Department of Science and Technology Institutionen för teknik och naturvetenskap
Linköping University Linköpings universitet
Indoor propagation modelling
at microwave frequencies in a
server environment
Andreas Joelsson
Jonathan Ohlsson
LiU-ITN-TEK-G--15/033--SE
Indoor propagation modelling
at microwave frequencies in a
server environment
Examensarbete utfört i Elektroteknik
vid Tekniska högskolan vid
Linköpings universitet
Andreas Joelsson
Jonathan Ohlsson
Handledare Adriana Serban
Examinator Magnus Karlsson
Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –
under en längre tid från publiceringsdatum under förutsättning att inga
extra-ordinära omständigheter uppstår.
Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,
skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för
ickekommersiell forskning och för undervisning. Överföring av upphovsrätten
vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av
dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,
säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ
art.
Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i
den omfattning som god sed kräver vid användning av dokumentet på ovan
beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan
form eller i sådant sammanhang som är kränkande för upphovsmannens litterära
eller konstnärliga anseende eller egenart.
För ytterligare information om Linköping University Electronic Press se
förlagets hemsida
http://www.ep.liu.se/Copyright
The publishers will keep this document online on the Internet - or its possible
replacement - for a considerable time from the date of publication barring
exceptional circumstances.
The online availability of the document implies a permanent permission for
anyone to read, to download, to print out single copies for your own use and to
use it unchanged for any non-commercial research and educational purpose.
Subsequent transfers of copyright cannot revoke this permission. All other uses
of the document are conditional on the consent of the copyright owner. The
publisher has taken technical and administrative measures to assure authenticity,
security and accessibility.
According to intellectual property law the author has the right to be
mentioned when his/her work is accessed as described above and to be protected
against infringement.
For additional information about the Linköping University Electronic Press
and its procedures for publication and for assurance of document integrity,
please refer to its WWW home page:
http://www.ep.liu.se/Abstract
The Link¨oping site is the first of Ericsson’s three information and communication tech-nology centres. This facility will house the company’s complete portfolio and use the latest cloud technology in order to enable the research and development engineers to more efficiently test and develop new technologies. In the test lab environment there is a high capacity microwave telecommunication system called MINI-LINK. These systems operate at much higher frequencies than more traditional telecommunication systems. In the test lab these systems are communicating with a cable interface instead of its intended air interface. The purpose of this thesis is to evaluate the potential leakage of this system in the test lab environment.
The evaluation of the leakage in the test lab is done by developing an empirical path loss model for the desired frequencies used by the equipment in the test lab. This model is later implemented in a leakage simulation tool designed in Matlab, which simulates and displays the leakage power in a 2D plane. This report mainly focuses on the process of determining the constants and the implementation of the path loss model.
We would like to thank Ericsson for providing the opportunity to make this project with them. We would also like to thank our supervisors and others involved by providing support and feedback along the way.
Contents
1 Introduction 1
1.1 Background . . . 1
1.1.1 Cellular Networks and MINI-LINK . . . 2
1.2 Limitations . . . 3 1.3 Requirements . . . 4 1.3.1 Pre-study Requirements . . . 4 1.3.2 Model Requirements . . . 5 1.3.3 Measurement Requirements . . . 5 1.3.4 Project Requirements . . . 5 2 Theoretical Background 6 2.1 Path Loss . . . 6
2.1.1 Free-space Path Loss . . . 6
2.1.2 Simplified Path Loss Model . . . 7
2.1.3 Empirical Path Loss Models . . . 7
2.2 Environmental Factors on Radio Propagation . . . 10
2.2.1 Multipath Propagation . . . 10 2.2.2 Fading . . . 10 2.3 Indoor Propagation . . . 11 3 Model Development 13 3.1 Model Selection . . . 13 3.1.1 Comparison of Models . . . 13 3.1.2 Initial Model . . . 15 3.2 Measurement Equipment . . . 15
3.2.1 Interface Loss Characterization . . . 15
3.2.2 SMA Antenna Characterization . . . 16
3.2.3 Equipment Parameters . . . 18
3.3 Path Loss Exponent . . . 18
3.3.1 Path Loss Exponent Measurements . . . 19
3.4 Rack Attenuation Factor . . . 21
3.4.1 Rack Attenuation Factor Measurements . . . 22
3.4.2 Solid Object Attenuation . . . 24
4 Model Verification and Correction 27
4.1 Verification Measurements . . . 27
4.1.1 Measurement Setup . . . 27
4.1.2 Model Deviation . . . 28
4.2 Model Correction . . . 29
4.2.1 Path Loss Exponent Correction . . . 30
4.2.2 Rack Attenuation Factor Correction . . . 30
4.2.3 Deviation after Model Correction . . . 31
5 Results 32 5.1 Final Model . . . 32
5.2 Leakage Simulations . . . 33
5.2.1 Maximum Allowed Radiated Power . . . 33
5.2.2 Leakage from MINI-LINK Setup . . . 33
5.2.3 Model Simulations . . . 34
6 Discussion and Conclusion 36 6.1 General Discussion . . . 36
6.2 Requirement Evaluation . . . 37
6.2.1 Pre-study Requirement Evaluation . . . 37
6.2.2 Model Requirement Evaluation . . . 38
6.2.3 Measurement Requirement Evaluation . . . 38
6.2.4 Project Requirement Evaluation . . . 39
6.3 Conclusion . . . 40
6.4 Future Work . . . 41
Appendices 45 A Plotted Measured Data 46 A.1 Open Space Plot for 23.0 GHz . . . 46
A.2 Open Space Plot for 18.0 GHz . . . 47
A.3 Rack Environment Plot for 23.0 GHz . . . 47
A.4 Rack Environment Plot for 18.0 GHz . . . 48
B Raw data 49 B.1 Path Loss Exponent Measurements . . . 49
B.2 Verification Data Measurements . . . 50
B.3 RAF measurements . . . 51
B.4 Solid Object Attenuation Measurements . . . 52
List of Figures
1.1 Illustration of cellular data forwarding . . . 2
2.1 Free-space path loss . . . 7
2.2 Example of multipath components of transmitted signal . . . 10
2.3 Illustration of destructive and constructive interference . . . 11
3.1 Comparison of path loss models . . . 14
3.2 Conversion connector loss characterization . . . 16
3.3 Loss characterization of Sucoflex 104 cable . . . 16
3.4 Characterization with reference antenna . . . 17
3.5 Characterization with antenna under test . . . 17
3.6 Measurement setup for determining the path loss exponent . . . 19
3.7 Measured path loss for an open space environment at 23.0 GHz . . . 20
3.8 Measured path loss for an open space environment at 18.0 GHz . . . 21
3.9 Measurement setup for determining the rack attenuation factor . . . 22
3.10 Line-of-Sight and Non-Line-of-Sight components in the RAF measurements 23 3.11 Measurement setup solid object attenuation seen from above . . . 24
3.12 Estimated path loss for open space at 23.0 GHz . . . 26
3.13 Estimated path loss for open space at 18.0 GHz . . . 26
4.1 Verification measurement locations in the test lab environment . . . 28
4.2 Deviating points . . . 29
5.1 Single source simulation . . . 35
5.2 Multiple source simulation . . . 35
A.1 Measured data for open space at 23.0 GHz . . . 46
A.2 Measured data for open space at 18.0 GHz . . . 47
A.3 Measured data for rack environment at 23.0 GHz . . . 47
A.4 Measured data for rack environment at 18.0 GHz . . . 48
C.1 Default layout of the leakage simulation tool . . . 53
1.1 Definition of requirement priority . . . 4 1.2 Pre-study requirements . . . 4 1.3 Model requirements . . . 5 1.4 Measurement requirements . . . 5 1.5 Project requirements . . . 5 3.1 Simulation parameters . . . 13 3.2 Interface losses . . . 16 3.3 Component properties at 23.0 GHz . . . 18 3.4 Component properties at 18.0 GHz . . . 18
3.5 Path loss exponents in different scenarios at 23.0 GHz . . . 20
3.6 Path loss exponents in different scenarios at 18.0 GHz . . . 21
3.7 Definition of rack densities . . . 22
3.8 Rack attenuation factor for different rack densities at 23.0 GHz . . . 23
3.9 Rack attenuation factor for different rack densities at 18.0 GHz . . . 23
3.10 Solid object attenuation for different separation distances at 23.0 GHz . 24 3.11 Solid object attenuation for different separation distances at 18.0 GHz . 25 4.1 Deviation of the LOS components . . . 28
4.2 Deviation of the NLOS components . . . 29
4.3 K for 23.0 GHz model . . . 31
4.4 K for 18.0 GHz model . . . 31
4.5 Corrected deviation of the LOS components . . . 31
4.6 Corrected deviation of the NLOS components . . . 31
5.1 Measured power of the leakage scenarios . . . 34
6.1 Analysis of Pre-study requirements . . . 37
6.2 Analysis of Model requirements . . . 38
6.3 Analysis of Measurement requirements . . . 38
List of Abbreviations
AF Antenna Factor AUT Antenna Under Test
EMC ElectroMagnetic Compatibility EMI ElectroMagnetic Interference FAF Floor Attenuation Factor
GSM Global System for Mobile communication ICT Information and Communication Technology LOS Line Of Sight
LTE Long-Term Evolution NLOS Non Line Of Sight PCB Printed Circuit Board RAF Rack Attenuation Factor RF Radio Frequency
Rx Reception Tx Transmission
Introduction
In this chapter the background, limitations and requirements of the project will be pre-sented.
1.1
Background
The Ericsson facility in Link¨oping is in the process of completing the construction of its test lab, which is the first of three global test centers. Ericsson’s reason for collecting its entire portfolio of products in three locations is to allow the engineers to develop and test new solutions remotely, while Ericsson estimate that they will reduce their total energy consumption of their test facilities with 40 %, with the aim for a more sustainable future. [8]
Since the equipment in the test lab is operated remotely, the engineer using the equipment in the lab need to know that it is working as intended. The test environ-ment engineer are responsible of safely integrating new equipenviron-ment and maintaining the hardware in the test lab, in order to prevent potential problems.
Problems facing the test environment engineers when integrating new equipment is generally concerning leakages from the equipment in the lab. Since a large portion of the products in the Ericsson portfolio is designed to operate in an outdoor environment, the RF leakages need to be characterized and controlled to be able to guarantee a fully functional test environment. These leakages, if the signal strength is large enough, can also provide a possible undesired access point to the system. The leakage can also pose a health risk to the test lab employees if the emitted energy is too high.
Accurately modelling and evaluating potential problems in any leakage scenario is im-portant when sensitive equipment is placed close to each other. This potential problem is more relevant when outdoor equipment is placed in an indoor test environment, using cables, connectors and waveguides instead of using the intended air interface. Knowl-edge on how to troubleshoot potential problems before they occur is critical knowlKnowl-edge when maintaining a large amount of equipment. Knowledge about the possible leakage scenarios and the ability to evaluate different equipment placement is a valuable asset for engineers responsible for maintaining the system.
sce-June 25, 2015 CHAPTER 1. INTRODUCTION narios, leakage strength and its propagation characteristics in a server environment. The tool will give engineers the ability to plan the placement of the equipment and increase awareness where different leakage scenarios will potentially cause a problem.
1.1.1
Cellular Networks and MINI-LINK
To be able to transmit data from one remote location to another, a connection between the two points are required. Since the devices may not be located in the same area, the transmitted data will be forwarded via a microwave link and/or Ethernet/fiber in order to connect the devices.
The increased demand for high data rates and availability of modern communication systems requires high performance forwarding for nodes in the communication network. The MINI-LINK system from Ericsson offers the possibility for high capacity transfer when forwarding data between network nodes. This system allows technologies such as 4G and LTE to be implemented in a cost efficient way, while preparing the wireless network for future technologies. [10]
In the following subsections the procedure of connecting the wireless device to the Internet will be explained. The forwarding procedure is illustrated in Figure 1.1.
Figure 1.1: Illustration of cellular data forwarding Wireless Device to Base Station:
The transmitting device connects to a base station in the telecommunication network using 4G or any other applicable technology. The data is sent from the transmitting device and received by the base stations antenna. The data is collected and repackaged from several devices connected to the same base station.
Base Station to Base Station Controller:
The collected data from several connected wireless devices is sent via microwave transmis-sion to the base station controller. The base station controller is responsible of collecting transmissions from several base stations and repackaging the data.
Base Station Controller to Switching Center:
The data from several base stations are collected and repackaged. This data is forwarded in the same way as between the base station and the base station controller, but may be forwarded to a switching center. The switching center sends the data to its destination using either Ethernet or optical fiber. [2]
1.2
Limitations
The main purpose of this project is to evaluate leakage power levels of the MINI-LINK setup in the test environment at the Ericsson test lab in Link¨oping. When investigating different leakage scenarios, a simulation model can be created to estimate leakage levels in the environment. There are several ways to construct and implement a model to be able to represent the specific scenario needed. The signal propagation can be modelled either by a deterministic or an empirical approach.
A deterministic approach uses the description of the physical material in the modelling environment. In order to get an accurate model, material parameters such as relative permittivity, permeability and conductivity has to be specified for the environment. This approach requires a huge amount of data to describe the modelling environment and a huge computational effort to determine the loss contribution from the environment.
An empirical approach is based on measurements instead of physical properties of the materials. This approach can generate models with less computational effort but with the drawback of being limited to the environment and parameters used in the data acquisition. [11]
Using this knowledge a model can be formed and appropriate measurements be exe-cuted to be able to verify the model. Since the model will be used to determine signal levels for leakage evaluation, the level of accuracy needed does not warrant a determin-istic model. An empirical modelling approach will be used to determine the signal levels since it can be verified and used to evaluate the leakage in a shorter time frame.
In the list below the limitations of the project is summarized. Furthermore, the requirements of the project is presented in section 1.3.
• The project duration is 20 weeks
• The model will be based on a empirical modelling method
• The quality and quantity of empirical data will lay the foundation of the model • The model will only be verified for a maximum of two operating frequencies • The model will only be verified in the Ericsson test lab in Link¨oping
June 25, 2015 CHAPTER 1. INTRODUCTION
• The quality of the model will be determined with standard deviation calculation compared to verification measurements
• The model will only be simulated in a 2D environment
1.3
Requirements
During the project certain criteria have to be met for the project. The requirements will be presented and ranked on a priority scale as seen in Table 1.1.
Table 1.1: Definition of requirement priority Priority Description
1 Critical importance 2 High importance 3 Low importance
1.3.1
Pre-study Requirements
During the pre-study, knowledge will be acquired in order to fully understand the prob-lem. This knowledge will be used to derive and implement an empirical model. The requirements for the pre-study can be seen in Table 1.2.
Table 1.2: Pre-study requirements
No. Description Priority
1 Appropriate knowledge about signal leakage 1
2 Appropriate knowledge of current EMC standards 2 3 Evaluation of indoor propagation models & methods 1 4 Measurement techniques at microwave frequencies 1
1.3.2
Model Requirements
In order to create an appropriate tool that implements the simulation model, several aspects of the model need to be considered. The requirements for the model can be seen in Table 1.3.
Table 1.3: Model requirements
No. Description Priority
6 Model will be implemented in Matlab 1
7 Tuning of model parameters in different environments 1 8 Tuning of model parameters at appropriate frequencies 1
9 Environment mapping for simulation model 1
10 Multiple assignable leakage sources in model 1
11 Random variance consideration of model 3
12 Simulation will alert if signal level will cause potential security risk 2 13 Simulation will alert if signal level is above recommended levels 2 14 Simulation implementation will have a graphical interface 2
1.3.3
Measurement Requirements
A series of measurements need to be performed during the project in order to be able to implement the empirical simulation model. The requirements for the measurements can be seen in Table 1.4.
Table 1.4: Measurement requirements
No. Description Priority
15 Measurements will be performed with appropriate equipment 1 16 Orientation of measurement and equipment will be considered 1
17 Height of measurement will be considered 2
18 Multiple frequencies will be measured 2
1.3.4
Project Requirements
A number of documents, meetings and presentations will be performed during the project. The requirements for the project can be seen in Table 1.5.
Table 1.5: Project requirements
No. Description Priority
19 Deliver a written report at the end of the project 1
20 Weekly meetings with supervisor at Ericsson 1
21 Monthly written report to supervisor at Ericsson 1
Chapter 2
Theoretical Background
In this chapter, the theoretical background of the project will be presented. The theory will be used in order to understand and implement the requirements of the project.
2.1
Path Loss
The path loss of a transmitted signal is the energy reduction of the propagating electro-magnetic wave. Understanding and accurately modelling the path loss of a propagating wave is important to be able to plan and evaluate the range of communications systems. This project uses path loss modelling to determine the leakage power at a certain distance from the source.
2.1.1
Free-space Path Loss
The simplest way to model path loss is where only the free-space path loss is taken into account. The free-space path loss model does not consider the environment it is propagating in, since the model is based on waves propagating in space. The free-space path loss is usually used as a starting point when forming a path loss model when only the line-of-sight component is required. The free-space path loss can be seen in (2.1) and (2.2).
PL(d) =
λ2
(4πd)2 (2.1)
PL(d) dB = 20 log10(d) + 20 log10(f ) − 27.55 (2.2)
where λ is the wavelength of the carrier frequency in meters, f is the carrier frequency in MHz and d is the distance in meter between the transmitter and receiver [1]. A plot of the free-space path loss for a source with an operational frequency of 23.0 GHz can be seen in Figure 2.1.
Figure 2.1: Free-space path loss
2.1.2
Simplified Path Loss Model
Modelling path loss in complex environments, such as dense urban areas, offices or other indoor environments, requires a adjustable model for accurate path loss modelling. The simplified path loss model uses adjustable factors which can be obtained with both em-pirical and deterministic modelling methods. This gives the simplified path loss model a wide range of applicable scenarios. The simplified path loss model is shown in (2.3).
PL(d) dB = −20 log10 λ 4πd0 + 10γ log10 d d0 + ψdB (2.3)
where λ is the wavelength of the carrier frequency, d0 is the reference distance for the
model in meters, γ is the path loss exponent tuned for a specific environment and ψdB is
a Gaussian distributed random variable. This variable is used to describe the variance of the model caused by environmental effects. These variations will be explained in section 2.2. The reference distance d0 is typically set to 1 m for indoor environments, and γ is
typically in the range of 1.6 to 3 for an indoor environment. [2]
2.1.3
Empirical Path Loss Models
Accurately modelling path loss in complex environments requires more than only free-space path loss. There are several path loss models based on empirical data collected for specific model parameters such as frequency and environment. In this section, a selection of these models are presented.
June 25, 2015 CHAPTER 2. THEORETICAL BACKGROUND
Hata model
The Hata model is an empirical model typically used for estimating path loss in outdoor urban environments. The Hata model is widely used since its able to model propagation of modern cellular systems with smaller cell sizes and higher frequencies [2]. The Hata model is described by (2.4).
PL,urban(d) dB = 69.55 + 26.16 log10(fc) − 13.82 log10(ht)
−a(hr) +
44.9 − 6.55 log10(ht)
log10(d) (2.4)
where fc is the carrier frequency, ranging from 150 MHz to 1500 MHz, hr and ht are
the height of the receiving and transmitting antenna, ranging from 1-10 m for hr and
30-200 m for ht. The distance d ranges from 1 km to 100 km, a(hr) is the correction
factor, which for small and medium sized cities is given by (2.5) and for larger cities given by (2.6). [2] a(hr) dB = 1.1 log10(fc) − 0.7 hr− 1.56 log10(fc) − 0.8 (2.5) a(hr) dB = 3.2 log10(11.75hr) 2 −4.97 (2.6)
COST 231 extension to Hata model
Since the Hata model does not cover frequencies over 1.5 GHz, the European cooperative for scientific and technical research extended the model to cover an higher range of frequencies. The model is described by (2.7).
PL,urban(d) dB = 46.3 + 33.9 log10(fc) − 13.82 log10(ht)
−a(hr) +
44.9 − 6.55 log10(ht)
log10(d) + CM (2.7)
where fc is the frequency, ranging from 1.5 GHz to 2 GHz, hr and ht are the height
of the receiving and transmitting antenna, ranging from 1-10 m for hr and 30-200 m for
ht. The distance d ranges from 1 to 10 km and a(hr) is the correction factor as in the
Hata model, as seen in (2.5) and (2.6). CM is a correction factor for the model with the
COST 231 Multi Wall
The COST 231 Multi Wall path loss model provides a path loss estimation for indoor environments and is applicable for frequencies between 900 MHz and 1800 MHz. Included in the model is a linear attenuation factor for the number of walls in the propagating path, as described by (2.8). PL(d) dB = L0+ 10n log10(d) + Lc+ W X i=1
NW,i· WAFW,i
(2.8) where L0 is the path loss in dB at 1 m for the modelled frequency, Lc is an empirically
derived constants depending on the environment in dB and n is the power decay index. NW,i is the number of walls in the transmitting path and WAFW,i is the wall attenuation
factor. The WAF is typically set to 3.4 dB for light walls and 6.9 dB for heavy walls. [4][13]
ITU-R path loss model
The radio section of the International Telecommunication Union, ITU-R, implements a combination of average path loss and site specific data for estimating path loss. Included in the model is floor attenuation to the transmitted signal. The basic ITU-R model is described by (2.9).
PL(d) dB = 20 log10(f ) + N log10(d) + Lf(n) − 28 (2.9)
where f is the frequency in MHz, N is the power loss coefficient ranging from 22 to 33 in an office environment in the frequency range of 700 MHz to 70 GHz. The floor loss factor, Lf, ranges from 9 dB at 900 MHz to 22 dB at 5.8 GHz for single floor loss. [12]
WINNER II model
In the WINNER II channel models, models for both indoor and outdoor environments are derived. The model for indoor environments is described by (2.10) for the line-of-sight component and (2.11) for the non-line-of-sight. [14]
PL,LOS(d) dB = 46.8 + 18.7 log10(d) + 20 log10
f 5
(2.10)
PL,N LOS(d) dB = 46.4 + 18.7 log10(d) + 20 log10
f 5
+ WAF · NW (2.11)
where f is given in GHz between 2 and 6 GHz and d is the distance ranging from 3 to 100 m. WAF is the wall attenuation factor in dB and NW is the number of walls in
June 25, 2015 CHAPTER 2. THEORETICAL BACKGROUND
2.2
Environmental Factors on Radio Propagation
An propagating electromagnetic wave may be subjected to several propagation effects, such as reflection, diffraction, scattering and object attenuation. These aspects need to be considered when developing propagation models. This section will look into these factors.
2.2.1
Multipath Propagation
In urban and indoor environments, the signal has several additional paths of propagation besides the line-of-sight component. Reflection of objects, diffraction and scattering will affect the signal strength at the receiver. This phenomena is known as multipath propagation. In Figure 2.2, the diffraction and reflection multipath components are illustrated.
Figure 2.2: Example of multipath components of transmitted signal
The multipath propagation problems can be solved using Maxwell’s equations, but the computational effort is generally to great for a practical implementation. Instead, the electromagnetic waves are represented as simple particles, thus reducing the com-putational complexity needed. Diffraction, reflection and scattering problems are solved with geometrical equations instead of solving the partial differential equations used in Maxwell’s equations. [2]
2.2.2
Fading
Transmitted signals usually suffer from random variations in the signal level at the re-ceiver. Shadow fading, or slow fading, is caused by reflection and scattering of objects in the signal path. Generally, when considering shadow fading in outdoor and urban envi-ronments, the objects causing the fading are large and the shadowing effect is considered constant over a given number of wavelengths. [7]
The most common model used is the log-normal shadowing model. It has been empirically confirmed to accurately model variation in received power for indoor and outdoor environments. When calculating the variance of the shadow fading, the difference
between the model and measured values at several distances is needed. The variance can then be calculated using (2.12). For outdoor environments, σψdB is typically in the range
of 4 dB to 13 dB. [2] σ2 ψdB = 1 N N X i=1 Mmeasured(di) − Mmodel(di) 2 (2.12) Fast fading variations behave in the same way as slow fading, but are generally caused by minor variation closer to the receiving antenna due to multipath components from the transmitter. The multipath components cause destructive or constructive interference, as seen in Figure 2.3, affecting the received signal power, due to the relative phase of the electromagnetic waves. For indoor environments, the fading can be considered to have fast fading characteristics. [7]
Figure 2.3: Illustration of destructive and constructive interference
2.3
Indoor Propagation
When modelling path loss, free-space path loss is not sufficient to accurately model the actual path loss in an indoor environment. The propagation environment is generally more complex with object causing attenuation, reflections and diffractions of the trans-mitted electromagnetic waves.
For indoor propagation there are three major ways to model the propagating wave; ray tracing, dominant path and direct path. Ray tracing is the most complex of the methods, as it often uses a deterministic modelling approach. The ray tracing method approximates the electromagnetic waves as rays to be able to determine the contribution
June 25, 2015 CHAPTER 2. THEORETICAL BACKGROUND of each multipath component. Modelling path loss using ray tracing methods requires more data and is more computationally demanding, but yields a more accurate result.
Direct path models are the most computational efficient, since the model only con-siders the direct path between the transmitter and receiver. Models using the direct path method does not only consider free-space loss, but may also consider loss caused by objects in the propagation path. Typically propagation through walls and floors is considered in these models, as described by (2.13).
PL,total = PL(d) + Nf X i=1 FAFi+ Nw X i=1 WAFi (2.13)
where FAF is the floor attenuation factor, WAF is the wall attenuation factor and PL(d) is the path loss from any given path loss model.
When analyzing different typical scenarios for propagation, it has been found that in many cases there are one path of propagation that has the largest contribution to the received signal strength. The dominant path models calculates the path loss contribution for a series of propagation paths. The model then disregards all paths but the dominant one. The dominant path models are less computational demanding than the ray tracing models and are more accurate than the direct path models.
Model Development
In this chapter, models described in section 2.1 will be compared. Then, an initial model adapted to the project specifications will be presented.
3.1
Model Selection
In the requirements previously presented in section 1.3.2, the model needs to work at specified frequencies. Since the main operational frequency of the MINI-LINK system used in the test lab is 23 GHz, the model need to be able to handle this frequency. A secondary frequency of 18 GHz was selected in order to extend the usage of the model. 18 GHz was selected since it is an operational frequency being used by some of the MINI-LINK radio units in the test lab.
3.1.1
Comparison of Models
In order to be able to select a model to implement, a preliminary simulation of the models in section 2.1 was performed. The parameters used for the simulations can be seen in Table 3.1. The simulations have all been performed with a carrier frequency of 23 GHz.
Table 3.1: Simulation parameters
Model Parameter Value
Simplified Path loss model d0 1 m
γ 2.5
ITU-R N 25
Hata & COST 231 ext. to Hata ht 1 m
hr 1 m
Cm 0
COST 231 Multi Wall L0 20 log10
λ 4π
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT A path loss comparison of the selected models can be seen in Figure 3.1. The com-parison only consider the line-of-sight component in the models.
Figure 3.1: Comparison of path loss models
The free-space model is accurate when the line-of-sight component is dominant and reflection is negligible. This model is not suitable in an indoor environment, since it does not take reflections, diffusion and object attenuation into account, which is not negligible in this kind of indoor environment.
The Hata and COST 231 extension to Hata models may be very useful for cellular coverage estimation for larger areas with the frequencies covered, but is not suited for modelling systems at the frequencies nor the environment needed for this project. This can be seen in Figure 3.1, as the path loss is significantly higher than for the compared models.
The COST 231 Multi Wall, ITU-R model and WINNER II gives an idea how indoor path loss models are designed. These models generally give environment properties for general cases but not for the environment required for this project.
The simplified path loss model is adjustable for any frequency and environment, and such is suitable for an indoor propagation model at a generic frequency as required by this project. Object attenuation needs to be taken into consideration when implementing the final model, which the basic simplified path loss model does not.
3.1.2
Initial Model
Using the conclusions of section 3.1.1, an initial model can be selected. The model is based on the simplified path loss model, with a RAF (Rack Attenuation Factor ) for attenuation due to objects in the propagation path. For indoor environments the reference distance is set to d0 = 1 m. The path loss and object attenuation will be empirically determined
for the model. The initial model is described by (3.1).
PL = −20 log10 λ 4πd0 + 10γ log10 d d0 + RAF + ψdB (3.1)
3.2
Measurement Equipment
To determine the path loss component and the RAF, the measured values need to be adjusted for gain and losses in the measurement equipment. The gains and losses will be characterized in section 3.2.1 and 3.2.2. The following measurement equipment was used:
• Signal Analyzer, Keysight PXA N9030A • Signal Generator, Keysight MXG N5183B
• Horn Antenna 18.0 - 26.5 GHz, A.H. Systems SAS-587 • PCB SMA connector (transmitter antenna), Rosenberger • Low loss cable, A.H. Systems SAC-26G
• RF cable, HUBER+SUHNER Sucoflex 104 • Precision 2.4 mm to SMA converters
The signal analyzer has a displayed noise floor of -100 dBm with a configured analysis and video bandwidth of 1 kHz. Using a bandwidth of 1 kHz allows the transmitted signal to be separated from adjacent signals, while accurately representing the received power. These settings will be used for all the measurements.
3.2.1
Interface Loss Characterization
To be able to determine the gain offset of the measurements, the equipment losses need to be characterized. This factor specified as LM easure in (3.5).
For the measurements, the SAC-26G cable was used as a reference since it has char-acterization data provided by A.H. Systems Inc. An illustration of the measurement setup can be seen in Figure 3.2 for the conversion connector loss characterization, and in Figure 3.3 for the characterization of the Sucoflex cable. The loss for each setup was determined by subtracting the transmitted power with the received power and the know losses. The measurement results are summarized in Table 3.2.
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT
Figure 3.2: Conversion connector loss characterization
Figure 3.3: Loss characterization of Sucoflex 104 cable
Table 3.2: Interface losses
Component 23.0 GHz 18.0 GHz
SAC-26G 1.90 dB 1.61 dB
Precision 2.4 mm to SMA converter 1.13 dB 0.86 dB
Sucoflex 104 0.81 dB 0.66 dB
3.2.2
SMA Antenna Characterization
To compare the results as predicted by the proposed model to the measured results, the antenna gains need to be characterized. The receiver antenna is a standard gain horn antenna, which has been characterized and calibrated by A.H. Systems Inc [6]. As a transmitter antenna, a PCB SMA connector was used. However, its gain is unknown and need to be characterized. The gain of the transmitter antenna can be characterized with a method called the gain comparison method. This method requires three antennas in order to calculate the gain of the Antenna Under Test (AUT).
The gain comparison method can be utilized by using a spectrum analyzer and a signal generator. First the reference antenna is connected to the spectrum analyzer and the ”Don’t Care”-antenna, marked as X in Figure 3.4, is connected to the signal generator. The ”Don’t Care”-antenna is an antenna which gain is not required for the characterization, but has to have a sufficient dynamic range in order to transmit to the
AUT. The reference antenna is removed and the AUT is connected to the spectrum analyzer as shown in Figure 3.5. [16]
Figure 3.4: Characterization with reference antenna
Figure 3.5: Characterization with antenna under test The received power for the described measurements is given by (3.2).
Pr,dBm = Pt,dBm−PL+ Gt,dBi+ Gr,dBi−LM easure (3.2)
where Pr,dBm and Pt,dBm are the received and transmitted power, PL is the path loss
between the transmitter and receiver, Gt,dBi and Gr,dBi are the transmitter and receiver
gains and LM easure is the interface losses.
Since the measurements will be carried out in the same environment and with the same interface losses, the equation from the two measurement setups can be subtracted to obtain the gain of the AUT, as shown in (3.3).
GAU T = Pr,REF −Pr,AU T + GREF (3.3)
where Pr,REF and Pr,AU T is the received power for the respective measurement. GREF
is the gain of the reference antenna, and GAU T is the gain of the AUT. The results of
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT
3.2.3
Equipment Parameters
Using the methods described in section 3.2.1 and 3.2.2, the component parameters were determined. The interface and antenna parameters for an operational frequency of 23.0 GHz are presented in Table 3.3 and in Table 3.4 for 18.0 GHz.
Table 3.3: Component properties at 23.0 GHz
Component Gain
Horn antenna 15.3 dBi
Signal Analyzer cables and connectors -3.03 dB
SMA Antenna 1.2 dBi
Signal Generator cables and connectors -1.97 dB
Table 3.4: Component properties at 18.0 GHz
Component Gain
Horn antenna 14.0 dBi
Signal Analyzer cables and connectors -2.47 dB
SMA Antenna 2.3 dBi
Signal Generator cables and connectors -1.52 dB
3.3
Path Loss Exponent
The path loss exponent of the initial model in (3.1) determines the rate the power de-creases over distance. Measurements need to be performed in different scenarios to be able to determine the environmental factor of the path loss. The path loss exponent will be determined for the following scenarios:
• Open space
• Populated rack environment
During the measurements, the transmitter antenna had a fixed position in the en-vironment, while the receiving antenna will be placed at different distances from the transmitting antenna. The measurements were performed in small in small enough spa-cial intervals for an accurate estimation of path loss characteristics. An illustration of the measurement method is shown in Figure 3.6.
Figure 3.6: Measurement setup for determining the path loss exponent
When performing measurements for the path loss exponent, additional factors need to be considered in the measurements. The following parameters need to be considered:
• Loss in cable interface • Gain of receiver antenna • Gain of transmitter antenna
• Orientation/polarization of the antennas
The collected data will be adjusted for relative gain of the measurement equipment and then plotted in Matlab.
3.3.1
Path Loss Exponent Measurements
The path loss exponent is assumed to be a scaling factor for a logarithmic function, as seen in (3.1). The measurements were performed with a distance interval of 0.5 m and a range of 1-10 m for the open space measurements and 1-6 m in the rack environment. Five series of measurements were performed for each frequency in order to give a good approximation of the path loss exponent.
The fit used to approximate the path loss exponent is based on least-square method, which fits a linear curve to the data as it minimizes the summed square of the residuals. The least-square linear fit is implemented in Matlab with the Curve Fitting Toolbox software. The equation for determining the sum of the squared residuals can be seen in (3.4). [18] S = n X i=1 yi−ybi 2 (3.4) where S is the square sum of the residuals, yi is i:th data point and byi is the fitted
data point for the i:th value.
In Figures 3.7 and 3.8, the path loss calculated by (3.5) is represented by a black dot. The calculated average path loss is represented by a red star, and the fitted line curve is represented by a blue line. The fitted curve is represented as a linear function in the presentation of the measured values because of the logarithmic scaled axes.
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT Using the method described in [17] gives the ability to extract the path loss exponent for the different scenarios. The measured data for the rack environment is calculated in the same way as the open space measurements. The plotted received power can be seen in Appendix A and the raw data can be found in Appendix B.
Path loss exponent at 23.0 GHz
The plotted data for the open space measurement for 23.0 GHz can be viewed in Figure 3.7 and are summarized in Table 3.5.
Figure 3.7: Measured path loss for an open space environment at 23.0 GHz
Table 3.5: Path loss exponents in different scenarios at 23.0 GHz
Scenario γ
Open space 1.86 Rack environment 1.82
Path loss exponent at 18.0 GHz
The plotted data for the open space measurement for 18.0 GHz can be viewed in Figure 3.8 and are summarized in Table 3.6.
Figure 3.8: Measured path loss for an open space environment at 18.0 GHz
Table 3.6: Path loss exponents in different scenarios at 18.0 GHz
Scenario γ
Open space 1.97 Rack environment 1.83
3.4
Rack Attenuation Factor
The RAF introduced in section 3.1.2, is initially assumed to be linearly proportional to the distance propagated through the rack. The RAF is also assumed to depend on the density of the server racks, where a more dense rack will attenuate the signal more than a lower dense rack. The initial RAF equation is given by (3.6).
RAF=
NR
X
i=i
ρi·di (3.6)
where ρi is the attenuation constant depending on the rack density, and di is the
distance in meters the signal travels through a rack and NRis the number of racks in the
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT To determine the RAF, a series of measurement have to be performed for different scenarios. The measurement scenarios to determine the attenuation constant ρ is illus-trated in Figure 3.9. The receiver antenna will be displaced vertically along the server rack. This combined with the path loss of the environment will determine the final estimate values for the RAF.
The measurement scenarios to determine the attenuation constant ρ is illustrated in Figure 3.9.
Figure 3.9: Measurement setup for determining the rack attenuation factor This measurement will be performed in the different density cases, since the server racks in the test lab are not equally populated. The population density will be a factor since the separation between the rack equipment will determine if the signal will have a line-of-sight component or not when propagating through the rack. Four different server rack densities have been defined to more accurately model de object attenuation, as shown in Table 3.7.
Table 3.7: Definition of rack densities
Scenario Population percentage
Lightly populated rack 10 - 40 % Medium populated rack 40 - 70 % Heavily populated rack 70 - 90 %
Solid object 90 - 100 %
3.4.1
Rack Attenuation Factor Measurements
To obtain an estimate value for the losses, measurements were carried out as illustrated in Figure 3.10. Three series of measurements with different receiver heights were performed for each rack density. The object attenuation was derived by subtracting the the equiva-lent open space path loss for each measurement. The use of different receiver heights is to
enable both line-of-sight and non-line-of-sight components to affect the average received power. This will give a more accurate average attenuation for the different rack densities.
Figure 3.10: Line-of-Sight and Non-Line-of-Sight components in the RAF measurements Rack attenuation factor at 23.0 GHz
The calculated average values for the ρ in dB/m at 23.0 GHz can be seen in Table 3.8. Table 3.8: Rack attenuation factor for different rack densities at 23.0 GHz
Scenario ρ
Light rack density 4.5 Medium rack density 6.0 High rack density 19.0 Rack attenuation factor at 18.0 GHz
The calculated average values for the ρ in dB/m at 18.0 GHz are shown in Table 3.9. Table 3.9: Rack attenuation factor for different rack densities at 18.0 GHz
Scenario ρ
Light rack density 8.4 Medium rack density 11.2 High rack density 17.6
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT
3.4.2
Solid Object Attenuation
In the case of propagation through a solid object, there will not be any line-of-sight component available, as illustrated in Figure 3.10. Therefore, the measurement setup was changed compared to the setup used in section 3.4.1. Instead of moving the receiving antenna vertically it is moved horizontally along the object. This method is used since the are no line-of-sight components available for the receiving antenna, as shown in Figure 3.10. In addition, the measurement were performed with three different separation distances, as seen in Figure 3.11. Different separation distances will indicate how the multipath components affect the received power. These measurements will be used in order to estimate a loss factor for solid objects.
For each separation distance a series of measurements were performed. The average attenuation value was calculated for each series of measurements. The results of the measurements are shown in Table 3.10 and 3.11.
Figure 3.11: Measurement setup solid object attenuation seen from above Rack attenuation factor for solid object at 23.0 GHz
The calculated average values for the ρ in dB/m for 23.0 GHz can be seen in Table 3.10. Table 3.10: Solid object attenuation for different separation distances at 23.0 GHz
Separation ρ 2.9 m 32.0 4.9 m 30.2 6.9 m 22.4
Rack attenuation factor for solid object at 18.0 GHz
The calculated average values for the ρ in dB/m for 18.0 GHz are shown in Table 3.11. Table 3.11: Solid object attenuation for different separation distances at 18.0 GHz
Separation ρ 2.9 m 27.7 4.9 m 26.1 6.9 m 19.7
The decreed object attenuation with increased separation distance can be explained by the multipath properties of indoor propagation. The increased number of available multipath components at the receiver will reduce the apparent attenuation of the solid object.
3.5
Proposed Model
After performing the necessary measurements, the initial model in (3.1) was modified in order to include the measured values in section 3.3 and 3.4. Moreover the model is rewritten to be frequency dependent instead of wavelength dependent as seen in (3.7).
PL= 20 log10(f ) + 10γ log10(d) − 20 log10
c 4π
+ RAF + ψdB (3.7)
where c is the speed of light in m/s and f is the operational frequency in Hz.
The path loss model can be reduced further as shown in (3.8). With the obtained results for the path loss exponent, the model can be compared with the fitted curve from the measurements in section 3.3.1. The comparison can be seen in Figure 3.12 for 23.0 GHz and Figure 3.13 for 18.0 GHz.
June 25, 2015 CHAPTER 3. MODEL DEVELOPMENT
Figure 3.12: Estimated path loss for open space at 23.0 GHz
Figure 3.13: Estimated path loss for open space at 18.0 GHz
This model will initially be used in the simulation model and and it will be further processed and evaluated in chapter 4.
Model Verification and Correction
In this chapter the preliminary model will be evaluated and optimized in order to create a more accurate model. Verification measurements were performed in order to evaluate and correct the path loss model.
4.1
Verification Measurements
In order to evaluate the accuracy of the model, verification measurements were performed in the test lab environment. This section will describe how the measurements were collected and show the deviation of the preliminary model described in (3.8).
4.1.1
Measurement Setup
The leakage simulation tool will implement the path loss model using a direct path modelling method. The verification measurements were performed using the direct path component of the transmitted signal. This method was used in order to more accurately represent the direct path modelling method. The verification measurements were per-formed for both LOS and NLOS locations in the test lab. This allows the final path loss model to be adjusted more accurately. The measurement locations are placed in a 2 meter by 2 meter grid in the test environment, illustrated in Figure 4.1.
June 25, 2015 CHAPTER 4. MODEL VERIFICATION AND CORRECTION
Figure 4.1: Verification measurement locations in the test lab environment
4.1.2
Model Deviation
In order to evaluate the accuracy of the model the deviation of the model needs to be calculated. The deviation of the model is assumed to be normally distributed with a mean of µ = 0 [2].
The normal distribution is symmetrical and has the property of containing a percent-age of its values within a set number of standard deviations from its mean. Within 2σ above and below the mean, 95% of the possible outcomes will be included in the distri-bution [5]. This characteristic will be used to evaluate the possible worst case scenario for the signal leakage levels.
Calculating the standard deviation for the LOS and NLOS components separately will indicate where the model needs to be corrected. The calculated deviation for the model can be seen in Table 4.1 for the LOS components and Table 4.2 for the NLOS components.
Table 4.1: Deviation of the LOS components
Frequency σ µ MAX
23.0 GHz 3.91 dB 1.27 dB 11.29 dB 18.0 GHz 4.81 dB 0.36 dB 12.25 dB
Table 4.2: Deviation of the NLOS components
Frequency σ µ MAX
23.0 GHz 16.11 dB -4.13 dB 43.10 dB 18.0 GHz 14.22 dB -8.70 dB 56.10 dB
4.2
Model Correction
Comparing the deviation between the LOS and NLOS components, it can be seen that the model is less accurate in the NLOS case. This is due to larger residual values in the calculation of the standard deviation. The modelled path loss combined with the object attenuations are higher than the measured path loss in the NLOS cases. The residuals exceeding 15 dB is represented with a blue circle in Figure 4.2.
Figure 4.2: Deviating points
The large deviation of the NLOS case can be partially explained by the result which were obtained in section 3.4.2. It indicated that with increased transmitter-receiver separation the apparent attenuation of the object is reduced. Alternatively this can be explanation by the dominant path aspect of the propagating signal.
June 25, 2015 CHAPTER 4. MODEL VERIFICATION AND CORRECTION
4.2.1
Path Loss Exponent Correction
The simulation of the preliminary path loss model, seen in (3.8), does not fully represent the data fit as seen in Figure 3.12 and 3.13. This offset can be explained by the mean value of the deviation calculations seen in Table 4.1, as the µ 6= 0. This offset is 3.4 dB for the 23.0 GHz model and 1.0 dB for the 18.0 GHz model. This offset is assumed to be frequency dependent, however with only two modelled frequencies this can not be further investigated.
The offset between the preliminary model and the data fit could be decreased by adjusting the value of the path loss exponent. This adjustment will reduce the mean error for shorter distance, however this adjustment will reduce the validity of the model with increased simulation distances. Since the standard deviation for the model is below 5.0 dB with the current implementation, the validity of the LOS components are considered to be acceptable. With these factors considered, the path loss exponent will not be adjusted in the final implementation of the path loss model.
4.2.2
Rack Attenuation Factor Correction
In the simulations of the preliminary implementation of the path loss model it can be seen that the deviation of the NLOS components are considerably larger than the LOS components. This can be seen in the standard deviation and the maximum deviation in Table 4.2.
The initial assumption of the linear property of the RAF may cause this increased deviation. The most troublesome locations are where there are multiple objects obstruct-ing the path between the transmitter and receiver. Observations on multiple floor losses shows that the apparent attenuation of consecutive floors are not linear compared to a single floor loss. In an office environment the attenuation of the second floor is roughly 30% of the first floor [12]. Adjusting for this observation in the calculation of the RAF may reduce the deviation of the NLOS components. The new implementation of the RAF calculations can be seen in (4.1).
RAF= ρ1 d1+ K · NR
X
i=2
ρi di (4.1)
The optimal value for K is derived by calculating the standard deviation and the mean deviation of the NLOS components with different K values. The results of these calculations can be seen in Table 4.3 for 23.0 GHz and Table 4.4 for 18.0 GHz. The initial implementation of the RAF calculations corresponds to K = 1 used in (4.1).
Table 4.3: K for 23.0 GHz model K σ µ MAX 0.1 9.05 dB 4.10 dB 18.47 dB 0.2 8.37 dB 3.18 dB 18.47 dB 0.3 8.04 dB 2.27 dB 18.47 dB 0.4 8.09 dB 1.36 dB 18.47 dB 0.5 8.51 dB 0.44 dB 18.47 dB 0.6 9.27 dB -0.47 dB 23.86 dB 0.7 10.27 dB -1.39 dB 27.17 dB 0.8 11.47 dB -2.30 dB 32.48 dB 0.9 12.80 dB -3.21 dB 37.79 dB 1.0 14.22 dB -4.13 dB 43.10 dB
Table 4.4: K for 18.0 GHz model
K σ µ MAX 0.1 8.38 dB 0.70 dB 18.60 dB 0.2 7.99 dB -0.35 dB 18.60 dB 0.3 8.02 dB -1.39 dB 18.60 dB 0.4 8.46 dB -2.44 dB 23.95 dB 0.5 9.26 dB -3.50 dB 27.64 dB 0.6 10.33 dB -4.52 dB 33.33 dB 0.7 11.60 dB -5.57 dB 39.02 dB 0.8 13.01 dB -6.61 dB 44.71 dB 0.9 14.53 dB -7.65 dB 50.40 dB 1.0 16.11 dB -8.69 dB 56.10 dB
4.2.3
Deviation after Model Correction
Changing the definition of the RAF improves the overall accuracy of the model. The results of introducing K can be seen on the µ and σ values in Table 4.3 and 4.4.
The final path model will be implemented using the originally derived path loss expo-nents and attenuation constants for the RAF calculations. The implementation of the RAF has been reworked from (3.6) to (4.1). Selecting a K value of 0.3 for the RAF calculations gives the best compromise between the standard and mean deviations for both frequencies, as shown in Table 4.3 and 4.4. The deviations for the final path loss model can be seen in Table 4.5 and 4.6.
Table 4.5: Corrected deviation of the LOS components
Frequency σ µ MAX
23.0 GHz 3.91 dB 1.27 dB 11.29 dB 18.0 GHz 4.81 dB 0.36 dB 12.25 dB
Table 4.6: Corrected deviation of the NLOS components
Frequency σ µ MAX
23.0 GHz 8.04 dB 2.27 dB 18.47 dB 18.0 GHz 8.02 dB -1.39 dB 18.60 dB
Chapter 5
Results
This chapter will present the results of the final model and its implementation. The aspects of different leakage scenarios will be investigated and simulated using the final path loss model.
5.1
Final Model
In the leakage simulation tool, the path loss model will be implemented in order to calculate the potential received power at any point in a 2D space. Using the leakage power and the path loss model will give the received power at any point in the 2D space. The simulation model can be seen in (5.1).
Pr(d) = Pleakage−20 log10(f ) − 10γ log10(d) + 147.5 − ρ1 d1+ 0.3 NR X i=2 ρi di ! (5.1) where Pr is the received power and Pleakage is the leakage power. The remaining
components are from the preliminary path loss model in (3.8).
In the worst case scenario the standard deviation of the NLOS case will be used to evaluate the signal strength. The standard deviation of the NLOS case is roughly 8.0 dB for both the 23.0 GHz and the 18.0 GHz model. Currently there are no known com-mercially used indoor path loss model at the microwave frequencies used in this report. Hence the implemented path loss model are compared with results from models designed for significantly lower frequencies. The resulting standard deviation for the NLOS com-ponents is roughly 8.0 dB for both 18.0 and 23.0 GHz, this result is be compared with standard deviations that ranges between 7 and 17 dB retrieved in [12][24][25]. Since the obtained standard deviation is in the lower range in that comparison, the final prop-agation model is assumed to be valid and can be applied to evaluate possible leakage scenarios.
5.2
Leakage Simulations
The simulation model in (5.1) will be used to evaluate different leakage scenarios in the MINI-LINK setup used in the test lab. The worst case leakage power will be used to evaluate the possible scenarios.
5.2.1
Maximum Allowed Radiated Power
Evaluating the worst case leakage scenarios, the maximum allowed radiated power is required. The maximum exposure limit set by the responsible authority (Str˚ als¨aker-hetsmyndigheten) in the frequency range between 2 GHz and 300 GHz is 10 W/m2
, or a field strength of 155.7 dBµV/m [19]. According to the MIL-STD-461F the radiated susceptibility of ground equipment in the frequency range 18-40 GHz is set to be able to handle 50 V/m, or a 154 dBµV/m [15]. This level is below the maximum level set by Str˚als¨akerhetsmyndigheten and will be used to calculate the maximum allowed power. Using the characterization data of the SAS-587 reference antenna and a 50 Ω system impedance, the maximum power can be calculated as seen i (5.2) [20].
PM AX = VLim−107 − AF + LM easure−Gr (5.2)
where PM AX is the maximum allowed power in dBm, VLim is the maximum allowed
field strength in dBµV/m, AF is the antenna factor of the antenna in dB/m, LM easure is
the interface loss in dB and Gr is the gain of the antenna in dBi.
The maximum allowed radiated power is -8.5 dBm for 23.0 GHz and -6.7 dBm for 18.0 GHz.
5.2.2
Leakage from MINI-LINK Setup
In order to understand the potential leakage scenarios, the equipment connecting the MINI-LINK radio units will be evaluated. The following equipment is used for transmit-ting the signals between the MINI-LINK radio units in the test lab:
• Waveguide
• Waveguide to coaxial converter • Coaxial cable
• Attenuators
The connection between the majority of the MINI-LINK radio units uses a Sucoflex 104 cable as listed in section 3.2. The shield used in this cable is a silver-plated copper tape with an additional silver-plated copper braid. This shield setup has a leakage attenuation of about 95 dB [21]. Some of the MINI-LINK radio units are connected with a waveguide instead of the Sucoflex cable. Typically, with gaskets and proper assembly, the waveguide has a leakage attenuation of about 90 dB [22]. This leaves the connector of the cable as the major leakage contributor [23].
June 25, 2015 CHAPTER 5. RESULTS Evaluating the leakage from the SMA connector, a 50 Ω termination was connected to the end of the Sucoflex cable. The power was measured with a transmitted power of 30 dBm at a distance of 2λ from the connector. This power was used since the maximum transmitted power of the MINI-LINK units is in the 30 dBm range. The measured power of this test can be seen in Table 5.1. The reference torque is 90 N cm with a Rosenberger 32W100-016 torque wrench.
Table 5.1: Measured power of the leakage scenarios
Scenario 23.0 GHz 18.0 GHz Torque wrench -53.4 dBm -50.3 dBm Finger tight -49.3 dBm -54.9 dBm 1 /4 revolution loosened -41.6 dBm -41.1 dBm 1/2 revolution loosened -29.5 dBm -34.1 dBm
With the measured power adjusted for the gain and losses in the measurement equip-ment, the leakage power is found to be roughly 70 dB below the transmitted power for the scenario with the connector1/
2revolution loosened. This leakage level is 20 dB higher
than the leakage from the shield and waveguide. This leakage scenario will be used for the leakage simulations in the test environment.
5.2.3
Model Simulations
Simulating the leakage in the test environment requires the simulation tool to accurately represent the actual power levels. The implemented leakage simulation tool is regulated by the requirements in Table 1.3.
In Figure 5.1 the environment layout used in section 4.1.1 is recreated in the simulation tool. A single leakage source of 0 dBm at 23.0 GHz is configured to replicate this scenario. This configuration was used in order to replicate the verification measurements.
Figure 5.1: Single source simulation
In Figure 5.2 the environment is the same as in Figure 5.1. Three leakage sources is configured with a frequency of 18.0 GHz and leakage power of -40 dBm. This leakage power of -40 dBm corresponds to a radio unit transmitting with a power of 30 dBm with the SMA connector loosened half a revolution, as seen in section 5.2.2.
Figure 5.2: Multiple source simulation
Chapter 6
Discussion and Conclusion
The primary purpose of the project was to investigate and evaluate potential leakage scenarios in the test lab. Two primary aspects of the leakage needed to be evaluated; the security and the health aspects. These parts are of critical importance to the test environment engineers to ensure the operational capability of the test lab.
6.1
General Discussion
During the process of developing the path loss model there have been no available studies or literature that cover the specific frequencies and environment used in this thesis. Hence, while deriving the path loss model comparisons were made with models designed for other frequencies and environments.
From section 3.3 it can be seen that the derived path loss exponent results in a reduced loss over a certain distance compared to free space loss. This indicates that there are lower losses at higher frequencies, compared to typical telecommunication frequencies, which is something that are not always observed or easily explained. Typically signals at higher frequencies are more affected by object attenuations, but less obstruction in the transmission path may lead to lower losses [12]. Comparing the derived path loss expo-nent with results from other path loss models designed for high microwave frequencies gives an indication of the same characteristics [12]. With this aspect the derived path loss exponents from section 3.3 is considered to be applicable in the desired environment. Modelling the attenuation of objects in the transmission path proved to result in a much higher deviation compared to the measurements conducted in the test lab. Initially the attenuations were assumed to be linear to the distance travelled through the object. This created large model deviations compared to the performed verification measure-ments. Adjustments of the particular attenuations and the path loss exponent proved to have a much higher impact on the characteristics of the model with only small im-provements to the overall accuracy. Scaling the object attenuation after the first object improved the standard deviation of the model. The ρ value for the specific rack densities is not changed in the current implementation of the RAF modelling method.
Some problems still remains, such as the mean of the deviation are not fully centred around zero. Since the amount of data used for these calculations are limited, performing
additional verifications measurements could solve this problem. Although this is consid-ered to be less relevant due to the relative small deviation and it will not be needed for the evaluation of the leakage levels. The final standard deviation of the model is in an acceptable range compared to other models [12][24][25].
In the final implementation of the leakage simulation tool, multiple leakage sources are implemented. These leakages are simulated separately and combined with fully con-structive interference in order to evaluate the largest potential signal power at any given point of the simulated environment.
6.2
Requirement Evaluation
At the start of the project, requirements were set in order to evaluate the results. This section will evaluate the requirements.
6.2.1
Pre-study Requirement Evaluation
In Table 6.1 the requirements of the pre-study are evaluated. Table 6.1: Analysis of Pre-study requirements No. Comment
1 The major leakage contributors were investigated in order to simulate the leak-age.
2 The MIL-STD-461F standard and guidelines from Str˚als¨akerhetsmyndigheten were used to evaluate potential risks of the leakage.
3 Indoor propagation models and modelling methods were investigated. A direct path modelling method was used to implement the path loss model.
4 Appropriate measurement methods were used to determine the aspects needed for the model.
5 The Matlab programming was investigated when needed and applied in the leakage simulation tool.
June 25, 2015 CHAPTER 6. DISCUSSION AND CONCLUSION
6.2.2
Model Requirement Evaluation
In Table 6.2 the requirements of the model are evaluated.
Table 6.2: Analysis of Model requirements No. Comment
6 The model was implemented in the leakage simulation tool using Matlab. 7 The leakage simulation tool has adjustable parameters for different
environ-ments.
8 The leakage simulation tool has adjustable frequency setting.
9 A user mappable environment has been implemented in the leakage simulation tool.
10 The leakage simulation tool is able to assign three independent leakage sources at one given frequency.
11 An analysis of the verification measurements has been performed to give the total accuracy of the given model. This is not implemented in the leakage simulation tool.
12 The leakage simulation tool does not alert the user when there is a potential security risk. This feature has not been implemented.
13 The leakage simulation tool will alert the user when the signal strength is above a set level.
14 The leakage simulation tool is implemented with a graphical user interface with all appropriate assignable parameters.
6.2.3
Measurement Requirement Evaluation
In Table 6.3 the requirements of the measurements are evaluated. Table 6.3: Analysis of Measurement requirements No. Comment
15 The equipment used for the measurements are of high quality and precision. 16 The orientation and placement of the measurement equipment have been
con-sidered and is defined in the individual experiments.
17 The elevation of the measurement equipment have been considered and the method is defined in the individual experiments.
6.2.4
Project Requirement Evaluation
In Table 6.4 the requirements of the project are evaluated.
Table 6.4: Analysis of Project requirements No. Comment
19 Written report is in the form of the current document.
20 Weekly meetings have not been performed, but meeting has been performed per need basis during the project.
21 Monthly written report have not been performed, but reports have been per-formed per need basis during the project.
22 Presentation of the results and the leakage simulation have been scheduled at both Ericsson and LiU.
June 25, 2015 CHAPTER 6. DISCUSSION AND CONCLUSION
6.3
Conclusion
The purpose of the project was to investigate the leakage from the MINI-LINK setup in the Ericsson test lab. This was accomplished with an empirical path loss model for the lab environment. This model is adjusted for frequencies and an environment that has not earlier been documented, as indicated by the initial pre-study. The path loss model is adapted to the environment with an empirical derived characteristics of the path loss and object attenuation.
Path loss modelling is an appropriate method to evaluate different leakage scenarios, if the model is able to accurately represent the environment. The deviation of the model used is important when the worst case scenarios are being evaluated. In the worst case scenario the simulated power will have to be increased by 2σ, or 16 dB for the path loss models constructed by this project. This additional contribution will cover over 95% of the possible leakage levels that the model will provide and give an accurate representation of the worst case scenario.
It was found that the most likely component to cause the majority of the signal leakage was the SMA connectors of the MINI-LINK radio units. In the worst case scenario it was found that the radiated power was below the limit by both Str˚als¨akerhetsmyndigheten and the MIL-STD-461F standard. Additionally a signal at this frequency attenuates significantly over distance and will not pose a potential security risk. However the sim-ulation tool will alert the user if the signal level is above a set limit anywhere in the simulation environment.
Signal transfer at microwave frequencies are more sensitive. In order to prevent sig-nal leakage from the systems using microwave frequency sigsig-nals, proper assembly with appropriate tools are essential. It was shown that not connecting the equipment appropri-ately may cause increased signal leakage and additional RF problems such as impedance mismatch and signal reflection.