Antenna Diversity Studies and Evaluation - Jeziorsaanteny.jeziorski.info/wp-content/uploads/2011/12/2004_Plicanic.pdf · Antenna Diversity Studies and Evaluation from the IE3D simulation

Department of Electroscience

Master of Science Thesis Vanja Plicanic

in cooperation withEricsson Mobile Communications AB

May 2004

Antenna Diversity Studiesand Evaluation

Antenna Diversity Studies and Evaluation

Abstract Stepping into the new generation of mobile communication it is of concern to increase the performance of the mobile terminals and their antennas to be able to answer to the demand of faster and more various communicational services. Still, the issue of signal fading in a multi-path environment stands as an obstacle today. The purpose of this master thesis is to introduce diversity techniques as a promising way of improving the performance at the mobile terminals. The goal is more profound, dealing with analysis of the correlation between received signals and the propagation environments that diversity improvement depends on, and putting together a repeatable method for measuring the improvement. The thesis work resulted in simulated and measured correlation and diversity performance for five different, small diversity antenna concepts created for the purpose of the thesis. The results confirmed equivalence between calculating correlation from far-field pattern and from S-parameters, presented the expected diversity improvement, illustrated the effects of correlation and propagation models on diversity improvement and finally outlined the simulations in IE3D and measurements in a Satimo measurement system as the suitable methods for collecting data for calculation of correlation. The method for measuring diversity improvement was completed by creating a calculation program folder Diversity Performance Program including the effects of correlation and propagation models.

i


ii


Acknowledgements The five months of working on my master thesis at Sony Ericsson have been very instructive and inspiring. I would like to thank my supervisors Thomas Bolin and Zhinong Ying at Sony Ericsson for their experience and time as well as giving me the opportunity to work in a, for me and for the company, new research area. I would also like to thank my supervisors Gerhard Kristensson and Anders Derneryd from Department of Electroscience at LTH for their commitment, support and knowledge. Finally, many thanks should go to the Antenna Section people: Kenneth Håkansson, André da Silva, Igor Egorov, Johan Andersson, Anders Dahlström and Peter Nordenström and the “outsiders” Ramadan Plicanic and Hamid Kami-Shirazi for their help and support and for making my time at Sony Ericsson a lot of fun.

iii


iv


Contents ABBREVIATIONS ................................................................................................................................ 1 1 INTRODUCTION ............................................................................................................................... 3

1.1 BACKGROUND................................................................................................................................. 3 1.2 PURPOSE AND GOAL OF THE THESIS................................................................................................. 3 1.3 METHOD ......................................................................................................................................... 3 1.4 LIMITATIONS................................................................................................................................... 4

2 DIVERSITY......................................................................................................................................... 5 2.1 BACKGROUND................................................................................................................................. 5 2.2 LINEAR DIVERSITY COMBINING....................................................................................................... 7 2.3 MECHANISMS FOR ACHIEVING DIVERSITY BRANCHES ................................................................... 8

2.3.1 Space diversity ........................................................................................................................ 8 2.3.2 Frequency diversity ................................................................................................................ 8 2.3.3 Angle diversity ........................................................................................................................ 8 2.3.4 Time and Multi-path Diversity................................................................................................ 8 2.3.5 Polarization diversity.............................................................................................................. 9 2.3.6 Pattern diversity...................................................................................................................... 9

2.4 DIVERSITY GAIN ............................................................................................................................. 9 2.4.1 Correlation coefficient.......................................................................................................... 10

3 ANTENNA PERFORMANCE......................................................................................................... 13 3.1 MEAN EFFECTIVE GAIN ................................................................................................................. 13

3.1.1 Propagation environments.................................................................................................... 13 3.1.2 Efficiency of the antenna....................................................................................................... 17 3.1.3 Mutual coupling.................................................................................................................... 17 3.1.4 Mean effective directivity...................................................................................................... 18

3.2 DIVERSITY SYSTEM GAIN .............................................................................................................. 18 4 DIVERSITY PERFORMANCE PROGRAM................................................................................. 19

4.1 DATA CODES ................................................................................................................................. 19 4.2 CORRELATION COEFFICIENT AND PROPAGATION MODELS ............................................................. 19 4.3 DIVERSITY PERFORMANCE............................................................................................................ 20

5 PROTOTYPES.................................................................................................................................. 21 5.1 TWO-SLOT ANTENNA SYSTEM ....................................................................................................... 21 5.2 CLAMSHELL ANTENNA SYSTEM..................................................................................................... 21 5.3 TWO-MONOPOLE ANTENNA SYSTEM.............................................................................................. 22 5.4 PIFA-SLOT ANTENNA SYSTEM ...................................................................................................... 22 5.5 TWO-FEED PATCH ANTENNA ......................................................................................................... 23

6 SIMULATIONS................................................................................................................................. 25 7 MEASUREMENTS........................................................................................................................... 27

7.1 SATIMO STARGATE 64 SYSTEM ..................................................................................................... 27 7.2 SCATTERED FIELD CHAMBER (SFC) .............................................................................................. 28

8 RESULTS........................................................................................................................................... 31 8.1 CORRELATION COEFFICIENT.......................................................................................................... 31

8.1.1 Far-field pattern and S-parameters ...................................................................................... 31 8.1.2 Satimo ................................................................................................................................... 33

v


vi

8.1.3 Network analyzer .................................................................................................................. 35 8.1.4 SFC ....................................................................................................................................... 36 8.1.5 T-dipoles ............................................................................................................................... 40 8.1.6 Handheld measurements....................................................................................................... 41

8.2 DIVERSITY PERFORMANCE RESULTS – A SUMMARY ...................................................................... 41 8.2.1 Simulated diversity performance .......................................................................................... 41 8.2.2 Measured diversity performance in the Satimo system......................................................... 47 8.2.3 Measured diversity performance in the SFC ........................................................................ 49

9 DISCUSSION..................................................................................................................................... 51 10 CONCLUSIONS.............................................................................................................................. 55 11 FUTURE WORK............................................................................................................................. 57 11 REFERENCES ................................................................................................................................ 59 APPENDIX A SOME DEFINITIONS .......................................................................................... 63

A.1 S-PARAMETERS ............................................................................................................................ 63 A.2 THE MARCUM Q-FUNCTION ......................................................................................................... 63

APPENDIX B PROTOTYPE CHARACTERISTICS................................................................. 65 B.1 TWO-SLOT ANTENNA SYSTEM....................................................................................................... 65 B.2 CLAMSHELL ANTENNA SYSTEM .................................................................................................... 66 B.3 TWO-MONOPOLE ANTENNA SYSTEM............................................................................................. 67 B.4 PIFA-SLOT ANTENNA SYSTEM...................................................................................................... 68 B.5 TWO-FEED PATCH ANTENNA SYSTEM ........................................................................................... 69

APPENDIX C HANDHELD MEASUREMENT RESULTS........................................................71


Abbreviations DCS -Digital Cellular System, terminology used for 1800 MHz frequency band DG -diversity gain DSG -diversity system gain DUT -device under test EGC -equal gain combining GSM –Global System for Mobile Communication, 900 MHz frequency band MED -mean effective directivity MEG -mean effective gain MoM -method-of-moments MRC -maximum ratio combining PCS -Personal Communications System, terminology used for 1900 MHz frequency band PIFA -Planar Inverted F Antenna RF -radio frequency SFC -scattered field chamber SNR -signal-to-noise ratio SMS -short message service SWR -standing wave ratio

1


2


1 Introduction 1.1 Background In the next generation of mobile communication systems there is a need for higher performance of the mobile terminals, accomplished by increasing capacity or reducing multi-path interference. Multi-path interference in the mobile communication environment has been and still is a major issue but there are ways of overcoming the problem. Signal fading due to the multi-path propagation can be reduced using antenna diversity techniques, most commonly introduced by two antenna functions at the mobile terminal. Being a quite unexplored area of antenna performance improvement, antenna diversity has been of great interest for both the Antenna Section at Sony Ericsson Mobile Communications AB in Lund and the Department of Electroscience at Lund Institute of Technology. This interest has resulted in a collaboration where this master thesis work is an attempt to analyze diversity parameters and evaluate diversity performance quality by creating a repeatable and adequate method for measuring diversity performance. At Sony Ericsson in Lund there is today an implementation of a method for measuring and calculating diversity performance in a scattered field chamber but it has not been evaluated yet. The master thesis work has taken place during the spring of 2004 at the Sony Ericsson Mobile Communications AB site in Lund.

1.2 Purpose and goal of the thesis Antenna diversity involves antenna interaction making correlation between incoming signals and mutual coupling between antennas play major roles in performance improvement. To fully understand the potential of antenna diversity, the propagation environment has also to be taken into account. The purpose of this master thesis is to study and evaluate diversity performance by evaluating the correlation between antennas as well as propagation models. The goal of the master thesis is except for the theoretical part of understanding the parameters affecting antenna diversity performance also to create a method for measuring the above mentioned parameters.

1.3 Method The studies and the evaluation of the diversity performance have been realized by creating a set of Matlab programs for calculation of signal correlation coefficient and diversity gain of the diversity antenna system, including the different propagation environments and using data from both measurements and simulations. The program folder is to be used in the future as the calculation program for diversity performance. The studies have been performed for mainly five different diversity antenna prototypes as well as an example of two close T-dipoles for some parts of the evaluation. Data processed in the diversity performance programs has been collected

3


from the IE3D simulation program and from measurements done in a scattered field chamber, in a Satimo measurement system and with a network analyzer.

1.4 Limitations The scientific field of antenna diversity is quite extensive. Given the time of 20 weeks for this master thesis some limitations were needed. The antenna diversity techniques chosen to work with are space and polarisation diversity without further exploration and study of the techniques themselves. Antenna diversity improvements are studied here for one signal combination method called switched combining. The centre frequency is at 2GHz but there are prototypes with resonances at frequencies below and above this as well. This choice of centre frequency is due to the fact that future mobile communication is all about data transmission. That is why the measurements have been done in free space and in data-mode (SMS-mode) for the measurement methods that required a specific terminal position. Although some measurements and evaluations were performed in this master thesis for handheld positions of the terminal involving the effect of the human hand on the performance, no extensive analysis has been made on the body interaction.

4


2 Diversity 2.1 Background RF signal transmission between two antennas commonly suffers from power loss in space that affects performance significantly. This power loss between transmitter and receiver is a result of three different phenomena: distance-dependent decrease of the power density called path loss or free space attenuation, absorption due to the molecules in the atmosphere of the earth and signal fading caused by terrain and weather conditions in the propagation path [31]. Atmospheric absorption is due to the electrons, uncondensed water vapor and molecules of various gases. It is frequency dependent with two quite high peaks at 60GHz and 21GHz for oxygen and water vapor respectively. Path loss is a theoretical attenuation which occurs under free-line-of-sight conditions and which increases with the distance between base station and mobile. Fading is an attenuation that varies between a maximum and minimum value in an irregular way. Terminals move through areas with obstacles of various sizes, such as mountains, buildings and tunnels. Occasionally, these obstacles will shadow or completely cut off the signal. Although the consequences of such shadowing effects will depend on the size of an obstacle and on the distance to it, the received signal strength will inevitably vary. This type of fading is referred to as shadow fading [30].

Figure 2.1 Shadow fading [31]

The effect of shadow fading is possible to decrease with some awareness in network planning. By placing base stations as high as possible or close to each other it is possible to avoid some obstacles in transmission.

Still, it is not the shadow fading that is the most unpredictable power loss. Rayleigh fading, multi-path fading or short term fading is a completely different kind of fading, involving irregular signal strength variations and it is problematic to overcome.

5


Figure 2.2 Multi-path propagation [31]

Rayleigh fading is a result of a reception of several signals at the receiver incoming reflected from many different objects and directions in the area (figure above). Due to their different traveling distances the signals are usually not in phase, reinforcing or extinguishing each other. The movement of the terminal causes continuous and unpredictable variations of the signal phases over time, making the attenuation very variable and extremely high at some points (fading dips). Rayleigh fading is most perceptible in urban areas. Dips will occur more frequently at higher frequencies and more rapid mobile movement. To avoid dips it is necessary to attain a sufficient fading margin. The average value of the signal must be at least as many decibels above the receiver sensitivity level as the strongest expected dip.

Figure 2.3 Rayleigh fading [31] The performance degradation shown in the difference between the mean signal-to-noise ratio (SNR) and non-fading signal can be solved by increasing the transmitting power and resizing the antenna, however the economical aspect of these solutions, especially for the small terminals, is not that attractive. To overcome these problems special reception techniques are used, namely the multiple receiver combining techniques known as diversity.

6


2.2 Linear diversity combining The first encounter with diversity occurred in experiments with spaced receiving antennas at high frequencies. It was discovered that multi-path fading variations of the signal received at one antenna have a tendency of being independent of fading variations at the other antenna, assuming a sufficient spacing between the two antennas. This observation lead to an experiment where a switch was used to derive the output signal from the stronger of the two antennas at the observed time and resulted in a reduction of the depth of the fading effects compared to signals from either antenna alone [24]. If the two signals are uncorrelated, it is seldom the case that both of the fading signals have a dip at the same time. The linear combining method of selecting an instantaneously stronger signal, selection combining, is not the only linear method used for accomplishing diversity. This could also be done by using switched combining, equal gain combining (EGC) and maximum ratio combining (MRC). Switched combining works by switching to the other antenna when the signal from one antenna falls below a certain threshold. EGC uses the signal phase, it works by adding the signals from the two antennas when they are co-phased. MRC is the most complicated one, using weights at each signal such that the output will be a sum of their signal-to-noise ratios [6]. The use of a specific combining method is not as much a technical question when it comes to small mobile terminals, as an economical issue. For all of the methods, except for switched combining, two receivers are required. Dual receivers are not, however, easily implemented in mobile terminals due to limitations in size, battery and complexity restrictions imposed by manufacturing [6]. That is the foremost reason for using the linear switched combining method in mobile terminals and also the reason why this thesis is only concentrating on this method of achieving diversity.

Figure 2.4 Selection and switched combining [28]

7


2.3 Mechanisms for Achieving Diversity Branches Independently fading signals, diversity branches, are not only achieved by spaced receiving antennas, space diversity. Other mechanisms are as common: frequency diversity, angle-of-arrival diversity, polarization diversity, time diversity, multi-path diversity and pattern diversity. The overall diversity performance is usually a result of a combination of different diversity mechanisms.

2.3.1 Space diversity The most common and probably the simplest mechanism for achieving diversity branches is space or spatial diversity. Using two antennas with a distance between them the phase delay makes multi-path signals arriving at the antennas differ in fading. The minimum spacing between the antennas at a mobile terminal required for sufficient low correlation between fading signals is usually some 0.5 wavelengths, calculated using the zero order Bessel function. This is not including the affect of the mutual coupling that can be significant. When spacing is smaller than 0.5 wavelengths other diversity mechanisms will have strong influence [6]. Space diversity is nowadays in focus because of the higher frequencies used for transmission making it possible to apply this kind of diversity mechanics in smaller terminals.

2.3.2 Frequency diversity Frequency diversity utilizes transmission of the same signal at two different, spaced, frequency carriers achieving two independently fading versions of a signal. It is a costly mechanism to use because of the difficulties to generate several transmitted signals and the combining signals received at several different frequencies simultaneously [24].

2.3.3 Angle diversity Signals arriving at the antennas are coming from different directions. Being independent in their fading variations these signals can be used for angle or angular diversity. At a mobile terminal angle diversity can be achieved using two omnidirectional antennas acting as parasitic elements to each other changing their patterns to manage the reception of signals at different angles [6]. Omnidirectional antennas on a mobile terminal might not be a good idea because of the field lobe blocked by the user.

2.3.4 Time and Multi-path Diversity Time and multi-path diversity are related mechanisms mostly applicable in the digital transmission. Time diversity is achieved by transmitting the same bit of information repetitively at short time intervals. Fading variations for these different repetitions of a signal will be independent. Multi-path diversity is using time diversity in multi-path environments getting the information from repetitive signals traveling different paths [24].

8


2.3.5 Polarization diversity The transmission of one polarization is depolarized by the propagation medium resulting in two orthogonal polarizations with uncorrelated fading variations. This diversity mechanism is very practical because of the very small size of antennas that can be used. It is also the mechanism that is mostly combined with space diversity [6].

2.3.6 Pattern diversity Pattern diversity can be achieved when patterns of two diversity antennas are compared. When using two different antennas, the radiation patterns are different so the signals arriving at the antennas will be from different directions and uncorrelated. Pattern diversity is never applied alone, it usually appears in addition to space diversity [25]. In this thesis mainly space diversity is being utilized, but also a combination of space and polarization diversity. The reason for this choice is that these mechanisms are simplest to implement, they do not have any complicated signal or antenna system requirements as other diversity mechanisms. Another merit is that the antennas used are usually of compact dimensions suitable for small mobile terminal applications.

2.4 Diversity Gain The effectiveness of diversity is usually presented in terms of diversity gain (DG). Diversity gain can be defined as the improvement in time-averaged signal-to-noise ratio (SNR) from combined signals from a diversity antenna system, relative to the SNR from one single antenna in the system, preferably the best one. This definition is conditioned by the probability that the SNR is above a reference level. The probability value is optional but usually set to 50% or 99% reliability [28]. The general mathematical expression for diversity gain is

)()/(1

1 dBDGscPc

c

Γ<

Γ

−Γ

=γγ

γγ (2.1)

where cγ is the instantaneous SNR of the diversity combined signal, Γ is the mean SNR of the combined signal,

c

1γ is the highest SNR of the diversity branch signals, 1Γ is the mean value of 1γ and Γ/sγ is a threshold or reference level. The probability P is dependent on the number of branches M in the diversity system. Assuming uncorrelated signals with Rayleigh distribution at the branches with equal noise the probability can be expressed as follows1 [1]:

1 Conditioned that we are using selected linear combining. Since switched combining is a form of selection combining this applies to this thesis as well [22].

9


Ms ))−scP exp(1()/(Γ

−=Γ<γ

γγ

(2.2)

his will give a theoretical diversity gain value of maximum 10 dB for 1% probability Tand approximately 2.5 dB for 50% probability, see figure 2.5.

Figure 2.5 Probability for different number of branches M and diversity gain for two

2.4.1 Correlation coefficient

p until this point, we have several times noted that the fading signals incoming to

branches (M=2)

Uthe receiver of the diversity system have to be independent of each other. This independency is seen in terms of a very low correlation, ideally having a correlation coefficient equal to zero 2 . In practice the faded signals are not completely independent, which affects the probability that the combined signal will be above the given threshold and in this way also the diversity gain. The independency can be measured using the general expression of the envelope cross correlation coefficient eρ , in angular domain3

( )∫ ∫

∫ΩΩΩ+ΩΩ⋅ΩΩΩ+ΩΩ

ΩΩΩΩ+ΩΩΩ=

∗∗

dPGPGXPRdPGPGXPR

dPEEPEEXPR

YYXX

YXYXe

))()()()(())()()()((

))()()()()()((2

φφθθφφθθ

φφφθθθρ (2.3)

here w ),( φθ=Ω , )()( ΩΩ= ∗

θθθ EEG , )(ΩXEθ and )(ΩYEθ are the θ polarized comple p X enna diversity ystem and

.sinx radiation atterns of antenna and ant Y in the s

θφθ ddd =Ω )(ΩθP and )(ΩφP represent incident power spectrum for both

2The correlation coefficient can vary between zero and one, one meaning that total dependency exists [30]. 3 All the field measurements, simulations and calculations throughout this thesis have been done in the angular domain for practical reasons.

10


φ

θ

x

z

y Azimuth

Elevation

polarizations and XPR (cross polar discrimination) is the ratio of time average vertical (θ ) power to time average horizontal (φ ) power [16]:

sρ

(

*21S

−

H

v

PP

XPR = (2.4)

The relation between horizontal and vertical Cartesian coordinates and the angular coordinates is shown in figure 2.6:

Figure 2.6 Relation between Cartesian and angular coordinates Correlation can also be expressed using the complex cross correlation coefficient

related to envelope correlation coefficient by:

2se ρρ = (2.5)

Computation of correlation coefficient above requires radiation pattern of antennas and involves integral calculation, which is experimentally and numerically very time- consuming. To make this process easier, a method for calculating the envelope correlation coefficient from S-parameters4 of the diversity antenna system has been derived [4]:

))1))((1( 2

122

222

212

11

2

2212*11

SSSS

SSSe

++−

+=ρ (2.6)

This expression assumes matched antennas and uniformly distributed incoming waves. In this thesis a practical comparison between the two methods for calculating correlation coefficient is performed.

11

4 see Appendix A.1


To provide sufficient diversity gain, the envelope correlation coefficient should be 0.7 or less. The lower correlation, the higher diversity gain will be achieved. The relation between correlation and diversity gain for two switched diversity branches taking into account the probability that the combined signal will be above the given threshold [24]:

( ) ( )

( ) ( )

−Γ

−Γ−⋅

Γ

−−

−Γ−Γ⋅

Γ

−−=<

21

222

21

221

12,

121exp

12,

12exp1)(

ss

s

s

s

s

c

xxQx

xxQxxP

ρρρ

ρρ

ργ

(2.7)

where Γ= /sx γ is the threshold, 1Γ and 2Γ are mean SNR values for antennas 1 and 2 respectively, in the diversity antenna system. The Q-function is the Marcum function, expressed in more detail in Appendix A.2. Expression (2.6) includes the effects that unequal power branches in the diversity system can have on the probability. Equal power branches mean better condition for receiving high diversity gain but in practice this condition can not always be met [6]. The expression above can fulfill this condition if 1Γ and 2Γ are set equal.

12


3 Antenna performance Antenna performance is strongly affected by the environment that the antenna is positioned in. The propagation possibilities in different terrain and surroundings can have both positive and negative effects on the performance. Presence of a user (its head and hand) is also a factor that generally makes a difference but in this thesis work the focus lies elsewhere. Propagation models covering the most common environments for mobile terminals are defined here to be included in a diversity performance program. Propagation effects can not be discussed without discussing the effective gain of the antenna, as its foremost performance characteristic.

3.1 Mean effective gain In multi-path environments the mean effective gain (MEG) defines the power received by the antenna. It is a characteristic that includes antenna radiation power pattern ( , ) ), antenna efficiency and propagation effects ( , ). It is defined by [16]:

)(ΩθG (ΩφG )(ΩθP )(ΩφP

Ω

ΩΩ

++ΩΩ

+= ∫ ∫ dPG

XPRPG

XPRXPRMEG

π π

φφθθ

2

0 0

)()(1

1)()(1

(3.1)

MEG is normalized by a normalization of the gains [20]: ∫ =ΩΩ+Ω πφθ 4))()(( dGG (3.2)

3.1.1 Propagation environments Both MEG values of the antennas in the antenna diversity system as well as correlation between signals arriving at the system are very dependent on the distribution of the incoming power and the orientation of the antennas relative to the radio environment. It is important to take into consideration the statistical propagation properties of both vertically and horizontally polarized incident radio waves. The propagation effects are included in (2.3) and (3.1) as the power spectrum of the polarized incoming waves ( )(ΩθP and )(ΩφP ) expressed as statistical distribution functions [22]:

)()(),()()()(),()(

φθφθφθφθ

φφφφ

θθθθ

PPPPPPPP

==Ω==Ω

(3.3)

where )(θθP and )(θφP are power spectrums in elevation and )(φθP and )(φθP are power spectrums in azimuth.

13


Figure 3.1 Incident radio wave arriving at the receiving antenna in a multi-path environment5[27]

The incident waves are diffracted over buildings, reflected off them and/or scattered by different objects in the terrain. Since there is no way of predicting the size, height, shape and materials of the objects that the incident waves come across propagating in the different environment structures, the incident waves´ direction varies. Moving an antenna randomly in this environment, it is said that the incident waves arrive over a random route or statistically they are uniform in the azimuth [27]. In this thesis, in three of four propagation models, the assumption is that the power spectra of both vertical and horizontal polarization are uniform in the azimuth direction. The power spectra in the elevation direction however are not uniformly distributed. There are different distributions describing the incoming polarized waves in elevation and the most common are Gaussian and Laplacian distributions. The Gaussian distribution is shown in figure 3.2 and expressed as follows:

( )[ ]( )

( )[ ]( )πθ

σπθ

θ

πθσ

πθθ

φφ

θθ

≤≤

−−−=

≤≤

−−−=

02

2/exp)(

02

2/exp)(

2

2

2

2

H

H

V

V

mAP

mAP

(3.4)

where and are mean elevation angles of the vertical and horizontal polarized wave distributions respectively, and and are the standard deviations of the vertical and horizontal wave distributions, respectively. A and are constants determined by the following conditions:

Vm Hm

Vσ Hσ

θ φA

14

5 P in the figure stands for polarization and not effect.


(3.5) 1)sin(),()sin(),(2

0 0

2

0 0== ∫ ∫∫ ∫

π π

φ

π π

θ φθθφθφθθφθ ddPddP

Figure 3.2 Gaussian distribution model of incident waves in elevation [27]

The Laplacian distribution is expressed in a similar way [27][16]:

( )

( )πθ

σπθ

θ

πθσπθ

θ

φφ

θθ

≤≤

−−−=

≤≤

−−−=

02/2

exp)(

02/2

exp)(

2

2

H

H

V

V

mAP

mAP

(3.6)

For the indoor case of incident waves the power spectra are not uniform. The environment is somewhat different from an outdoor transmission model since most of the signal energy is coming through building openings, and for that reason it is more directional. Gaussian and Laplacian distribution models in azimuth have been investigated but it was found that neither fit experimental data very well. The best fit was found using an elliptical distribution model [16]:

)2/(),2/()(sin

)(

)2/(),2/()()(

)2/(),2/()(

22

2

0

0

πφππφπα

α

πφππφπφφ

πφππφπφ

ψα

ψαψψ

φφφφ

θθθ

≤≤−≤≤−+

=

≤≤−≤≤−−=

≤≤−≤≤−=

ss

AP

baAP

aAP

(3.7)

where is a measure of the angular spread in the ψαs αsin -domain for both elevation and azimuth of the polarized wave distributions. 0θa and are the relative amplitude of the uniformly distributed low energy scatters in azimuth. b is the linear decay factor for the horizontal polarized wave distribution in azimuth.

0φa

φ

In this thesis the following propagation models have been used to cover as much as possible of the indoor and outdoor environment situations:

15


Mod

el

Ele

vatio

n/az

imut

h

Gaussian/ Uniform

Laplacian/Uniform

Elliptical

Uniform (isotropic)

Stat

istic

al d

istr

ibut

ion

( )[ ]( )

( )[ ]( )

−−−=

−−−=

2

2

2

2

22/

exp)(

22/

exp)(

H

H

V

V

mAP

mAP

σπθ

θ

σπθ

θ

φφ

θθ

1)(1)(

==

φφ

φ

θ

PP

( )

( )

−−−=

−−−=

H

H

V

V

mAP

mAP

σπθ

θ

σπθ

θ

φφ

θθ

2

2

2/2exp)(

2/2exp)(

1)(1)(

==

φφ

φ

θ

PP

22

2

22

2

)(sin)(

)(sin)(

θθ

θθ

φθ

φθφφ

θθ

θθθθ

+=

+=

ss

AP

ssAP

44444 344444 21)()(

)(

0

0

φφ

φ

φφφφ

θθθ

baAP

aAP

−=

=

high directivity, 1/8 of the globe

4444 34444 2122

2

22

2

)(sin)(

)(sin)(

φφ

φφ

φφ

φφφφ

θφ

θφθθ

+=

+=

ss

AP

ss

AP

low directivity, ½ of the globe

1)(1)(

==

θθ

φ

θ

PP

1)(1)(

==

φφ

φ

θ

PP

Scen

ario

par

amet

ers [

4][1

3]

Isot

ropi

c6

Indo

or

O

utdo

or

°=°=°=°=

=

15151010

5

H

V

H

V

mm

dBXPR

σσ

°=°=°=°=

=

30302020

1

H

V

H

V

mm

dBXPR

σσ

∞=∞=°=°=

=

H

V

H

V

mm

dBXPR

σσ

00

1

°=°=°=°=

=

15151010

5

H

V

H

V

mm

dBXPR

σσ

°=°=°=°=

=

30302020

1

H

V

H

V

mm

dBXPR

σσ

∞=∞=°=°=

=

H

V

H

V

mm

dBXPR

σσ

00

1

11.07.016.018.144.006.129.0

5.5

0

0

=

==

=

=

==

=

φ

φ

θ

φφ

θφ

φθ

θθ

baassss

dBXPR

Table 3.1 Propagation models used in the thesis work

6 For Gaussian/Uniform and Laplacian/Uniform the isotropic environment is an omidirectional isotropic environment

16


3.1.2 Efficiency of the antenna The quality of the transmission and reception of an antenna is strongly dependent on the losses at the input terminals and within the structure of the antenna. The total efficiency of an antenna has to take into account losses due to conduction and dielectric properties of the antenna as well as losses caused by the reflections due to mismatch between the transmission line and the antenna [3]. (3.8) radrefant eee ⋅=where

=ante total antenna efficiency

=refe reflection (mismatch) efficiency 2111 S−=

== dcrad eee conduction and dielectric efficiency = antenna radiation efficiency For a diversity antenna system the concept of efficiency has to be expanded to include the effects of mutual coupling. Diversity efficiency can be defined: dive couplingbranchstrongestradrefdiv eeee ⋅⋅= )( (3.9)

3.1.3 Mutual coupling When two antennas are positioned near each other, and one or both of them are excited, some energy tends to end up at the other antenna. This interchange of energy is called mutual coupling and is a mechanism of great significance for antenna system performance. It depends on radiation characteristics of the antennas, relative separation between them and their orientation. When calculating diversity gain from the field pattern mutual coupling is easily included. The received voltage V at the antenna port X (figure 3.3) is proportional to the total radiation field function of the coupled antennas X and Y when antenna X is excited. The same goes for voltage V for the case when antenna Y is excited [14].

X

Y

V

(3.10) XYYX

totYY

YYXXtotXX

IEIEEV

IEIEE

),(),(),(

),(),(),(

φθφθφθ

φθφθφθ

+=∝

+=∝

17


Figure 3.3 Equivalent circuit of a two antenna system [14]

3.1.4 Mean effective directivity In this thesis the concept of MEG has to be redefined for practical reasons. All the radiation simulations concern fields and their directivity without taking into account the antenna efficiency. The relation between MEG and mean effective directivity (MED) is quite simple

MEDeMEG rad ⋅= (3.11) MED is more attractive to use in the simulation data calculations in this thesis. The directivity of an antenna is defined as the ratio of the radiation intensity in a given direction to the radiation intensity averaged over all directions [27].

3.2 Diversity system gain As a figure-of-merit, diversity gain is a very good tool for evaluating the diversity performance improvement, but it might not express the reality. In reality, the antenna performance encounters different losses, accumulated in the antenna efficiency. An attempt to create a better figure-of-merit for the improvement is done by defining effective diversity gain [15]: (3.12) branchstongestradeff eDGDG )(⋅= To create consistence in the analysis and get an overall picture of the diversity performance, we define the diversity system gain (DSG) (3.13) diveDGDSG ⋅= including not only the radiation efficiency but also the mismatch of the strongest branch in the system.

18


4 Diversity Performance Program7 Antenna diversity gain as well as diversity system gain is not directly measurable. They are the results from the calculations performed on simulated or measured data for the antenna diversity systems. To get a complete repeatable and adequate method for measuring diversity performance it is necessary to include the calculations in the method. In this thesis a diversity performance program has been created based on the theory presented in chapter 2 and 3. Diversity Performance Program is rather a folder of Matlab programs for evaluation of the diversity performance of mobile terminals, than one single program. It consists of the code for processing the collected data, measured and simulated, the code for calculating and evaluation of the correlation, the code including different propagation environments and finally the diversity gain program.

4.1 Data codes Calculation of diversity gain can be performed on both measured and simulated data and is in this case limited to far-field pattern and S-parameters data. Data codes are created for sorting data from both far-field pattern and S-parameters, and then calculating parameters necessary for further calculation. There are three data templates created in the Diversity Performance Program folder: template_simpatport1.m, template_simpatport2.m and templateSpar.m. The first two templates are for sorting the field pattern data when each of the antenna branches is excited and it is explicitly for simulated data. For measured data only a small change is done in the template to consider the file format from the Satimo system. This template can be found in the Matlab folder Performance Evaluation among other codes that have been created and used for studies and evaluations done in this thesis. The third template is for calculations of the correlation, using simulated or measured S-parameters and the expression (2.6).

4.2 Correlation coefficient and propagation models The correlation from field pattern data can not be calculated directly in the templates. A propagation environment has to be chosen first. In the Diversity Performance Program folder a program called Propagation.m is created for calculations of the correlation coefficient, considering different propagation models, see table 3.1. The propagation models that can be used are: Gaussian/Uniform, Laplacian/Uniform, Elliptic and Uniform. For each of the models except for the uniform model an environment can be chosen: outdoor, indoor or isotropic. For the uniform case directivity has to be chosen: high or low, see table 3.1. The propagation program also calculates the mean effective gain for each of the branches (3.1) needed together with the correlation for calculation of diversity gain (2.7).

7 Matlab folders and code are not included in this report because they are the property of Sony Ericsson Mobile Communication AB in Lund

19


4.3 Diversity Performance Finally, knowing the correlation coefficient and the mean effective gain for antenna branches from the far-field pattern, diversity gain and diversity system gain can be calculated in a program called Antenna Diversity.m using expression (2.7) and expression (2.1). Not having the mean SNR:s accessible, assumptions were made that the noise levels are equal and neglectable for both the branches and that the signals can be replaced by the mean effective gains of the antennas. This simplifies the use of expression (2.7). Using S-parameters diversity gain can be calculated solely for the case of unbalanced antennas (mean effective gains are equal for both the branches) because of no access to the antenna efficiencies. The results from running the Diversity Performance program folder are plots of the correlation and the diversity gain, as well as figures for the correlation, the mean effective gains, the diversity gain and the diversity system gain calculated from the far-field pattern data and the correlation calculated from S-parameters data. No interface has been created for the program folder, all the files are run from the Matlab workspace. One of the reasons why the interface has not been in the focus is that the collected data files, especially from the simulation program IE3D are unfortunately not compatible with Matlab requiring some adjustments before running the programs.

20


5 Prototypes All the simulations and measurements presented in this report have been done on five different prototypes created especially for this thesis work regarding diversity performance.

5.1 Two-slot antenna system

Figure 5.1 Measurement prototype Figure 5.2 Simulation prototype The two-slot antenna system is the simplest of the prototypes with the size 45x90 mm2 and slot size of 2mm cut out from the ground plane. The resonance frequency of the antenna system is 2GHz. The matching is quite good and coupling is low (see appendix B.1). The limit for a good matching is in this thesis work return loss of -10dB or less. The diversity technique used here is both space and polarization diversity. Though the spacing between the slot antennas is more then half a wavelength the space diversity is the dominant diversity mechanism. 5.2 Clamshell antenna system Figure 5.3 Measurement prototype Figure 5.4 Simulation prototype

21


The clamshell antenna system consists of a two-slot antenna system with an extended ground plane. The total size is 45x190 mm2 including the gap for connection between the planes. The antenna system is created for a 2GHz resonance with similar characteristics as for the two-slot antenna system, see appendix B.2.

5.3 Two-monopole antenna system

Figure 5.5 Measurement prototype Figure 5.6 Simulation prototype

This prototype has the antenna system placed on two T610 antenna carriers, 10mm above the ground plane. The size of the prototype is also 45x90 mm2 and the resonance frequency is near 2GHz. The diversity technique used here is space diversity. The antennas in the system are well matched but the coupling is higher than for the two previous prototypes.

5.4 PIFA-slot antenna system

Figure 5.7 Measurement prototype, PIFA antenna Figure 5.8 Simulation prototype

22


Figure 5.9 Measurement prototype, slot antenna As the name reveals this antenna diversity system consists of a PIFA type antenna and a slot antenna. The PIFA antenna is a common antenna found in many late models of Sony Ericsson mobile terminals and it comprises GSM (900 MHz band), DCS (1800 MHz band), PCS (1900 MHz band) and a Bluetooth antenna (2,4 GHz band). This causes several resonance frequencies and affects the 2GHz frequency performance of the slot antenna (see appendix B.4). The slot antenna is half a wavelength long (75 mm) and cut out from the ground plane. The PIFA antenna is 10mm above the ground plane placed on a T610 antenna carrier. Both antennas are relatively well matched but the coupling is, not surprisingly, the strongest of all the prototypes because of the very small spacing between the antennas.

5.5 Two-feed patch antenna

Figure 5.10 Measurement prototype Figure 5.11 Simulation prototype

The patch antenna used in this thesis is created by Anders Derneryd at Ericsson in Mölndal. It is a two orthogonally fed antenna using the spacing and orthogonal placements of feeding points as a diversity technique. The size of the antenna is 33x33 mm2 with a parasitic metal plate (size 39x39 mm) placed about 11 mm above the ground plane for an increased bandwidth. The antenna system resonance frequency is at 2.7 GHz and it is a very well matched with a very low mutual coupling, see appendix B.5.

23


24


6 Simulations As seen in chapter 2, the calculation of correlation and then the diversity performance can be performed by using either far-field patterns of the antennas in the diversity system, or by using the S-parameters of the diversity system itself. To compare the results from these two different parameters, before doing the practical measurements, simulations have been performed on the simulation prototypes seen in chapter 5. The simulation program used was the IE3D electromagnetic simulation program from Zeland. IE3D is a full-wave, method-of-moments (MoM) simulator solving the integral equations through the use of the Green’s function resulting in the current distribution in 3D and multilayer structures of a general shape. Quite quick and easy to use, the program has a Windows based graphic interface where 3D metallic and multilayer structures can be built using a set of polygons and manipulating vertices [31]. Most of the simulation prototypes in this thesis are made as 3D simulation files to be comparable with the realistic 3D measurements. They lack a dielectric loading. Only the patch antenna is made with an infinite ground plane limiting the simulation to radiation in the upper hemisphere. In this way it is easier and faster to simulate. In reality there is also diffracted pattern on the other side of the antenna, which can affect the comparison between simulation and measurement results. The principle of simulation of field pattern in IE3D is that one antenna at a time is excited while the other is terminated with a 50 ohm load. For S-parameter measurements the antenna system (both antennas) was simulated as a whole. At first the simulations have been performed for 10 frequencies for far-field pattern and 201 for S-parameters. Unfortunately the file format from IE3D is not compatible with Matlab which made it very time consuming to arrange the collected data from field pattern before the calculation. Therefore the simulation frequency number was cut to five. S-parameter data is much easier to handle, causing no need for reducing the number of frequency points.

25


26


7 Measurements The practical measurements of the prototypes to collect the data for calculation of correlation and diversity performance have been done for both far-field pattern and S-parameters. While S-parameters measurements were quite simple and quick to perform, calibrating and then using a network analyser, the measurements of the field patterns have been a lot more time consuming. Far-field pattern measurements have been performed in a Satimo measurement system but also a scattered field chamber has been used for measuring the performance.

7.1 Satimo Stargate 64 system The Satimo measurement system is a real time spherical near field antenna testing system that is built up in an anechoic chamber with an array of antenna probes encircling a rotation table for the device to be tested on. The concept of the system is based on a technique that identifies the signal from each probe in an array by perturbing the electromagnetic properties of the probe. The perturbation results in a modulation frequency component, in the output signal, which is directly related to the amplitude and phase of the incident field at the location of the probe. By modulating each probe, sequentially, the measurement is performed in “real time”. The near-field measurement data is transformed to a far-field radiation pattern using well known numerical techniques based on spherical wave expansion [13].

Courtesy of Satimo

Figure 7.1 Anechoic chamber of the Satimo measurement system [32]

Measurements in the Satimo system can only be done for one antenna at a time, since there is only one channel on the rotation table. This means that only one antenna at the time is excited and its far-field pattern is measured and then the field patterns from the antennas are combined in the Diversity Performance Program. This is the same procedure as in the case of simulating the field pattern for the prototypes.

27


Measurements have been performed for all the prototypes by letting one antenna be excited while the other one is terminated with a 50 ohm load or left open or short circuited. Then the antennas are switched to repeat the procedure. The prototypes are measured in free space, data-mode position, and for the case of termination with a 50 ohm load measurements were performed in handheld position as well. The handheld position for the two-feed patch antenna was not of any interest.

Figure 7.2 The hand substitute used for the handheld measurements

7.2 Scattered field chamber (SFC) Scattered field chamber is a Sony Ericsson version of a reverberation chamber, or as it is also called, mode stirred chamber. It is a metal cavity with three mechanical paddles that rotate inside the chamber and reflect and scatter the electromagnetic field. The electromagnetic field inside a metal cavity bounces back and forth between the metal walls and builds up standing waves. These waves are effectively randomized by the paddles so that different interference patterns arise and a uniform multi-path environment is created. The statistical power that is measured in the cavity has a chi2 distribution with two degrees of freedom, it is independent of the location and the orientation and in that way it is a homogenous and isotropic field [22].

Two kinds of measurements have been performed in the SFC. The first one is of the same principle as simulations and Satimo measurements: one antenna at a time is measured while the other is either terminated with a 50 ohm load or left open or short circuited. Free space and handheld positions were used; data-mode was not of any importance since the SFC has an isotropic propagation environment. The second measurement method is more of a diversity measurement method. While there are two channels on the rotation table, where the test device is placed, it is possible to measure the receiving powers at both antennas at the same time. The following set up is used for both of the measurement versions in the SFC, it is only slightly modified for the first version.

28


Courtesy of Sony Ericsson Mobile Communications AB

Figure 7.3 Set up for a diversity measurement in the SFC

Working as receivers the antennas on the prototypes in the SFC are exposed with the power radiated from a Cobra antenna [29] with 0dBm controlled output power. The measurement, whether using only one or using both the channels, is performed by a Matlab program Meas.m created by Johan Andersson of the Antenna Section in Sony Ericsson. The analysis of the data can be done by using the correlation function in Matlab and for the diversity case using diversity_best_data.m also by Johan Andersson. Using diversity_best_data.m diversity gain can be calculated [2], see figure 7.4. This method is to be compared with simulations and Satimo measurements to find which the simplest repeatable method of measuring diversity performance is. Before the measurements were performed it was necessary to verify that the mean gain received at both channels were the same, since the measurement method has not been verified before, and there were incidents of different results despite the identical test device, same length of the cables and identical matching at both channels. The test was done by using two identical T-dipoles, connecting both of them to one channel each, and then switching the channels after the first measurement. The distances between the antennas were 1cm and 5cm. The measurements of two identical T-dipoles showed that there was a difference between the gains of the channels in the SFC of maximum 0.5dB. This is acceptable and enough to consider the channels to be the same.

29


Figure 7.4 Calculation of diversity gain [2]

Figure 7.5 A comparison between the mean gains of the T-dipoles for two different cases of connection with the SFC channels. Antenna 1 case 1 should be compared with antenna 1 case 2 and antenna 2 case 1 should be compare with antenna 2 case 2. The maximum difference between the cases is 0.5dB. In every diversity measurement in the SFC, a large number of the data samples are collected [2]. Starting with 600 data samples per frequency, the evaluation could not be done correctly since the values were not consistent due to the large statistical variations. Optimizing the time for the measurement and the quality of it the final number of samples chosen was 2400.

30


8 Results

8.1 Correlation coefficient

8.1.1 Far-field pattern and S-parameters The correlation coefficient is an important parameter in calculation of the diversity performance, as well as an early indication of the same. The two methods of calculating the correlation coefficient, with far-field pattern data and with S-parameters, are compared to see if they are equivalent. The comparison is made on simulation data which gives the following results:

Figure 8.1 Simulated envelope correlation coefficients for the two-slot antenna system.

Figure 8.2 Simulated envelope correlation coefficients for the clamshell antenna system.

31


Figure 8.3 Simulated envelope correlation coefficients for the two-monopole antenna system.

Figure 8.4 Simulated envelope correlation coefficients for the PIFA-slot antenna system.

Figure 8.5 Simulated envelope correlation coefficients for the two-feed patch antenna system.

32


The collected simulation data is directivity data. It is used in calculation of the correlation coefficient in expression (2.3). The expression (2.3) involves gain and not directivity but this has not affected the results for the prototypes with low correlation. For the case of the PIFA-slot antenna system the results are affected. That is why a simulation, including efficiency, was performed for this case i.e. gain data was used in the calculation of the correlation coefficient.

Figure 8.6 Simulated envelope correlation coefficients for the PIFA-slot antenna system with very high conductive antennas i.e. calculation made with gain

8.1.2 Satimo For the case with the Satimo measurements, only one antenna at the time was able to be excited. The following results for the case with a 50 ohm termination load are presented here:

Figure 8.7 Envelope correlation coefficient for the two-slot antenna system measured in Satimo chamber for termination 50ohm

33


Figure 8.8 Envelope correlation coefficient for the clamshell antenna system measured in Satimo chamber for termination 50ohm

Figure 8.9 Envelope correlation coefficient for the two-monopole antenna system measured in Satimo chamber for termination 50ohm

Figure 8.10 Envelope correlation coefficient for the PIFA-slot antenna system measured in Satimo chamber for termination 50ohm

34


Figure 8.11 Envelope correlation coefficient for the two-feed patch antenna system measured in Satimo chamber for termination 50 ohm

Measurements with the cases open and short circuited have also been performed in the Satimo system. They gave quite low correlation coefficient values, which is something that should be further investigated. This is unfortunately additional work that did not fit in the time plan for this master thesis.

8.1.3 Network analyzer Measurements of the S-parameters for the prototypes with a network analyzer were quite simple and fast, still unfortunately not all the results are presented here. For two-slot and two-monopole antenna systems the S-parameter data contained some distortion that caused strange irregularities in the correlation coefficients. Results for these two prototypes are not included here, but since the results of the three other prototypes look satisfying the method is acceptable.

Figure 8.12 Correlation calculated the for clamshell antenna system from S-parameters measured with a network analyzer

35


Table 8.13 Correlation calculated for the PIFA-slot antenna system from S-parameters measured with a network analyzer

Table 8.14 Correlation calculated for the two-feed patch antenna system from S-parameters measured with a network analyzer

8.1.4 SFC Correlation coefficients were measured for one antenna at the time with second antenna terminated with a 50 ohm load, open and short circuit.

Figure 8.15 Envelope correlation coefficient for the two-slot antenna system measured in SFC with a 50 ohm termination, open and short circuit.

36


Figure 8.16 Envelope correlation coefficient for the clamshell antenna system measured in SFC with a 50 ohm termination, open and short circuit.

Figure 8.17 Envelope correlation coefficient for the two-monopole antenna system measured in SFC with a 50 ohm termination, open and short circuit.

Figure 8.18 Envelope correlation coefficient for the PIFA-slot antenna system measured in SFC with a 50 ohm termination, open and short circuit.

37


Figure 8.19 Envelope correlation coefficient for the two-feed patch antenna system measured in SFC with a 50 ohm termination, open and short circuit. Measurements in SFC with both channels simultaneously gave the following results:

Figure 8.20 Envelope correlation coefficient for the two-slot antenna system measured in SFC.

Figure 8.21 Envelope correlation coefficient for the clamshell antenna system measured in SFC.

38


Figure 8.22 Envelope correlation coefficient for the two-monopole antenna system measured in SFC.

Figure 8.23 Envelope correlation coefficient for the PIFA-slot antenna system measured in SFC.

Figure 8.24 Envelope correlation coefficient for the two-feed patch antenna system measured in SFC.

39


8.1.5 T-dipoles

Figure 8.25 Simulated envelope correlation coefficient for the two T-dipoles with a distance of 1cm.

Figure 8.26 Envelope correlation coefficient for the two T-dipoles with a distance of 1 cm measured in the SFC

40


8.1.6 Handheld measurements As discussed in chapter 1.4 the body loss effects on diversity performance have not been in focus in this thesis. The measurements done on the handheld prototypes both in the SFC and in the Satimo system are interesting for the simple indication that there are some effects to consider in some additional work concerning diversity studies and evaluation. The results can be seen in Appendix C.

8.2 Diversity performance results – a summary

8.2.1 Simulated diversity performance

Antenna efficiency (%)Antenna 1 Antenna 2

Radiation efficiency (%) Antenna 1 Antenna 2

Two-slot antenna system 2GHz

97.0

98.7

99.2

99.0

Clamshell antenna system 2GHz

86.8

86.0

99.5

99.3

Two-monopole antenna system 2GHz

79.7

79.7

99.5

99.5

PIFA-slot antenna system 2GHz

61.2

24.5

100

99

Two-feed patch antenna system 2.7 GHz

93.3

93.3

93.7

93.7

Table 8.1 System parameters for simulated prototypes at the resonance frequency

41


Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%, 50%

Diversity System Gain

(dB) 1%, 50 %

Outdoor

-17.7

-19.4

310−

9.1, 1.6

8.8, 1.5

Indoor

-14.7

-16.9

410−

8.6, 1.72

8.3, 1.7

Gau

ssia

n/U

nifo

rm

Isotropic

-14.2

-16.8

610−

8.8, 1.6

8.6, 1.5

Outdoor

-17.7

-19.6

0.002

9.0, 1.5

8.8, 1.5

Indoor

-14.8

-16.9

0.001

8.9, 1.8

8.2, 1.8

Lap

laci

an/U

nifo

rm

Isotropic

-14.2

-16.8

610−

8.8, 1.6

8.9, 1.5

High directivity

-14.3

-18.3

0.022

7.4, 1.0

7.2, 1.0

Elli

ptic

Low directivity

-13.0

-18.9

0.008

7.2, 0.4

7.0, 0.4

Uniform

-3.1

-5.7

610−

9.0, 1.3

8.7, 1.3

Table 8.2 Diversity performance results for the two-slot antenna system at 2GHz

42


Propagation models MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ

Diversity Gain (dB)

1%,50%


(dB) 1%,50%

Outdoor

-16.2

-15.8

0.011

9.6, 2.4

8.3, 2.1

Indoor

-14.1

-13.2

0.014

9.4, 1.9

8.1, 1.7

Gau

ssia

n/U

nifo

rm

Isotropic

-14.1

-12.9

0.015

9.3, 1.8

8.0, 1.5

Outdoor

-16.2

-15.8

0.01

9.6, 2.4

8.3, 2.1

Indoor

-14.1

-13.3

0.02

9.5, 2.0

8.2, 1.7

Lap

laci

an/U

nifo

rm

Isotropic

-14.1

-12.9

0.015

9.3, 1.8

8.0, 1.5

High directivity

-14.8

-13.0

0.002

9.2, 1.6

7.9, 1.4

Elli

ptic

Low Directivity

-15.5

-13.1

0.026

9.1, 1.4

7.8, 1.2

Uniform

-3.0

-1.8

0.015

9.0, 1.7

7.7, 1.5

Table 8.3 Diversity performance results for the clamshell antenna system at 2GHz

43


Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%,50%


(dB) 1%,50%

Outdoor

-14.6

-14.7

0.015

10.0, 2.6

7.9, 2.1

Indoor

-13.9

-14.0

0.023

8.9, 2.4

4.4, 1.9

Gau

ssia

n/U

nifo

rm

Isotropic

-14.1

-14.3

0.030

9.9, 2.6

7.9, 2.1

Outdoor

-14.6

-14.7

0.02

10.0, 2.6

7.9, 2.1

Indoor

-13.8

-14.0

0.02

9.8, 2.4

7.8, 1.9

Lap

laci

an/U

nifo

rm

Isotropic

-14.1

-14.3

0.030

9.9, 2.6

7.9, 2.1

High directivity

-14.1

-14.5

0.103

9.8, 2.4

7.8, 1.9

Elli

ptic

Low Directivity

-14.3

-14.4

0.11

9.9, 2.5

7.9, 2.0

Uniform

-3.0

-3.2

0.03

9.7, 2.2

7.7, 1.7

Table 8.4 Diversity performance results for the two-monopole antenna system at 2GHz

44


Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%,50%


(dB) 1%,50%

Outdoor

-15.3

-15.1

0.87

5.7, 0.9

1.4, 0.2

Indoor

-14.1

-13.9

0.85

5.8, 1.1

1.4, 0.3

Gau

ssia

n/U

nifo

rm

Isotropic

-14.1

-14.1

0.84

5.8, 1.7

1.3, 0.3

Outdoor

-15.3

-15.1

0.87

5.7, 0.9

1.4, 0.2

Indoor

-14.1

-13.8

0.85

5.8, 1.0

1.4, 0.2

Lap

laci

an/U

nifo

rm

Isotropic

-14.1

-14.1

0.84

5.8, 1.7

1.3, 0.3

High directivity

-13.9

-14.6

0.85

5.5, 0.9

3.4, 0.5

Elli

ptic

Low Directivity

-14.5

-14.8

0.84

5.7, 1.1

3.5, 0.7

Uniform

-3.0

-3.0

0.84

6.1, 1.1

1.5, 0.2

Table 8.5 Diversity performance results for the PIFA-slot antenna system at 2GHz

45


Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%,50%


(dB) 1%,50%

Outdoor

-18.1

-18.2

510−

9.7, 2.2

9.1, 2.1

Indoor

-14.4

-14.4

710−

10.0, 2.7

9.3, 2.5

Gau

ssia

n/U

nifo

rm

Isotropic

-14.4

-14.4

710−

10.0, 2.7

9.3, 2.5

Outdoor

-18.1

-18.3

310−

9.8, 2.2

9.1, 2.1

Indoor

-14.9

-14.9

610−

10.1, 2.8

9.4, 2.6

Lap

laci

an/U

nifo

rm

Isotropic

-14.4

-14.4

710−

10.0, 2.7

9.3, 2.5

High directivity

-11.1

-14.3

0.001

8.0, 1.1

7.4, 1.0

Elli

ptic

Low Directivity

-12.4

-12.6

410−

9.6, 2.2

9.0, 2.0

Uniform

-3.3

-3.3

810−

9.8, 2.3

9.1, 2.2

Table 8.6 Diversity performance results for the two-feed patch antenna system at 2.7GHz

46


8.2.2 Measured diversity performance in the Satimo system Diversity system gain is presented here because the antenna efficiencies have not been collected at the same time as measurements in the Satimo system were performed.

Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%,50% Outdoor

-5.5

-9.4

0.27

7.0, 0.7

Indoor

-6.7

-9.5

0.21

8.1, 1.0

Gau

ssia

n/U

nifo

rm

Isotropic

-8.3

-9.8

0.21

9.1, 1.5

Outdoor

-8.5

-9.9

0.33

9.1, 1.5

Indoor

-8.7

-9.9

0.19

9.4, 1.8

Lap

laci

an/U

nifo

rm

Isotropic

-8.3

-9.8

0.21

9.1, 1.5

High directivity

-7.9

-9.3

0.31

8.9, 1.4

Elli

ptic

Low Directivity

-10.0

-10.6

0.30

8.5, 2.2

Uniform

-3.0

-4.5

0.21

9.1, 1.5

Table 8.7 Diversity performance results for the PIFA-slot antenna system at 2GHz

47


Propagation models

MEG Antenna 1

(dB)

MEG Antenna 2

(dB) eρ Diversity

Gain (dB)

1%,50% Outdoor

-0.2

-1.8

0.02

8.6, 1.9

Indoor

-5.1

-4.4

0.04

9.8, 2.4

Gau

ssia

n/U

nifo

rm

Isotropic

-8.3

-8.1

0.03

9.7, 2.2

Outdoor

-12.3

-11.2

0.03

9.0, 1.9

Indoor

-9.2

-9.6

0.03

9.8, 2.5

Lap

laci

an/U

nifo

rm

Isotropic

-8.3

-8.1

0.03

9.7, 2.2

High directivity

-10.9

-9.7

0.03

9.5, 2.2

Elli

ptic

Low Directivity

-11.5

-9.9

0.04

8.7, 2.0

Uniform

-3.0

-2.9

0.03

9.6, 2.2

Table 8.8 Diversity performance results for the two-feed patch antenna system at 2.7GHz

48


8.2.3 Measured diversity performance in the SFC

Environment

Diversity Gain (dB) 1%


SFC, uniform

6.81

0.945

Table 8.9 Diversity performance results for the two-slot antenna system at 2GHz, measured in SFC

Environment Diversity Gain

(dB) 1%


SFC, uniform

7.56

0.67

Table 8.10 Diversity performance results for the clamshell antenna system at 2GHz, measured in SFC

Environment Diversity Gain

(dB) 1%


SFC, uniform

7.95

1.15

Table 8.11 Diversity performance results for the two-monopole antenna system at 2GHz, measured in SFC

Environment



SFC, uniform

3.73

0.26

Table 8.12 Diversity performance results for the PIFA-slot antenna system at 2 GHz, measured in SFC

Environment



SFC, uniform

8.14

0.47

Table 8.13 Diversity performance results for the two-feed patch antenna system at 2.7 GHz, measured in SFC

49


50


9 Discussion Being an important parameter for the diversity performance, focuses in this master thesis on evaluation of the antenna signal correlation and the way of collecting the parameter for indication and calculation of the diversity improvement. Calculating the correlation coefficient using far-field pattern or S-parameters is confirmed to be equivalent for matched antennas in the diversity system. It is of importance to include antenna efficiency in the calculations of the field pattern. Using gain values, and not directivity values, is something that can easily be overlooked when using simulation data. This is actually not critical for very low correlation coefficients ( <eρ 0.1) but as in the case of the PIFA-slot antenna system with higher correlations a difference is noticed, see figure 8.4. In figure 8.4 calculation of correlation coefficient has been done using directivity values of the far-field patterns. To get rid of the difference, a simulation has been performed on the prototype after changing the metal properties of the antenna system to be highly conductive (very high efficiency). For very high efficiencies, when directivity and gain of the antennas are practically the same, the results were according to the figure 8.6 showing good agreement between far-field pattern and S-parameters. Initially, the data collection of the far-field pattern in the Satimo measurement system was not satisfying in its values. For correlation coefficients less than 0.1 the results agreed with the simulated ones quite well. For higher correlation coefficients the results are confusing and not at all agreeing with the simulated ones. Of course there is a difference between the simulated models and real prototypes, especially if one antenna is not well matched in the case of the real prototype, see appendix B.4. Still, the effect seems to be greater than the actual difference. Measurements with the network analyzer showed, however, that it might be just prototype related properties that make it harder to compare simulation- and measurement results, see figures 8.10 and 8.13. Collecting correlation data by measuring S-parameters seems to match collection from Satimo measurements better than simulation even if not all prototypes can be compared because of the distortion when measuring S-parameters. Since the prototypes where created on the basis of the position and the size of the antennas in the diversity systems, the correlation coefficients of the prototypes could not be absolutely predicted. Only one prototype, the PIFA-slot antenna system ended up with higher correlation coefficients. That is why further evaluations on the high correlation prototypes should be performed to confirm the discussion above. However, on the basis of the results on the PIFA-slot antenna system, both Satimo and network analyzer measurements can be considered as acceptable methods of collecting correlation data. Using a network analyzer is a quick way of getting an indication of diversity but only an indication. Calculating signal correlation from S-parameters can not be used for calculating the diversity improvement alone. Efficiencies as well as mean effective gain values for the antennas in the diversity system are necessary for the calculation but they can not be measured with a network analyzer. Measurements of far-field pattern in the Satimo system provide these parameters automatically which makes the Satimo measurements more useful in this context.

51


The correlation coefficient was also calculated from data measured from the SFC. The performance of the measurements was the same as for the Satimo system but the data collected was actually received antenna signal samples. As the results in chapter 8.1.4 show for low correlation coefficient the behavior is the same as for the Satimo- and simulation data so it is once again better to concentrate on the prototype case with high correlations, i.e. the PIFA-slot antenna system. In this case the signal correlation curve in figure 8.18 is quite different from the simulated one, and from curves from the Satimo and the network analyzer data. It shows very low correlation which is not the case for the prototype. The possible explanation why this might have occurred is that unlike measurements in the Satimo system, the measurements in SFC are not repeatable. The paddles and rotation table are not set to the same start position at every measurement causing, together with the statistical measurement method, results that are kind of mean values. This method is not acceptable for diversity performance evaluations. For the case of using both channels simultaneously for the antenna system in the SFC the results are much better. Still, the correlation curve in figure 8.23 differs from the curves from the Satimo-, the network analyzer and simulation data making the method questionable. Since the real high correlation prototype might be a source of errors, a set of T-dipoles have been used for the same measurements and simulations. The results in selection 8.1.5 show there might be a resemblance between the SFC results and simulation results. There is a need for further measurements and evaluations before this method may be accepted for collecting data for diversity performance calculations. It may be used as an indicator for low or high correlation but not quite for absolute correlation values yet. Diversity calculations have been summarized in selection 8.2 for all the prototypes and methods of collecting the data. For the case of diversity calculations with the Satimo system, only two-feed patch and PIFA-slot antenna systems were taken into consideration due to the large and time consuming data files. The diversity results show quite strongly that it is not only the signal correlation that defines the antenna diversity improvement, even low correlations can in different environments result in lower improvement due to the effect the environments have on mean effective gain of the antennas in a diversity system. The propagation models used in this thesis work seem to affect the diversity performance in a similar way. For most of the prototypes, the largest difference between the diversity gain values for different propagation models is 1-2 dB. The most similar are Gaussian/Uniform and Laplacian/Uniform models, especially for the outdoor and isotropic environments. The elliptic model that is considered suitable for indoor environments does not show any clear differences relative to the other models throughout the prototype cases. For some cases the model shows lower diversity gain than the other models and for some cases it is similar to the rest. Since the area of propagation models is a subject of intense research, the choice of the propagation parameters in this thesis is neither general nor fixed. They are the cases taken from the papers where someone else has tested them to work in some non-general environments. That is why the calculation programs in this thesis are open for further changes of the propagation parameters in the future.

52


There is a noticeable difference when comparing the diversity performance results for different methods of collecting the data. The difference between results for the simulation and the Satimo data for the PIFA-slot antenna system is probably the same difference that could be seen when comparing the correlation curves due to the properties of the real prototype. For the two-feed patch antenna system the results agree with deviations of at most 1-2 dB. Especially for the uniform case, results agree quite well. The diversity results from the SFC are lower than from simulations and the Satimo data, a deviation of at least 2 dB is present. A comparison is made between the results from the uniform propagation model for all the cases and no agreement could be seen. The possible reason for this might lie in the measurement and calculation method for the SFC, but since it has not been verified at Sony Ericsson either in this thesis or before, this can not be confirmed. Since the maximum theoretical reference for the diversity gain is 10 dB at 99% reliability (see figure 2.5) the diversity gain results seem to be quite reasonable. The values of the diversity gain and the diversity system gain for 99% reliability indicate an evident improvement of signal received at the antenna ports. The improvement results are valid for high efficiency antenna systems but also for some lower ones as in the case of the PIFA-slot antenna, see table 8.1 and table 8.5. For the PIFA-slot case the difference between diversity gain and diversity system gain due to the efficiency is obvious. Considering that the diversity system gain is the most realistic figure-of-merit for the diversity improvement the values in this case are quite low. On the other hand, considering that the antenna efficiency requirements for a consumer mobile product should be higher, the diversity improvement would probably not reach those low values. For reliability of 50 %, even higher antenna efficiencies should be considered. Finally, some reflections on the handheld measurements done in this thesis work: The results are in appendix C and they are showing a great change in the mean gain for the antenna that is closer to the hand. For the cases of the clamshell and the PIFA-slot antenna systems there is no significant change because the antennas lie outside the hand coverage. The change of the gain of the antenna should affect the diversity performance, probably proportionally.

53


54


10 Conclusions Diversity performance is a large, interesting and promising area within the antenna theory that can be transferred into practical applications not only in base station antenna systems, but also in the area of small mobile terminals. This has been shown in this master thesis by applying diversity performance techniques on five different small prototype terminals. After many simulations, measurements and calculations on the prototypes, the final diversity results have shown an evident improvement in the received signal, being about 7-9 dB for the two antenna systems with high antenna efficiencies and low correlations. Even for antenna systems with lower efficiencies and higher correlations the improvement is quite obvious.

The studies and evaluations in this master thesis have confirmed that calculation of the correlation coefficients using far-field pattern or S-parameters is equivalent for matched antennas in an antenna diversity system and assuming uniformly incoming waves. Collecting data for the calculation can be done by simulations in IE3D or measurements in the Satimo measurement system or with network analyzer. For further calculations of the diversity performance, only simulations and measurements done in the Satimo system can be considered, as correlation and diversity results showed. For a repeatable and adequate method of the measuring diversity performance it is recommended to use one of the mentioned methods of collecting far-field pattern (IE3D and the Satimo system) and S-parameters (IE3D) data and then processing the data in the Matlab based program folder Diversity Performance Program, created in this master thesis for calculation of diversity gain and diversity system gain considering different propagation models. However, the S-parameter method is only valid for the isotropic propagation model. The diversity performance of the antenna system is dependent on the environment but since the results have shown that most of the models affect the diversity performance similarly, no model has attracted any special attention. The environment parameters can not be considered as general.

55


56


11 Future work For the near future and in connection to this master thesis, the following tasks on antenna diversity performance could be of interest:

• An evaluation and a verification of diversity performance measurement in the SFC at Sony Ericsson. Method for measuring diversity gain in the SFC has been proposed by Paul Hallbjörner [9] and Per-Simon Kildal [13].

• How does the user’s presence affect diversity performance? FDTD

simulations and measurements are suggested for this investigation. • A study on propagation models. What parameters are to be used? • An analysis of the effect of antenna loading on the gain and diversity

improvement. • A study on effects of spacing between the antennas on the correlation, mutual

coupling and diversity performance, see [7] and [8].

57


58


12 References [1] Ahlin L., Zander J., Digital radiokommunikation-system och metoder, Studentlitteratur, 1992 [2] Arai H., Measurement of Mobile Antenna Systems, Artech House, 2001 [3] Balanis C. A., Antenna Theory, Second Edition, John Wiley&Sons, 1997 [4] Blanch S., Romeu J., Corbella I., Exact representation of antenna system diversity performance from input parameter description, Electronic Letters, Vol. 39, No. 9, May 2003, pp.705-707 [5] Braun C., Engblom G., Beckman C., Evaluation of Antenna Diversity Performance for Mobile Handsets Using 3-D Measurement Data, IEEE Trans. on Antennas and Propagation, Vol. 47, No.11, November 1999, pp.1736-1738 [6] Brown W.C. T., Antenna Diversity for Mobile Terminals, PhD Dissertation, University of Surrey 2002 [7] Derneryd A., Kristensson G., Antenna Signal correlation and its relation to the impedance matrix, Electronic Letters, Vol. 40, 2004, pp.401-402 [8] Derneryd A., Kristensson G., Signal correlation including antenna coupling, Electronic Letters, Vol. 40, 2004, pp.157-159 [9] Douglas M. G., Okoniewski M., Stuchly M.A., A Planar Diversity Antenna for Handheld PCS Devices, IEEE Trans. on Vehicular Technology, Vol.47, No.3, August 1998, pp.747-754 [10] Fuhl J., Molisch A. F., Bonek E., Unified Channel Model for Mobile Radio Systems with Smart Antennas, IEE Proc.-Radar, Sonar Navig., Vol. 145, No. 1, February 1998, pp.32-41 [11] Hallbjörner P., Measuring Terminal Diversity Performance with a Mode Stirred Chamber, Internal Report UA/H-2000:051, Ericsson Mobile Communication AB, June 2000 [12] Hallbjörner P., Madsén K., Terminal Antenna Diversity Characterization Using Mode Stirrer Chamber, Electronic Letters, Vol. 37, No. 5, March 2001, pp. 273-274 [13] Iversen P. O., Garreau P., Englund K., Pasalic, E., Edvardsson, O., Engblom, G., Real-time spherical near-field antenna test facility for personal communications applications, IEEE AP2000 Conference, Davos April 2001 [14] Kildal P., Rosengren K., Electromagnetic Analysis of Effective and Apparent Diversity Gain of Two Parallel Dipoles, IEEE Antennas and Wireless Propagation Letters, Vol. 2, 2003, pp.9-13

59


[15] Kildal P., Rosengren K., Byun J.,Lee J., Definition of Effective Diversity Gain and How to Measure It in a Reverberation Chamber, Microwave and Optical Technology Letters, Vol. 34, No. 1, July 2002, pp.56-59 [16] Knudsen M. B., Antenna Systems for Handsets, ATV-Industrial PhD Project EF-755, Aalborg University 2001 [17] Leather P. S. H., Parsons D., Antenna Diversity for UHF Handportable Radio, Electronic Letters, Vol. 39, No. 13, June 2003, pp.946-948 [18] Leifer M. C., Signal Correlations in Coupled Cell and MIMO Antennas, Antennas and Propagation Society International Symposium, Vol. 3, 2002, pp.194-197 [19] Meksamoot K., Krairiksh M., Takada, J., A Polarization Diversity PIFA on Portable Telephone and Human Body Effects on Its Performance, IEICE Trans. Commun. Vol. E84-B, No. 9, September 2001, pp.2460-2467 [20] Ogawa K., Matsuyoshi T., Monma K., An Analysis of the Performance of a Handset Diversity Antenna Influenced by Head, Hand and Shoulder Effects at 900 MHz:Part I-Effective Gain Characteristics and Part II-Correlation Characteristics, IEEE Trans. On Vehicular Technology, Vol.50, No.3, May 2001, pp.830-853 [21] Ogawa K., Takada J., Concept of Diversity Antenna Gain, COST 273 TD(03)142, May 2003 [22] Pedersen G. F., Andersen J. B., Handset Antennas for Mobile Communications: Integration, Diversity and Performance, Review of Radio Science1996-1999, August 1999, pp. 119-137 [23] Rosengren, K., Measuring Radiation Efficiency of Terminal Antennas in Reverberation Chamber, Technical Report No 388L, Department of Electromagnetics, Chalmers University of Technology, Gothenburg 2001 [24] Schwartz M., Bennet W.R., Stein, S., Communication Systems and Techniques, Vol.4, McGraw-Hill Book Company, US 1966 [25] Sulonen K., Evaluation of Handset Antenna Configurations, Dept. of Electrical and Communications Engineering, Helsinki University of Technology, March 1999 [26] Sundström L., Jönsson G., Börjeson H., Radio Electronics, Department of Electroscience, Lund University 2003 [27] Taga T., Analysis for Mean Effective Gain of Mobile Antennas in Land Mobile Radio Environments, IEEE Transactions on Vehicular Technology. Vol. 39, No. 2, May 1990, pp.117-131 [28] Vaughan R., Andersen J. B., Channels, Propagation, and Antennas for mobile Communications, IEE, London 2003

60


[29] Ying Z., Andersson J., An ultra-broad band cobra antenna, Antenn03, Kalmar 2003. [30] Young P.H., Electronic Communication Techniques, 4th edition, Prentice Hall, 1999 [31] www.ericsson.com [32] www.zeland.com [33] www.mathworld.wolfram.com [34] www.satimo.com

61


62


Appendix A Some definitions

A.1 S-parameters

Figure A.1 A two-port network with incident and reflected waves The two-port network considered in the figure above has an incident wave voltage

at the input port that contributes to the reflected wave , but is also scattered through the two-port and contributes to the reflected waveb . The same principle is valid for the output port. Since incident waves can scatter from one port to another four different parameters can be defined that characterize the two-port. These are S-parameters and are defined as [23]:

1a 1b

2

2221212

2121111

aSaSbaSaSb

+=+=

(A.1)

A.2 The Marcum Q-function The basic mathematical definition and elementary properties of the Marcum Q-function are [23]:

[ ]

1)0,(2

exp),0(

)2()(exp)2,2(

)(2

exp),(

2

0

0

22

=

−=

+−=

+−=

∫

∫∞

∞

aQ

bbQ

dyayIyabaQ

xdxaxIxabaQ

b

b

(A.2)

where is the modified Bessel function of zero:th order. 0I

63


64


Appendix B Prototype characteristics

B.1 Two-slot antenna system

Figure B.1 Measured and simulated SWR for the two-slot antenna system

Figure B.2 Measured and simulated matching and mutual coupling8 for the two-slot antenna system

65

8 Mutual coupling = S12 in the diagram.


B.2 Clamshell antenna system

Figure B.3 Measured and simulated SWR for clamshell antenna system

Figure B.4 Measured and simulated matching and mutual coupling for clamshell antenna system

66


B.3 Two-monopole antenna system

Figure B.5 Measured and simulated SWR for the two-monopole antenna system

Figure B.6 Measured and simulated matching and mutual coupling for the two-monopole antenna system

67


B.4 PIFA-slot antenna system

Figure B.7 Measured and simulated SWR for the PIFA-slot antenna system

Figure B.8 Measured and simulated matching and mutual coupling for the PIFA-slot antenna system

68


B.5 Two-feed patch antenna system

Figure B.9 Measured and simulated SWR for the two-feed patch antenna system

Figure B.10 Measured and simulated matching and mutual coupling for the two-feed patch antenna system

69


70


Appendix C Handheld measurement results

Table C.1 Mean gain measurements in SFC for free space and handheld two-slot antenna system

Table C.2 Mean gain measurements in SFC for free space and handheld clamshell antenna system

Table C.3 Mean gain measurements in SFC for free space and handheld two-monopole antenna system

71


72

Table C.4 Mean gain measurements in SFC for free space and handheld PIFA-slot antenna system

Documents

Antenna Diversity Studies and Evaluation - Jeziorsaanteny.jeziorski.info/wp-content/uploads/2011/12/2004_Plicanic.pdf · Antenna Diversity Studies and Evaluation from the IE3D simulation