 dB and CSI
requires JS RRF ≈3.09 dB.

69
 In Figure 4.11 are shown the BER versus Eb /No results for the best jammers found:
MTJ, BPSK, CSI. The NBN jammer for an optimum bandwidth W J ≈8.86 KHz is presented
as a reference due to it presents similar performance to the optimized FHI, RFMI and LFMI

jammer. The receiver’s performance with no jamming is shown to compare results with the
jamming’s degradation as the Eb /No varies. The simulations consider a fixed JS RRF =3.1
dB to reach a BER’s degradation to Pb =1.4 × 10−2 which is approximately equivalent to
the receiver’s performance in absence of jamming for an Eb /No =4.5 dB. The plot shows
that, for an Eb /No =7 dB (S NRRF =-8.1 dB) and JS RRF =3.1 dB, indicated with a segmented
vertical red line, the receiver under the incidence of the most effective jammers degrades

its performance to a BER=1.4 × 10−2 or one order-of-magnitude in respect to the BER for
Eb /No = 7 dB with no jamming. It is observed that NBN jammer does not achieve the BER

degradation, due to it requires JS RRF =4.9 db consistent with previous results. However,

for the best jammers as the Eb /No > 7 dB there is a slight reduction of BER due to average

noise power decreases. However, above Eb /No =9 dB, as the noise decreases considerably,
its impact on BER becomes negligible in comparison to the errors caused by the jamming

power, generating a fixed BER that trends to a Pb =10−2 . This result is because at high Eb /No
and a fixed JSR the average noise power N trends to zero and the JSR becomes dominant.
Considering the theoretical BER for DSSS QPSK receiver according to Equation (3.14)
with equivalent noise power spectral density Noe = No + Jo yields :

 √( )
 
Pb ≈ Q 
2E b  ,
No + Jo 
(4.5)

as the average noise power N trend to zero, the Pb for high Eb /No , the Pb goes
asymptotically to:
√  √   √( )
 2Eb   E s   · 
s 
 = Q 
S T  .
lim =⇒ Pb ≈ Q   = Q 
J/W ss 
 (4.6)
N→0 Jo Jo

70
 0
10
Pulsed BPSK
MTJ
CSI

−1
10
BER (dB)

−2
10

−3
10
−15 −10 −5 0 5 10 15
JSRRF (dB)

Figure 4.10: Pulsed BPSK, MTJ and CSI, the most effective jammers for Eb /No =7 dB.

As a general analysis, the JS RDS values result in a variable reduction with respect to

the input JS RRF , depending on the jamming waveform, due to desired signal is despread

conserving its average energy with typical power losses after filtering that approximate to
10% comparing to the average energy before the bandpass filter, that normally in PSK

modulation corresponds approximately to 90% of the average power for the null-null
bandwidth [33]. On the other hand, the jamming signal is spread across the sequence
bandwidth limiting its average power not only because the processing gain but also because
of the despreading filter effect. This characteristic of DSSS explain that JSR decreases
reducing the jammings energy in favor of signal’s energy. As the JSR improvement

factor decreases it suggest the jamming waveform is more effective. The simulations
with optimum bandwidth for broad-band jammers in Table 4.3 denoted that the jamming

71
 0
10
DSSS/QPSK no jamming
MTJ
−1 Pulsed BPSK
10
NBN
CSI

−2
10
BER

−3
10

−4
10

−5
10
0 5 10 15 20 25
Eb/No (dB)

Figure 4.11: BER performance for the optimized jammers at JS RRF =3.1 dB.

improvement factor resulted in approximately 3 dB better than the values presented in

Table 4.1 for the non optimized jammers. Besides, the jamming improvement factors are
consistent with the most effective jammers found.
The bandwidth consideration is key to affect the performance of the simulated DSSS

QPSK receiver with matched filtering implementation. However there is not a specific
optimum bandwidth for each waveform and it requires simulations and comparisons. Under
this perspective the pulsed BPSK required less parameter variations to affect the receiver
performance, considering a variable duty cycle in function of the JSR.

72
 V. Conclusions

5.1 Research Contributions

This research provides a comprehensive characterization of intentional interference

and the evaluation of the effects in a DSSS QPSK receiver as a function of the JSR
degrading reference BER (Pb ) by one order-of-magnitude. Also jamming parameter
variation such as bandwidth and duty cycle of pulsed interferers was explored. Results
delivered give a reference for required simulated JSR to degrade a DSSS QPSK receiver
considering traditional jamming techniques available in open literature. Also this research

gives a simulated perspective of varying jamming waveform parameters and how the
jamming in DSSS can be optimized. The receiver model utilized for jamming assessment

was developed with the idea of implementing and varying parameters such as spreading

sequences or testing M-ary PSK and DPSK modulations with low complexity and good
flexibility. The receiver model has been validated with respective analytical Pb expressions

for binary and 4-ary DPSK and PSK cases.

5.2 Results Found

The DSSS QPSK simulated receiver result in a degraded performance of approxi-

mately 0.5 dB as a consequence of the RF and despreading filter incorporated to analyze

the bandwidth effects on the receiver before and after despreading the received signal. This

effect is due to filter coloration and signal attenuation. The SNR improvement from the
input (S NR sim ) to the output of the spreading filter (S NRDS ) was approximately 19.8 dB
and the processing gain measured from after the RF filter(S NRRF ) to the despreading filter
output (S NRDS ) was 14.87 dB, very close to the analytical processing gain for Gold code
of Nc =31 chips which is 10 log 10(31) ≈ 14.91dB.

73
 The tone jamming and the BPSK jammer were the least effective techniques with a
required JS RRF ≈ 13.3 and JS RRF ≈13.48 dB respectively.
Broadband jammers resulted in a better performance in comparison with the tone

jammers with slight variations, depending on the waveform simulated, because they
required JS RRF varied between 6-7.18 dB to degrade the BER to Pb =1.4 × 10−2 reference.
That is the case of LFMI, FHI, MTJ, RFMI and CSI. Regarding to noise jamming, the effect
of taking a smaller fraction of the noise bandwidth such as W J =W ss /8 and W J =W ss /16
reduce the required JS RRF by approximately 2 dB.

The pulsed BPSK jammer obtained the best performance between the jammers
analyzed with a required JS RRF ≈3.05 dB. However pulsed BBN required JS RRF ≈7.33
dB to reach the threshold that is even more than the broadband interferers discussed with

maximum values around JS RRF ≈7.1 dB. Generating pulsed jamming waveforms using

tone jamming significantly improved the jamming performance by approximately 9.6 dB,

whereas for pulsed LFMI improvement was just 0.24 dB. Regarding to pulsed noise the
simulation resulted in a poorer performance compared with NBN with a required JS RRF

close to 7.33 dB that is approximate 2.1 dB more.

Considering the differences between broad-band jammers and the best case of pulsed

jammers which was near to JS RRF ≈3 dB, the improvements achieved for noise jammers

by reducing the bandwidth and also the high JSR required for tone jamming, the effect
of bandwidth variation for broad-band jammers was explored. The analysis showed that
for W J ≈10 KHz to W J ≈40 KHz it was possible to optimize the jamming performance
and the jammers that generated more degradation were MTJ using a bandwidth W J ≈17.7
KHz and CSI with a bandwidth W J ≈22.1 KHz obtaining a required JS RRF ≈3.03 dB and
JS RRF ≈3.09 dB respectively. In regard to LFMI, RFMI the optimum bandwidth found was
W J ≈17.7 KHz with similar required JS RRF ≈5.08 dB and JS RRF ≈5.12 dB respectively.

74
With respect to FHI and NBN the optimum bandwidths required were W J ≈13.3 KHz and
W J ≈8.86 KHz respectively for a required JS RRF ≈4.9 dB for both jammers.
The effect of varying the duty cycle ρ for the pulsed jammers showed that ρ=0.21 was

the optimum duty cycle. However in terms of JS RRF the results do not vary significantly in
respect to the worst case pulsed jamming which has a duty cycle that changes as the inverse
of the E s /Jo ratio.
As a general conclusion the most effective jamming technique between those analyzed
in this research was pulsed BPSK. The jamming effectiveness from the baseline AWGN

interference denoted that broadband jamming waveforms can increase the degradation
when the bandwidth is chosen wisely.Using optimum bandwidth allowed decreasing the
required JSR by approximately 2 dB for BBN, RFMI, LFMI and FHI relative to the

same jammers with sub-optimal bandwidth. But also it was demonstrated, in a simulated

environment, that MTJ and CSI with optimum bandwidth result in a reduction in the

required JSR of approximately 4.1 dB and 3.6 dB respectively matching the best jammer
found, that is, pulsed BPSK with a variable duty cycle that follows the worst case analytical

expression in Equation (4.4) for pulsed jamming. Also in this research was demonstrated
that the processing gain expressed as a comparison between the JSR at the despreading
filter with respect to the JSR at the output of the RF filter increases when the jammer are
optimized, agreeing with the most effective jammers and it is approximately 3 dB lower for
the least effective jammers.
Finally it can be concluded that jammer bandwidth choice is the most important
parameter to degrade the performance of the DSSS QPSK with matched filtering
implementation. However the optimum bandwidth depends on the waveform chosen. The
optimum results implemented with Gold sequence Nc =31 chips require at least JS RRF ≈3
dB to degrade the receiver’s BER by one order-of-magnitude; this was the case for pulsed

BPSK, MTJ and CSI jammer with optimum bandwidths.

75
5.3 Recommendations for Future Research.

In this research several jamming waveform have been analyzed considering a DSSS
QPSK receiver using Gold code spreading. Besides, the interference assessment was
performed under the assumption of a perfect channel, synchronized receiver and supposing
a case where the receiver is stationary and the interferer approaches the receiver to induce
increased jamming power. Future research areas that can be considered are:

1. Analysis of the effect of comparing different spreading sequences for the jammers
evaluated in this research and the extension for long sequence to explore if the

optimum bandwidth and jamming improvement factor for broad-band jammers keeps

similar proportion for large sequences in respect to shorter sequences.

2. Analysis of the effect of error correction and interleaving over the most effective

jammers assessed in this research and what codes are more effective for particular

waveforms.

3. Study of repeater jammers and the effects on synchronization [21], assuming that the

jammer represent a delay copy of the desired signal. This topic involves developing

code and carrier acquisition and tracking module in order to estimate code delay and
frequency shit of interference signals.

4. Another area of research could include interference assessment under flat fading or

selective fading. It implies channel estimation under the cases that the channel phase
response is linear or when the channel gain and phase response varies with time and

frequency. This area of research requires adding channel estimation and additional
techniques for interference due to multipath such as channel equalization or rake
receiver implementation. In [34] is explored the performance of DSSS over flat
Rayleigh fading channels in single-tone interference further research could include
other jamming waveforms analyzed in this thesis. Another reference for performance
evaluation of PSK system in selective fading is studied in [35].

76
 Bibliography

[1] R.A. Scholtz, “The Origins of Spread-Spectrum Communications,” IEEE

Transactions on Communications, vol. 30, no. 5, pp. 822–854, 1982.

[2] R. Price, “Further Notes and Anecdotes on Spread-Spectrum Origins,” IEEE

Transactions on Communications, vol. 31, no. 1, pp. 85–97, 1983.

[3] M. J. Marcus, “Early Civil Spread Spectrum History,” [Online]. Available:

http://www.marcus-spectrum.com/page4/SSHist.html, 2011.

[4] A. Viterbi, “Spread Spectrum Communications - Myths and Realities,” IEEE

Communications Magazine, vol. 17, no. 3, pp. 11–18, 1979.

[5] B. Sklar, Digital Communications: Fundamentals and Applications, Prentice Hall

Communications Engineering and Emerging Technologies Series. Prentice-Hall PTR,
2001.

[6] G.R. Cooper and C.D. McGillem, Modern Communications and Spread Spectrum,
Series in Electrical Engineering: Communications and Signal Processing. McGraw-
Hill, 1986.

[7] D.J. Torrieri, Principles of Spread-Spectrum Communication Systems, Springer,

2005.

[8] S.S. Haykin, Communication systems, Wiley, 2001.

[9] M. Skolnik, Radar Handbook, Third Edition, Electronics Electrical Engineering.

McGraw-Hill Education, 2008.

[10] C.E. Shannon, “Communication in the Presence of Noise,” Proceedings of the IRE,
vol. 37, no. 1, pp. 10–21, 1949.

[11] R. Poisel, Modern Communications Jamming: Principles and Techniques,

Communications engineering. Artech House, Incorporated, 2004.

[12] M.K. Simon, Spread Spectrum Communications, Number v. 1 in Electrical

engineering communications and signal processing series. Computer Science Press,
1985.

[13] R. L. Peterson, R.E. Ziemer, and D.E. Borth, Introduction to Spread Spectrum
Communications, Prentice Hall International, 1995.

[14] A. Wood and J.A. Stankovic, “Denial of Service in Sensor Networks,” Computer,
vol. 35, no. 10, pp. 54–62, 2002.

77
[15] Wenyuan Xu, W. Trappe, Y. Zhang, and T. Wood, “The Feasibility of Launching and
Detecting Jamming Attacks in Wireless Networks,” in Proceedings of the 6th ACM
international symposium on Mobile ad hoc networking and computing, New York,
NY, USA, 2005, MobiHoc ’05, pp. 46–57, ACM.
[16] Yao Liu, Peng Ning, Huaiyu Dai, and An Liu, “Randomized Differential DSSS:
Jamming-Resistant Wireless Broadcast Communication,” in Proceedings IEEE
INFOCOM, 2010, pp. 1–9.
[17] M. Wilhelm, I. Martinovic, J. B. Schmitt, and V. Lenders, “Short Paper: Reactive
Jamming in Wireless Networks: How Realistic is the Threat?,” in Proceedings of the
Fourth ACM Conference on Wireless Network Security, New York, NY, USA, 2011,
WiSec ’11, pp. 47–52, ACM.
[18] Y. Bo, Sh. Gao-ping, and S. Hong-sheng, “An Automatic Interference Recognition
Method in DSSS Communication System Based on SVM,” in 2nd International
Conference on Computer Engineering and Technology (ICCET), 2010, vol. 3, pp.
V3–4–V3–8.
[19] Kewu Huang and Ran Tao, “Interference Suppression in Spread Spectrum Systems
Using the CR-Fractional Fourier Transform,” in First International Conference on
Pervasive Computing Signal Processing and Applications (PCSPA), 2010, pp. 560–
563.
[20] M. Tilal and R. Minhas, Effects of Jamming on IEEE 802.11p Systems, M.S. Thesis,
Chalmers Univ. of Technology, Gothenburg, Sweden, November 2010.
[21] J. Guo H.Wang and Z. Wang, “Evaluation of Security for DSSS Under Repeater
Jamming,” in IEEE International Conference on Communications, ICC ’07., 2007,
pp. 5525–5530.
[22] A. Yousaf and A. Loan, “Effect of Jamming Technique on the Performance of Direct
Sequence Spread Spectrum Modem,” in 13th International Conference on Advanced
Communication Technology (ICACT), 2011, pp. 874–877.
[23] S.M. Kay, Fundamentals of Statistical Signal Processing: Detection Theory,
Prentice Hall Signal Processing Series. Prentice-Hall PTR, 1998.
[24] The MathWorks Inc., “Communications System Toolbox Documentation,
R2013b,” [Online]. Available: http://www.mathworks.com/help/comm/ug/digital-
modulation.html, 2013.
[25] M. Temple, “Class Project, Synchronous DSSS Interference Suppression , EENG673
Spread Spectrum Communications,” School of Engineering and Management, Air
Force Institute of Technology, Wright-Patterson AFB OH, Sumer Quarter 2013.
[26] J.G. Proakis, Digital Communications, McGraw-Hill Series in Electrical and
Computer Engineering. McGraw-Hill Education, 2001.

78
[27] M. K. Simon and M.S. Alouini, “A Unified Approach to the Probability of Error
for Noncoherent and Differentially Coherent Modulations over Generalized Fading
Channels,” IEEE Transactions on Communications, vol. 46, no. 12, pp. 1625–1638,
1998.

[28] M.K. Simon and M.S. Alouini, Digital Communication over Fading Channels: a
Unified Approach to Performance Analysis, Wiley series in telecommunications and
signal processing. John Wiley & Sons, 2000.

[29] L. Pralon, B. Pompeo, G. Beltrao, H. Cioqueta, B. Cosenza, and J.M. Fortes, “Ran-
dom Phase/Frequency Modulated Waveforms for Noise Radar Systems Considering
Phase Shift,” in 9th European Radar Conference (EuRAD), Oct 2012, pp. 314–317.

[30] M. Temple and S. Stone, “Class handout, EENG673 Spread Spectrum Communica-
tions,” School of Engineering and Management, Air Force Institute of Technology,
Wright-Patterson AFB OH, Sumer Quarter 2013.

[31] Sune R J Axelsson, “Noise Radar Using Random Phase and Frequency Modulation,”
IEEE Transactions on Geoscience and Remote Sensing,, vol. 42, no. 11, pp. 2370–
2384, Nov 2004.

[32] N. Levanon and E. Mozeson, Radar Signals, Wiley, 2004.

[33] D.A. Guimaraes, Digital Transmission: A Simulation-Aided Introduction with

VisSim/Comm, Signals and Communication Technology. Springer, 2010.

[34] Z. Guo-zhen, G. Yuan-Yuan, and M. Jing, “Performance Analysis of the Cooperative

DS/SS Systems in Single-tone Interference Over Flat Rayleigh Fading Channels,” in
International Conference on Communications, Circuits and Systems (ICCCAS), July
2010, pp. 126–130.

[35] T.L. Staley, R.C. North, Jianxia Luo, W.H. Ku, and J.R. Zeidler, “Performance
Evaluation for Multichannel Reception of Coherent MPSK over Frequency Selective
Fading Channels,” IEEE Transactions on Vehicular Technology, vol. 50, no. 4, pp.
877–894, Jul 2001.

79
 Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704–0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including
suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704–0188), 1215 Jefferson Davis Highway,
Suite 1204, Arlington, VA 22202–4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection
of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.

1. REPORT DATE (DD–MM–YYYY) 2. REPORT TYPE 3. DATES COVERED (From — To)

27–03–2014 Master’s Thesis Oct 2013–Mar 2014
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER

5b. GRANT NUMBER

Simulated Assessment of Interference Effects in Direct Sequence Spread
Spectrum (DSSS) QPSK Receiver
5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

Rojas, Luis S., Captain, Chilean Air Force
5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT

NUMBER
Air Force Institute of Technology
Graduate School of Engineering and Management (AFIT/EN)
2950 Hobson Way AFIT-ENG-14-M-64
WPAFB, OH 45433-7765
9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

Intentionally Left Blank 11. SPONSOR/MONITOR’S REPORT

NUMBER(S)

12. DISTRIBUTION / AVAILABILITY STATEMENT

Distribution Statement A: Approved for Public Release; Distribution Unlimited
13. SUPPLEMENTARY NOTES
This work is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
14. ABSTRACT
This research developed and validated a generic simulation for a direct sequence spread spectrum (DSSS), using
differential phase shift keying (DPSK) and phase shift keying (PSK) modulations, providing the flexibility for assessing
intentional interference effect using DSSS quadrature phase shift keying receiver (QPSK) with matched filtering as a
reference. The evaluation compares a comprehensive pool of jamming waveforms at pass-band that include continuous
wave (CW) interference, broad-band jamming, partial-band interference and pulsed interference. The methodology
for jamming assessment included comparing the bit error rate (BER) versus required jamming to signal ratio (JSR)
for different interferers using the Monte Carlo approach. This thesis also analyzes the effect of varying the jammer
bandwidth for broad-band jammers including broad-band noise (BBN), frequency hopping interference (FHI), comb-
spectrum interference (CSI), multi-tone jamming (MTJ), random frequency modulated interference (RFMI) and linear
frequency modulated interference (LFMI). Also, the effect of changing the duty cycle for pulsed CW waveforms is
compared with the worst case pulsed jamming equation. After the evaluation of different interferers, the research
concludes that pulsed binary phase shift keying (BPSK) jamming is the most effective technique, whereas the CW
tone jamming and CW BPSK interference result are least effective. It is also concluded that by finding an optimum
bandwidth, FHI and BBN improves the required JSR by approximately 2.1 dB, RFMI and LFMI interference by 0.9 and
1.5 dB respectively. Alternately, MTJ and CSI improves their effectiveness in 4.1 dB and 3.6 dB respectively, matching
the performance of the pulsed BPSK jammer.
15. SUBJECT TERMS
DSSS, QPSK, Jamming, Interference assessment, BBN, FHI, CSI, LFM, RFMI.
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
ABSTRACT OF Dr. Richard K. Martin (ENG)
a. REPORT b. ABSTRACT c. THIS PAGE PAGES
19b. TELEPHONE NUMBER (include area code)
U U U UU 95 (937)785-3636 xx4625 [email protected]
Standard Form 298 (Rev. 8–98)
Prescribed by ANSI Std. Z39.18