Department of Science and Technology Institutionen för teknik och naturvetenskap Linköping University Linköpings universitet

gnipökrroN 47 106 nedewS ,gnipökrroN 47 106-ES

LiU-ITN-TEK-A--18/027--SE

Methods for locating signaljammers with a UAV

Andreas Höggren

Love Lindmark

2018-06-13

LiU-ITN-TEK-A--18/027--SE

Methods for locating signaljammers with a UAV

Examensarbete utfört i Elektroteknikvid Tekniska högskolan vid

Linköpings universitet

Andreas HöggrenLove Lindmark

Handledare Magnus KarlssonExaminator Qin-Zhong Ye

Norrköping 2018-06-13

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –under en längre tid från publiceringsdatum under förutsättning att inga extra-ordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat förickekommersiell forskning och för undervisning. Överföring av upphovsrättenvid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning avdokumentet kräver upphovsmannens medgivande. För att garantera äktheten,säkerheten och tillgängligheten finns det lösningar av teknisk och administrativart.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman iden omfattning som god sed kräver vid användning av dokumentet på ovanbeskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådanform eller i sådant sammanhang som är kränkande för upphovsmannens litteräraeller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press seförlagets hemsida http://www.ep.liu.se/

Copyright

The publishers will keep this document online on the Internet - or its possiblereplacement - for a considerable time from the date of publication barringexceptional circumstances.

The online availability of the document implies a permanent permission foranyone to read, to download, to print out single copies for your own use and touse it unchanged for any non-commercial research and educational purpose.Subsequent transfers of copyright cannot revoke this permission. All other usesof the document are conditional on the consent of the copyright owner. Thepublisher has taken technical and administrative measures to assure authenticity,security and accessibility.

According to intellectual property law the author has the right to bementioned when his/her work is accessed as described above and to be protectedagainst infringement.

For additional information about the Linköping University Electronic Pressand its procedures for publication and for assurance of document integrity,please refer to its WWW home page: http://www.ep.liu.se/

© Andreas Höggren, Love Lindmark

Abstract

Wireless communication today is a modern way to transport data from one location to another.One of the drawbacks of this feature is that a signal jammer can disrupt communications betweenthe receiver and transmitter since the radio waves travel in the open air. This drawback can beexploited in both military and civilian applications.

This thesis will aim on how to locate this kind of transmitting signal jammer over an open fieldusing an Unmanned Aerial Vehicle (UAV) that searches the designated area with the assumptionof line of sight between the UAV and the transmitting jammer. Two different methods will beinvestigated with different techniques, antennas and flight patterns.

iii

Acknowledgments

We would like to thank FOI for giving us the opportunity to do our master thesis at FOI at thedepartment of Radio Electronic Warfare Systems. We would like to thank our supervisors LeifFestin and Henrik Eriksson and their colleagues Patrik Hedström, Per Brännström, Greger Hast,Anders Johansson, Rolf Jonsson, Joakim Rydell, Michael Tulldahl and all the other great mindsat FOI. We would also like to thank our supervisor Magnus Karlsson and our examiner Qin-ZhongYe at Linköping University for their time.

iv

Contents

Abbreviations viii

1 Introduction 1

1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Signal jammers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.2 Usage of signal jammers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.3 Usage of multirotor drones . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Related work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Radio jamming interference 5

2.1 Different jamming methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1.1 Frequency-hopping and sweeps . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2 Jammer for field tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Localization methods 7

3.1 Method 1: RSSI using dipole antenna only . . . . . . . . . . . . . . . . . . . . . . . 73.2 Method 2: RSSI positioning using directional antenna only . . . . . . . . . . . . . 8

3.2.1 Estimating the jammer direction . . . . . . . . . . . . . . . . . . . . . . . . 93.2.2 Filtering out noisy rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.2.3 Triangulation ground test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Radio propagation theory 13

4.1 Attenuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.1.1 Friis equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4.2 Reflection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.3 Multipath propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4.3.1 Rician fading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.3.2 Rician K-factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4.4 Received Signal Strength Indication . . . . . . . . . . . . . . . . . . . . . . . . . . 144.5 Radiation pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144.6 Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5 Simulation and measurement environments 17

5.1 Introduction to the user interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175.2 Antenna profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175.3 Radiation pattern for the UAV platform and antennas . . . . . . . . . . . . . . . . 18

5.3.1 Reflection test at FOIs courtyard . . . . . . . . . . . . . . . . . . . . . . . . 185.3.2 Anechoic test chamber tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

6 Reflection measurements 21

6.1 Dipole antenna measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216.2 Directional antenna measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

7 Equipment for RSSI measurements 27

7.1 The directional antenna on the UAV . . . . . . . . . . . . . . . . . . . . . . . . . . 277.2 The dipole antenna on the UAV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277.3 MikTran platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

7.3.1 MikTran software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287.4 Raspberry Pi 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

8 Test flights 31

8.1 UAV equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318.2 First flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328.3 Second flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348.4 Third flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

v

9 Results 39

9.1 Antenna performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399.1.1 Dipole antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399.1.2 Directional antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

9.2 GPS accuracy and reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409.3 Performance of localization methods . . . . . . . . . . . . . . . . . . . . . . . . . . 40

9.3.1 LS algorithm using dipole antenna . . . . . . . . . . . . . . . . . . . . . . . 409.3.2 Triangulation method in the courtyard . . . . . . . . . . . . . . . . . . . . . 419.3.3 Triangulation method during flight . . . . . . . . . . . . . . . . . . . . . . . 42

10 Conclusion 45

10.1 Localization based on real flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4510.2 Reliability of the user interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4510.3 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

11 Future work 47

11.1 Frequency-hopping implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 4711.2 UAV improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4711.3 RSSI using directional and dipole antennas . . . . . . . . . . . . . . . . . . . . . . 4811.4 Autonomous navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4811.5 Further UAV reflection tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4811.6 Avoid hostile danger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

vi

List of Figures

1 Radiation pattern and appearance for the Katherein diple antenna . . . . . . . . . 52 Flight principle of the method 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Localization algorithm plot in MATLAB . . . . . . . . . . . . . . . . . . . . . . . . 84 Principle of the triangulation method . . . . . . . . . . . . . . . . . . . . . . . . . . 95 UAV with the mounted directional antenna . . . . . . . . . . . . . . . . . . . . . . 96 Example of the 3/4 dB method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Comparison of a clean and noisy rotation measurement . . . . . . . . . . . . . . . 108 Different antenna radiation patterns . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Cross-polarization from test chamber investigation . . . . . . . . . . . . . . . . . . 1510 The user interface in MATLAB while simulating . . . . . . . . . . . . . . . . . . . 1711 Improvised outdoor test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1812 UAV in the anechoic test chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . 1913 3D radiation plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2114 Dipole rotation test without extra absorbing material . . . . . . . . . . . . . . . . 2215 Comparing dipole rotation tests with and without absorbing material. Elevation

angle 10◦. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2216 Comparing dipole rotation tests with and without absorbing material. Elevation

angle 45◦. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2317 Dipole antenna elevation sweep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2318 Final modification for the dipole antenna . . . . . . . . . . . . . . . . . . . . . . . 2419 Directional antenna elevation sweep . . . . . . . . . . . . . . . . . . . . . . . . . . 2420 EMC chamber test with the directional antenna rotated on full rotation in azimuth 2521 The directional antenna and installation on the UAV . . . . . . . . . . . . . . . . . 2722 Radiation patterns of the directional antenna . . . . . . . . . . . . . . . . . . . . . 2723 The appearance of the dipole antenna . . . . . . . . . . . . . . . . . . . . . . . . . 2824 MikTran . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2825 Rasberry Pi 3 Model B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2927 Block diagram of UAV hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3228 GPS positions from the first flight . . . . . . . . . . . . . . . . . . . . . . . . . . . 3329 The UAVs antenna equipment for the first flight . . . . . . . . . . . . . . . . . . . 3330 The fixed position of the dipole antenna for the second flight . . . . . . . . . . . . 3431 GPS positions from the second flight with dipole antenna . . . . . . . . . . . . . . 3532 GPS positions from the second flight with directional antenna . . . . . . . . . . . . 3633 UAV with the mounted directional antenna . . . . . . . . . . . . . . . . . . . . . . 3634 The two flight routes done on the 7th of June . . . . . . . . . . . . . . . . . . . . . 3735 Directional antenna RSSI drop graph compared to Friis . . . . . . . . . . . . . . . 3936 MikTran GPS test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4037 UAV RSSI from the second flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4138 Polar plot of yaw-data heading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4139 Triangulation test at FOIs courtyard . . . . . . . . . . . . . . . . . . . . . . . . . . 4240 Third flight triangulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4341 Final flight GPS log . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

vii

Abbreviations

Notation Description

ADC Analog-to-Digital ConverterAPI Application Programming InterfaceARM Advanced RISC Machine

CCW CounterclockwiseCW Continuous Wave

EMI Electromagnetic InterferenceEW Electronic Warfare

FOI Swedish Defence Research Agency

GPS Global Positioning SystemGUI Graphical User Interface

LIDAR Light Detection And RangingLOS Line Of SightLS Least-Square

MAVLink Micro Air Vehicle LinkMP Mission Planner

PDF Probability Density FunctionPDOP Position Dilution of Precision

RF Radio FrequencyRSSI Received Signal Strength Indication

SDR Software-defined RadioSNR Signal to Noise RatioSoC System-on-a-ChipSSH Secure Shell

UAV Unmanned Aerial VehicleUDP User Datagram Protocol

viii

1 Introduction

This master thesis is performed for the Swedish Defence Research Agency (FOI) at the departmentof Radio Electronic Warfare Systems. This section describes the background of signal jammersand this master thesis goals and limitations.

1.1 Background

Wireless communication today is expanding fast as the need for a connection between larger dis-tances for communicating systems or individuals is essential. This thesis work will investigate ifa UAV equipped with the proper equipment can localize a signal jammer that transmits a knownfrequency and a constant radiated power.

1.1.1 Signal jammers

A signal jammers purpose is to disrupt a communication between two communicating devices. Asignal jammer is a device that’s being used to transmit a signal that has the same frequency asthe wireless communication system which is the victim of this operation.

1.1.2 Usage of signal jammers

The catch with this type of communication is that the wireless version travels through the air,which leaves the traveling signal vulnerable to interference. This interference is an electromagneticinterference (EMI). Signal jammers are often used in combat zones to disrupt communicationsof the opposing force. An example of this is the Russian developed vehicle Borisoglebsk-2 whichdisrupts wireless HF, UHF and mobile communications [1]. This is called Electronic Warfare (EW)and this vehicle has been involved in the annexation of Crimea in Ukraine [1].There are other cases than military purposes, an example of a civil case is the Swedish police inStockholm, Sweden, reports illegal use of signal jammers by thieves to block the signal from thecar keys, leaving the vehicle unlocked and therefore easy to break into [2].

1.1.3 Usage of multirotor drones

The different applications of UAVs in terms of drones, has evolved fast through the years, both froma military and civil perspective. Military use of multirotor UAVs is rapidly growing. Since they arelight, relatively cheap and easy to use they are perfect for tasks such as reconnaissance, sendingor receiving signals from a higher ground, and even for reducing morale by sending propagandaand misinformation. UAVs has through the years performed certain deliveries from one place toanother, companies such as Amazon is involved in this area among others. Another example isthat Apple has an idea in how to improve Apple Maps after a several year long struggle to get theapplication to work, which is to let drones capture aerial images [3].

1.2 Goals

The goal of this master thesis is to investigate different UAV-related methods for determining theposition of a signal jammer, which in this scenario is a stationary continuous wave (CW) signalsource. The transmitting signal is at a fixed frequency and an unknown power.

As a summary, this thesis will cover:

• Performance investigation of antennas and considerations.

• Development of a simulation environment to test positioning methods.

• Evaluation and comparison of signal source positioning methods.

1.3 Limitations

Some limitations are present in this thesis, such as implementation requirements placed by FOI.These are described in detail below.

1

Signal jammer

A signal generator is used to generate a sine CW signal at a frequency of 1877 MHz and 17 dBmradiated power which is transmitted through a dipole antenna. A weaker signal shrinks the regionwhere the UAV can detect and find the signal jammer accordingly.

UAV Platform

An existing UAV platform is to be used with some limitations regarding weight and physical size ofthe payload. The flight controller is a Pixhawk Mini [4] and any communication with it is thereforelimited to the Micro Air Vehicle Link (MAVLink) interface [5].

Receiver

One of the requirements of the master thesis is that the receiver uses a Software-Defined Radio(SDR) such as FOI’s MikTran platform.

Since the UAV platform is already designed, no fabrication is required. The antennas for thethesis are either already acquired or bought from third-party sources. The hardware platform thatwill perform the recordings from the UAV deck is also already developed and is called MikTran.The hardware part will therefore mostly consist of choosing appropriate antennas and mountingthem with minimal interference.

1.4 Related work

There has previously been similar research regarding this field where a UAV is equipped withantennas in order to acquire received signal strength indication (RSSI) data and give headingestimations. UAVs has also been equipped with other gear than antennas, such as 3D LightDetection And Ranging (LIDAR) sensors to investigate points of interest.

1. Körner et al. [6] discusses the improvement of locating radio tagged wild animals by aUAV with the mounted directional antenna. Their method offers a number of advantagescompared to search from the ground, including better line-of-sight (LOS) signal reception,terrain-independence and faster localization.

2. FOI has performed projects by using UAVs several times. One of them consist of evaluatingthe benefits and capabilities of high resolution 3D data by a UAV equipped with LIDAR.Application examples is a way to fast perform a geometric documentation of an interestingarea [7].

1.5 Method

This thesis work process can be described by the following five steps:

Background

A theoretical background is an essential vantage point of the understanding in how Radio Frequency(RF) behaves in the environment and this thesis aim of focus. Therefore a background studyregarding this field is necessary, see Chapter 2.

Theory

Includes theory for antenna radiation patterns and equations regarding wireless communicationsystems. This is presented in Chapter 3 and Chapter 4.

Implementation

When the pre-study phase is passed, implementation of algorithms and equations are added to theMATLAB Graphical User Interface (GUI), and are tested in real flights. See the relevant Chapter3, Chapter 5, Chapter 6 and Chapter 7.

2

UAV flights

Every UAV test flight during the entire thesis project is presented in Chapter 8. It covers informa-tion regarding UAV equipment, flight patterns as different methods of localization and thoughtsof improvement.

Results

When the UAV-simulation GUI in MATLAB is advanced enough, the received power characteristicswill be compared of that of an actual flight with a signal jammer in the field. The UAV will be testedwith different antennas and give position estimations. The RSSI measurements are presented inChapter 9 and discussion regarding the results are presented in Chapter 10 as conclusions. Finallysome thoughts regarding improvements as future work are presented in Chapter 11.

3

2 Radio jamming interference

Radio jamming occurs when a transmitting device gets overpowered by a signal jammer that istuned to the same frequency as the targeted transmitter. The receiving antenna will then beinterfered and therefore the communication between a receiver and transmitter is disrupted.

2.1 Different jamming methods

There are several kinds of signals when it comes to signal jamming, for example continuous wavesignals and modulated signals. A continuous wave is defined as a constant frequency and power asa simple sine wave while modulated signals vary in amplitude or frequency. Modulated signals cantherefore be harder to detect since the jamming signal may appear as a regular communication toa receiver. This thesis will focus solely on localization of CW signal jammers.

2.1.1 Frequency-hopping and sweeps

Frequency-hopping is a method for transmitting a signal at different frequency bands by adjustingthe carrier frequency at a programmed pace and in a pseudorandom order. The order of frequencyshifts are pre-programmed for both transmitter and receiver. This is a way to avoid a communica-tion being interrupted or eavesdropped by a third part. There are frequency-hopping systems thatbehave more like a frequency sweep rather than a random pattern. These electronics has a usagefor civil purposes while more advanced systems like military radios attempt a more pseudorandompattern [8]. However, there are jammers that counter this sort of techniques and solves the systemsrandom pattern algorithm.

2.2 Jammer for field tests

The signal generator that’s being used to generate a Continuous Wave (CW) jamming signal isthe R&S SMC100A signal generator [9]. The signal generator transmits a sine-wave signal at 1877MHz with a transmitted power of 17 dBm. The R&S signal generator is connected to a Katherein738 449 dipole antenna mounted vertically 1.5 meters above ground on a tripod [10]. See Figure1a for the jamming antennas radiation pattern along the vertical plane. The appearance of thejamming antenna is presented in Figure 1b.

(a) (b)

Figure 1: The Katherein dipole antenna. (a) The radiation pattern of the Katherein antenna that’sbeing used in combination with the R&S signal generator [10]. (b) The Katherein antenna thatcan be mounted on top of a tripod for field tests [10]

5

3 Localization methods

Two different methods for localizing the jammer were evaluated.The first method uses a dipoleantenna to provide RSSI data, while the second method uses a directional antenna instead. Bothof these methods require a certain flight navigation to determine a possible signal source location.The reference as a true location of the signal jammer was with a smartphones Global PositionSystem (GPS) coordinate by the jammer. This method of the true jamming location was used forall flights.

3.1 Method 1: RSSI using dipole antenna only

The first method of jammer localization was the dipole antenna as receiver that was mounteddirected downwards underneath the UAV and the UAVs flight pattern was similar to a wide zigzagpattern across the field of interest. See Figure 2. The idea of this method is that it could provide anexact position if the general area the jammer is located in is known. Two different dipole antennamountings were evaluated in this thesis. The satellite imagery of FOIs courtyard are acquired fromEniro [11].

Figure 2: The marked cross represent an example of the signal jammer location while the whitedrawn line illustrates the flight pattern of the first method.

This method has previously been used for indoor localization and uses the optimization algo-rithm from [12] which estimates the source position based on the estimated distance to the sourceat a number of anchor nodes. This algorithm is from here on called the unconstrained least-square(LS) estimate. In this case, the anchor nodes is the UAV’s positions along the flight path and theestimated distance is the RSSI value at each position. Equation 1 gives the estimated position qas a 2× 1 matrix.

q =1

2(AS)+At, A = S′PwWPd (1)

Where (AS)+ is the pseudoinverse of AS, Pd is I(N − 1) − dd′

d′d, S is a 2 × N matrix of all

position coordinates, W is I(N − 1), t is R2 − d2, R is the baseline distance of all N positionsfrom the first position, d is the reported distance (RSSI) difference of all positions from the firstposition and I is the identity matrix.

It is not possible to make absolute distance estimations from RSSI values if the jamming signalstrength is not known, since a received signal of a particular strength could be a weak signal froma short distance or a strong signal from a long distance. The received power is the key elementwhen estimating the jammer location. There are many kinds of algorithms regarding estimationsof a signal source when using several sensors or anchor nodes as they also are called. These anchornodes will correspond to the UAVs measurement points during flight. From the received signal

7

strength variation the distance variation can be estimated and the location of said estimation canbe found from the GPS-coordinate log. See Figure 3 of a MATLAB simulation of the algorithm.

Figure 3: The unconstrained LS estimate with 25 measurement points. The jammer is locatedwith perfect precision since all distance measurements are ideal.

3.2 Method 2: RSSI positioning using directional antenna only

For the second method, the UAV is equipped with a directional (panel) antenna mounted un-derneath at a vertical angle of 39◦. A vertical angle between 30◦ to 45◦ is desired to make thedirectional antenna able to receive signals from a large distance span with minimum attenuation.While logging the current heading (yaw) from the flight-controller as well as the GPS-position andreceived signal strength from MikTran, the UAV rotates one full rotation along its yaw-axis whileremaining stationary. The Analog-to-Digital Converter (ADC) on the MikTran platform measures1500 RSSI values approximately 8 times a second and logs the max value of these 1500 values.The Raspberry Pi 3 supplies the current UAV heading at a sample rate of 10 samples per second.

The yaw-data gives coordinates from 0◦ to 359◦, where 0◦ is North and 90◦ is East and so on,all related to true North. Flying to a second point perpendicular to the direction of the jammerand performing another yaw-rotation will give another jammer direction. These two directions willintersect at the jammers position. See Figure 4 of the triangulation method.

Since the directional antenna will be the first area of contact to the ground when performinglandings, a set-up of boxes were placed in a square on the ground for the UAV to land flatly onto avoid tilting and break any of its propellers. See Figure 5 for the mounted directional antennawhile the UAV is airborne.

8

Figure 4: This is the principle of the triangulation method that is described. The UAV willcommence its flight by performing one full rotation while receiving RSSI data from the directionalantenna. The UAV will then navigate towards a second coordinate and perform the same rotationonce more. The crossing of the two RSSI headings will give the jamming location.

Figure 5: UAV with the mounted directional antenna while the drone is airborne at a low altitude.

The UAV will perform a full rotation in the air at an average rotation speed of approximately20◦/sec and save the heading which receives the highest RSSI. The rotation pace of the UAV isset to the lowest possible by the UAV pilot. When the rotation is accomplished and one headingnoted, the UAV will travel to another GPS coordinate perpendicular to the heading found in thefirst rotation and repeat the full rotation scan. These two lines with their respective locationand direction will eventually cross at the jammers approximate location. This kind of localizationmethod fall under the triangulation category and is a typically known localization method inposition estimations from RSSI.

3.2.1 Estimating the jammer direction

Ideally, the heading of the maximum RSSI will be the direction to the jammer since the radiationpattern of the directional antenna is strongest in the straight forward direction. In reality, thispeak can be found anywhere between −30◦ to +30◦ of the main lobe. A more reliable method isto use the mean or median between the two angles where the signal is -3/-4 dB lower than themax, since the main lobe is much wider and more defined than the absolute peak [13]. See Figure6.

9

Figure 6: Example of the various methods of estimating the direction of the jammer. Note thatthe max measurement direction differs quite a lot from the true direction and that the -3/-4 dBmethod provide better accuracy.

3.2.2 Filtering out noisy rotations

While the UAV is airborne and performing a full rotation, wind can cause it to tilt, rotate atuneven speed and gain or lose altitude rapidly. Therefore some of the rotations may become noisyand as a result provide less accurate estimates of the jammer’s direction. By setting a maximumallowed RSSI variance of at least 5 in the -4 dB region the noisy and inaccurate rotations can befiltered out. See Figure 7.

(a) (b)

Figure 7: Comparison of a clean rotation measurement and a noisy rotation measurement. (a) Theclean rotation. (b) The noisy rotation, which is filtered out during jammer location estimationsbecause of the large variance found in the -4 dB region.

10

3.2.3 Triangulation ground test

While developing the triangulation method, a series of ground tests were performed in the court-yard. The triangulation method with the directional antenna mounted on the UAV was tested onFOIs courtyard with the UAV held at approximate 1.5 meters above ground. The jamming dipoleantenna transmitted the CW signal while the UAV was displaced on foot at different locations androtated at a slow rotation speed. The RSSI-data is then evaluated in MATLAB to investigate if themaximum power peak is having the correct bearing towards the jamming source. The evaluationof the RSSI data are presented in Chapter 9.3.1.

11

4 Radio propagation theory

Background knowledge for multipath propagation is needed for the simulation environment as wellas for understanding the measured results. This chapter will describe some parameters that areincluded in our simulations for the measurements in radiated power.

4.1 Attenuation

The power reduction based on the distance in free space is know to be 1/R2 path loss factor fromFriis equation. Reflected signals from the terrain may increase the path loss to 1/Rn. Where thepotency n can vary from 2 to 5 or 6 when exposed to many lossy obstructions [14]. The decreaseof signal power caused by losses in propagation paths is called attenuation. This loss depends onthe distance between receiving and transmitting antennas, interference, multipath propagation andreflections for this thesis. Other factors include atmosphere and precipitation [14].

4.1.1 Friis equation

Friis equation expresses the received signal as a function of transmitted power, antenna gain, rangeand frequency and is essential in wireless communication systems. See Equation 2 for the powerdensity radiated by an isotropic antenna. Pt is the transmitted power and R is the distance betweenthe two antennas. The power density Savg is defined in W/m2 and is the free space path loss [14].

Savg =Pt

4πR2W/m2 (2)

The received power, Pr to the receiving antenna can be derived from Friis equation. Ae is theeffective aperture area of receive antenna, it can also be called as "capture area". See Equation 3.

Pr = AeSavg =GtPtAe

4πR2W (3)

4.2 Reflection

Transmitted electromagnetic waves have a tendency to have its signal reflected, which leads to thatthe signal takes different paths to the corresponding receiver. This phenomenon is depending onwhat kind of surface the signal reaches. A communication on a short distance will most likely beaffected by metallic objects or big environment objects surfaces, such as buildings. Wet objectssuch as oceans provides reflection at longer ranges [15].

4.3 Multipath propagation

Multipath propagation is a phenomenon that occurs when signals reflect on various obstacles inthe environment which causes reflected signals to take multiple paths to the receiver, and thereforearrive with different delays and signal strengths. This can cause constructive or destructive inter-ference depending on the path delays of the environment. Destructive multipath propagation iscalled multipath fading [16]. Accurate multipath propagation simulation requires a detailed mapof the surrounding landscape and a lot of computational power, however there are a couple ofstatistical approximations which are quicker to calculate. For LOS-signals, Rician fading is thebest fit.

4.3.1 Rician fading

For LOS signals, the fading can be statistically modeled by Rician channel fading which is basedon the Rician probability density function (PDF). The Rican fading model takes a dominant signalwave into account, typically the LOS signal wave since it has no blocking obstacle. The signal isaffected by reflections from the environment and can cause constructive- or destructive interference.I0 is the zeroth-order of the Bessel function of the first kind. See Equation 4 [17].

f(r) =r

σ2exp (−

r2 + k2d2σ2

)I0rk2dσ2

, r ≥ 0 (4)

13

Equation 5 can be used to estimate the Rician K-factor, where A is defined as the amplitudegiven for an incoming signal, I1 is the first order modified function of the first kind, E[A] is theaverage amplitude and E[A2] is the average of the amplitude squared [17].

E[A]

E[A2]=

√

π

4(K + 1)exp (−

K

2× [(K + 1)I0(

K

2) +KI1(

K

2)] (5)

4.3.2 Rician K-factor

The K-factor describes the Rician distribution as the ratio between deterministic signal power(LOS) and indirect paths. The amount of fading can be estimated in dB. See Equation 6 [17].

K(dB) = 10log10k2d2σ2

(6)

kd is a constant that relates to the direct components strength and σ is the standard deviation.The K-factor can also vary if the receiver is in motion relative to the transmitter. But this isneglected since the UAV will keep a low velocity.

4.4 Received Signal Strength Indication

RSSI is a relative measurement of received power and indicates the signal strength between atransmitter and receiver. The RSSI has an important role regarding this thesis since these mea-surements can be used for the localization of the signal source. The Signal to Noise Ratio (SNR)is useful for RSSI estimations when running background noise to the simulation environment ofthe GUI. It gives the ratio of the signal energy and noise energy level. See Equation 7 [18].

SNR(dB) = 10log10Psignal

Pnoise

(7)

The SNR(dB) is a value in decibels, Psignal determines the average power of the signal andPnoise is the average power of the noise.

4.5 Radiation pattern

The field strength pattern of a transmitting or receiving antenna can be illustrated in a plot basedon either Cartesian coordinates or polar coordinates. In this thesis the radiation patterns areshown in polar plots. These plots gives information on how the radiation expands in gain (dBi)for different directions around the antenna and these areas are named lobes. A low gain antennawill typically emit a radiation pattern uniformly in all directions, while a high gain antenna willfocus its main lobe in a particular direction. The lobe(s) with the highest gain is called the mainlobe and most of the antennas radiated energy will be directed in this direction. Good examplesof high- and low-gain antennas are the directional- and dipole antennas respectively. See Figure8a of a dipole antenna’s radiation pattern and 8b of a directional antenna’s radiation pattern. The3D pattern illustration of the directional antenna is based on the given H- and E-fields from thedistributors data-sheet [19], see Figure 22.

14

(a) (b)

Figure 8: The simulated 3D radiation pattern plots of the two antennas used on the UAV. Both aredrawn with the same point of view with a 27◦ elevation rotation and both are vertical polarized andthe colorbars represents the power in dB-scale. The simulations are from the MATLAB toolboxkit. (a) The dipole antenna. (b) The directional antenna.

4.6 Polarization

Two communicating antennas should both keep the same placement related to polarization in orderto avoid any cross-polarization that drastically reduces the received power [14]. If one of the twoantennas is placed sideways, cross-polarization occurs [20]. Cross-polarization is investigated inthe anechoic test chamber at FOI as a test. See Figure 9 of a case when cross-polarization occursas the antennas both are vertically polarized but the receiving directional antenna is turned 90degrees sideways and therefore affects the performance of receiving signals as the elevation sweepgoes approximately 90◦. It is important that the receiving antenna is not cross-polarized with thetransmitting antenna or the signal will be extremely attenuated.

Figure 9: This is a plot of the directional antenna that is placed 90◦ sideways. The directionalantenna is then rotated approximately 90◦ in elevation. The negative peak that decreases instrength by over -20 dBm is due to the cross-polarization phenomenon.

15

5 Simulation and measurement environments

A simulation environment in MATLAB is made before a real flight with a UAV and its equipment.This is done to test localization methods before trying them in real flight, e.g. predict which flightroute gives the best performance in the shortest time.

5.1 Introduction to the user interface

The GUI-field is a landscape of size 500x500 pixels that has a jammer location given and a UAVthat moves accordingly to a pre-programmed route or by dragging it with the mouse cursor, seeFigure 10. The jammer and UAV also have given heights relative to the ground. The jammeris affected by its radiation pattern to determine how much of the peak power is radiated to theUAV’s coordinate. The amount of received power also decreases quadratically with the distancebetween jammer and UAV. This fading pace is due to Friis equation, which says that the receivedpower decreases with 1/R2, where R is the radial distance from transmitting antenna [14].

Figure 10: The simulation GUI with a heatmap that shows the radiation pattern of the jammer.In this case the jammer is equipped with the Kathrein dipole antenna. The dots ahead of the UAVdefines its heading. In the title text the position, received power, distance, antenna fading andangle of the UAV is shown.

5.2 Antenna profiles

Since the UAV is moving around the jammer in all three dimensions, the antenna radiation patternsmust be taken into account to simulate the signal strength accurately. This was done by calculatingthe angle between the UAV and the jammer in the ZX-plane and the XY-plane and then referencingthe antenna’s radiation pattern strength in that direction.

17

5.3 Radiation pattern for the UAV platform and antennas

An investigation in order to find out if the UAV platform affects the RSSI measurements viareflections on the dipole and directional antennas mounted on the UAV is described here. The firstimprovised test was performed outdoors in FOI’s courtyard and later on a more professional testis performed in an anechoic test chamber at FOI.

5.3.1 Reflection test at FOIs courtyard

A basic test was performed outdoor in a partially open field, rotating the UAV on a short distanceof approximately nine meters from the transmitting jammer with bare hands. By using a hand-held R&S FSH3 spectrum analyzer, the RSSI strength proved to deviate from different azimuth(horizontal) angles while the R&S signal generator used the same settings as it would during a realflight [21]. The azimuth angle was rotated at 45◦ increments as the spectrum analyzer measuredthe received signal peak. A signal reflection test in the anechoic test chamber is necessary in orderto reduce undesired noise and investigate the antenna further. See Figure 11 of the set-up.

Figure 11: The improvised outdoor test where the performance of the dipole antenna mounted onthe UAV was evaluated with a spectrum analyzer. The UAV is held upside down.

5.3.2 Anechoic test chamber tests

FOI in Linköping owns two anechoic test chambers and one of them is used to investigate thereflections from the UAV to the dipole antenna on board. The UAV is placed on a table thatcan rotate automatically both in elevation (vertical) and azimuth angles, while a signal generatordelivers a signal for the UAV to receive. The Anritsu M54644A network analyzer will be used tolog the measured power for a certain angle on the table for the antennas separately on the UAV.The rotation sweeps are rotated counterclockwise (CCW). The tests in the anechoic chamber tookplace in the 26th of April 2018. A Yagi-antenna was acting as the jammer during this test andsends CW signals vertically polarized from 700 MHz to 3000 MHz for every angle while the tablethat is six meters away. The antennas on the UAV are connected to the network analyzer torecord the received power data. The antenna was placed on the table upside down in order toacquire the correct relation in angles between the receiver and transmitter, as it would during areal flight. This test performed RSSI measurements when the UAV on the table was fix at 10◦

and 45 ◦ respectively, while performing a full rotation in the azimuth plane. Elevation sweeps werealso performed but with a limitation to approximate 100◦ span. See the mounting of the UAV inFigure 12.

18

Figure 12: The UAV mounted upside down in the test chamber with the dipole antenna attachedclearly visible. The table on which the UAV is placed could rotate in both azimuth- and elevationangles.

The received signal with the dipole antenna attached while the UAV rotated at different ele-vation angles can be seen in Figures 14, 15 and 16 in Chapter 6. There is a possibility that theUAV platform reflects the incoming signal and affects the received radiated power on the dipoleantenna. This ripple fades somewhat when the UAV rotates so that the dipole antenna is closerto the signal source than the UAV platform.

19

6 Reflection measurements

This chapter presents the measurements from the anechoic test chamber for both UAV antennasat FOI.

6.1 Dipole antenna measurement

This section describes the simulations from anechoic test chamber experimentation, when eventualreflection occur by the UAV platform from the transmitted signal to receiving dipole antenna.The network analyzer in the test chamber generated a signal of 700 MHz to 3000 MHz for everymeasurement angle on the rotating table on which the UAV was placed. See Figure 13 for thesimulation of the entire 700-3000 MHz frequency spectrum of the dipole antenna reflection test.

Figure 13: Three dimensional plot for the dipole antenna with a fixed elevation angle of 10◦ fromthe anechoic test chamber across the entire frequency span of 700-3000 MHz with correspondingrotation angles in azimuth and RSSI values.

All the measurement graphs from the anechoic test chamber are plotted at 1875 MHz frequencysince the network analyzer transmits the CW with an increment of 5 MHz. The dipole antennais used to receive the transmitted signals in order to investigate any reflections. See Figure 12for UAV placement on the rotating table. The reason why the UAV was placed upside down onthe table is because that it gave the desired relation in angles between transmitting and receivingantennas just as in an actual flight. Figure 14 shows the reflection behaviour after the UAV withthe dipole antenna was rotated one full rotation.

21

Figure 14: UAV rotated 360◦ along the azimuth axis while the elevation was fixed at 10◦ and 45◦

respectively. The frequency was set to 1875 MHz.

Due to the ripple in Figure 14, a new test was performed in the anechoic test chamber, since ametallic frame below the transmitting antenna for another irrelevant project might have affectedthe previous measurements. Figures 15 and 16 shows the comparison between with or without theabsorption material.

Figure 15: UAV rotated 360◦ along the azimuth axis while the elevation was fixed at a 10◦ elevationangle. The first test is without any absorbing material covering the metallic frame below thetransmitting antenna, while the second test is with the absorbing material applied. The frequencywas set to 1875 MHz.

22

Figure 16: Same test as in Figure 15 but with a fixed elevation angle of 45◦ instead. The frequencywas set to 1875 MHz.

The purpose of the dipole antenna mounted on the UAV is to receive the transmitted signal re-lated to method one, compared to the directional antenna which works in long range measurements.Since the reflection simulations in the anechoic chamber at FOI proved to give rather unstable graphplots with ripple, the dipole antenna was displaced to another location that wouldn’t risk the sameamount of reflections. Figure 17 shows the dipole antennas performance when the azimuth wasfixed and the elevation angle was tilted.

Figure 17: This graph shows the received power of the dipole antenna while the rotation tablerotated 96◦ in elevation. The frequency was set to 1875 MHz.

The RSSI ripple for the dipole antenna with the different elevation angles is possibly influencedby the UAV platform and the rotating table which is conducting. No further experimentation inwhat may inflict ripple on the measurements was be performed, since this was not the main focus.To minimize platform interference, the dipole antenna was relocated to a more suitable positionfor less exposure to reflection. See Figure 18 for the new dipole antenna location on the UAV.

23

Figure 18: This is the final position of the dipole antenna on the UAV. The UAV with the dipoleantenna headed downwards at a more central location than previously. This installation was nevertested or used in flights.

6.2 Directional antenna measurement

The directional antenna was used for the second methods position estimations. Figure 19 showsthe elevation sweep of the directional antenna.

Figure 19: This graph shows the smooth curve of the received power while the rotation tablerotated in elevation from –5◦ to 91◦. The frequency was set to 1875 MHz.

The directional antenna was also rotated along the azimuth axis one full rotation and the RSSIis drawn as a polar plot that can be seen in Figure 20.

24

Figure 20: Measured radiation pattern of the directional antenna in the anechoic chamber. Thefrequency was set to 1875 MHz and the scale of the plot is in dBm. The elevation angle was keptconstant at 0◦ while performing the azimuth sweep.

25

7 Equipment for RSSI measurements

This section covers both the hardware and software that are required in order to perform thedesired RSSI measurements.

7.1 The directional antenna on the UAV

The directional antenna is the Pro-1000, GSM panel antenna that will be mounted on the UAVis a panel antenna bought from Televes (formerly Macab) and antenna characteristics is given ontheir website. The frequency spectra is 1710-2500 HMz among others and 1877 MHz was usedin the field as well as 1875 MHz for the anechoic chamber tests at FOI. Maximum power is 100W and the antenna will be connected to the RF input port on the MikTran-platform [19]. Thisantenna will only be used to receive signals. See Figure 21a for the panel antenna, which is calledthe directional antenna in this thesis and the mounting on the UAV on Figure 21b.

(a) (b)

Figure 21: The directional antenna used for triangulating the signal jammer. (a) As it appears onthe distributors website [19]. (b) Installed on the UAV.

The radiation patterns are illustrated straight from the data-sheet in Figure 22. Note that thevertical radiation pattern corresponds to the 3D radiation pattern of the directional antenna inFigure 13. The vertical opening angle is at 55◦ for the relevant frequency band at 1710-2500 MHz.

Figure 22: The polar graphs of the radiation patterns as horizontal and vertical respectively. Thefrequency for the radiation patterns here is at 2100 MHz [19].

7.2 The dipole antenna on the UAV

The provided dipole antenna that is mounted on the UAV has been purchased from Round Solutionsand is named Screwable Small Stubby Antenna with 1 Cable 2G/3G [22]. See Figure 23.

27

Figure 23: The dipole antenna that is installed on the UAV during the flights related to the firstmethod [22].

7.3 MikTran platform

The MikTran platform has an onboard AD9361 ADC, a GPS module and a Xilinx Zynq AdvancedRISC Machine (ARM) System-on-a-chip (SoC). The AD9361 Application Programming Interface(API) and the libgps library allows software to run on the Xilinx Zynq with real-time data fromthe ADC and the GPS module. When this platform is used in the field on the UAV it is sealedinside a metal box that functions both as protection and as a heatsink, see Figure 24.

The two receiving antennas is connected one at a time with an SMA connection to the MikTranin order to record its RSSI measurements. A non-magnetic GPS antenna provided by FOI is alsoconnected to the unit on a separate SMA connector in order to pair the correct measurement tothe current GPS coordinate of the UAV. The MikTran platform is also connected to a RaspberryPi 3 using an ethernet cable.

Figure 24: The size of the MikTran compared to a five SEK coin.

7.3.1 MikTran software

The Xilinx Zync SoC onboard the MikTran platform runs a custom application for doing measure-ments, written in C. This application uses the AD9361 API to retreive 1500 RSSI measurements,gets the latest GPS location using the libgps library with the onboard ORG4572 chip from Orig-inGPS, and listens for User Datagram Protocol (UDP) packets from the Raspberri Pi on a separatethread. The applications logs the maximum value of the 1500 RSSI measurements, the latest GPS-coordinates and the latest heading-value that was received from the Raspberry Pi to a text fileapproximately 8 times a second.

28

7.4 Raspberry Pi 3

The UAV platform comes with a Raspberry Pi 3 unit installed, see Figure 25. This unit is usedto communicate with other hardware present on the UAV and is connected to the Pixhawk flightcontroller via USB. Over this connection, the MAVLink protocol feeds the Raspberry Pi withconstant updates from the Pixhawk flight controller. Since the heading of the UAV is neededfor determining in which direction the jammer is relative to the UAV, the heading data from thePixhawk is forwarded to MikTran via UDP using an ethernet cable between the two.

By

Make

Magazi

n[C

CBY-S

A4.0

(htt

ps://cr

eative

com

mons.org

/lice

nse

s/by-

sa/4.0

)],from

Wik

imed

iaCom

mons

Figure 25: Top view of a Raspberry Pi 3 Model B unit.

29

8 Test flights

This chapter contains the outcomes of each test flight with thoughts and improvements for upcom-ing flights.

8.1 UAV equipment

This sub-chapter describes all hardware material and arrangements that are relevant for the flights.The UAV is navigated manually by a pilot from FOI via radio telemetry hardware. Mission Planner(MP) is the interface that gives information directly from the UAV during the flights from differentsensors such as altitude above ground. See Table 2 for the list of equipment that where presentduring all flights. The UAVs main platform and arms are made of carbon fibre and designed atFOI.

Table 2: This table covers the equipment that were used during the UAV flights, excluding antennasfor RSSI measurements.

UAV EquipmentSpecific hardware Model QuantityVehicle Hardware

Propellers T-Motor CF 15"x4.8" (diameter x inclination) 4Motors T-Motor kV 580 (MN3508-20) 4Battery Dynamo LiPo 4S, 14.8 V, 10000 mAh 1Electronic Speed Controller XRotor-pro 25 A 4Flight Computer Pixhawk Mini, Arducopter software 1RC Radio Futaba 1Vehicle Telemetry

Telemetry 3DR Wifi Telemetry Radio 1Ground Control Hardware

WiFi router (with external antenna) FOI 1Computer with PuTTY FOI 1

With the components in Table 2, the UAV is estimated to be able to perform flight missionsfor 18 minutes ±1-3 minutes depending on the wind and temperature in the air. The appearanceof the UAV for the thesis project is presented in Figure 26a and Figure 26b. A detailed blockdiagram of the UAV hardware is shown in Figure 27.

(a) (b)

Figure 26: The UAV with attached components, such as MikTran, GPS, dipole antenna, RaspberryPi and Wifi antenna. Note that the propellers are removed for these pictures. (a) The UAV seenfrom underneath. The microprocessor MikTran is clearly visible at the centre of the UAV, capsuledin its protective box. (b) The UAV seen from above. This side reveals the strapped on battery atthe centre, Raspberry Pi 3 and GPS (upper left arm).

31

Figure 27: Block diagram of the UAV system hardware.

8.2 First flight

The first test flight took place on the 27th of March 2018 in an open field environment on thecountryside of Linköping. The aim of this test flight was to measure radiated power data from astationary dipole antenna while the UAV was controlled manually by a pilot who took the flightroute as seen in Figure 28. The satellite image was taken from Eniros website and the dots wereplaced using the MATLAB mapping toolbox and colored based on the received signal strength[23].The MikTran was connected to a small dipole antenna hanging from underneath the UAV dur-ing flight and communications with the MikTran on-board the UAV was done via a Secure Shell(SSH) tunnel over Wi-Fi. During this flight the UAV began the flight by increasing its altitudeto approximately 20 meters above ground and then it flew according to the plot in Figure 28. Inthe end of the flight above the jammers location the altitude was decreased to a few meters aboveground. The LOS between the antennas lasted during the entire flight.

32

Figure 28: The recorded data of GPS coordinates by the GPS antenna stored in the MikTransmemory. The dots show the flight path of this test and the color indicates the measured radiationstrength (white - strong signal, red - weak signal).

The thoughts for the next flight were to increase the sample rate of the MikTran and to installthe dipole antenna in a fixed position so that it does not hang freely. The appearance of the UAVduring its first flight can be seen in Figure 29.

Figure 29: The UAVs vertically hanging dipole antenna for the first flight with RSSI measurements.

33

8.3 Second flight

The second test flight took place on the 20th of April 2018. The altitude was decreased to approx-imately 13 meters above ground due to strong winds above the surrounding treetops. Figure 30shows the new, fixed mounting of the dipole antenna. The dipole flight route is shown in Figure31.

Figure 30: The dipole antenna steadily mounted on one of the UAV arms is facing downwards.

The UAV landed on the ground after the flight with the dipole antenna so that the dipoleantenna could be replaced by the directional antenna. The flight route with the directional antennacan be seen in Figure 32 and Figure 33 illustrates the set up with the directional antenna belowthe UAVs platform. Since the directional antenna under the UAV is the first area of contact whenthe UAV performs a takeoff or landing, four plastic boxes were placed under each arm as a takeoff-and landing-pad.

34

Figure 31: The recorded data from the flight stored in the MikTrans memory. The dots show theflight path of this test and the color indicates the measured received signal strength with the dipoleantenna (white - strong signal, red - weak signal).

For the test with the directional antenna were four different 360-degree rotations performed atdifferent locations ranging from approximately 30 meters up to 180 meters away from the signaljammer. The signal generator that powered the signal jammer was turned off between the rotationswhen the UAV was moving to the next position. The signal jammer was left transmitting the entiretime during the flight back towards the signal jammer after the fourth spin.

35

Figure 32: The flight route with the directional antenna. The UAV performed four rotations at thedifferent white rotation markings. The distance away from the jammer at the different locationsare I) 32.12, II) 52.2, III) 108.0 and IV) 180.8 meters.

Figure 33: UAV with the mounted directional antenna while the drone is airborne.

Thoughts after the second flight was to change localization method rather than modify thealgorithm that uses RSSI data based on the dipole antenna only. The reason to change thelocalization method is because that the RSSI measurements gave two peaks rather than one thatwas desired. See Figure 37 of the RSSI (dBm) data plotted in a 3D graph where the x-axis andy-axis is the landscape of the field area.

36

8.4 Third flight

The third test flight took place on the 7th of June 2018. Due to unstable wind conditions, thealtitude was held at approximately 13 meters just as the second test flight in April. Only thedirectional antenna was installed during this flight since the purpose was to test the triangulationmethod. The two flight routes are shown in Figure 34.

(a) (b)

Figure 34: The two flights done on the third flight day. The first route is shown in (a), the secondroute in (b).

The UAV performed two flights where the first one consist of seven azimuth rotations andthe second flight of five rotations. Two flights were performed due to strong wind and unstablerotations during the first flight but unfortunately the second flight had similar conditions. Theperformance of the RSSI measurement is affected by how much the UAV sways and how stable thealtitude can be kept when the azimuth rotations are performed.

37

9 Results

In this chapter the performance of the antenna equipment, GPS positioning and the differentlocalization methods will be presented.

9.1 Antenna performance

This sub-chapter will go through the performance of the two antennas mounted on the UAV.

9.1.1 Dipole antenna

Since the reflection simulations in the anechoic chamber at FOI proved to give rather unstablegraph plots with ripple, the dipole antenna was displaced to another location that wouldn’t riskthe same amount of reflections, see Figure 18. Figure 17 shows the dipole antennas performancewhen the azimuth was fix and the elevation angle was tilted.

The RSSI ripple for the dipole antenna with the different elevation angles is possibly influencedby the UAV platform and the rotating table which is made of metal. No further experimentationin what may inflict ripple on the measurements was be performed since it wasn’t the main focusof the thesis. However, the name of the dipole antenna was provided after the anechoic chambermeasurements. The manufacturer at Round Solutions doesn’t have a radiation pattern on theirwebsite for this dipole antenna. Other websites has similar antennas of the same appearance andthe radiation patterns provided shows a unstable pattern for similar frequencies [24]. This meansthat the RSSI ripple for the dipole antenna in the anechoic chamber could also be affected by thereceiving antennas radiation pattern.

9.1.2 Directional antenna

The directional antenna was not affected meaningfully by any reflections during the test in theanechoic chamber, as seen in Figure 19. The measured radiation patterns of the directional antennahad no significant ripple or discontinuities as seen in Figures 19 and 20. With the large verticalopening angle of approximately 40◦, the directional antenna can receive signals without muchattenuation over a large span of distances when mounted with a vertical angle between 39◦.

The directional antennas performance of an increasing distance between the UAV and jammerduring a flight can be seen in Figure 35. The UAV kept a fix altitude between the differentdistances. The altitude was kept at approximate 13 meters above ground instead of 20 meters, dueto weather conditions during the flight.

Figure 35: This graph is based on RSSI data from the second UAV flight when the UAV performedrotations at different locations with the directional antenna mounted. The graph plots the RSSImaximum values from the different rotations at a certain distance and is compared to a quadraticplot based on Friis formula. The maximum power peaks are obtained when the directional antennais aimed directly at the jammer. LOS during the entire test phase.

Figure 35 reveals that the RSSI measurement of the LOS signal aligns very well with thequadratic curve which is related to Friis equation. The reason the RSSI decreases as it does

39

with increased distance is due to the free-space path loss model [25]. Since the RSSI decreasesdepending on the distance and the current directional antenna, the antenna should be considered tobe replaced by an even higher gain directional antenna if RSSI measurements are to be performedat even larger distances. However, a higher gain comes at a cost of a smaller beamwidth of themain lobe.

9.2 GPS accuracy and reliability

To increase the update rate of GPS coordinates, a custom application was written using the libgpslibrary in the C programming language [26]. This software update was added to the MikTranplatform and was tested with the UAV and GPS antenna. The route of the UAV was performedon foot at FOIs courtyard, see Figure 36. The GPS achieved a Position Dilution of Precision(PDOP) of 1.4 and had an accuracy of within one meter when compared to satellite imagery ofthe area. More importantly, as seen in the figure the GPS secures an accurate fix within secondsof heading outside the building and tracks the walked trail accurately.

Figure 36: Test with the GPS logging at FOI’s courtyard. The movement of the GPS is based onthe taken course on foot and are marked as white dots on the map. The dots are very accurateand the satellite imagery are acquired from Eniros website [11].

9.3 Performance of localization methods

This sub-section will cover the performance of the different methods of localization methods.

9.3.1 LS algorithm using dipole antenna

After the second flight was performed and RSSI data examined, it proved to reveal two maximumpeaks since the transmitter is a dipole antenna with the radiation pattern as in Figure 8a. Sincethe receiving antenna also is a dipole antenna, the received power right above the signal jammeris significantly lower. See Figure 37 for a view at the two peaks.

Due to this phenomenon the algorithm that’s being used will point out one of the two peaksas a transmitting source, which is not especially accurate since the jammer is located at the valleybetween the two maxima. This resulted in a margin of error of at least thirty meters from thecorrect GPS-spot, and the UAV was in presence of the jammer by at least one hundred meters.

40

(a) (b)

Figure 37: Surface plot of the radiated power over the area covered by the UAV’s flight. The twopeaks on either side of the jammer due to the radiation pattern are clearly visible. The X-axis andY-axis represent normalized coordinates, the Z-axis shows the RSSI. (a) Shows a perspective view,(b) shows a top view.

9.3.2 Triangulation method in the courtyard

In order to improve the RSSI directon estimate at the different rotations, the -4 dB span of theRSSI is used. This span is useful since the maximum RSSI may not be the correct heading of thesignal source, and the median angle of the -4 dB region is more accurate to the correct bearing.See Figure 38 for the 3-4 dB span related to the ground test on the courtyard of the directionalantenna.

Figure 38: Measured RSSI from the triangulation test in the courtyard. The two plots show theresults from the rotation of the UAV at the a) and b) point in Figure 39, respectively.

The courtyard test with the hand held UAV was performed at five different locations, see Figure39. Three of the rotations identified the direction to the jammer accurately.

41

Figure 39: The results from the triangulation test in the courtyard. Almost all of the five triangleheadings are directed at the estimated direction of the jammer based on the -4 dB boundary witha ±5◦ spread. The white cross shows the jammers true location. a) and b) represents two differentrotation locations and can be referred to Figure 38 a) and b).

9.3.3 Triangulation method during flight

The same triangulation principle as in section 9.3.2 was used with the UAV airborne. Wind madethe UAV unstable during some of the rotations, causing noisy measurements. With the filteringmethod described in section 3.2.2, most of these measurements were removed while the moreaccurate measurements remained. Three of the four final direction estimates cross each other ata point approximately two meters from the true position of the jammer. See Figure 40b for theestimated jammer directions of the rotations.

42

(a) (b)

Figure 40: Triangulation results from the third flight with each direction estimate drawn as a 2◦

wide cone. The cross marks the true location of the jammer. Due to strong winds several of themeasurements were inaccurate. Using the filtering method described in section 3.2.2 the sevenmeasurements shown in (a) were reduced to the four shown in (b).

43

10 Conclusion

In this chapter the conclusions based on the results in Chapter 9 is presented.

10.1 Localization based on real flights

Two different methods were evaluated, each with its own antenna, flight strategy and technique.The idea of the first method is that it could provide an exact position if the general area of thejammer is known. This method will make the UAV eventually pass over the jammer and receiveone maximum RSSI peak per transmitting lobe it passes. These two peaks will be further apartthe higher the altitude of the UAV and will cause the algorithms estimation to settle on the wronglocation. Since the goal of this method was high-precision positioning in a smaller area it is notdeemed suitable.

The RSSI maximum power peaks at different distances are presented in Figure 35 and its al-most perfectly satisfactory with the quadratic decrements of RSSI from the theory predicted inFriis equation. This localization method that performs two rather quick position estimations ismore efficient than the first method, which uses a dipole antenna. This is due to the fact thatthe first method searches a field with the wide zigzag pattern that is more time consuming thanthe second method consisting of the directional antenna. Both of these methods works to verifyif there is in fact a transmitting signal jammer in an area of interest. When it comes to accuracy,the directional antenna is considerably better.The both flights on the 7th of June verifies that the second method with triangulation in combi-nation with the median RSSI values with noisy rotation filtering gives a position estimation errorthat’s less than 3 meters.The location of the true jammer location spot was only estimated once per flight using a smart-phone. The accuracy of this reference location could be improved by using several GPS devices orusing a more accurate device. According to the satellite imagery, the real jamming location wasplaced a bit closer to the small truck road that’s visible in Figure 40, making the position estimateof the third flight even more accurate.

10.2 Reliability of the user interface

The performance of the developed simulation environment GUI gives a simplified estimation onwhat to expect from a real UAV flight with RSSI data. Since it does not take landscape reflectionsfrom objects such as mountains, trees or water into account, it differs from measurements in thereal world. It is not very usable for testing the triangulation method, since it will yield a perfectresult every time. In other words, the GUI is more suitable for simulating the first method withthe dipole antenna since the main problem with this method comes from the transmitting antennasradiation pattern which it does take into account.

45

10.3 Goals

There are three goals regarding this thesis that has been evaluated:

• Performance investigation of antennas and considerations.

This goal has been accomplished. The directional antenna RSSI proved to be suitable forthe purpose and wasn’t affected by any reflections from the UAVs platform and maintaineda high directional accuracy, compared to that of the dipole antenna which proved unreliable.

• Development of a simulation environment to test position estimation methods.

This goal has been accomplished. The developed GUI was implemented with similar radiationpattern characteristics and yielded similar results with peaks in a circle around the jammerand a valley right above the jammer.

• Evaluation and comparison of signal source positioning methods.

This goal has been accomplished. The two methods, the LS algorithm method using thedipole antenna and the triangulation method using the directional antenna were implementedand tested.

46

11 Future work

In this chapter some ideas of future improvements and developments regarding this master thesisarea in how to locate a transmitting jammer with a UAV are presented.

11.1 Frequency-hopping implementation

The UAV antenna equipment for this thesis will only search for jammers at a CW with a fixed fre-quency. However there are more complicated jammers that work in frequency-hopping. Localizingthese require broadening the received spectrum and may also require analysis of the signal.

11.2 UAV improvements

The UAV platform in this thesis is quite strongly affected by wind, mostly due to its size andweight. Using a bigger UAV will likely make measurements more stable and reliable in windierconditions. A landing gear will also make taking off and landing easier since without landing gearsthe directional antenna is the first point of contact with the ground.

A replacement of the directional antenna to an antenna with a suitable beamwidth is useful if themeasurements are to be done at longer ranges than this thesis, which ranged from approximately30-200 meters. As Figure 35 shows, the RSSI peak values decreases as the distance increases.There is no antenna that will evade this outcome, but a directional antenna with a higher gainthan the panel antenna that was used will receive RSSI data at longer ranges before merging withthe noise levels.

The sample rate of the UAV’s flight computer can also be increased, this will allow more sampleswhile doing the triangulation rotations if the RSSI measurement sample rate is also increased.

Figure 41: This is the UAV flights that were performed on the 7th of June. The tight line clusterreveals the UAV rotations for the RSSI measurements and the alternating altitude along the z-axisshows that the UAV was affected by the winds.

47

11.3 RSSI using directional and dipole antennas

In the thesis there are two methods that has been tested and validated. However there is a thirdmethod, which is a combination of the two other previously tested jamming location methods.This method that combines both the directional antenna and dipole antenna is thought to letthe UAV commence its search for the jammer by completing two full rotations at two differentlocations at the same rotation rate as the previously tested method, and after the estimation ofa crossing between the two measurements, the UAV will head to that specific direction. Whenthe UAV is getting close to the jammer the dipole antenna will start to receive signals from thejammer on the ground. When the short distance measurement is accomplished an estimation ofsignal source position can be done. The dipole antenna continually measures the RSSI during thefly-over and will record the aforementioned two peaks on either side of the jammer created by itsvertical components of the main lobe. The jammer will therefore be in the middle of the two peakswhere the dipole antenna receives the lowest amount of power.

11.4 Autonomous navigation

Since a UAV that is being controlled manually it requires a wireless connection that may beinterfered by the jammer it is trying to locate, an autonomous UAV will provide better reliabilitysince it’s not dependent on any external signal. The UAV will perform its search based on anyof the mentioned methods and return to its take-off point and provide the users with the data ofjamming coordinates. Since the MAVLink interface is already setup for communications betweenthe flight controller and MikTran, this implementation will not be difficult software wise but mayrequire a lot of testing. An idea regarding the autonomous navigation is related to Section 11.3that could eventually use the antenna combination for the autonomous flight.

11.5 Further UAV reflection tests

Although there isn’t a radiation pattern plot in the data sheet of the dipole antenna used, thereare some radiation patterns from similarly sized antennas in the same frequency regions, showinglarge variances in the horizontal plane at the frequency used in this thesis. It is possible that thisis what caused the majority of the ripple in the RSSI measurements with the dipole antenna butfurther investigation can be done to verify this.

11.6 Avoid hostile danger

Since the purpose of this UAV is to locate a signal jammer, it’s in the opposing parts interest tohinder it from completing this task. An easy way to ensure higher safety is to increase the altitudeand/or distance of the UAV while estimating the jammers position since it decreases the likelihoodof it being detected and being shot down. With higher altitude and distance, antenna gain andpositioning becomes more important. Optimally, an antenna with a narrow horizontal lobe and awide vertical lobe should be used since it allows for greater precision at long distances while stillallowing a wide array of vertical angles to the jammer.

48

References

[1] S. Sukhankin, “Russian Electronic Warfare in Ukraine: Between Real and Imaginable,” May24 2017. [Online; accessed 18-May-2018].

[2] A. Andersson, “Polisen varnar för störsändare på stora parkeringsplatser,” Oct 11 2017. [On-line; accessed 18-May-2018].

[3] T. Lee, “Apple using drones to improve apple maps,” Oct 10 2018. [Online; accessed 18-May-2018].

[4] Pixhawk project, “Pixhawk mini documentation.” http://docs.px4.io/en/flight_

controller/pixhawk_mini.html. [Online; accessed 15-June-2018].

[5] QGroundControl, “Mavlink protocol.” http://qgroundcontrol.org/mavlink/start. [On-line; accessed 15-June-2018].

[6] Fabian Körner et al., “Autonomous Airborne Wildlife Tracking Using Radio Signal Strength,”October 2010. [Online; accessed 18-May-2018].

[7] Michael Tulldahl et al., “Application and capabilities of lidar from small UAV,” 2016. [Online;accessed 18-May-2018].

[8] Petar Popavski et al., “Frequency hopping.” "https://ieeexplore.ieee.org/document/

4052302/". [Online; accessed 15-June-2018].

[9] “R&S SMC100A Signal Generator.” "https://www.rohde-schwarz.com/pt/product/

smc100a-productstartpage_63493-10181.html". [Online; accessed 14-June-2018].

[10] Katherein, “Datenblatt 9361696/c.” "https://www.kathrein.de/fileadmin/media/

produkte/imports/mca/9361696c.pdf". [Online; accessed 18-May-2018].

[11] Eniro, “Satellite image over FOIs courtyard.” "https://kartor.eniro.se/?c=58.392384,

15.575631&z=18&l=aerial". [Online; accessed 14-June-2018].

[12] J. L. Petre Stoica, “Source localization from range-difference measurements,” pp. 1–2, Novem-ber 2006. [Online; accessed 18-May-2018].

[13] Adrien Perkins et al., “Antenna Characterization for UAV Based GPS Jammer Localization,”[Online; accessed 18-May-2018].

[14] D. M. Pozar, Microwave and RF Design of Wireless Systems. 2000.

[15] “Rf reflection.” "http://www.radio-electronics.com/info/propagation/em_waves/

electromagnetic-reflection-refraction-diffraction.php". [Online; accessed 18-May-2018].

[16] “Multipath propagation.” "http://www.radio-electronics.com/info/propagation/

multipath/multipath-propagation-basics-tutorial.php". [Online; accessed 19-June-2018].

[17] A. Doukas and G. Kalivas, “Rician K Factor Estimation for Wireless Communication Systems,”IEEE, p. 3, 2006. [Online; accessed 18-May-2018].

[18] “Signal to noise ratio.” "http://www.scholarpedia.org/article/Signal-to-noise_

ratio". [Online; accessed 18-May-2018].

[19] “Antenn, 3G/4G, Pro-1000, 698-960MHz/1710-2700MHz, 7dBi.” "https://site.televes.

com/se/650202-antenn-3g-4g-pro-1000-698-960mhz-1710-2700mhz-7dbi.html". [On-line; accessed 19-June-2018].

[20] Ian Poole, “Antenna polarisation or polarization.” http://www.radio-electronics.com/

info/antennas/basics/polarisation-polarization.php. [Online; accessed 16-June-2018].

49

[21] R&S FSH3, “Spectrum analyzer.” "https://www.rohde-schwarz.com/us/product/

fsh3-6-18-productstartpage_63493-358532.html". [Online; accessed 15-June-2018].

[22] Round Solutions, “Dipole antenna.” "https://www.roundsolutions.com/en/products/

antennas/stubby-antenna/9109/-screwable-small-stubby-antenna-with-1-cable-2g/

3g/". [Online; accessed 08-June-2018].

[23] Eniro, “Satellite image over the flight location.” "https://kartor.eniro.se/?c=58.258865,15.571693&z=18&l=aerial". [Online; accessed 14-June-2018].

[24] Jinchang, “Jinchang JCG411 Combo Antenna.” http://www.jinchanggps.com/index.php?

c=content&a=show&id=496. [Online; accessed 16-June-2018].

[25] Ian Poole, “Free Space Path Loss: Details, Formula, Calcula-tor.” http://www.radio-electronics.com/info/propagation/path-loss/

free-space-formula-equation.php. [Online; accessed 19-June-2018].

[26] Eric S. Raymond, Charles Curley, “libgps service library.” http://www.catb.org/gpsd/

libgps.html. [Online; accessed 19-June-2018].

50