23.05.2014 Views

Title page - Swedish Institute of Space Physics - Uppsala

Title page - Swedish Institute of Space Physics - Uppsala

Title page - Swedish Institute of Space Physics - Uppsala

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Title</strong> <strong>page</strong><br />

This is the title<strong>page</strong> dummy. This <strong>page</strong> should be substituted for real title<strong>page</strong>.<br />

The real title<strong>page</strong> is obtained from the Electronic Publishing Centre after having<br />

submitted the template for electronic posting; "spikbladsmallen" to the<br />

Electronic Publishing Centre at espik@ub.uu.se. Spikningsmallen (the posting<br />

template) can be downloaded from<br />

http://publications.uu.se/forauthors/templates.<br />

More information about the publishing routines can be found at<br />

http://publications.uu.se/forauthors


Dissertation at <strong>Uppsala</strong> University to be publicly examined in Polhemsalen, 10134, Ångström<br />

Laboratory, Friday, September 30, 2005 at 14:00 for the Degree <strong>of</strong> Doctor <strong>of</strong> Philosophy. The<br />

examination will be conducted in English<br />

Abstract<br />

Karlsson, R. 2005. Theory and Applications <strong>of</strong> Tri-axial Electromagnetic Field Measurements.<br />

. Acta Universitatis Upsaliensis. <strong>Uppsala</strong> Dissertations from the Faculty <strong>of</strong> Science and Technology<br />

. vi, 63 pp. <strong>Uppsala</strong>. ISBN 91-554-5436-4<br />

Polarisation, which was first studied in optics, is a fundamental property <strong>of</strong> all electromagnetic<br />

fields. A convenient way to describe the polarisation <strong>of</strong> 2D electromagnetic fields is given by<br />

the Stokes parameters.<br />

This thesis deals with different aspects <strong>of</strong> wave polarisation and electromagnetic field measurements.<br />

A generalisation <strong>of</strong> the Stokes parameters to three dimensions is presented. The<br />

theory has been used to develop methods and systems for obtaining the polarisation parameters<br />

<strong>of</strong> electromagnetic waves. The methods can be applied for a wide range <strong>of</strong> electromagnetic<br />

fields, measured both on ground and onboard satellites. The applications include, e.g. directionfinding,<br />

polarisation analysis, radio telescopes, radar, and several examples in the field <strong>of</strong> wireless<br />

communication.<br />

Further applications are given in the analysis <strong>of</strong> satellite data, where a whistler wave is considered.<br />

Whistlers are circularly polarised electromagnetic waves propagating in the magnetosphere<br />

along the geomagnetic field. Dispersion in the magnetospheric plasma make the whistler<br />

frequency components travel at different speeds and the signal takes the form <strong>of</strong> a chirp. From<br />

instantaneous polarisation analysis <strong>of</strong> the whistler’s magnetic wave field, the normal to the polarisation<br />

plane is obtained and found to precess around the geomagnetic field.<br />

A statistical analysis <strong>of</strong> ionospheric stimulated electromagnetic emissions (SEE) is also presented.<br />

SEE is generated by injecting a powerful high frequency radio wave into the ionosphere.<br />

It is shown that the SEE features have a statistical behaviour indistinguishable from the amplitude<br />

and phase distributions <strong>of</strong> narrow-band Gaussian noise. The findings suggests that SEE<br />

cannot be explained by simple coherent processes.<br />

An expression for the complex Poynting theorem is derived for the general case <strong>of</strong> anharmonic<br />

fields. It is found that the complex Poynting theorem, in contrast to the Poynting theorem<br />

for real fields and sources, is not a conservation law <strong>of</strong> the imaginary part <strong>of</strong> electromagnetic<br />

energy.<br />

Keywords: polarisation, polarization, radio waves, Stokes parameters, antennas, SEE, whistler<br />

waves, direction-finding, LOIS, statistics, Poynting theorem<br />

Roger Karlsson, Department <strong>of</strong> Astronomy and <strong>Space</strong> <strong>Physics</strong>. <strong>Uppsala</strong> University. Box 515,<br />

SE-751 20 <strong>Uppsala</strong>, Sweden<br />

c○ Roger Karlsson 2005<br />

ISSN 1651-6214<br />

ISBN 91-554-5436-4<br />

urn:nbn:se:uu:diva-3344 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-3344)


List <strong>of</strong> Papers<br />

This thesis is based on the following papers, which are referred to in the text<br />

by their Roman numerals.<br />

I<br />

II<br />

III<br />

IV<br />

V<br />

Parameters characterizing electromagnetic wave<br />

polarization,<br />

T. D. Carozzi, R. Karlsson, and J. Bergman,<br />

Phys. Rev. E 61, 2024-2028, 2000.<br />

Method and system for obtaining direction <strong>of</strong> an<br />

electromagnetic wave,<br />

J. Bergman, T. D. Carozzi, and R. Karlsson,<br />

Metod och system för att erhålla riktning för en elliptiskt<br />

polariserad elektromagnetisk vågutbredning, <strong>Swedish</strong> patent No<br />

512 219, 2000; US Patent 6,407,702, 2002.<br />

Multipoint antenna device,<br />

J. Bergman, T. D. Carozzi, and R. Karlsson,<br />

Antennanordning för användning av tredimensionell elektromagnetisk<br />

fältinformation inherent i en radiovåg, <strong>Swedish</strong> patent No<br />

517 524, 2002; International Patent Publication WO03/007422,<br />

2003.<br />

Three-channel digital radio vector field sensor: Description<br />

and demonstration,<br />

R. Karlsson, W. Puccio, J. Bergman, T. D. Carozzi, and B. Thidé,<br />

Proceedings <strong>of</strong> the HF–04, Nordic Shortwave Conference, Fårö,<br />

Sweden, 2004.<br />

Present and future applications <strong>of</strong> the Information Dense<br />

Antenna,<br />

J. Bergman, T. D. Carozzi, and R. Karlsson,<br />

Proceedings <strong>of</strong> the HF–04, Nordic Shortwave Conference, Fårö,<br />

Sweden, 2004.<br />

v


VI<br />

VII<br />

VIII<br />

Precession <strong>of</strong> the whistler polarisation plane normal observed<br />

on Freja,<br />

R. L. Karlsson, T. D. Carozzi, J. E. S. Bergman, and A. I. Eriksson,<br />

Submitted to Geophysical Research Letter, 2005.<br />

Statistical properties <strong>of</strong> ionospheric stimulated<br />

electromagnetic emission,<br />

R. L. Karlsson, T. D. Carozzi, J. E. S. Bergman, L. Norin, B.<br />

Thidé,<br />

Submitted to Annales Geophysicae, 2005.<br />

Complex Poynting theorem as a conservation law,<br />

T. D. Carozzi, J. E. S. Bergman, and R. L. Karlsson,<br />

Submitted to Journal <strong>of</strong> Mathematical <strong>Physics</strong>, 2005.<br />

Reprints were made with permission from the publishers.<br />

Papers not included in the thesis<br />

Method for Three-Dimensional Evaluation,<br />

J. Bergman, T. D. Carozzi, and R. Karlsson,<br />

System för tredimensionell utvärdering av ett elektroniskt vektorfält, <strong>Swedish</strong><br />

patent No 523 086, 2004; International Patent Publication WO03/067710,<br />

2003.<br />

Three-channel digital radio receiver for simultaneous reception in three<br />

orthogonal dimensions,<br />

R. Karlsson and W. Puccio,<br />

Proceedings <strong>of</strong> the HF–01, Nordic Shortwave Conference, Fårö, Sweden,<br />

2001.<br />

vi


Contents<br />

1 Introduction 1<br />

2 Plasma 5<br />

2.1 The magnetosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6<br />

2.2 The ionosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6<br />

3 Electromagnetic fields and wave polarisation 11<br />

3.1 Electromagnetic fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />

3.2 Polarisation and Stokes parameters . . . . . . . . . . . . . . . . . . . . . 13<br />

3.3 The coherency tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />

3.4 The spectral tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

3.5 Wave polarisation in 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

4 Electromagnetic field measurements 21<br />

4.1 Method for estimating polarisation parameters . . . . . . . . . . . . . 21<br />

4.2 System for estimating polarisation parameters . . . . . . . . . . . . . 23<br />

4.3 The high frequency band . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23<br />

4.4 LOIS—the LOFAR Outrigger In Scandinavia . . . . . . . . . . . . . 25<br />

4.5 Wireless telecommunication . . . . . . . . . . . . . . . . . . . . . . . . . . 27<br />

5 <strong>Space</strong> applications 31<br />

5.1 Antennas in space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31<br />

5.2 Satellite projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32<br />

6 Whistler analysis using polarisation 37<br />

7 Statistical approach to stimulated electromagnetic emissions 41<br />

7.1 Characteristics and models for SEE . . . . . . . . . . . . . . . . . . . . . 41<br />

7.2 Statistical tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42<br />

8 The complex Poynting theorem as a conservation law 45<br />

8.1 The instantaneous Poynting theorem . . . . . . . . . . . . . . . . . . . . 45<br />

8.2 The complex Poynting theorem . . . . . . . . . . . . . . . . . . . . . . . . 46<br />

8.3 The time-harmonic complex Poynting Theorem . . . . . . . . . . . . 47<br />

vii


9 Summary in <strong>Swedish</strong> 49<br />

10 Conclusion and Outlook 53<br />

References 57<br />

viii


1. Introduction<br />

The existence <strong>of</strong> radio waves was first predicted by James Clerk Maxwell<br />

in the 1860’s. In 1886, Heinrich Hertz managed to generate and detect these<br />

waves for the first time and they were first named Hertzian waves. Hertz did<br />

not follow up his work by using the waves for sending and receiving information.<br />

Instead, that important step was taken by a number <strong>of</strong> radio pioneers in<br />

the 1890’s.<br />

In 1893, Nicola Tesla demonstrated the first wireless communication and<br />

described the equipment in articles. Tesla filed his first radio patent applications<br />

in 1897 and three years later they were granted 1 . Tesla continued his<br />

work on radio and in 1898 he demonstrated the first radio-controlled vessel 2 .<br />

In 1895, Guglielmo Marconi presented a radio and made his first demonstrations<br />

<strong>of</strong> radio communication. Marconi’s first American patent application,<br />

filed in 1897, was turned down because <strong>of</strong> prior art <strong>of</strong> Tesla and other<br />

inventors, and so were the succeeding applications. But in 1904, the US patent<br />

<strong>of</strong>fice changed its decision and granted Marconi US patent 3 for the invention<br />

<strong>of</strong> the radio. Marconi was also awarded the 1909 Nobel prize in <strong>Physics</strong>.<br />

Among the radio pioneers were also Alexandr Stephanovitch Popov and<br />

Edouard Branly. Already in 1885 Popov suggested that electromagnetic waves<br />

could be used to transmit messages and in 1896 he demonstrated it.<br />

For long time, Marconi was considered as the inventor <strong>of</strong> radio, but in 1943,<br />

the US supreme court decided to uphold Tesla’s radio patent. Unfortunately,<br />

Tesla never experienced the success. He died nine months earlier.<br />

Even though Marconi cannot be considered as the inventor <strong>of</strong> radio, he<br />

made important contributions to the development. In 1899, Marconi transmitted<br />

radio signals across the English channel and in 1901, he was the first<br />

person who succeded in transmitting radio wave across the Atlantic.<br />

During the first time <strong>of</strong> radio transmission, spark machines were used to<br />

generate Morse code. The Alexanderson alternator, named after its inventor<br />

Ernst Alexanderson, made it possible to generate high-frequency oscillations<br />

and allowed transmission <strong>of</strong> continuous radio waves. Reginald Fessenden invented<br />

the heterodyne principle and the amplitude modulation (AM) <strong>of</strong> radio<br />

1 US Patents 645,576 and 649,621<br />

2 US Patent 613,809<br />

3 US Patent 763,772<br />

1


waves. In 1900, Fessenden managed to transmit voice the first time, but it<br />

was not until 1906 that Fessenden made the first broadcast <strong>of</strong> voice by using<br />

the Alexanderson alternator. Among the many inventions made by Ernst<br />

Alexanderson, a tuning device for radios must be mentioned 4 . This device is<br />

fundamental for radios also today.<br />

On both sides <strong>of</strong> the Atlantic, many radio telegraph stations were built for<br />

the transatlantic communication. These facilities were equipped with Alexanderson<br />

alternators and huge antennas. All but one <strong>of</strong> these radio facilities are<br />

gone today. The only remaining station is located at Grimeton, close to Varberg,<br />

Sweden, see Figure 1.1.<br />

Figure 1.1: Radio station Grimeton (SAQ) close to Varberg, Sweden.<br />

A very important step in the development <strong>of</strong> radio was when Lee De Forest<br />

in 1906 invented the triode, a three-electrode vacuum tube. The triode made<br />

amplification <strong>of</strong> radio signals possible and soon became a fundamental part <strong>of</strong><br />

all radio equipment.<br />

Radio transmission requires that the transmitted signal can be detected by<br />

the receiving antenna. In other words, the transmitting antenna should be in<br />

line <strong>of</strong> sight with the receiving antenna. But how could it work for transatlantic<br />

radio transmissions? The explanation was given in 1902 when Heaviside and<br />

Kennelly predicted that the radio waves could be reflected <strong>of</strong>f by an ionised<br />

layer in the upper atmosphere, making them propagate to other continents.<br />

4 US Patent 1,008,577<br />

2


Although the most fundamental radio inventions already were made, the<br />

following decades <strong>of</strong>fered a lot <strong>of</strong> new inventions and applications. In the<br />

late 1920’s TV broadcast started and the 1930’s the radar (radio detection<br />

and ranging) was invented. Sir Robert Alexander Watson-Watt is considered<br />

to be the inventor <strong>of</strong> radar in the mid 1930’s. Watson-Watt also coined the<br />

phrase ionosphere for the ionised upper part <strong>of</strong> the Earth’s atmosphere. Radar<br />

can operate at any radio frequency, but the applications vary with frequency.<br />

Common for all radar applications is that a radio wave is transmitted and the<br />

echo is received, either by the same antenna or a separate antenna. The time<br />

difference between the transmission and the echo specifies the distance to the<br />

target and the signature <strong>of</strong> the echo can provide valuable information about<br />

the target.<br />

In the beginning, low frequencies were used for radio but with time, higher<br />

and higher frequencies have come into use. Today, the radio frequency spectrum<br />

is densely populated from the Long Wave (LW) radio band at 150 kHz<br />

up to about 50 GHz. Frequencies below 150 kHz is used in, e.g., submarin<br />

communication. Above 50 GHz, damping caused by rain or fog limits the applications.<br />

The radio spectrum is used for radio and television broadcasting,<br />

amateur radio, airplanes, ships, taxi, police, military communication, cellular<br />

telephones, Bluetooth, Wireless LAN, satellite communication, radar, and<br />

much more.<br />

During the last 25 years, cellular telephone has developed strongly. The<br />

first commercial systems were analogue (NMT 1981 in the Nordic countries;<br />

ETACS and NTACS 1985 in the rest <strong>of</strong> Europe; AMPS 1983 in the US). In<br />

1991 the first digital system was introduced (GSM 1991 in Europe and elsewhere;<br />

PDC 1994 in Japan; IS, PCS (1996) in US), and recently the second<br />

generation digital systems (3G) have started to operate.<br />

Other mobile services have also been (or will soon be) introduced, for example<br />

Bluetooth, WLAN (wireless local area network, the term Wi-Fi, wireless<br />

fidelity is also used), Ultra wideband (UWB), ZigBee, and WiMax.<br />

In many cases <strong>of</strong> radio broadcast and communication, only the amplitude<br />

<strong>of</strong> the signal is measured. Since both the phase and the amplitude are equally<br />

important for the characterisation <strong>of</strong> an electromagnetic wave field, half <strong>of</strong><br />

the information in the radio signal is then never used. Furthermore, in most<br />

radio applications, only one component <strong>of</strong> the electromagnetic wave field is<br />

measured. For most cases, that is enough, e.g., in receiving a signal with a<br />

transistor radio. There are applications where two or even three components<br />

<strong>of</strong> the wave field are measured, but it is implemented in far from all systems<br />

which would benefit from it. By measuring two, or preferrably three, components<br />

<strong>of</strong> the electromagnetic field it is possible to make use <strong>of</strong> its polarisation.<br />

This thesis deals mainly with polarisation and merges science and technology<br />

into applications beneficial for both <strong>of</strong> them. In Chapter 2, a short intro-<br />

3


duction to plasma and the plasma environment surrounding the Earth is given.<br />

Chapter 3 deals with electromagnetic fields and wave polarisation. Electromagnetic<br />

field measurements on Earth and in space are considered in Chapters<br />

4 and 5, respectively. In Chapters 6, polarisation is shown to be useful in<br />

analysing satellite data and in Chapter 7, the statistical nature <strong>of</strong> radio wave<br />

induced ionospheric radio emissions are investigated. Chapter 8 is devoted to<br />

the complex Poynting theorem and finally, in Chapter 9 a summary in <strong>Swedish</strong><br />

is presented.<br />

4


2. Plasma<br />

When a gas is heated to a sufficiently high temperature, the molecules or atoms<br />

in the gas start to dissociate and positively charged ions and free electrons are<br />

formed. If the ionisation level is sufficiently high, this gas <strong>of</strong> charged particles<br />

is a plasma.<br />

Plasma is usually considered the fourth state <strong>of</strong> matter after solid, liquid<br />

and gas. Much energy is needed to ionise a gas and this rarely happen naturally<br />

on Earth. The only exception is lightning where the gas particles in the<br />

conducting lightning channel become ionised. Plasma also exist in, e.g. manmade<br />

devices for industrial processes, light tubes, and plasma displays. Even<br />

though plasma is rare on Earth, it is abundant in space and 99% <strong>of</strong> the known<br />

matter in universe is in plasma state.<br />

Plasma behaves much more complicated than neutral gas because the free<br />

charges interact electromagnetically over large distances. This leads to a lot<br />

<strong>of</strong> phenomena which do not take place in a neutral gas. The physics needed to<br />

describe many <strong>of</strong> the processes is also complicated. In general, the equations<br />

describing a plasma cannot be solved exactly and approximations have to be<br />

used.<br />

Any ionised gas does not fulfill the requirements <strong>of</strong> being a plasma. The required<br />

amount <strong>of</strong> ionisation can be estimated if we define a plasma in the<br />

following way [10]: “A plasma is a quasineutral gas <strong>of</strong> charged and neutral<br />

particles which exhibits collective behavior.” With collective behaviour is<br />

meant that the motion <strong>of</strong> the plasma is affected not only by the local plasma,<br />

but due to the long ranged electromagnetic force, also on remote properties.<br />

The quasineutral requirement means that, locally the plasma is not neutral,<br />

but globally, we can have approximately as many electrons as ions, n e ≃ n i .<br />

We must allow for small variations in the plasma density. Otherwise, all the<br />

electromagnetic forces would cancel.<br />

In presence <strong>of</strong> only a homogeneous magnetic field B, a charged particle<br />

with mass m and charge q will be affected by the Lorentz force qv × B and<br />

the particle will move in a circular orbit. The frequency <strong>of</strong> the motion is the<br />

gyr<strong>of</strong>requency<br />

ω c = qB m<br />

(2.1)<br />

Because <strong>of</strong> differencies in masses, the electron gyr<strong>of</strong>requency is much higher<br />

5


than the ion gyr<strong>of</strong>requencies.<br />

Plasma particles are influenced not only by magnetic fields. Presence <strong>of</strong>, e.g.<br />

electric fields, gravitaional fields, and magnetic field gradients cause plasma<br />

particles to drift.<br />

A plasma has a natural oscillation frequency which depends on the plasma<br />

density n = n e = n i . This frequency is the plasma frequency<br />

ω 2 p = nq2<br />

ε 0 m<br />

(2.2)<br />

which is very important in plasma physics.<br />

2.1 The magnetosphere<br />

From the upper part <strong>of</strong> the Sun’s atmosphere, a flow <strong>of</strong> plasma is emerging at<br />

a supersonic speed <strong>of</strong> around 400 km/s. This is the solar wind and it carries a<br />

magnetic field. When the solar wind comes close Earth, it feels the presence<br />

<strong>of</strong> Earth’s magnetic field. A shock front is formed and behind the shock front<br />

the solar wind flow is subsonic. A few Earth radii behind the shock front, the<br />

magnetopause is located. The magnetopause is the boundary where the solar<br />

wind pressure is balanced by the magnetic pressure <strong>of</strong> Earth’s magnetic<br />

field, p m = B0 2/(2µ 0), where B0 2 is the geomagnetic field. Inside <strong>of</strong> the magnetopause,<br />

the magnetosphere is located. A picture <strong>of</strong> the Magnetosphere is<br />

shown in Figure 2.1, where different regions have been labelled. Charged particles<br />

move under the influence <strong>of</strong> a magnetic field and form currents. The<br />

flowing currents, on the other hand, generate magnetic fields. The magnetospheric<br />

currents and magnetic fields are also shown in Figure 2.1. The subsolar<br />

point on the Magnetopause is located about ten Earth radii from Earth, but because<br />

<strong>of</strong> the dynamic nature <strong>of</strong> the solar wind, the distance varies with time.<br />

The magnetosphere functions as a shield and protects Earth from the solar<br />

wind. Mars, on the other hand, has a very weak magnetic field and the solar<br />

wind impinges directly on Mars’ atmosphere. Probably Mars once had a much<br />

denser atmosphere and the interaction between the solar wind the atmosphere<br />

is believed to be a contributing factor to the loss <strong>of</strong> Mars’ atmosphere.<br />

2.2 The ionosphere<br />

The magnetosphere has a rather low plasma density, but the density increases<br />

as we move closer to Earth. The plasma region <strong>of</strong> high density between about<br />

60 km and 1000 km altitude is called the ionosphere. Figure 2.2 shows the<br />

density pr<strong>of</strong>ile <strong>of</strong> the ionosphere. The shape <strong>of</strong> the pr<strong>of</strong>ile is caused by an in-<br />

6


Figure 2.1: The Magnetosphere. From Kivelson and Russel [15].<br />

creasing particle density as we approach the Earth from above in combination<br />

with an ionisation—mostly by ultraviolett (UV) radiation—from the sun, but<br />

also from cosmic rays and meteorites. The maximum density at about 300<br />

km altitude can be explained as follows. At very high altitude the atmosphere<br />

is very thin and even though the UV radiation is strong, there are few particles<br />

which can be ionised. The plasma density is therefore low at these high<br />

altitudes. For low altitudes, the particle density is high, but most <strong>of</strong> the UV<br />

radiation have already been absorbed at higher altitudes, giving a low plasma<br />

density. At mid-altitude regions, we have quite high particle density and still a<br />

strong UV radiation, conditions which are favourable for a high plasma density.<br />

The different layers <strong>of</strong> the ionosphere have been named in alphabetic order<br />

starting with the E-layer for the layer first discovered. The names <strong>of</strong> the layers<br />

are shown in Figure 2.2.<br />

The ionised ionosphere can be used as a mirror for transmitting radio signals<br />

around the Earth. For vertically injected signals, the frequency must be<br />

lower than a critical frequency, which is equal to the plasma frequency <strong>of</strong> the<br />

densest part <strong>of</strong> the ionosphere. Frequencies higher than the critical frequency<br />

will penetrate the ionosphere and be lost. By increasing the ionospheric angle<br />

<strong>of</strong> incidence, high frequencies, which otherwise would have passed through<br />

7


Figure 2.2: The electron density <strong>of</strong> the ionosphere. From Kivelson and Russel [15].<br />

the ionosphere can be made to reflect <strong>of</strong>f, as shown in Figure 2.3. Long distance<br />

radio transmissions are therefore possible in a wider frequency range.<br />

Detailed descriptions <strong>of</strong> radio wave propagation in plasma are given by Budden<br />

and Ginzburg [8, 12].<br />

The plasma density <strong>of</strong> the ionosphere can be measured by utilising the ionosphere’s<br />

ability to reflect radio waves <strong>of</strong> lower frequency than the critical frequency.<br />

This is routinely done by ionosonds. A radio beam is then transmitted<br />

into the ionosphere with a frequency which is swept from below 1 MHz up to<br />

about 20 MHz. For a certain frequency, the time for the echo to return gives<br />

the altitude for the reflection, while the frequency, according to Equation (2.2)<br />

is a measure <strong>of</strong> the plasma density. In this way, only the plasma density in<br />

the lower part <strong>of</strong> the ionosphere can be measured. The upper pr<strong>of</strong>ile must be<br />

measured with satellites equipped with top-side ionosonds.<br />

8


Figure 2.3: Radio waves <strong>of</strong> different incident angles reflecting <strong>of</strong>f the ionosphere.<br />

From The ARRL handbook [13]<br />

9


3. Electromagnetic fields and wave<br />

polarisation<br />

Charges placed close to one another interact and the situation is similar for<br />

currents, which are formed by moving charges. By introducing electromagnetic<br />

fields, the description <strong>of</strong> the interaction between charges and currents<br />

can be simplified. In this chapter, we consider the electromagnetic fields and<br />

focus on the polarisation, which is an important characteristic <strong>of</strong> electromagnetic<br />

fields.<br />

3.1 Electromagnetic fields<br />

The electromagnetic fields consist <strong>of</strong> the electric field E(t, x) and the magnetic<br />

field B(t, x), which both in general depend on time and space. These<br />

fields are coupled and related to the electric charge density ρ(t, x) and the<br />

electric current density J(t, x) by Maxwell’s equations<br />

∇ · E = ρ/ε (3.1)<br />

∇ · B = 0 (3.2)<br />

∇ × E = − ∂B<br />

∂t<br />

(3.3)<br />

∇ × B = µJ + µε ∂E<br />

∂t<br />

(3.4)<br />

In a non-conducting and source free medium, Maxwell’s equations simplifies<br />

to<br />

∇ · E = 0 (3.5)<br />

∇ · B = 0 (3.6)<br />

∇ × E = − ∂B<br />

∂t<br />

(3.7)<br />

∇ × B = µε ∂E<br />

∂t<br />

(3.8)<br />

A wave equation can be derived from Maxwell’s equations by first taking<br />

the curl <strong>of</strong> Equation (3.7) and using the vector identity ∇ × (∇ × a) =<br />

11


∇·(∇·a)−∇ 2 a, where a is an arbitrary vector. Equation (3.8) then becomes<br />

∇×(∇×E) = ∇·(∇·E)−∇ 2 E = −∇ 2 E = − ∂B<br />

∂t = −µε∂2 E<br />

∂t 2 (3.9)<br />

or simplified<br />

∇ 2 E − µε ∂2 E<br />

∂t 2 = 0 (3.10)<br />

This is the wave equation for the electric field propagating in a medium with<br />

permittivity ε, and permeability µ. The wave equation for B is found in a<br />

similar way by taking the curl <strong>of</strong> Equation (3.8). This equation is <strong>of</strong> exactly<br />

the same form as the wave equation for the electric field. All components ψ<br />

<strong>of</strong> E and B then satisfy the scalar wave equation<br />

where<br />

∇ 2 ψ − 1 v 2 ∂ 2 ψ<br />

∂t 2 = 0 (3.11)<br />

v = 1 √ µε<br />

(3.12)<br />

is the speed by which the pertubation propagates. A solution <strong>of</strong><br />

Equation (3.11) is<br />

ψ = ψ(t, ˆn · x) (3.13)<br />

where ˆn is a unit vector <strong>of</strong> unspecified direction. Defining ζ as the projection<br />

<strong>of</strong> x along ˆn, ζ = ˆn · x, we have ∇ = ˆn∂/∂ζ, so that<br />

∇ 2 ψ = ∂2 ψ<br />

∂ζ 2 (3.14)<br />

and the wave equation becomes<br />

∂ 2 ψ<br />

∂ζ 2 − − 1 ∂ 2 ψ<br />

v 2 ∂t 2 = 0 (3.15)<br />

This equation has a general solution <strong>of</strong> the form<br />

ψ(t, ζ) = ψ 1 (ζ − vt) + ψ 2 (ζ + vt) (3.16)<br />

where ψ 1 represents a pertubation propagating along increasing ζ and ψ 2 a<br />

pertubation propagating along decreasing ζ. Assuming harmonic time varia-<br />

12


tion, the solution becomes<br />

ψ(t, ζ) = ψ 0 e i(±kζ−ωt) (3.17)<br />

where k = ω/c is the wave number. By introducing the wave vector<br />

k = kˆn = kˆk = ω c ˆk = √ µε ωˆk (3.18)<br />

Equation (3.17) becomes<br />

ψ(t, ζ) = ψ 0 e i(k·x−ωt) (3.19)<br />

Since ψ denotes any components <strong>of</strong> E or B, the solution for the electromagnetic<br />

fields can be written<br />

E(t, x) = E 0 e i(k·x−ωt) (3.20)<br />

B(t, x) = B 0 e i(k·x−ωt) (3.21)<br />

These fields represent a plane wave propagating along ˆk = ˆn with E and<br />

B perpendicular to each other and to the direction <strong>of</strong> wave propagation. The<br />

fields E(t, x) nd B(t, x) are complex, while the observable physical fields<br />

are real. The physical fields are found by taking the real parts <strong>of</strong> the complex<br />

fields:<br />

E physical (t, x) = Re{E 0 e i(k·x−ωt) } (3.22)<br />

B physical (t, x) = Re{B 0 e i(k·x−ωt) } (3.23)<br />

In the following, we will work with complex fields and in order to obtain the<br />

physical fields, we have to remember to take the real part <strong>of</strong> the complex field.<br />

3.2 Polarisation and Stokes parameters<br />

Polarisation is an important property <strong>of</strong> an electromagnetic wave field. It describes<br />

how the electric and magnetic field vectors change with time. As a<br />

plane wave propagates, the tips <strong>of</strong> the field vectors describe figures in the<br />

polarisation plane, which is the plane perpendicular to the direction <strong>of</strong> wave<br />

propagation. In general, the figures are ellipses. There are two special cases,<br />

where the major axes <strong>of</strong> the ellipse is equal and a circle is traced, and where<br />

one <strong>of</strong> the major ellipse axes is zero and a straight line is traced. These three<br />

cases correspond to elliptical, circular, and linear polarisation, respectively. In<br />

all three cases, we will use the general term polarisation ellipse for the traced<br />

figure.<br />

The sense in which the polarisation ellipse is traced determines the sense<br />

13


<strong>of</strong> polarisation. If the field vector sweeps clock-wise, as seen along the direction<br />

<strong>of</strong> propagation, the sense is right-handed and if the field vector sweeps<br />

counterclock-wise, the sense is left-handed.<br />

A convenient way to describe the state <strong>of</strong> polarisation is in terms <strong>of</strong> the<br />

Stokes parameters, which were introduced in 1852 by G. G. Stokes [28]. The<br />

Stokes parameters are defined for quasi-monochromatic wave fields and we<br />

will suppose that the wave field only contains frequencies within a narrow<br />

(angular) frequency band ∆ω, centered around a mean frequency ¯ω. For a<br />

quasi-monochromatic field, the bandwidth is small compared to the mean frequency:<br />

∆ω<br />

¯ω<br />

= ∆ν<br />

¯ν ≪ 1 (3.24)<br />

The Stokes parameters are defined for a two-dimensional wave field and we<br />

choose the direction <strong>of</strong> wave propagation to be along the z axis. An electromagnetic<br />

field F(t, x) is then be represented by<br />

( ) (<br />

)<br />

F x (t, x) F x,0 e i(kz−ωt+δx)<br />

F(t, x) =<br />

=<br />

F y (t, x) F y,0 e i(kz−ωt+δy) (3.25)<br />

where F x,0 and F y,0 are real valued and δ x and δ y are real valued phase factors<br />

<strong>of</strong> the field components. The Stokes parameters are [7]<br />

I = 〈F x F ∗ x + F y F ∗ y 〉 (3.26)<br />

Q = 〈F x F ∗ x − F yF ∗ y 〉 (3.27)<br />

U = 2 Re{〈F x F ∗ y 〉} (3.28)<br />

V = 2 Im{〈F x F ∗ y 〉} (3.29)<br />

where the brackets symbolise time average and ∗ complex conjugation. The<br />

parameters satisfy the relation<br />

Q 2 + U 2 + V 2 ≤ I 2 (3.30)<br />

Here, I symbolises the intensity <strong>of</strong> the wave field. The identity in Equation<br />

(3.30) only holds for a monochromatic wave field.<br />

The degree <strong>of</strong> polarisation [7] is defined by<br />

√<br />

Q 2 + U 2 + V 2<br />

p =<br />

I<br />

(3.31)<br />

The degree <strong>of</strong> polarisation varies between zero and unity and a value p = 0<br />

represents a totally unpolarised wave field and p = 1 a completely polarised.<br />

The parameters Q and U represent linear polarisation and V circular polarisa-<br />

14


tion.<br />

A geometrical representation <strong>of</strong> the Stokes parameters, the Poincaré<br />

sphere, was introduced in 1892 by H. Poincaré. In this representation, the<br />

Stokes parameters describe a point with coordinates (Q, U, V ) on, or inside,<br />

the Poincaré sphere. The point lies on the sphere for a completely polarised<br />

wave field, inside the sphere for a partially polarised, and at the centre <strong>of</strong> the<br />

sphere for a totally unpolarised wave field. The Poincaré sphere is shown in<br />

Figure 3.1. The radius <strong>of</strong> the sphere is I and the degree <strong>of</strong> polarisation, p, is,<br />

in general, less than I.<br />

V<br />

p<br />

I<br />

2χ<br />

U<br />

2ψ<br />

Q<br />

Figure 3.1: The Poincaré sphere visualises the state <strong>of</strong> polarisation with the Stokes<br />

parameters Q, U, and V representing the coordinates <strong>of</strong> p, the degree <strong>of</strong> polarisation.<br />

Three <strong>of</strong> the Stokes parameters can also be expressed in terms <strong>of</strong> the angles<br />

ψ and χ, shown in the Poincaré sphere:<br />

Q = √ Q 2 + U 2 + V 2 cos 2χ cos 2ψ (3.32)<br />

U = √ Q 2 + U 2 + V 2 cos 2χ sin 2ψ (3.33)<br />

V = √ Q 2 + U 2 + V 2 sin 2χ (3.34)<br />

The angle χ is a measure <strong>of</strong> the degree <strong>of</strong> circular polarisation, or the ellipticity<br />

<strong>of</strong> the polarisation ellipse, while ψ specifies the orientation <strong>of</strong> the polarisation<br />

15


ellipse and can be found from<br />

tan 2ψ = U Q<br />

(3.35)<br />

Figure 3.2 shows the polarisation ellipse, where ψ specifies the orientation <strong>of</strong><br />

the ellipse, and a and b are the two major axes <strong>of</strong> the ellipse.<br />

y<br />

b<br />

a<br />

2Fy,0<br />

ψ<br />

x<br />

2F x,0<br />

Figure 3.2: The polarisation ellipse.<br />

If we consider a completely polarised state, linear polarisation is represented<br />

by points on the equator <strong>of</strong> the Poincaré sphere, circular polarisation<br />

by the poles <strong>of</strong> the sphere, where the north pole represents left-hand circular<br />

polarisation and the south pole right-hand circular polariation. All other points<br />

on the Poincaré sphere represent elliptical polarisation with left-hand polarisation<br />

on the northern hemisphere and right-hand on the southern. Figure 3.3<br />

shows a 2D projection <strong>of</strong> the polarisation on the Poincaré sphere.<br />

3.3 The coherency tensor<br />

The Stokes parameters can also be obtained from the coherency tensor, constructed<br />

from the wave field F(t, x) in Equation (3.25). It is defines as<br />

(<br />

)<br />

J = 〈FF † 〈Fx,0 2 〉 =<br />

〉 〈F x,0F y,0 e −iδ 〉<br />

〈F x,0 F y,0 e iδ 〉 〈Fy,0 2 〉 (3.36)<br />

where † symbolises Hermitean conjugate and δ = δ y − δ x . Using the definitions<br />

<strong>of</strong> Stokes parameters in Equations (3.26)–(3.29), the coherency tensor<br />

16


2χ<br />

2ψ<br />

0 ◦<br />

−90 ◦ 0 ◦<br />

90 ◦ 180 ◦<br />

−180 ◦ 90 ◦<br />

−90 ◦<br />

Figure 3.3: A 2D projection <strong>of</strong> the polarisation on the Poincaré sphere. Points on the<br />

equator represent linear polarisation, while the two poles represent circular polarisation.<br />

Points between the poles and the equator have elliptical polarisation. Points in<br />

the upper hemisphere are left-hand polarised and in the lower right-hand polarised.<br />

The orientation <strong>of</strong> the polarisation ellipse is specified by the angle 2ψ. After one turn<br />

around the equator <strong>of</strong> the Poincaré sphere, the polarisation ellipse has only turned<br />

180 ◦ , but the state <strong>of</strong> polarisation is back where it started.<br />

can be expressed in terms <strong>of</strong> the Stokes parameters as<br />

(<br />

)<br />

J = 1 I + Q U − iV<br />

2 U + iV I − Q<br />

(3.37)<br />

The total intensity I is equal to the trace <strong>of</strong> the coherency tensor, I = Tr(J).<br />

The coherency tensor has some interesting properties. In 1930, N. Wiener<br />

[33] showed that the unit tensor and the three Pauli spin matrices form a base<br />

in which the coherency tensor can be expanded. U. Fano [11] later showed that<br />

the coefficients in this expansion are the Stokes parameters. The expansion is<br />

J = 1 2 (I1 2 + Uσ 1 + V σ 2 + Qσ 3 ) (3.38)<br />

[ ( ) ( ) ( ) ( )]<br />

= 1 1 0 1 0<br />

0 1 0 −i<br />

I + Q<br />

+ U + V<br />

2 0 1 0 −1 1 0 i 0<br />

17


3.4 The spectral tensor<br />

The Stokes parameters can also be defined in the frequency domain. Let<br />

˜F(ω, x) denote the Fourier transform <strong>of</strong> F(t, x). Then the spectral density<br />

tensor is<br />

S d = ˜F˜F † =<br />

( )<br />

˜Fx ˜F<br />

∗<br />

x<br />

˜Fx ˜F<br />

∗<br />

y<br />

˜F y ˜F<br />

∗<br />

x<br />

˜Fy ˜F<br />

∗<br />

y<br />

(3.39)<br />

When the spectral density tensor is used the fields do not have to be quasimonochromatic.<br />

We can define the spectral Stokes parameters I, Q, U, and<br />

V, which are analogous to the ordinary Stokes parameters.<br />

If the wave field has a small bandwidth ∆ω, we can integrate over that<br />

bandwidth by applying the bandwidth operator (Paper I):<br />

¯B =<br />

∫ ¯ω+∆ω/2<br />

¯ω−∆ω/2<br />

dω (3.40)<br />

This operator integrates over the small bandwidth ∆ω, centered around the<br />

mean frequency ¯ω. Applying the bandwidth operator to the spectral density<br />

tensor gives the spectral tensor:<br />

S = ¯BS d =<br />

∫ ¯ω+∆ω/2<br />

¯ω−∆ω/2<br />

S d dω (3.41)<br />

3.5 Wave polarisation in 3D<br />

The Stokes parameters were introduced to describe the polarisation <strong>of</strong> light. In<br />

optics, the location <strong>of</strong> the source is known because it can easily be observed<br />

and therefore only two field components in the polarisation plane are necessary<br />

to describe the transverse electromagnetic wave field. If, on the other<br />

hand, the location <strong>of</strong> the source is unknown, as for most radio waves, we must<br />

consider all three field components.<br />

Using the Fourier transform <strong>of</strong> a 3D wave field, ˜F(ω, x), the spectral density<br />

tensor becomes<br />

⎛<br />

S d = ˜F˜F † =<br />

⎜<br />

⎝<br />

⎞<br />

˜F x ˜F<br />

∗<br />

x<br />

˜Fx ˜F<br />

∗<br />

y<br />

˜Fx ˜F<br />

∗<br />

z<br />

˜F y ˜F<br />

∗<br />

x<br />

˜Fy ˜F<br />

∗<br />

y<br />

˜Fy ˜F<br />

∗ ⎟<br />

z ⎠ (3.42)<br />

˜F z ˜F<br />

∗<br />

x<br />

˜Fz ˜F<br />

∗<br />

y<br />

˜Fz ˜F<br />

∗<br />

z<br />

In Paper I, we have noted that the Pauli spin matrices, which were used<br />

in the expansion <strong>of</strong> the coherency tensor in Equation (3.38) are generators <strong>of</strong><br />

the special unitary group SU(2). For 3D, the natural choice is to expand the<br />

18


spectral density tensor in the generators <strong>of</strong> the symmetry group SU(3). This<br />

has been done in Paper I and the unit tensor and the eight generators described<br />

by Gell-Mann and Ne’eman [20] were used. The coefficients in the expansion<br />

are nine generalised polarisation parameters. If the generalised polarisation<br />

parameters are denoted by Λ i , the unit tensor in 3D by 1 3 , and the generators<br />

<strong>of</strong> the SU(3) group by λ i , the spectral density tensor becomes<br />

S d = 1 3 1 3 + 1 8∑<br />

Λ i λ i (3.43)<br />

2<br />

i=1<br />

where the normalisation 1/3 has been chosen to give the spectral intensity I<br />

the same meaning as in the 2D case.<br />

The explicit expression <strong>of</strong> the spectral density tensor is presented in Paper<br />

I, where some interesting properties <strong>of</strong> the spectral density tensor also is<br />

described. One such property is the pseudo vector V, formed from the antisymmetric<br />

part <strong>of</strong> the spectral density tensor:<br />

⎛<br />

⎞<br />

−Im{ ˜F y ˜F<br />

∗<br />

z }<br />

⎜<br />

V = 2 ⎝ Im{ ˜F x ˜F<br />

∗<br />

z }<br />

−Im{ ˜F x ˜F<br />

∗<br />

y }<br />

⎟<br />

⎠ (3.44)<br />

This pseudo vector is normal to the plane <strong>of</strong> the polarisation ellipse. The<br />

pseudo vector V is also parallel to the vector described by Means [22]. The<br />

magnitude <strong>of</strong> V is coupled to the spectral Stokes parameter V by the relation<br />

|V| = |V| (3.45)<br />

A complex electromagnetic wave field can be fully described in terms <strong>of</strong><br />

three amplitudes and three phases, so only at most six <strong>of</strong> the generalised polarisation<br />

parameters can be independent. We also note that the spectral density<br />

tensor does not change if an arbitrary phase is added to all three field components.<br />

Therefore, only five <strong>of</strong> the nine generalised polarisation parameters can<br />

be independent.<br />

19


4. Electromagnetic field measurements<br />

In this chapter, a method and a system for vector measurements <strong>of</strong> electromagnetic<br />

field are described. Applications are considered in Section 4.3 for the HF<br />

band (3-30 MHz), in Section 4.4 for a radio telescope, and in Section 4.5 for<br />

wireless telecommunication in the microwave regime. <strong>Space</strong> applications are<br />

presented in the next chapter.<br />

4.1 Method for estimating polarisation<br />

parameters<br />

A method for estimation <strong>of</strong> polarisation parameters is described in Paper I,<br />

Paper II, and Paper V. The method requires knowledge <strong>of</strong> an electromagnetic<br />

vector field, such as E or B, and is based on the spectral density tensor. Expressed<br />

in terms <strong>of</strong> the Fourier transform Ẽ <strong>of</strong> the electric field, the spectral<br />

density tensor is<br />

⎛<br />

⎞<br />

S d = ẼẼ† =<br />

⎜<br />

⎝<br />

Ẽ x Ẽ ∗ x<br />

Ẽ y Ẽ ∗ x<br />

Ẽ x Ẽ ∗ y<br />

Ẽ y Ẽ ∗ y<br />

Ẽ x Ẽ ∗ z<br />

Ẽ y Ẽ ∗ z<br />

⎟<br />

⎠ (4.1)<br />

Ẽ z Ẽ ∗ x<br />

Ẽ z Ẽ ∗ y<br />

Ẽ z Ẽ ∗ z<br />

The spectral density tensor contains five independent polarisation parameters,<br />

which can be used to extract information from the wave field. The first<br />

<strong>of</strong> these is the spectral intensity I, which is equal to the trace <strong>of</strong> the spectral<br />

density tensor:<br />

I = Tr(S d ) = |Ẽx| 2 + |Ẽy| 2 + |Ẽz| 2 (4.2)<br />

The next thing to consider is the pseudo vector V, which is obtained from<br />

the antisymmetric (imaginary) part <strong>of</strong> the spectral density tensor:<br />

{<br />

}<br />

V = (V x , V y , V z ) = −2 Im ˆxẼyẼz − ŷẼxẼz + ẑẼxẼy (4.3)<br />

This pseudo vector contains three independent parameters. Two <strong>of</strong> them specifies<br />

the direction <strong>of</strong> the surface normal <strong>of</strong> the polarisation ellipse. Using simple<br />

vector analysis, they can be expressed as the polar and azimuthal angle in<br />

21


a sperical coordinate system:<br />

cos θ = V y<br />

V x<br />

(4.4)<br />

tan φ = V z<br />

|V|<br />

(4.5)<br />

The third independent parameter contained in V is the degree <strong>of</strong> circular polarisation,<br />

defined by<br />

r c = V I<br />

(4.6)<br />

where the magnitude <strong>of</strong> V is given by Equation (3.45). Unless some additional<br />

information about the wave field is available, the sign <strong>of</strong> V cannot be determined.<br />

This means that there is an 180 ◦ ambiguity in the determination <strong>of</strong> the<br />

surface normal <strong>of</strong> the polarisation ellipse. The pseudo vector V points in the<br />

direction given by a right-hand polarised wave field, but the measured wave<br />

could equally well be a left-hand polarised wave propagating in the opposite<br />

direction.<br />

There are a number <strong>of</strong> ways to solve the ambiguity. For example, if the signal<br />

appear to come from a direction below ground, it is likely that it comes<br />

from the opposite direction above ground. Apart from measuring, in our case,<br />

three electric field components, knowledge <strong>of</strong> one <strong>of</strong> the magnetic field components<br />

will in most cases give the answer. A better method is, <strong>of</strong> course, to<br />

measure both the electric and magnetic fields, E and B, and then calculate<br />

the Poynting vector N = 1<br />

µ 0<br />

E × B, which points in the direction <strong>of</strong> wave<br />

propagation. Another method is to use several vector field antennas to form<br />

an array.<br />

So far, we have found four independent parameters. The fifth is the tilt angle<br />

<strong>of</strong> the polarisation ellipse, given by Equation (3.35). From that equation, it is<br />

clear that the tilt angle is related to linear polarisation by the spectral Stokes<br />

parameters Q and U. Instead <strong>of</strong> expressing the fifth independent parameter as<br />

the tilt angle, the total degree <strong>of</strong> polarisation, defined in Equation (3.31) can<br />

be used.<br />

Since the method for direction-finding is built on the normal vector <strong>of</strong> the<br />

polarisation ellipse, the accuracy <strong>of</strong> the method scales with the degree <strong>of</strong> circular<br />

polarisation. For linear polarisation, the ellipse has collapsed to a line<br />

and the normal to that line can lie anywhere in a plane perpendicular to the<br />

line. Reflections also affect the accuracy <strong>of</strong> the method, specially for the polar<br />

angle, or the elevation angle. The azimuthal angle is not so sensitive to<br />

reflections. A discussion about reflections is given in Section 2.2 <strong>of</strong> Paper V.<br />

22


4.2 System for estimating polarisation<br />

parameters<br />

A system for estimation <strong>of</strong> polarisation parameters is described in Paper II.<br />

The system is general in the sense that it can operate at any radio frequency<br />

and can use any means for measuring an electromagnetic vector field. The system<br />

outputs polarisation parameters such as the direction <strong>of</strong> arrival or the direction<br />

<strong>of</strong> wave propagation, the intensity <strong>of</strong> the wave field at any frequency in<br />

the band <strong>of</strong> operation, the degree <strong>of</strong> circular polarisation, and so on. The means<br />

used for receiving the vector electormagnetic field is not specified. Any kind<br />

<strong>of</strong> antennas can be used for extracting the electric or magnetic vector fields, or<br />

both <strong>of</strong> them. Since an orthogonalisation can be applied to the received vector<br />

fields, the antennas do not have to be orthogonal either.<br />

In order to measure the electromagnetic fields for a particular frequency<br />

band, suitable antennas must be used. In Figure 4.1, two antenna configurations<br />

for measuring the electric or magnetic vector field is presented. In a)<br />

three orthogonal electric antennas are shown and in b) three orthogonal magnetic<br />

antennas. In c) the antennas in b) has been assembled to form a tri-axial<br />

antenna, or a magnetic vector field sensor. Figure 4.2 shows two orientations<br />

<strong>of</strong> the same configuration <strong>of</strong> electric vector field antennas. In the first case,<br />

the coordinate system becomes simple, but the reflections from ground will<br />

be different for vertical and horisontal antennas. In the second case, all antennas<br />

are oriented with the same angle to ground and the electric fields will<br />

experience the same reflection coefficients for all three antennas.<br />

4.3 The high frequency band<br />

Electromagnetic wave propagation in the high frequency (HF) band (3-30<br />

MHz) and below is complex because <strong>of</strong> reflections from the ionosphere. The<br />

ionosphere sometimes behaves like a perfect mirror and sometimes it is completely<br />

irregular. In the first case, an electromagnetic wave is reflected from<br />

the ionosphere mainly from one point, making it possible to obtain a good estimate<br />

<strong>of</strong> the angle <strong>of</strong> arrival and the polarisation <strong>of</strong> the wave. When the ionosphere<br />

is irregular, the received waves has been reflected from many points <strong>of</strong><br />

the ionosphere, making the direction <strong>of</strong> arrival fluctuate. In this second case,<br />

averaging over time is required for an accurate estimate.<br />

The accuracy <strong>of</strong> direction-finding systems in the HF-band are dependent on<br />

the state <strong>of</strong> the ionosphere, the amount <strong>of</strong> reflections from ground and other<br />

objects near the receiving antenna, and the equipment used to measure the<br />

wave field.<br />

Paper IV presents a vector field receiver developed at the <strong>Swedish</strong> <strong>Institute</strong><br />

23


z<br />

x<br />

y<br />

x<br />

z<br />

y<br />

y<br />

x<br />

z<br />

(a)<br />

(b)<br />

Figure 4.1: Antenna configurations, which can be used to measure the vector fields.<br />

In a) three orthogonal electric dipole antennas with pre-amplifiers are shown. These<br />

antennas will be mounted with a common centre. In b) three orthogonal loops with<br />

pre-amplifiers are displayed and in c) the three loops has been put together with a<br />

common origin.<br />

(c)<br />

Figure 4.2: Electric vector field sensors with two different orientations <strong>of</strong> the antennas<br />

with respect to ground. The first one has been used for field test and the second is used<br />

for the LOIS test station, see Section 4.4.<br />

24


<strong>of</strong> <strong>Space</strong> <strong>Physics</strong> in <strong>Uppsala</strong>, Sweden, see Figure 4.3. It is a fully digital receiver<br />

for the frequency band 0–20 MHz and it samples the antenna signals directly<br />

after the anti-aliasing filers and amplifiers. Digital mixers down-convert<br />

the signals to complex base band in-phase (I) and quadrature (Q). The vector<br />

receiver outputs a digital stream with 16+16 bits resolution and a bandwidth<br />

between 1.5 kHz and 82 kHz. The vector receiver can be controlled remotely<br />

and data can be streamed over the internet to the user.<br />

DIGITAL MIXER CHIP<br />

I x<br />

Input<br />

x channel<br />

cos<br />

High decimation<br />

(LP) filter<br />

FIR (LP) filter<br />

Preamp<br />

Anti−aliasing<br />

(LP/BP) filter<br />

Diff amplifier<br />

10 dB gain<br />

010<br />

ADC<br />

Phase/frequency<br />

generator<br />

High decimation<br />

sin (LP) filter<br />

Scale<br />

Factor<br />

(digital<br />

gain)<br />

FIR (LP) filter<br />

Qx<br />

DIGITAL MIXER CHIP<br />

I y<br />

Control<br />

Input<br />

y channel<br />

cos<br />

Unit<br />

010<br />

Scale<br />

Factor<br />

DSP<br />

LAN<br />

sin<br />

Qy<br />

FPGA<br />

PC<br />

DIGITAL MIXER CHIP<br />

Iz<br />

Input<br />

z channel<br />

cos<br />

Scale<br />

010<br />

Factor<br />

sin<br />

Qz<br />

Figure 4.3: Block diagram <strong>of</strong> a three-channel digital vector field receiver. The antennas<br />

are external and can be exchanged. The three inputs <strong>of</strong> the instrument are marked with<br />

arrows.<br />

4.4 LOIS—the LOFAR Outrigger In<br />

Scandinavia<br />

The Low frequency array (LOFAR) will be built in the Netherlands, see Figure<br />

4.4. It is the first <strong>of</strong> a new generation radio telescopes, which instead <strong>of</strong><br />

being composed <strong>of</strong> a small number <strong>of</strong> huge radio telescopes, will consist <strong>of</strong><br />

25


small and simple antennas located over an area covering the Netherlands and a<br />

part <strong>of</strong> northern Germany. In total, 15,000 crossed 2 m long V-shaped electric<br />

dipole antennas will be used. The sensitivity will be 100 times better than the<br />

present radio telecopes and the amount <strong>of</strong> collected data will require computer<br />

power <strong>of</strong> tens <strong>of</strong> Tera FLOPS.<br />

LOFAR will be sensitive to see objects very far away and thereby it will<br />

also see very far back in time, back to the early universe. LOFAR will also be<br />

used to observe the sun, the solar wind, and the highest energy cosmic rays. Its<br />

data transport network and computing facilities will make it possible to also<br />

host sensors for geoscience, bioscience and weather research.<br />

Figure 4.4: The locations <strong>of</strong> the radio telescopes LOFAR and LOIS.<br />

LOIS is an extension to the Low frequency array. The LOIS will be located<br />

in the southern Sweden with Växjö as a hub, Figure 4.4. It will contribute to<br />

LOIS with up to a few thousands simple electric dipole antennas. These antennas<br />

will be grouped three and three to form 3D vector sensors, as shown in the<br />

right-hand part <strong>of</strong> Figure 4.2. LOIS will also contribute with a radar for active<br />

experiments. The first step towards building LOIS has been to set up a test station<br />

in Risinge outside <strong>of</strong> Växjö, see Figure 4.5. The test station is currently<br />

26


housing five <strong>of</strong> the vector receivers described in Section 4.3. These receivers<br />

are running continously and on request streaming data over the internet to the<br />

user. S<strong>of</strong>tware for accessing these vector receivers can be downloaded from<br />

(http://www.lois-space.net/s<strong>of</strong>tware.html).<br />

Figure 4.5: LOIS test station in Risinge outside <strong>of</strong> Växjö, Sweden. The figure only<br />

show four tri-axial antennas, but a fifth has been installed since the photo was taken.<br />

Photo by Jan Bergman.<br />

4.5 Wireless telecommunication<br />

For long time, polarisation has been underestimated in telecommunication.<br />

Only two kinds <strong>of</strong> polarisations have been considered for base station antennas:<br />

vertical and horisontal. Actually, a better formulation would be vertical or<br />

horisontal. The reason is that when base station antennas utilise polarisation<br />

diversity, only the strongest <strong>of</strong> the two components is selected.<br />

The purpose <strong>of</strong> Paper III and Bergman et al. [6] was to introduce a new<br />

dimension in telecommunication with focus on a full use <strong>of</strong> the polarisation.<br />

This was achieved by, adding the third polarisation and also making proper<br />

use <strong>of</strong> all three measured signals.<br />

By measuring all three polarisations, the direction to the mobile terminals<br />

(telephones) can be estimated. Even in a reflective environment, the estimation<br />

<strong>of</strong> the direction to a mobile terminal could be <strong>of</strong> importance. Such examples<br />

are, e.g., capability <strong>of</strong> directing the transmitted power mainly in the direction<br />

<strong>of</strong> the mobile terminal, putting a null <strong>of</strong> the receiving antenna field pattern in<br />

the direction <strong>of</strong> a mobile terminal in order to prevent it from disturbing the<br />

27


communication to another mobile terminal at the same frequency, and higher<br />

accuracy in positioning mobile terminals when information about the direction<br />

to a terminal can be used together with time differences measured by several<br />

base stations.<br />

To measure more than one polarisation in mobile applications has also implications<br />

on the received power. If two polarisations were measured instead<br />

<strong>of</strong> only one, the received signal would, on the average, be twice as strong or<br />

3 dB stronger. By also measuring the third polarisation, the received signal is<br />

three times stronger, or 4.8 dB stronger. This means that weaker signals can<br />

be detected and the signal to noise ratio increases.<br />

The importance <strong>of</strong> polarisation diversity was demonstrated by a research<br />

team from Bell Labs, which transmitted a Miro painting in colour over three<br />

channels <strong>of</strong> orthogonal polarisation [3]. They demonstrated that the capacity<br />

<strong>of</strong> a wireless transmission can be trippeled by using three orhogonal polarisations<br />

<strong>of</strong> the electric field, or even incresed six times by using the three<br />

polarisations <strong>of</strong> the magnetic field as well.<br />

In an experiment performed by Redsnake Radio Technology 1 , Stockholm,<br />

three parallel vertically oriented dipoles were compared with three orthogonal<br />

dipoles, see Figure 4.6. The measurements were performed at 2.45 GHz in a<br />

reflective indoor environment. The result was that the number <strong>of</strong> antennas determines<br />

the capacity independently <strong>of</strong> how they are oriented. Three mutually<br />

perpendicular antennas give the same capacity as three parallel antennas [1].<br />

1 Redsnake Radio Technology AB was founded by Jan Bergman, Tobia Carozzi, Anders Engström,<br />

Anders Eriksson, Roger Karlsson, and Johan Warnström in 2000. The business idea<br />

was to improve the use <strong>of</strong> bandwidth in cellular systems by making use <strong>of</strong> the last information<br />

carrying dimension, the polarisation. In 2002, the company was closed down.<br />

28


Figure 4.6: The antennas used in the capacity measurements performed by Redsnake<br />

Radio Technology AB [1]. The antenna, placed in the stand on the left-hand side, was<br />

used to transmit three orthogonal polarisations. The three antennas on the right-hand<br />

side were used as receiving antennas. Either the three vertical elements were used<br />

for the antenna array with equal polarisations, or the antenna with three orthogonal<br />

polarisations were used alone.<br />

29


5. <strong>Space</strong> applications<br />

In this chapter, the conditions for antennas in space will be considered, followed<br />

by a presentation <strong>of</strong> satellite projects for which the vector field sensor<br />

has been selected as payload.<br />

5.1 Antennas in space<br />

The conditions for measuring electromagnetic fields with satellites differ in<br />

many ways from ground-based measurements. In space there is no ground or<br />

other bodies in the vicinity <strong>of</strong> the spacecraft and the field pattern is relatively<br />

undisturbed and free from reflections. The only disturbance is the structure<br />

<strong>of</strong> the spacecraft itself. The importance <strong>of</strong> this effect depends on the size <strong>of</strong><br />

the spacecraft in comparison to the antennas. If the body <strong>of</strong> a spacecraft is<br />

large, i.e., larger than λ/2, the spacecraft has the same function as the ground<br />

for antennas placed above the Earth. For a spacecraft much smaller than the<br />

wavelength, i.e.,


electric antennas are not a source <strong>of</strong> instability. Long antennas can still be a<br />

problem because rapid change <strong>of</strong> the attitude or the orbit can cause them to<br />

start oscillating, which in turn might disturb the stability <strong>of</strong> the spacecraft.<br />

In general, antennas for measuring the magnetic field can be made much<br />

smaller than electric antennas and they do not either require a cumbersome<br />

deploy mechanism as is the case for most electric antennas. Because <strong>of</strong> the<br />

small size <strong>of</strong> the magnetic field antennas, all three field components can relatively<br />

easy be measured. Consequently, measurements <strong>of</strong> the magnetic field in<br />

space do not suffer from the same restrictions as electric field measurements.<br />

5.2 Satellite projects<br />

Instruments built at the <strong>Swedish</strong> <strong>Institute</strong> <strong>of</strong> <strong>Space</strong> <strong>Physics</strong> in <strong>Uppsala</strong> (IRF-<br />

U) for measuring the vector fields in the HF band have already been selected<br />

as payloads. It is also under consideration for a number <strong>of</strong> other satellites. All<br />

projects are listed below.<br />

Obstanovka-1<br />

The vector field sensor has been selected for the Radio Frequency Analyzer<br />

(RFA), which is a part <strong>of</strong> the Obstanovka-1 instrument package [16] onboard<br />

the International <strong>Space</strong> Station (ISS). The <strong>Space</strong> Research <strong>Institute</strong> (IKI) in<br />

Moscow, Russia, is responsible for Obstanovka-1, which consists <strong>of</strong> seven<br />

different instruments and is planned to be launched 2006/2007. Obstanovka is<br />

Russian for environment and the instruments are going to monitor the plasma<br />

environment <strong>of</strong> ISS.<br />

The Obstanovka-1 instruments will be mounted on the Russian part <strong>of</strong> ISS.<br />

Figure 5.1 shows ISS and the Russian modules are closest to the observer. On<br />

the nearest <strong>of</strong> the thicker modules, the instruments will be mounted on the<br />

side facing the Earth, as illustrated in Figure 5.2. An explanation <strong>of</strong> what the<br />

instrument acronymes in Figure 5.2 represent is given in Table 5.1.<br />

The RFA will measure the electromagnetic fields up to 18 MHz and is being<br />

built by the <strong>Space</strong> Research Centre (CBK) <strong>of</strong> the Polish academy <strong>of</strong> sciences,<br />

Warsaw, Poland, and IRF-U. The main unit consists <strong>of</strong> two parts, a<br />

wave recorder (WRC) from Poland and a vector receiver (VRX) built by IRF-<br />

U. The WRC has two receiver channels and samples the signals directly to<br />

the memory. The three-channel VRX can perform frequency sweeps and data<br />

reduction by digital down-conversion to base band.<br />

An illustration <strong>of</strong> the RFA antennas is shown in Figure 5.3. These antennas<br />

is under construction by CBK and will measure the vector electric and<br />

32


Figure 5.1: The international space station (ISS). The Russian modules (RS ISS) are<br />

closest to the observer. Photo from NASA.<br />

Figure 5.2: A sketch <strong>of</strong> how the Obstanovka-1 instrument packages CWD-1 and<br />

CWD-2 are going to be mounted on the second closest ISS module in Figure5.1.<br />

The RFA antennas are shown in the lower right corner <strong>of</strong> the figure and the three<br />

crossed orthogonal electric dipole antennas are labeled AD and the three orthogonal<br />

loop antennas AM. Figure from IKI.<br />

33


Table 5.1: Obstanovka-1 scientific instrument packages CWD1-1 and CWD1-2<br />

Unit Name Country<br />

LP1,-2 Langmuir probe Bulgaria<br />

CWS-1,-2 Combined wave sensor Ukraine<br />

DFM-1 Flux gate magnetometer Russia<br />

DFM-2 Flux gate magnetometer Ukraine<br />

RFA Radio frequency analyser Poland, Sweden<br />

CORES Correlating electron spectrograph UK<br />

DP-1,-2 <strong>Space</strong>craft potential monitor Bulgaria<br />

SAS3 Signal analyser and sampler Hungary<br />

magnetic fields. A switch inside <strong>of</strong> the main unit allows selection <strong>of</strong> any combination<br />

<strong>of</strong> the six field components to the five receivers.<br />

Figure 5.3: The RFA antennas with the three crossed dipoles to the left and the three<br />

orthogonal loops to the right. The length <strong>of</strong> the electric dipoles is 1 m. Figure from<br />

CBK.<br />

COMPASS-2<br />

A one-channel version <strong>of</strong> the Obstanovka-1 Radio frequency analyser has<br />

been selected as payload onboard the Russian microsatellite COMPASS-2 [2].<br />

COMPASS-2 is a disaster warning satellite designed to monitor and warn for<br />

man-made and natural disasters, such as earthquakes, and to study the electrodynamic<br />

coupling between the atmosphere, ionosphere and magnetosphere.<br />

COMPASS-2 has been built by the <strong>Institute</strong> <strong>of</strong> Terrestial Magnetism,<br />

Ionosphere and Radio Wave Propagation <strong>of</strong> the Russian Academy <strong>of</strong> Science<br />

(IZMIRAN), Russia. The mass <strong>of</strong> COMPASS-2 is 85 kg and it will be<br />

34


launched on 15 September from a Russian submarine into a circular orbit<br />

<strong>of</strong> 400 km altitude and 79 ◦ inclination. Figure 5.4 shows the COMPASS-2<br />

microsatellite.<br />

Figure 5.4: COMPASS-2 microsatellite. Figure from IZMIRAN.<br />

Microlink-1<br />

Microlink-1, Figure 5.5, is a <strong>Swedish</strong> nano satellite which will be built<br />

by the Ångström Aerospace Corporation (ÅAC) in <strong>Uppsala</strong>. The purpose<br />

<strong>of</strong> Microlink-1 is to demonstrate advanced multifunctional micro- and<br />

nanotechnology for space applications. With advanced technology is<br />

meant that traditional aluminium structure will be replaced by multichip<br />

modules built in silicon and containing most <strong>of</strong> the electronics and new<br />

micropropulsion systems. The payload will be minaturised and sensors will<br />

be developed with new technology and lightweight materials.<br />

Microlink-1 has a mass <strong>of</strong> 10 kg and in its deployed configuration, excluding<br />

the antennas, it has the size <strong>of</strong> approximately 100×100×10 cm 3 .<br />

Microlink-1 is planned to carry in total four scientific instruments, listed in<br />

Table 5.2.<br />

Table 5.2: Microlink-1 scientific instruments. The EFVS antennas has been extracted<br />

25 cm <strong>of</strong> maximum 100 cm.<br />

Instrument Measures Range<br />

Electric field vector sensor 3D electric field Few kHz–20 MHz<br />

Langmuir probe Density and temp. DC–10 kHz<br />

Anisotropic magnetoresistive mag. 3D magn. field DC–10 kHz<br />

Flux gate magnetometer 3D magn. field DC–100 Hz<br />

The Electric field vector sensor (EFVS), the Langmuir probe, and the<br />

Anisotropic magnetoresistive magnetometer will all be developed by IRF-U,<br />

while the Flux gate magnetometer will be built by Alfvén lab at the Royal<br />

35


<strong>Institute</strong> <strong>of</strong> Technology in Stockholm. The electric field antennas <strong>of</strong> the EFVS<br />

and the Langmuir probe booms are placed symmetrically at the corners <strong>of</strong> the<br />

solar panels and span all three spatial dimensions. This antenna configuration<br />

keeps the moment <strong>of</strong> inertia concentrated to the spin plane.<br />

The time schedule depends on the financing and according to the present<br />

situation, launch is planned 2007/2008.<br />

Figure 5.5: Microlink-1. Figure from Ångström <strong>Space</strong> Technology Center (ÅSTC).<br />

Future satellite projects<br />

Apart from the projects mentioned above, which all are at advanced stages,<br />

there are a number <strong>of</strong> projects possible for the future. The first <strong>of</strong> these is<br />

Canopus-Volcano, which is a follow-up to COMPASS-2 with launch 2006.<br />

Another project is Chandrayaan-1 which is an Indian moon mission. IRF-U<br />

has together with an Indian research institute and an Indian university, both in<br />

Pune, submitted a proposal for the Electromagnetic vector information sensor<br />

(ELVIS). ELVIS measures the vector electric field from a few kHz up to<br />

18 MHz and one component <strong>of</strong> the magnetic field from a few Hz up to 100<br />

kHz. A transient detection mode is also implemented for electric fields up to<br />

200 MHz. The Indian space board, ISRO, has not yet come to decision about<br />

ELVIS.<br />

Discussions are also taking place with the Taiwanese space board about an<br />

instrument for a future Taiwanese satellite planned to be launched in 2008.<br />

Additional contacts have also been made for satellite projects in Turkey, Mexico,<br />

and Russia. IRF-U are also participating in a working group for development<br />

<strong>of</strong> infrastructure and instruments for a future radio observatory on the<br />

Moon (2020–2030).<br />

36


6. Whistler analysis using polarisation<br />

Whistler waves are circularly polarised electromagnetic waves propagating<br />

along the geomagnetic field in the magnetosphere. They are generated by<br />

lightning which forms a broadband electromagnetic pulse. The pulse starts to<br />

propagate along the geomagnetic field. When the wave reaches the conjugate<br />

ionosphere, it is reflected and propagates back again, see Figure 6.1. The magnetospheric<br />

plasma is dispersive causing high frequencies to propagate faster<br />

than low frequencies. The whistlers therefore aquire a typical time-frequency<br />

signature—they sound like a whistle.<br />

Figure 6.1: A whistler wave is generated by lightning (A). The whistler propagates<br />

upward along the geomagnetic field, is reflected at the conjugate ionosphere (B) and<br />

propagates back to A. Whistlers can be reflected many times between the hemispheres.<br />

From Krall and Trivelpiece [17].<br />

In Paper VI, a study <strong>of</strong> a whistler is presented. The whistler was measured<br />

by the <strong>Swedish</strong>-German Freja satellite [19] with three AC magnetic field components<br />

b(t), sampled at 4096 samples/s. Three components <strong>of</strong> the DC magnetic<br />

field B 0 were measured simultaneously. The data was taken on June 23,<br />

1993, at 21:49:29.2 UT, at an altitude <strong>of</strong> 1402 km and magnetic local time<br />

15:18 and invariant latitude 49 ◦ .<br />

A method for estimating the instantaneous polarisation properties <strong>of</strong> the<br />

whistler has been used. From b(t), the windowed Fourier transform ˜b was<br />

calculated using a Gaussian window <strong>of</strong> length 256 samples and a with full-<br />

37


width half-maximum <strong>of</strong> 29.95 samples, and an overlap <strong>of</strong> 255 samples. A<br />

spectrogram <strong>of</strong> the whistler is shown in Figure 6.2. The figure displays the<br />

total spectral intensity, which is the sum <strong>of</strong> the intensity <strong>of</strong> the three field<br />

components.<br />

2<br />

1.8<br />

1.6<br />

Frequency, [kHz]<br />

1.4<br />

1.2<br />

1<br />

0.8<br />

0.6<br />

0.4<br />

0.2<br />

0<br />

0.05 0.1 0.15 0.2 0.25 0.3<br />

Time τ, [s]<br />

−6 −5 −4 −3 −2 −1<br />

Spectral intensity, log (|b| 2 ) [nT 2 /Hz]<br />

10<br />

Figure 6.2: Spectrogram showing the total intensity <strong>of</strong> the whistler. Note how the<br />

wave frequency varies with time due to dispersion in the magnetosphere. Figure from<br />

Paper VI.<br />

All combinations <strong>of</strong> auto- and cross-spectra are obtained by forming the<br />

spectral density tensor in Equation (3.42) and Equation (2) in Paper VI. From<br />

the spectral density tensor we obtain the normal to the polarisation plane<br />

}<br />

V = 2 Im<br />

{−ˆx˜b y˜b∗ z + ŷ˜b x˜b∗ z − ẑ˜b x˜b∗ y (6.1)<br />

This (pseudo) vector is parallel (anti-parallel) to the wave vector k for a righthand<br />

(left-hand) polarised wave field. The magnitude <strong>of</strong> V is, according to<br />

Equation (3.45), equal to the magnitude <strong>of</strong> the spectral Stokes parameter V.<br />

The whistler is known to be circularly polarised and can be separated from<br />

the background by its high degree <strong>of</strong> circular polarisation. We denote the normal<br />

to the whistler’s polarisation plane V w .<br />

A comparison between the geomagnetic field B 0 and V w shows that they<br />

are approximately anti-parallel. The sign <strong>of</strong> the scalar product V w · B 0 gives<br />

the sign <strong>of</strong> the spectral Stokes parameter V. The sign is found to be negative<br />

and the whistler is therefore right-hand polarised and must be propagating<br />

38


antiparallel to B 0 .<br />

The normal to the polarisation plane deviates, on average, from the direction<br />

antiparallel to B 0 with 6 ◦ and precesses in total four turns around the direction<br />

antiparallel to B 0 . The precession frequency is estimated to about 20 Hz,<br />

which is 10–100 times lower than the wave frequency. Figure 6.3 illustrates<br />

the precession.<br />

Three more whistlers have been studies and they also show precession <strong>of</strong><br />

the normal to the polarisation plane, although not as clear as in this example.<br />

Theory can not give an unambiguous explanation to the observed precession.<br />

Further studies are needed to see if the precession is a general feature <strong>of</strong><br />

whistlers. Three-dimensional ray tracing studies may also provide an answer<br />

to how common the precession is.<br />

Precession at 20 Hz<br />

V w<br />

b(t)<br />

Wave<br />

frequency<br />

B 0<br />

Polarisation plane<br />

Figure 6.3: The whistler is propagating antiparallel to B 0 . The field vector b(t)<br />

sweeps out the polarisation ellipse, which defines the plane <strong>of</strong> polarisation. The normal<br />

<strong>of</strong> the polarisation plane, V w precesses around the direction antiparallel to B 0 at<br />

a frequency <strong>of</strong> 20 Hz.<br />

39


7. Statistical approach to stimulated<br />

electromagnetic emissions<br />

Transmitting a powerful radio wave vertically into the ionosphere gives rise<br />

to secondary HF radiation from the ionosphere. The radiation has repeatable<br />

characteristics and the power spectral density (PSD) is weak, asymmetric, and<br />

has well structured sidebands around the injected frequency [30]. The secondary<br />

radiation has been given the name stimulated electromagnetic emissions<br />

(SEE). A comprehensive description <strong>of</strong> SEE is given in [18].<br />

7.1 Characteristics and models for SEE<br />

A power spectrum from the Sura ionospheric HF pumping facility close to<br />

Vasil’sursk, Russia is shown in Figure 7.1. The spectrum displays SEE features<br />

in a frequency band <strong>of</strong> ±120 kHz from the pump frequency <strong>of</strong> 6760<br />

kHz. From right, the features are broad up-shifted maximum (BUM), upshifted<br />

maximum (UM), narrow continuum (NC), down-shifted maximum<br />

(DM), and second down-shifted maximum (2DM). The figure also shows a<br />

structure which is displayed at the same frequency as where the broad downshifted<br />

maximum (BDM) usually appear. This structure we denote tentative<br />

BDM. The signals identified as interferences are marked by shading.<br />

Most models <strong>of</strong> SEE are based on parametic processes, which are phase<br />

coherent [31]. These processes also conserve energy and momentum. For a<br />

three-wave process, the conservation <strong>of</strong> energy and momentum implies that<br />

ω mother = ω daughter1 + ω daughter2 (7.1)<br />

k mother = k daughter1 + k daughter2 (7.2)<br />

Letting φ denote the phase <strong>of</strong> a wave, we have [32]<br />

ω = − ∂φ<br />

∂t<br />

(7.3)<br />

k = ∇φ (7.4)<br />

which means that the total phase also is conserved for the interaction:<br />

φ mother = φ daughter1 + φ daughter2 (7.5)<br />

41


−20<br />

Interference<br />

−40<br />

PUMP<br />

Power [dBm]<br />

−60<br />

−80<br />

Tentative BDM<br />

2DM<br />

DM<br />

NC<br />

UM<br />

BUM<br />

−100<br />

−120<br />

−100 −50 0 50 100<br />

Frequency shift relative to the pump [kHz]<br />

Figure 7.1: The SEE spectrum with the visible SEE components indicated. Interferences<br />

from other radio sources are marked by shading. The data was taken at the Sura<br />

ionospheric HF pumping facility, Russia at 15:46LT on September 22, 1998. From<br />

Paper VII.<br />

Since the pump wave is monochromatic and very phase coherent, the decay<br />

products should also are phase coherent. From the PSD, it is impossible to say<br />

if the SEE features are caused by coherent signals.<br />

7.2 Statistical tests<br />

A statistical investigation <strong>of</strong> the spectral SEE amplitudes and phases should<br />

reveil the nature <strong>of</strong> the SEE characteristics. Such a study has been made in<br />

Paper VII.<br />

The opposite <strong>of</strong> a coherent signal is coloured narrow band Gaussian noise<br />

which has Rayleigh distributed [5] amplitudes and uniformly distributed<br />

phases. If the noise contains a sinusoidal, the amplitudes are instead Rice<br />

distributed [27].<br />

The stochastic/deterministic nature <strong>of</strong> the SEE features are investigated<br />

with statistical tests <strong>of</strong> the null hypothesis:<br />

Null Hypothesis (H 0 ) SEE is coloured complex Gaussian noise.<br />

If SEE is generated by coherent processes, the null hypothesis will be<br />

rejected. The following tests have been made:<br />

42


Test 1 (T 1 ) χ 2 test to determine if the spectral component amplitudes are<br />

Rayleigh distributed.<br />

Test 2 (T 2 ) χ 2 test to determine if the spectral component phases are uniformly<br />

distributed.<br />

If no deviation from the Rayleigh distribution and the uniform distribution<br />

can be found, we cannot reject H 0 and the SEE spectral components cannot<br />

be distinguished from noise. If, on the other hand, any <strong>of</strong> the two tests show<br />

significant deviation from noise, we perform a third test for sinusoidal components,<br />

which is a sign <strong>of</strong> coherency.<br />

Test 3 (T 3 ) Test for sinusoidal components.<br />

In order to obtain estimates <strong>of</strong> the spectral amplitudes and phases, the data<br />

was first divided into segments <strong>of</strong> 1024 samples. A windowed Fourier transform<br />

was applied to each <strong>of</strong> these segment, giving in total 11,718 estimates <strong>of</strong><br />

amplitude and phase per frequency bin, or spectral component. The spectral<br />

amplitudes and phases were binned into histograms with linear spacing and<br />

the number <strong>of</strong> estimates per bin was compared to the predictions according to<br />

the Rayleigh and uniform distributions.<br />

The results <strong>of</strong> T 1 and T 2 are presented in Figure 7.2. Large values <strong>of</strong> the<br />

test parameters indicate deviation from coloured Gaussian noise. If the test<br />

parameter is larger than the p value <strong>of</strong>, say 5%, that spectral component is with<br />

95% certainty not noise. Panel a) shows the result <strong>of</strong> T 1 and none <strong>of</strong> the SEE<br />

features, except for the tentative BDM, show any significant deviation from<br />

the amplitude distribution <strong>of</strong> narrow band Gaussian noise, while the tentative<br />

BDM, the pump, and the interferences deviates significantly from noise. In<br />

panel b), which displays the result <strong>of</strong> the phase test T 2 , only the pump and<br />

one <strong>of</strong> the interferences deviate from the phase distribution <strong>of</strong> narrow band<br />

Gaussian noise. Panel c) is shown to facilitate the identification <strong>of</strong> the SEE<br />

features.<br />

From the reults <strong>of</strong> T 1 and T 2 , we conclude that H 0 cannot be rejected for<br />

the SEE features BUM, UM, DM, and 2DM. In other words, they behave<br />

statistically as narrow band Gaussian noise.<br />

For the tentative BDM, the pump and the interferences, it is interesting<br />

to perform also the last test T 3 . The result <strong>of</strong> this test is shown in Figure 7<br />

in Paper VII, from which we conclude that only the pump and one <strong>of</strong> the<br />

interferences contain sinusoidals. The tentative BDM can therefore not be a<br />

coherent feature. In fact, the tentative BDM appears to be a pump induced<br />

disturbance and not a real SEE feature.<br />

Since no coherent signatures were found for any <strong>of</strong> the SEE features, they<br />

cannot be the results <strong>of</strong> simple parametric processes. The SEE features may<br />

43


χ 2 amplitude test parameter<br />

10 3<br />

10 2<br />

a)<br />

p=10 −4 %<br />

p=0.1%<br />

p=5%<br />

χ 2 phase test parameter<br />

10 3<br />

10 2<br />

b)<br />

p=10 −4 %<br />

p=0.1%<br />

p=5%<br />

−60<br />

Spectrum, power [dBm]<br />

−70<br />

−80<br />

−90<br />

−100<br />

−110<br />

c)<br />

−120 −100 −80 −60 −40 −20 0 20 40 60 80 100 120<br />

Frequency shift [kHz]<br />

Figure 7.2: The result <strong>of</strong> the amplitude and phase tests. Panel a) shows the T 1 spectral<br />

ampliude test parameter, panel b) the T 2 spectral phase test parameter, and panel c)<br />

the average SEE power spectrum with vertical dotted lines marking the SEE features.<br />

The power spectrum is identical to Figure 7.1. The p values <strong>of</strong> 5%, 0.1%, and 10 −4 %<br />

are marked with the horizontal dashed lines in panels a) and b). From Paper VII.<br />

still be generated by processes which are coherent on a small scale, but when<br />

integrated over a large region they may give a random result.<br />

44


8. The complex Poynting theorem as a<br />

conservation law<br />

The Poynting theorem is the energy conservation theorem in electrodynamics.<br />

It is <strong>of</strong> fundamental importance and is described carefully in all books on the<br />

subject, e.g. Jackson [14]. In Section 8.1 will consider the original form <strong>of</strong><br />

the Poynting theorem, the instantaneous Poynting theorem and show that it<br />

describes the energy balance in an electrodynamic system. In Section 8.2, the<br />

complex Poynting theorem (CPT) is derived and it is shown that the real part<br />

<strong>of</strong> the CPT is a balance equation for real energy. The imaginary part <strong>of</strong> the<br />

CPT is not a balance equation for reactive energy and we explain why.<br />

8.1 The instantaneous Poynting theorem<br />

The instantaneous Poynting theorem will be derived for real fields E(t, x)<br />

and B(t, x), and a real current density J(t, x). We start with the two vector<br />

Maxwell’s equations<br />

∂E<br />

∇ × B =µ 0 J + µ 0 ɛ 0<br />

∂t<br />

∇ × E = − ∂B<br />

∂t<br />

(8.1)<br />

(8.2)<br />

and take the scalar product <strong>of</strong> E and (8.1) and subtract the scalar product <strong>of</strong><br />

B and (8.2). We obtain<br />

E · (∇ × B) − B · (∇ × E) = µ 0 E · J + µ 0 ε 0 E · ∂E<br />

∂t + B · ∂B<br />

∂t<br />

(8.3)<br />

Dividing with µ 0 and rewriting the left-hand side as −∇ · (E × B), gives<br />

(<br />

ε 0 E · ∂E<br />

∂t + 1 B · ∂B )<br />

+ 1 ∇ · (E × B) = −E · J (8.4)<br />

µ 0 ∂t µ 0<br />

Integration over the entire volume containing the sources and using the divergence<br />

theorem results in the instantaneous Poynting theorem<br />

∫ (<br />

1 ∂<br />

ε 0 E · E + 1 )<br />

B · B dV + 1 ∮<br />

∫<br />

(E × B)·ˆnda = − E·JdV<br />

2 ∂t<br />

µ 0 µ 0<br />

V<br />

S<br />

V<br />

45


(8.5)<br />

which states that the energy is conserved within the volome V . The first term<br />

in Equation 8.5 represents the rate <strong>of</strong> change <strong>of</strong> the electromagnetic field energy,<br />

the second term the energy flowing out <strong>of</strong> the volume V , and the right<br />

hand-side the total work done by the fields on the sources.<br />

8.2 The complex Poynting theorem<br />

In this section, the analytic signal will be used and we define the fields and<br />

sources E(t, x), B(t, x), and J(t, x) to be complex. Now we take the scalar<br />

product <strong>of</strong> E and the complex conjugate <strong>of</strong> (8.1) and subtract the scalar product<br />

<strong>of</strong> B ∗ and (8.2) we obtain a complex form <strong>of</strong> the Poynting theorem:<br />

1<br />

2<br />

(ε 0 E · ∂E∗<br />

∂t<br />

+ 1 µ 0<br />

B ∗ · ∂B<br />

∂t<br />

)<br />

+ 1<br />

2µ 0<br />

∇ · (E × B ∗ ) = − 1 2 E · J∗ (8.6)<br />

The real part <strong>of</strong> Equation 8.6 is<br />

1 ∂<br />

(ɛ 0 E · E ∗ + 1 )<br />

B · B ∗ + 1 ∇ · (E × B ∗ + E ∗ × B)<br />

4 ∂t<br />

µ 0 4µ 0<br />

= − 1 4 (E · J∗ + J · E ∗ ) (8.7)<br />

The left-hand term in Equation (8.7) has the form <strong>of</strong> a time derivative <strong>of</strong> a<br />

conserved quantity, the energy density<br />

u c (t, x) = 1 (ɛ 0 |E| 2 + 1 )<br />

|B| 2 (8.8)<br />

4<br />

µ 0<br />

Since the second term is the Poynting vector in complex form, Equation (8.7)<br />

is a balance equation for real energy density u c .<br />

The imaginary part <strong>of</strong> Equation (8.6) is<br />

1<br />

4i<br />

[ɛ 0<br />

(E · ∂E∗<br />

∂t<br />

− E ∗ · ∂E<br />

∂t<br />

)<br />

− 1 µ 0<br />

(B · ∂B∗<br />

∂t<br />

− B ∗ · ∂B )]<br />

∂t<br />

+ 1<br />

4iµ 0<br />

∇ · (E × B ∗ − E ∗ × B) = − 1 4i (E · J∗ − J · E ∗ ) (8.9)<br />

The temporal derivative in the term in the big paranthesis cannot be moved out<br />

<strong>of</strong> the paranthesis and Equation (8.9) can therefore not be written as an energy<br />

balance equation. In Paper VIII, this conclusion is motivated in a stricter form.<br />

By defining the big paranthesis in Equation (8.9) as a temporal derivative <strong>of</strong><br />

a quantity U, it is still possible to write Equation (8.6) in the form <strong>of</strong> a balance<br />

46


equation, which is one form <strong>of</strong> the complex Poynting theorem:<br />

∂(u c + iU)<br />

∂t<br />

+ ∇ · N c = − 1 2 E · J∗ (8.10)<br />

The conserved quantities are the real energy density u c and U, for the real<br />

and imaginary parts, respectively. It should be noted that U is not the reactive<br />

energy density.<br />

8.3 The time-harmonic complex Poynting<br />

Theorem<br />

If we assume harmonic time variation for the fields and the source, i.e.,<br />

E = E ω e −iωt (8.11)<br />

B = B ω e −iωt (8.12)<br />

J = J ω e −iωt (8.13)<br />

we can show that the time-harmonic CPT can be derived from Equation (8.10).<br />

With the electric and magnetic energy densities defined by<br />

u e = ɛ 0<br />

4 E ω · E ∗ ω (8.14)<br />

u m = 1 B ω · B ∗ ω<br />

4µ 0<br />

(8.15)<br />

the time-harmonic CPT can be written<br />

∫<br />

∫<br />

1<br />

E ω · J ∗ ω<br />

2<br />

d3 x + 2iω (u e − u m )d 3 x + 1 (E ω × B<br />

V<br />

2µ 0<br />

∮S<br />

∗ ω ) · ˆnda = 0<br />

V<br />

This is the form <strong>of</strong> the CPT found in the standard literature [14, 4].<br />

The conclusion we make is the following. Although the instantaneous<br />

Poynting theorem and the CPT are balance equations for (real) energy, the<br />

imaginary part is not a balance equation for reactive energy. So, the CPT is<br />

not a balance equation for electromagnetic energy density.<br />

47


9. Summary in <strong>Swedish</strong><br />

I denna avhandling belyses hur polarisationen hos elektromagnetiska vågor<br />

kan användas inom både vetenskap och radioteknologiutveckling. En elektromagnetisk<br />

våg antar olika skepnad beroende på dess våglängd eller frekvens.<br />

Lägst frekvens har radiovågor, sedan kommer med stigande frekvens mikrovågor,<br />

infrarött, synligt ljus, ultraviolett, röntgen och gamma. Med elektromagnetiska<br />

vågor kan man förmedla information över stora avstånd och detta används<br />

framför allt inom radioområdet.<br />

Synligt ljus är den form av elektromagnetisk vågrörelse som är lättast att<br />

föreställa sig. Vi ser ljuset och kan avgöra dess frekvens genom dess färg. Vi<br />

har också förmåga att skilja på olika ljusstyrkor, eller intensitet. Polarisationen<br />

är en viktig egenskap hos en elektromagnetisk våg, men våra ögon kan inte<br />

särskilja ljus med olika polarisation. Vi kan dock förvissa oss om att ljuset<br />

innehar denna egenskap genom att sätta på oss ett par polaroidglasögon och<br />

titta på solljus reflekterat fran en vattenyta eller en blöt väg. En ännu bättre<br />

illustration av ljusets polarisation får man om man tittar genom två polaroidglas<br />

och observerar att ljuset släcks ut helt då glasens polarisationsriktningar<br />

är vinkelräta.<br />

En elektromagnetisk våg är en vektorvåg, som har både storlek och riktning.<br />

Den bär med sig ett elektriskt och ett magnetiskt fält, som båda är vinkelräta<br />

mot utbredningsriktningen och definierar det s k polarisationsplanet. Det är<br />

tidsvariationen av dessa fält i polarisationsplanet som utgör vågens polarisation.<br />

För ljus är det uppenbart varifrån det kommer och detta ger då vågens utbredningsriktning.<br />

Det räcker då med två fältkomponenter för att beskriva vågen<br />

i dess polarisationsplan. Polarisationen beskrivs därför i två dimensioner<br />

och ett praktiskt sätt att göra detta är att använda Stokes parametrar.<br />

Polarisationen har också stor betydelse för radiovågor, men för dessa går<br />

det inte att se varifrån de kommer. För att då ge en generell beskrivning av<br />

polarisationen krävs därför tre fältkomponenter. En generalisering av Stokes<br />

parametrar till tre dimensioner är därför önskvärd och det är detta denna<br />

avhandling börjar med.<br />

Nedan beskrivs översiktligt de arbeten som denna avhandling bygger på.<br />

49


Arbete I: Parametrar för att karakterisera<br />

elektromagnetisk vågpolarisation<br />

I detta arbete beskrivs hur de vanliga tvådimensionella Stokes parametrarna<br />

kan generaliseras till tre dimensioner och hur elektromagnetiska fält kan<br />

karakteriseras på alternativa sätt. Ett av dessa karakteristika är polarisationsplanets<br />

normalvektor, som <strong>of</strong>tast är parallell eller antiparallell med vågens<br />

utbredningsriktning. Ytterligare två är de spektrala Stokes parametrarna som<br />

förknippas med intensiteten och graden av cirkulärpolarisation. Arbetet<br />

baseras på spektraltäthetstensorn, som utvecklas i en bas med nya generella<br />

polarisationsparametrar som koefficienter. I två dimensioner är denna bas<br />

Pauli-spinnmatriserna, som är generatorer till SU(2) gruppen, och koefficienterna<br />

är Stokes parametrar. I tre dimensioner används på liknande sätt en bas<br />

bestående av generatorerna till symmertigruppen SU(3). Koefficienterna i<br />

denna utveckling är nya generella spektrala polarisationsparametrar. De nya<br />

polarisationsparametrarna är härledda i frekvensdomänen, vilket möjliggör<br />

tillämping på bredbandiga elektromagnetiska vågor.<br />

Arbete II: Metod och system för att<br />

erhålla riktning för en elliptiskt polariserad<br />

elektromagnetisk vågutbredning<br />

I detta patent presenteras en metod och ett system för att bestämma riktningen<br />

till en radiokälla och dess polarisationsparametrar, såsom den spektrala intensiteten<br />

och den spektrala graden av cirkulärpolarisering. Minst tre antenner<br />

används för att ta emot det elektriska eller magnetiska fältet i minst en punkt<br />

i rummet. Metoden separerar den mottagna signalens frekvenser och ger en<br />

riktning per frekvenskomponent.<br />

Arbete III: Antennanordning för användning<br />

av tredimensionell elektromagnetisk<br />

fältinformation inherent i en radiovåg<br />

Här presenteras ett flertal antennanordningar för att mäta det tredimensionella<br />

elektromagnetiska fältet. Det uppmätta fältet kan sedan bearbetas för att minimera<br />

fädning, minimera interferens mellan mottagarkanalerna, etc. Antennanordningen<br />

möjliggör också styrning av antennens strålningskänslighet i<br />

olika riktningar.<br />

50


Arbete 4: Trekanalig digital vektorfältsensor:<br />

Beskrivning och demonstration<br />

Detta arbete presenterar ett fungerande system för att i HF-området (3-30<br />

MHz) mäta det elektriska eller magnetiska fältet och utifrån detta spektralt<br />

bestämma det mottagna fältets polarisationsparametrar, såsom har beskrivits i<br />

Arbete II. Systemet bygger på helt digital nedblandning av radi<strong>of</strong>rekvenserna<br />

till basband. Den nedblandade signalen skickas sedan till operatörens dator där<br />

processandet av data sker. Det grafiska användargränssnittet presenteras och i<br />

detta visas alla de beräknade polarisationsparametrarna. Systemet kan också<br />

demodulera den mottagna radiosignalen och gör det därför också möjligt att<br />

lyssna på långvågs-, mellanvågs- och kortvågsradio.<br />

Arbete 5: Nuvarande och framtida<br />

tillämpningar av den informationstäta antennen<br />

I detta papper ges en sammanfattning över teorin som presenterats i Arbete<br />

I och II. Vidare beskrivs en rad tillämpningar av teorin och mätmetoden.<br />

Tillämpningarna innefattar exempelvis riktningsbestämning för både bredbandiga<br />

och icke-stationära signaler, modulationstekniker, signalkombinering<br />

i MIMO (multiple input multiple output) system, riktningsbestämning genom<br />

att använda brusmätningar, bestämning av absolut fas, radar, polarisation som<br />

hjälpmedel för att detektera signaler gömda i brus och användning av ickeplana<br />

vågor för att undgå detektion. Avslutningsvis ges en presentation av<br />

forskningsprojekt som baseras på teorin: HF spektrometer på den internationella<br />

rymdstationen, LOIS radioteleskop i södra Sverige och nanosatelliten<br />

Nanospace-1, som numera heter Microlink-1.<br />

Arbete 6: Precession av normalen till en<br />

visslarvågs polarisationsplan observerat i<br />

Frejadata<br />

Precession hos normalen till en visslarvågs polarisationsplan har observerats<br />

i satellitdata från Freja. Visslarvågor är cirkulärpolariserade och pro<strong>page</strong>rar<br />

i magnetosfären utmed det geomagnetiska fältet. Magnetometerdata samplat<br />

med 4096 sampel per sekund har studerats och en instantan analysmetod har<br />

applicerats på en uppmätt visslare. Metoden gör det möjligt att studera visslarens<br />

instantana polarisationsegenskaper. Således har normalen till visslarens<br />

polarisationsplan studerats och denna är i detta fall approximativt antiparallel<br />

med det geomagnetiska fältet med en avvikelse på i medeltal 6 ◦ . Nor-<br />

51


malen uppvisar en precession kring riktningen motsatt det geomagnetiska fältet.<br />

Normalen precesserar totalt fyra varv och dess precessionsfrekvens är 20<br />

Hz, vilket är väsentligt lägre än visslarens vågfrekvens.<br />

Arbete 7: Statistiska egenskaper hos<br />

jonosfärisk stimulerad elektromagnetisk<br />

emission<br />

En statistisk studie av de spektrala amplituderna och faserna hos stimularad<br />

elektromagnetisk emission (SEE) har genomförts. Studien visar att<br />

SEE-komponenterna BUM, UM, DM och 2DM inte signifikant skiljer sig<br />

från amplitud- och fasfördelningarna för färgat smalbandigt brus. Det data<br />

studien baseras på uppmättes vid Sura jonosfärsmodifieringsanläggning<br />

utanför Vasil’sursk i Ryssland, efter att jonosfären under en tid bestrålats<br />

med en kraftig radiovåg på 6760 kHz. Statistiska test har sedan applicerats<br />

på de sekundära emissionerna från denna bestrålade jonosfär. Resultatet är<br />

inte vad som förväntas enligt teoretiska modeller för SEE, där faskoherenta<br />

parametriska processer ligger som grund. Eftersom den kraftiga utsända<br />

radiovågen är helt koherent borde något av koherensen kvarstå även hos<br />

sönderfallsprodukterna av de parametriska processerna. Den förmodade<br />

förklaringen till att ingen koherens kunde påvisas är att processerna kan<br />

vara koherenta på en liten skala. När man tar hänsyn till hela den bestrålade<br />

volymen summeras bidragen från alla delar och koherensen går förlorad.<br />

Arbete 8: Komplexa Poyntingteoremet som en<br />

konserveringslag<br />

Poyntingteoremet beskriver energibalansen i ett elektrodynamiskt system. Då<br />

reella storheter används för att representera de elektromagnetiska fälten och<br />

dess källor, beskriver Poyntingteoremet bevarandet av reell energi. Om istället<br />

komplexa storheter används för att representera fälten och källorna, erhålls<br />

det komplexa Poyntingteoremet. Dess realdel beskriver, precis som Poyntingteoremet<br />

för reella storheer, bevarande av energi. I standardlitteraturen på<br />

området behandlas inte det generella uttrycket för imaginärdelen av det komplexa<br />

Poyntingteorement. Istället ansätter man ett harmoniskt tidsberoende<br />

hos källor och fält och visar att imaginärdelen är proportionell mot skillnaden<br />

mellan de elektriska och magnetiska energitätheterna. I Arbete VIII ges en<br />

beskrivning av varför imaginärdelen av det komplexa Poyntingteoremet inte<br />

utgör någon bevarelselag.<br />

52


10. Conclusion and Outlook<br />

With the increasing use <strong>of</strong> the radio spectrum, bandwidth is becoming more<br />

and more important and also more expensive. It is therefore desirable to use<br />

the bandwidth as effectively as possible. In the latest digital cellular telephone<br />

systems, utilising Code division multiple access (CDMA), the usage <strong>of</strong> bandwidth<br />

is enhanced by making the signal spectrum resemble a Gaussian process<br />

and thereby approaching the theoretical maximum capacity over the radio<br />

channel. In those systems, signals are coded and made to share frequencies<br />

with other signals sent over the same channel. At the receiver, the signals are<br />

identified by their code.<br />

The capacity can be increased further by using Multiple input multiple output<br />

(MIMO) systems, where the signal is transmitted with more than one antenna<br />

and also received with more than one antenna. MIMO systems are in use<br />

today in, e.g., the latest systems for cellular telephone, where spatial diversity<br />

or polarisation diversity is implemented. By using spatial and polarisation diversity<br />

together in MIMO systems, the capacity may increase further. In many<br />

other areas <strong>of</strong> radio communication, the shortage in bandwidth should eventually<br />

make it necessary to use the polarisation.<br />

A number <strong>of</strong> new applications should be made possible in science, radar,<br />

communication, and signal intelligence by measuring the full vector electromagnetic<br />

field. In science the use <strong>of</strong> polarisation as a diagnostic can provide<br />

useful information about the underlying physical processes. This was, for example,<br />

demonstrated in the first polarimetric study <strong>of</strong> stimulated electromagnetic<br />

emission [9]. In the areas <strong>of</strong> communication and signal intelligencewe<br />

may see applications such as to use the polarisation to detect signals hidden by<br />

noise (spread spectrum signals); use <strong>of</strong> non-planar waves to propagate signals<br />

which make direction-finding techniques provide a false direction.<br />

53


Acknowledgement<br />

First <strong>of</strong> all I would like to thank my supervisor Bo Thidé. I also thank my tutor<br />

Tobia Carozzi for good cooperation and valuable assistance, and Jan Bergman<br />

who has been an invaluable part <strong>of</strong> our research team.<br />

I am grateful to ALL the people at the <strong>Institute</strong> for <strong>Space</strong> <strong>Physics</strong> and the<br />

Department <strong>of</strong> Astronomy and <strong>Space</strong> <strong>Physics</strong>. Apart from being colleagues,<br />

they are also very good friends. A special thank to those who have supported<br />

me much in my work for this thesis: Walter Puccio, Anders I. Eriksson, Sven-<br />

Erik Jansson, Lars Norin, David Sundkvist, and Lennart Åhlén. I thank also<br />

Bengt Holback for helping me with my personal mechanicar problems.<br />

A thank to Christer Wahlberg, Vladimir Pavlenko, and Bo Thidé for giving<br />

me the possibility to lecture advanced undergraduate courses. That experience<br />

has given me a lot <strong>of</strong> confidence for the future. I thank Vladimir Pavlenko<br />

also for being a source <strong>of</strong> inspiration with his kindness and deep knowledge<br />

in physics.<br />

I thank Erkki Brändas, Sven Kullander, Inger Eriksson, and Susanne<br />

Söderberg at the Advanced Instrumentation and Measurement (AIM)<br />

graduate school. I gratefully acknowledge AIM for financial support and<br />

for finding me friends at the other departments <strong>of</strong> <strong>Uppsala</strong> university. I am<br />

also greateful to AIM for pushing me to find my external supervisor Erland<br />

Cassel, with whom I have spent many fruitful days together with, talking and<br />

building antennas.<br />

The time working with and for Redsnake Radio Technology AB was very<br />

rewarding and I thank my colleagues Jan Bergman, Tobia Carozzi, Anders<br />

Engström, Anders B. Eriksson, and Johan Warnström.<br />

I am grateful to Mikael Lindholm and colleagues at Scandinova Systems<br />

AB and I thank them for giving me the possibility to finish this thesis.<br />

I have also enjoyed the time outside <strong>of</strong> work. All my friends are important<br />

to me and I would like to thank specially Mattias Georgson and Urban Simu,<br />

and also the former and present players <strong>of</strong> IBS Nabla.<br />

A special thank to Paul, Isabel, and Titi. Paul, something fun always happens<br />

when you are around and I will never forget how fun it was to have you<br />

as colleague.<br />

Finally I big thank to my family for all the support I have got through the<br />

years.<br />

55


References<br />

[1] Measurement specification emil phase c and measurement report emil phase c.<br />

Technical report on the cacpacity <strong>of</strong> the indoor IDA MIMO radio channel.<br />

2002. Redsnake Radio Technology AB, Unpublished report, contact Roger<br />

Karlsson.<br />

[2] V. A. Alekseev, V. A. Danilkin, G. A. Dmitriev, V. S Dokukin, V. D. Kuznetsov,<br />

and Yu. Ya. Ruzhin. The general conception <strong>of</strong> the microsatellite compass to<br />

study <strong>of</strong> the earthquakes forerunners. 2005. Unpublished Draft.<br />

[3] M. R. Andrews, P. P. Mitra, and R. Decarvalho. Tripling the capacity <strong>of</strong> wireless<br />

communications using electromagnetic polarization. Nature, 409:316–318,<br />

2001.<br />

[4] C. A. Balanis. Advanced Engineering Electromagnetics. John Wiley & Sons,<br />

New York, 1989.<br />

[5] J. S. Bendat and A. G. Piersol. Random data: Analysis and measurement<br />

procedures. Wiley-Interscience, New York, 1971.<br />

[6] J. Bergman, T. D. Carozzi, and R. Karlsson. System for tredimensionell<br />

utvärdering av ett elektroniskt vektorfält. <strong>Swedish</strong> patent, No 523 086, 2004.<br />

<strong>Swedish</strong> patent No 523 086, 2004; Method for three-dimensional evaluation,<br />

International Patent Publication WO03/067710, 2003.<br />

[7] M. Born and E. Wolf. Principles <strong>of</strong> Optics. Cambridge University Press,<br />

Cambridge, 6th edition, 1998.<br />

[8] K. G. Budden. The Propagation <strong>of</strong> Radio Waves. Cambridge University<br />

Press, Cambridge, 1985.<br />

[9] T. D. Carozzi, B. Thidé, T. B. Leyser, G. Komrakov, V. Frolov, S. Grach, and<br />

E. Sergeev. Full polarimetric measurements <strong>of</strong> stimulated electromagnetic emission:<br />

First results. J. Geophys. Res., 106:21,395–21,407, 2001.<br />

[10] F. F. Chen. Introduction to Plasma <strong>Physics</strong> and Controlled Fusion, Volume<br />

1: Plasma <strong>Physics</strong>. Plenum, New York, 1984.<br />

[11] U. Fano. A stokes-parameter technique for the treatment <strong>of</strong> polarization in quantum<br />

mechanics. Phys. Rev., 93:121–123, 1954.<br />

57


[12] V. L. Ginzburg. The propagation <strong>of</strong> electromagnetic waves in plasmas.<br />

Pergamon Press, Oxford, 1970.<br />

[13] The American Radio Relay League Inc. The A. R. R. L. Antenna Book. West<br />

Hartford, CT, 1955.<br />

[14] J. D. Jackson. Classical Electrodynamics. John Wiley & Sons, New York,<br />

1999.<br />

[15] M. G. Kivelson and C. T. (ed.) Russel, editors. Introduction to <strong>Space</strong> <strong>Physics</strong>.<br />

Cambridge University Press, Cambridge, 1995.<br />

[16] S. I. Klimov, V. E. Korepanov, Yu. V. Lissakov, O. V. Lapshinova, I. V. Sorokin,<br />

G. A. Belyaev, G. A. Stanev, K. Georgieva, B. Kirov, M. P. Gough, H. S.<br />

C. K. Alleyne, M. Balikhin, J. Lichtenberger, Cs. Ferencz, L. Bodnar, K. Szego,<br />

S. Szalai, J. Juchniewicz, H. Rothkaehl, and K. Stasiewicz. “obstanovka” experiment<br />

onboard international space station for space weather research. 2005.<br />

Unpublished draft.<br />

[17] N. A. Krall and A. W. Trivelpiece. Principles <strong>of</strong> Plasma <strong>Physics</strong>. McGraw-<br />

Hill, New York, 1973.<br />

[18] T. B. Leyser. Stimulated electromagnetic emissions by high-frequency electromagnetic<br />

pumping <strong>of</strong> the ionospheric plasma. <strong>Space</strong> Sci. Rev., 98:223–328,<br />

2001.<br />

[19] R. Lundin, G. Haerendel, and S. Grahn. Introduction to special section: The<br />

freja mission. J. Geophys. Res., 103:4119–4123, 1998.<br />

[20] Y. Ne’eman M. Gell-Mann. The Eight-fold Way. Benjamin, New York, 1964.<br />

[21] R. Manning, N Vernet, and J. L. Steinberg. The impedance <strong>of</strong> a dipole in the<br />

ionosphere, 1, experimental study. Radio Sci., 10:517–527, 1975.<br />

[22] J. D. Means. Use <strong>of</strong> three-dimensional covariance matrix in analyzing the polarization<br />

properties <strong>of</strong> plane waves. J. Geophys. Res., 77:5551–5559, 1972.<br />

[23] P. Meyer and N Vernet. Impedance <strong>of</strong> a short antenna in a warm magnetoplasma.<br />

Radio Sci., 9:409–416, 1974.<br />

[24] P. Meyer and N Vernet. The impedance <strong>of</strong> a dipole antenna in the ionosphere, 2,<br />

comparison with theory. Radio Sci., 10:529–536, 1975.<br />

[25] N Meyer-Vernet. On natural noises detected by antennas in plasma. J. Geophys.<br />

Res., 84:5373–5377, 1979.<br />

[26] N Meyer-Vernet and C. Perche. Tool kit for antennae and thermal noise near the<br />

plasma frequency. J. Geophys. Res., 94:2405–2415, 1989.<br />

58


[27] S. O. Rice. Mathematical analysis <strong>of</strong> random noise. Bell Syst. Tech. J., 23:282–<br />

332, 1944. Bell Syst. Tech. J., 24, 46–156, 1945. Reprinted in Selected Papers<br />

on Noise and Stochastic Processes, ed. N. Wax, Dover, New York, 1954.<br />

[28] G. G. Stokes. On the composition and resolution <strong>of</strong> streams <strong>of</strong> polarized light<br />

from different sources. Trans. Cambr. Phil. Soc., 9:399, 1852. Reprinted in his<br />

Mathemathical and Physical Papers, Vol III (Cambridge University Press,<br />

1901), p 233.<br />

[29] J. Stoltz. On the design <strong>of</strong> electric antennas on nanosat for measuring radio and<br />

plasma waves in the ionosphere. Linköpings university, Linköping, LiTH-<br />

IFM-EX-903, 2000.<br />

[30] B. Thidé, H. Kopka, and P. Stubbe. Observations <strong>of</strong> stimulated scattering <strong>of</strong> a<br />

strong high-frequency radio wave in the ionosphere. Phys. Rev. Lett., 49:1561–<br />

1564, 1982.<br />

[31] J. Weiland and H. Wilhelmsson. Coherent Non-Linear Interaction <strong>of</strong> Waves<br />

in Plasmas. Pergamon Press, Oxford, 1977.<br />

[32] G. B. Whitham. Linear and nonlinear waves. John Wiley & Sons, New York,<br />

1974.<br />

[33] N. Wiener. Generalized harmonic analysis. Acta Math., 55:117–258, 1930.<br />

59

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!