Department of Radio Science and Engineering Measurement-Based Millimeter-Wave Radio Channel Simulations and Modeling Jan Järveläinen DOCTORAL DISSERTATIONS
This thesis focuses on mm-wave channel modeling for future 5G wireless communication systems. The main contributions of the work include simulations tools and insights acquired through channel measurements. The work emphasize the need for more detailed descriptions of the model frameworks and environment descriptions at mm-waves compared to lower frequencies.
Aalto-D
D 16
4/2
016
9HSTFMG*agjhbf+
ISBN 978-952-60-6971-5 (printed) ISBN 978-952-60-6972-2 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Electrical Engineering Department of Radio Science and Engineering www.aalto.fi
BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS
Jan Järveläinen M
easurement-Based M
illimeter-W
ave Radio C
hannel Simulations and M
odeling A
alto U
nive
rsity
2016
Department of Radio Science and Engineering
Measurement-Based Millimeter-Wave Radio Channel Simulations and Modeling
Jan Järveläinen
DOCTORAL DISSERTATIONS
Aalto University publication series DOCTORAL DISSERTATIONS 164/2016
Measurement-Based Millimeter-Wave Radio Channel Simulations and Modeling
Jan Järveläinen
A doctoral dissertation completed for the degree of Doctor of Science (Technology) to be defended, with the permission of the Aalto University School of Electrical Engineering, at a public examination held at the lecture hall S1 of the school on 30 September 2016 at 12.
Aalto University School of Electrical Engineering Department of Radio Science and Engineering
Supervising professors Prof. Pertti Vainikainen Prof. Antti Räisänen Assist. Prof. Katsuyuki Haneda Thesis advisor Dr. Aki Karttunen Preliminary examiners Prof. Henry Bertoni, New York University, USA Prof. Wout Joseph, Ghent University, Belgium Opponents Prof. Henry Bertoni, New York University, USA Prof. Preben Mogensen, Aalborg University, Denmark
Aalto University publication series DOCTORAL DISSERTATIONS 164/2016 © Jan Järveläinen ISBN 978-952-60-6971-5 (printed) ISBN 978-952-60-6972-2 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) http://urn.fi/URN:ISBN:978-952-60-6972-2 Unigrafia Oy Helsinki 2016 Finland
Abstract Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi
Author Jan Järveläinen Name of the doctoral dissertation Measurement-Based Millimeter-Wave Radio Channel Simulations and Modeling Publisher School of Electrical Engineering Unit Department of Radio Science and Engineering Series Aalto University publication series DOCTORAL DISSERTATIONS 164/2016 Field of research Radio Engineering Manuscript submitted 21 April 2016 Date of the defence 30 September 2016 Permission to publish granted (date) 10 June 2016 Language English
Monograph Article dissertation Essay dissertation
Abstract The spectrum shortage at microwaves has necessitated the use of millimeter-wave (mm-wave) frequencies in future fifth generation (5G) wireless communication systems. In order to evaluate the performance of 5G networks at mm-waves, propagation channels in various scenarios must be properly characterized and modeled. This thesis aims at providing important insights, methods and tools for mm-wave channel modeling.
Deterministic field prediction tools, previously used mainly for coverage analysis, are increasingly used also for stochastic channel model parametrization or for evaluating system performance. The prediction accuracy of such tools, e.g., ray tracing, may be compromised due to missing details in the environment databases. In this work, a novel field prediction tool relying on accurate environmental information in the form of point clouds is developed. The prediction method parameters are tuned by measurements, and the performance is validated in several indoor environments by comparing predicted and measured channels in terms of power, delay and angular metrics. The results demonstrate excellent prediction accuracy in both line-of-sight (LOS) and non-line-of-sight links.
As the upcoming 5G mm-wave deployment will be made in new scenarios, channel sounding is essential to acquire knowledge about the propagation characteristics and the applicability of existing channel model frameworks. This work presents insights obtained from channel sounding results conducted in environments such as offices and a shopping mall in the 60- and 70-GHz bands. Unlike existing channel models, clustering was not found apparent in the large indoor environments, allowing a simpler spatio-temporal channel model structure to be developed. A parametrization of the WINNER II model at 60 GHz and a study on the depolarization at mm-waves is also presented. The results show that polarization is better preserved at mm-waves than at lower frequencies. Moreover, a novel method to evaluate LOS probability based on point clouds is demonstrated.
The last part of the thesis deals with the use of stochastic and site-specific channel models in evaluating the performance of mm-wave wireless systems. The point cloud-based simulation tool is used to study the mutual orthogonality of mm-wave links equipped with large antenna arrays. The result shows that compared to microwave frequencies, the number of active users should be smaller at mm-waves to guarantee efficient spatial multiplexing.
Keywords Millimeter-wave, channel modeling, prediction, point cloud ISBN (printed) 978-952-60-6971-5 ISBN (pdf) 978-952-60-6972-2 ISSN-L 1799-4934 ISSN (printed) 1799-4934 ISSN (pdf) 1799-4942 Location of publisher Helsinki Location of printing Helsinki Year 2016 Pages 167 urn http://urn.fi/URN:ISBN:978-952-60-6972-2
Tiivistelmä Aalto-yliopisto, PL 11000, 00076 Aalto www.aalto.fi
Tekijä Jan Järveläinen Väitöskirjan nimi Mittauksiin perustuva millimetriaaltoalueen radiokanavasimulointi ja -mallinnus Julkaisija Sähkötekniikan korkeakoulu Yksikkö Radiotieteen ja -tekniikan laitos Sarja Aalto University publication series DOCTORAL DISSERTATIONS 164/2016 Tutkimusala Radiotekniikka Käsikirjoituksen pvm 21.04.2016 Väitöspäivä 30.09.2016 Julkaisuluvan myöntämispäivä 10.06.2016 Kieli Englanti
Monografia Artikkeliväitöskirja Esseeväitöskirja
Tiivistelmä Mikroaaltotaajuuksien tiheä käyttö tietoliikenteessä on johtanut radiospektrin puutteeseen, jonka seurauksena tulevassa viidennen sukupolven (5G) langattomissa järjestelmissä myös millimetriaaltoaluetta on hyödynnettävä. Jotta 5G-verkkojen suorituskykyä olisi mahdollista arvioida millimetriaalloilla, etenemiskanavaa on mallinnettava eri ympäristöissä. Tämän väitöskirjan tavoitteena on tarjota tärkeitä havaintoja, menetelmiä ja työkaluja millimetriaaltoalueen radiokanavamallinnusta varten.
Deterministiset radioaaltojen etenemisen ennustusmenetelmät, joita aikaisemmin on käytetty verkkojen kuuluvuuden tutkimiseen, ovat yhä enemmän varteenotettavia työkaluja myös stokastisten kanavamallien parametrisoinnissa ja radiojärjestelmien suorituskykyarvioinnissa. Ennustusmenetelmien, kuten säteenseurannan, ennustustarkkuus saattaa kärsiä ympäristön tietokantojen huonosta tarkkuudesta johtuen. Tässä työssä on kehitetty uusi ennustusmenetelmä, jossa käytetään erittäin tarkkoja pistepilviä kuvaamaan ympäristöä. Menetelmän parametrit kalibroidaan mittausten avulla ja toimintakykyä arvioidaan useissa sisätilaympäristöissä vertaamalla simuloituja ja mitattuja kanavia eri tehon, viiveen ja kulma-alueen mittareilla. Tulokset osoittavat ennustustarkkuuden erinomaiseksi sekä näköyhteyden ollessa vapaa että sen puuttuessa.
Koska millimetriaaltoalueen 5G-verkkoja sijoitetaan uusiin ympäristöihin, kanavamittaukset ovat oleellisia radiokanavien ominaisuuksien ja olemassa olevien kanavamallien käyttökelpoisuuden tutkimiseksi. Tässä työssä esitetään havaintoja kanavamittauksista, joita on suoritettu esim. toimistotiloissa ja kauppakeskuksessa 60 ja 70 GHz:n taajuusalueilla. Vastoin yleisiä kanavamallinnusperiaatteita, mittaukset osoittavat että monitiekomponenttien ryhmittyminen ei ole voimakasta tutkituissa ympäristöissä, mikä sallii yksinkertaisemman rakenteen tilassa ja ajassa määritellylle kanavamallille. Lisäksi WINNER II-mallille lasketaan parametrit 60 GHz:llä ja polarisaation kääntymistä tutkitaan sisätilaympäristöissä. Tulokset osoittavat polarisaation säilyvän paremmin millimetriaalloilla alempiin taajuuksiin verrattuna. Työssä esitetään myös uusi menetelmä näköyhteysreitin todennäköisyyden laskemiseksi pistepilviä hyödyntäen.
Työn viimeinen osio käsittelee determinististen ja stokastisten kanavamallien sovelluskohteita langattomien järjestelmien suorituskykyarvioinnissa. Pistepilveen pohjautuvaa ennustusmenetelmää käytetään tutkimaan isoilla antenniryhmillä varustettujen millimetriaaltolinkkien keskinäiskorrelaatiota. Tuloksissa todetaan, että aktiivisten käyttäjien määrän on oltava millimetriaaltotaajuuksilla pienempi mikroaaltotaajuuksiin verrattuna, jotta tilallista limitystä (eng. spatial multiplexing) voidaan hyödyntää mahdollisimman hyvin. Avainsanat Millimeteriaallot, radiokanavamallinnus, ennustaminen, pistepilvi ISBN (painettu) 978-952-60-6971-5 ISBN (pdf) 978-952-60-6972-2 ISSN-L 1799-4934 ISSN (painettu) 1799-4934 ISSN (pdf) 1799-4942 Julkaisupaikka Helsinki Painopaikka Helsinki Vuosi 2016 Sivumäärä 167 urn http://urn.fi/URN:ISBN:978-952-60-6972-2
Sammandrag Aalto-universitetet, PB 11000, 00076 Aalto www.aalto.fi
Författare Jan Järveläinen Doktorsavhandlingens titel Mätningsbaserad simulation och modellering av utbredningskanaler på millimetervågsområdet Utgivare Högskolan för elektroteknik Enhet Institutionen för radiovetenskap och radioteknik Seriens namn Aalto University publication series DOCTORAL DISSERTATIONS 164/2016 Forskningsområde Radioteknik Inlämningsdatum för manuskript 21.04.2016 Datum för disputation 30.09.2016 Beviljande av publiceringstillstånd (datum) 10.06.2016 Språk Engelska
Monografi Sammanläggningsavhandling
Sammandrag Den begränsade tillgängligheten av radiospektrum på mikrovågsområdet har lett till att framtida femte generationens (5G) nätverk också måste använda sig av millimetervågsområdet (mm-vågsområdet). För att utvärdera prestationsförmågan av nätverk på mm-vågsområdet måste utbredningskanalen karakteriseras och modelleras i olika omgivningar. Denna doktorsavhandling strävar efter att erbjuda viktiga iakttagelser, metoder och redskap för kanalmodellering på mm-vågsområdet.
Traditionellt har redskap för att förutspå utbredningskanaler använts främst för att analysera hörbarheten i mobilnätverk, men nuförtiden behövs dessa redskap också för att parametrisera stokastiska kanalmodeller samt för att utvärdera prestandan i trådlösa system. Estimeringsprecisionen för dylika redskap, som tex. strålföljning (eng. ray tracing), kan försämras av att omgivningsmodellen inte är tillräcklig detaljerad. I detta arbete presenteras en ny metod för att förutspå utbredningskanaler som baserar sig på noggranna omgivningsmodeller i form av punktmoln. Modellparametrarna kalibreras med hjälp av radiokanalmätningar och funktionaliteten bekräftas i flera interiörer genom att jämföra simulerade och mätta radiokanaler beträffande effekt, fördröjning och vinklar. Resultaten tyder på en hög estimeringsprecision både med fri siktlinje samt med siktlinjen blockerad.
Eftersom 5G-nätverk på mm-vågsområdet kommer att placeras i flera nya omgivningar är det nödvändigt att utföra radiokanalmätningar för att skaffa sig information om utbredningskanalens egenskaper och användbarheten av existerande kanalmodeller. I detta arbete presenteras insikter som erhållits via radiokanalmätningar i tex. kontorsutrymmen och ett köpcentrum i frekvensbanden kring 60 och 70 GHz. I motsats till populära kanalmodeller som grupperar radiovågskomponenter i kluster, påvisar mätningarna ingen klustring vilket möjliggör en enklare kanalmodell i rum och tid. Mätningarna används också till att parametrisera WINNER II kanalmodellen vid 60 GHz samt till att studera hur polariseringen förändras under utbredningen. Resultaten visar att polariseringen bevaras bättre på mm-vågsområdet jämfört med lägre frekvenser. Vidare presenteras en ny metod för att utvärdera sannolikheten för fri siktlinje med hjälp av punktmoln.
Avhandlingen sista del ger en översikt av olika användningsändamål för både deterministiska och stokastiska kanalmodeller. Estimeringsmetoden som baserar sig på punkmoln används för att studera korrelationen mellan användare på mm-vågsområdet som är utrustade med massiva gruppantenner. Resultaten tyder på att antalet aktiva användare bör vara lägre än på mikrovågsområdet för att garantera effektiv rumsmultiplexering.
Nyckelord Millimetervågor, modellering av utbredningskanaler, estimering ISBN (tryckt) 978-952-60-6971-5 ISBN (pdf) 978-952-60-6972-2 ISSN-L 1799-4934 ISSN (tryckt) 1799-4934 ISSN (pdf) 1799-4942 Utgivningsort Helsingfors Tryckort Helsingfors År 2016 Sidantal 167 urn http://urn.fi/URN:ISBN:978-952-60-6972-2
Dei Gratia
9
10
Preface
The work for this thesis begun five years ago, in a time when I felt I didn’t
know anything. The hopes for learning something useful for the future
have truly been fulfilled, thanks to interesting topics, patient colleagues
and the passing of time. I want to express my gratitude to the late Prof.
Pertti Vainikainen for guiding me into the wonders of wireless communi-
cations. I will always be amazed by Pertti’s extraordinary combination of
expertise and kindness.
My journey from a struggling doctoral student to a confident researcher
would not have been possible without the help of my supervisor, Prof.
Katsuyuki Haneda. Thank you for teaching me about logical reasoning
and for expanding my horizons. I am also very grateful to Dr. Veli-Matti
Kolmonen, Dr. Mikko Kyrö and Dr. Aki Karttunen for instructing and
encouraging me during my work. I appreciate your lessons in leadership
and practical research work. Many thanks to our research group and the
whole RAD department for creating a warm and optimistic atmosphere
that enables top-level research. Especially, I want to thank all the profes-
sors for always having time with even the most trivial questions.
Besides the colleagues in our department, I have had the opportunity
to work with many great experts during research visits and projects. I
would like to thank Prof. Jun-ichi Takada, Prof. Vittorio Degli-Esposti,
Dr. Kenichi Takizawa, Mr. Pekka Kyösti, Mr. Tommi Jämsä and Mr. Jyri
Putkonen for valuable discussions and insights.
The preliminary examiners of this thesis, Prof. Henry Bertoni and Prof.
Wout Joseph, deserve my deepest gratitude for their contribution in mak-
ing the thesis more comprehensible. In addition I want to acknowledge
the effort of Dr. Joni Vehmas for the proofreading.
Furthermore, I am thankful for the funding received by Aalto ELEC
Doctoral School, the Walter Ahlström Foundation, the Scandinavia-Japan
11
Preface
Sasakawa Foundation, Svenska tekniska vetenskapsakademien i Finland,
the COST Action IC1004 and the HPY Research Foundation.
Last, I want to give my most sincere thanks to all my friends for their
support and “unbelievably deep” interest in my research topic. Pappa,
mamma, Hanna och Henrik, jag är oerhört tacksam för ert stöd och er
omsorg om mig. Freja, tack för din kärlek och din stora vishet. Filip,
tack för att du vill ha mig med i dina lekar. Till sist vill jag tacka Gud
för Hans stora nåd och förlåtelse, och för alla människor som möjliggjort
denna avhandling.
Helsinki, August 22, 2016,
Jan Järveläinen
12
Contents
Preface 11
Contents 13
List of Publications 16
Author’s Contribution 18
List of Abbreviations 20
List of Symbols 22
1. Introduction 24
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.2 Objectives and contribution of the thesis . . . . . . . . . . . . 25
2. The Wireless Propagation Channel 28
2.1 Propagation mechanisms at mm-waves . . . . . . . . . . . . 28
2.1.1 Reflection and transmission . . . . . . . . . . . . . . . 28
2.1.2 Scattering . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.1.3 Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2 Characterization of wireless channels . . . . . . . . . . . . . 33
2.2.1 The double-directional channel . . . . . . . . . . . . . 33
2.2.2 Channel characterization metrics . . . . . . . . . . . . 34
2.2.3 Depolarization . . . . . . . . . . . . . . . . . . . . . . . 35
3. Site-Specific Channel Modeling 36
3.1 Ray tracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Point cloud-based propagation prediction . . . . . . . . . . . 39
3.2.1 Point cloud-based prediction of diffusive propagation
channels in small rooms . . . . . . . . . . . . . . . . . 39
13
Contents
3.2.2 Shadowing detection . . . . . . . . . . . . . . . . . . . 41
3.2.3 Prediction of overall channel using point cloud data . 42
3.3 Contribution of the thesis . . . . . . . . . . . . . . . . . . . . 43
3.3.1 Validation of propagation prediction based on diffuse
scattering . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.2 Influence of point cloud density . . . . . . . . . . . . . 44
3.3.3 Shadowing detection . . . . . . . . . . . . . . . . . . . 45
3.3.4 Validation of overall channel prediction tool . . . . . . 46
4. Stochastic Channel Modeling 48
4.1 WINNER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 COST 2100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 IEEE 802.11ad . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.4 METIS channel model . . . . . . . . . . . . . . . . . . . . . . 51
4.5 Other mm-wave channel modeling works . . . . . . . . . . . 52
4.6 Contribution of the thesis . . . . . . . . . . . . . . . . . . . . 52
4.6.1 Spatio-temporal channel model for large indoor envi-
ronments . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.6.2 Parametrization of WINNER channel model in shop-
ping mall at 60 GHz . . . . . . . . . . . . . . . . . . . 53
4.6.3 Characterization of cross-polarization at 70 GHz . . . 53
4.6.4 Line-of-sight probability at millimeter-wave frequen-
cies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5. Millimeter-Wave Channel Sounding 57
5.1 Narrowband measurements . . . . . . . . . . . . . . . . . . . 58
5.2 Wideband channel measurements . . . . . . . . . . . . . . . 58
5.2.1 Measurements in the delay domain . . . . . . . . . . . 58
5.2.2 Measurements in the frequency domain . . . . . . . . 59
5.3 Directional channel measurements . . . . . . . . . . . . . . . 59
5.3.1 Rotation of directional antenna . . . . . . . . . . . . . 59
5.3.2 Antenna array measurements . . . . . . . . . . . . . . 60
5.4 Polarization measurements . . . . . . . . . . . . . . . . . . . 60
5.5 Millimeter-wave channel sounding campaigns for 5G sce-
narios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.6 Contribution of the thesis . . . . . . . . . . . . . . . . . . . . 61
6. Applications for Channel Models 63
6.1 The use of stochastic channel models . . . . . . . . . . . . . . 63
14
Contents
6.1.1 Coding and modulation . . . . . . . . . . . . . . . . . . 63
6.1.2 Mobile terminal antenna design . . . . . . . . . . . . 63
6.1.3 Network design . . . . . . . . . . . . . . . . . . . . . . 64
6.1.4 Capacity and throughput evaluation . . . . . . . . . . 64
6.2 The use of site-specific channel models . . . . . . . . . . . . . 64
6.2.1 Coverage analysis . . . . . . . . . . . . . . . . . . . . . 64
6.2.2 Base station antenna design . . . . . . . . . . . . . . . 65
6.3 Contribution of thesis . . . . . . . . . . . . . . . . . . . . . . . 65
7. Summary of Publications 66
8. Conclusions 69
References 71
Errata 88
Publications 89
15
List of Publications
This thesis consists of an overview and of the following publications which
are referred to in the text by their Roman numerals.
I J. Järveläinen and K. Haneda, “Sixty Gigahertz Indoor Radio Wave
Propagation Prediction Method Based on Full Scattering Model,”
Radio Science, vol. 49, no. 4, pp. 293-305, April 2014.
II J. Järveläinen, M. Kurkela, and K. Haneda, “Impacts of Room Struc-
ture Models on the Accuracy of 60 GHz Indoor Radio Propagation
Prediction,” IEEE Antennas and Wireless Propagation Letters, vol.
14, pp. 1137-1140, January 2015.
III J. Järveläinen, K. Haneda, and A. Karttunen, “Indoor Propagation
Channel Simulations at 60 GHz Using Point Cloud Data,” IEEE
Transactions on Antennas and Propagation, Accepted for publica-
tion in a future issue, 2016.
IV A. Karttunen, J. Järveläinen, A. Khatun, and K. Haneda, “Radio
Propagation Measurements and WINNER II Parameterization for a
Shopping Mall at 60 GHz,” In Proceedings of the IEEE 81st Vehicular
Technology Conference (VTC Spring), Glasgow, UK, pp. 1-5, May
2015.
V K. Haneda, J. Järveläinen, A. Karttunen, M. Kyrö, and J. Putko-
nen, “A Statistical Spatio-Temporal Radio Channel Model for Large
Indoor Environments at 60 and 70 GHz,” IEEE Transactions on An-
tennas and Propagation, vol. 63, no. 6, pp. 2694-2704, June 2015.
VI A. Karttunen, K. Haneda, J. Järveläinen, and J. Putkonen, “Polari-
sation Characteristics of Propagation Paths in Indoor 70 GHz Chan-
16
List of Publications
nels,” In Proceedings of the IEEE 9th European Conference on An-
tennas and Propagation (EuCAP), Lisbon, Portugal, pp. 1-4, May
2015.
VII J. Järveläinen, S. L. H. Nguyen, K. Haneda, R. Naderpour, and U. T.
Virk, “Evaluation of Millimeter-Wave Line-of-Sight Probability With
Point Cloud Data,” IEEE Wireless Communications Letters, vol. 5,
no. 3, June 2016.
VIII S. L. H. Nguyen, K. Haneda, J. Järveläinen, A. Karttunen, and J.
Putkonen, “On the Mutual Orthogonality of Millimeter-Wave Mas-
sive MIMO Channels,” In Proceedings of the IEEE 81st Vehicular
Technology Conference (VTC Spring), Glasgow, UK, pp. 1-5, May
2015.
17
Author’s Contribution
Publication I: “Sixty Gigahertz Indoor Radio Wave PropagationPrediction Method Based on Full Scattering Model”
The author had the main responsibility for implementing and validating
the prediction tool, and preparing the manuscript. Prof. Haneda per-
formed the measurements and supervised the work.
Publication II: “Impacts of Room Structure Models on the Accuracyof 60 GHz Indoor Radio Propagation Prediction”
The author had a leading role in developing the idea and content for the
paper. Mr. Kurkela assisted in point cloud handling and in the paper
writing. Prof. Haneda supervised the work.
Publication III: “Indoor Propagation Channel Simulations at 60 GHzUsing Point Cloud Data”
The author had the main responsibility of the prediction tool implemen-
tation and validation, and preparation of the manuscript. The measure-
ments were planned and performed together with Dr. Aki Karttunen.
Prof. Haneda supervised the work.
Publication IV: “Radio Propagation Measurements and WINNER IIParameterization for a Shopping Mall at 60 GHz”
Dr. Karttunen had the leading role in preparing the content for the manu-
script. The channel measurements were performed by the author and Dr.
18
Author’s Contribution
Karttunen. Dr. Khatun assisted in data processing and Prof. Haneda
supervised the work.
Publication V: “A Statistical Spatio-Temporal Radio Channel Modelfor Large Indoor Environments at 60 and 70 GHz”
Prof. Haneda had the main responsibility of developing the the channel
model and writing the paper. The author had a leading role in performing
the channel measurements together with Dr. Kyrö and Dr. Karttunen.
The author, Dr. Kyrö, Dr. Karttunen and Mr. Putkonen assisted in writ-
ing the manuscript.
Publication VI: “Polarisation Characteristics of Propagation Paths inIndoor 70 GHz Channels”
Dr. Karttunen had the leading role in preparing the content for the paper
and writing the manuscript together with Prof. Haneda. The author was
the main responsible person in conducting the channel measurements and
assisted in the paper writing. Mr. Putkonen assisted in the paper writ-
ing.
Publication VII: “Evaluation of Millimeter-Wave Line-of-SightProbability With Point Cloud Data”
The author had the main responsibility in developing the content for
the manuscript. Dr. Nguyen contributed to contents of the manuscript.
Mr. Naderpour and Mr. Virk assisted in point cloud handling and Prof.
Haneda supervised the work.
Publication VIII: “On the Mutual Orthogonality of Millimeter-WaveMassive MIMO Channels”
Dr. Nguyen had the main responsibility for developing the content and
writing the manuscript together with Prof. Haneda. The author imple-
mented the channel prediction tool, simulated the channel data and as-
sisted in the paper writing. Dr. Karttunen and Mr. Putkonen assisted in
the paper writing.
19
List of Abbreviations
3GPP Third generation partnership project
4G Fourth generation
5G Fifth generation
5GCM 5G channel model
A/D Analog-to-digital
ASA Azimuth spread of arrival
ASD Azimuth spread of departure
BER Bit error rate
CDF Cumulative distribution function
CIR Channel impulse response
co-pol Co-polarized
COST European Cooperation in Science and Technology
CTF Channel transfer function
D2D Device-to-device
DS Delay spread
ER Effective roughness
FCC Federal communications commission
FDTD Finite-difference time-domain
GSCM Geometry-based stochastic channel model
IEEE Institute of Electrical and Electronics Engineers
IMT International Mobile Telecommunications
ITU-R International Telecommunication Union, Radiocommunication Sector
LOS Line-of-sight
LSP Large-scale parameter
LTE Long term evolution
METIS Mobile and wireless communications Enablers for the
Twenty-twenty Information Society
MIMO Multiple-input multiple-output
MiWEBA Millimetre-Wave Evolution for Backhaul and Access
20
List of Abbreviations
mmMAGIC Millimetre-wave based mobile radio access network
for fifth generation integrated communications
mm-wave Millimeter-wave
MoM Method of moments
MS Mobile station
MU Multiuser
NLOS Non-line-of-sight
OLOS Obstructed line-of-sight
PADP Power angular delay profile
PAS Power angular spectrum
PDP Power delay profile
PL Path loss
QoE Quality of experience
QuaDRiGa Quasi deterministic radio channel generator
RF Radio frequency
rms Root mean square
Rx Receiving
SCM Spatial channel model
SCME Spatial channel model extension
SF Shadow fading
SIR Signal-to-interference ratio
SINR Signal-to-interference-and-noise ratio
SR Specular reflector
SSP Small-scale parameter
Tx Transmitting
UIR Ultrasonic inspection room
UTD Uniform theory of diffraction
VNA Vector network analyzer
WINNER Wireless world initiative new radio
WLAN Wireless local area network
XPD Cross-polarization discrimination
x-pol Cross-polarized
XPR Cross-polarization ratio
21
List of Symbols
A Path loss exponent
Att Penetration loss
B Path loss intercept
D Diffraction coefficient
D1,...,4 Coefficients for calculating diffraction
d3D Three-dimensional link distance
dave Average distance to neighboring points
dS Area of surface element
E0 Electric field at transmitter
Ei Incident electric field vector
Es Scattered electric field vector
EUTD Amplitude of diffracted electric field
f Frequency
FαR Scaling coefficient
GRx,Tx Antenna gain at receiver, transmitter
H Channel transfer function
h Channel impulse response
k Wavenumber
L Number of paths
lp Path length
n Normal vector
Nn Number of neighboring points
p0,spec Specular reflection point
PDP Power delay profile
PDPmeas Measured power delay profile
PDPpred Predicted power delay profile
PL Path loss
R0,n Reflection coefficients
ri Distance from scatterer to transmitter
22
List of Symbols
ri Direction vector for incident electric field
Rpara Reflection coefficient for parallel polarization
Rperp Reflection coefficient for perpendicular polarization
rr Direction of specular reflection
Rrough Effective reflection coefficient
rs Distance from scatterer to receiver
rs Direction vector for scattered electric field
Rsmooth Fresnel reflection coefficient
S Scattering coefficient
s Distance from wedge to receiver
s′ Distance from transmitter to wedge
Sφ Azimuth spread
Sϑ Elevation spread
α Amplitude of multipath
αR Width of scattering lobe
δ(·) Dirac delta function
δmax Maximum plane depth
δplane Plane depth
εr Relative permittivity
θi Incident angle
θr Outgoing angle
ϑRx,Tx Elevation angle at receiver, transmitter
λ Wavelength
μ Mean value
μφ Center of gravity for azimuth angle
ξ Phase randomly distributed over [0 2π)
σ Standard deviation
σh Standard deviation of height distribution
τ Delay
τm Mean delay
τrms Root mean square delay spread
ϕ Phase of multipath
φRx,Tx Azimuth angle at receiver, transmitter
ψR Angle between reflection and scattering
ΩRx,Tx Angle vector at receiver, transmitter
23
1. Introduction
1.1 Background
Over the past decade, the world has witnessed a tremendous growth in
mobile traffic and the number of connected devices. For twenty years, the
mobile traffic was dominated by voice, but since the commercial success
of smart devices in 2007, the data traffic has surpassed voice traffic and
continues to rise exponentially, as seen from Figure 1.1 [1–3]. In 2015,
the global mobile traffic grew by 74%, largely driven by increased video
consumption that already accounts for over half of all mobile traffic [4].
The huge availability of video content accompanied by livestreaming will
further expand the share of video traffic, which is forecast to make up over
70% of the global mobile data traffic in 2021 [1,5].
The constantly growing mobile data volumes have resulted in a band-
width shortage, and the microwave frequency bands, which have been
used in the earlier generations of mobile communications, have become
very crowded [6, 7].1 Consequently, the spectrum of the upcoming fifth
generation (5G) wireless systems has been extended to cover frequency
bands in the millimeter-wave (mm-wave) range [8, 9]. Mm-waves have
been of particular interest due to the availability of large bandwidths in,
for instance, the 57–64 GHz, the 71–76 GHz and the 81–86 GHz bands,
allowing multigigabit data rates [10, 11]. Recently, the federal commu-
nications commission (FCC) proposed that also the 28 GHz and 39 GHz
bands as well as the 64-71 GHz band would be used for high-throughput
small cell deployment [12]. Yet, using mm-waves brings new challenges,
e.g., due to the high attenuation by building materials [13] and high power
consumption in analog-to-digital (A/D) converters [14]. A successful de-
1In this work, the term “microwave frequency” refers to a frequency below 6 GHz.
24
Introduction
Figure 1.1. Mobile voice and data traffic between 2008 and 2015 [3].
ployment of 5G is expected to introduce extremely low latencies, very high
peak data rates and improved quality of experience (QoE) [8,9].
To evaluate the performance of 5G networks, proper models for the ra-
dio channel are required [15]. A comparison between the mm-wave and
microwave propagation channels reveals a few distinct differences, such
as the significantly higher shadowing and diffraction losses at mm-wave
frequencies [16–19]. As a result, the mm-wave channel is more dominated
by specular paths and is presumed to use highly directive antennas and
beamforming [20–23]. These differences, and the fact that many new en-
vironments are considered for mm-wave deployment, necessitate the need
for new channel models at mm-wave frequencies [24,25]. The urge for new
channel models has been aided by research projects involving both indus-
try and academia, such as METIS [26], mmMAGIC [27], MiWEBA [28]
and 5GCM [29].
1.2 Objectives and contribution of the thesis
The objective of the thesis is to participate in the mm-wave channel mod-
eling activities by the following contributions:
1. Simulation tools for accurate site-specific channel modeling and line-
of-sight probability evaluation.
2. A simplified stochastic channel model structure.
3. Channel measurements for providing essential insight about mm-
25
Introduction
wave channels and parametrization of existing stochastic channel
models.
4. Examples of radio link performance evaluation with site-specific chan-
nel models.
First, the relevant properties of mm-wave wireless propagation chan-
nels are reviewed in Chapter 2, including propagation mechanisms and
characterization metrics of wireless channels. Based on material rough-
ness parameters it is concluded that surface materials are usually quite
smooth in indoor environments, but rougher in outdoor scenarios. The im-
portant channel characterization metrics such as path loss, delay spread
and angular spread are also defined.
Using the theory behind propagation mechanisms to predict wireless
propagation channels, which commonly goes under the term site-specific
channel modeling, is discussed in Chapter 3. Ray tracing, which is the
most widespread field prediction tool in wireless communications, is pre-
sented, and the prediction accuracy compared to mm-wave measurements
is studied based on results found in the literature. The results suggest
that the mean values in terms of path loss and delay spread can be well
predicted, but a good agreement in angular domain is more difficult to
achieve. Looking at the agreement between measurements and ray trac-
ing for individual links and their power delay profiles, results indicate
that it is necessary to include also diffuse scattering in the simulation
tool. Furthermore, it is found that models of the environment usually
lack important details. To solve the issue regarding insufficient database
accuracies, a propagation prediction tool relying on accurate environmen-
tal information in the form of point clouds, obtained with laser scanning,
is proposed. As the data format is different from what is used in ray trac-
ing, new methods to consider the propagation mechanisms are developed.
The validity of the simulation tool in both diffuse- and specular-dominated
indoor environments is demonstrated in [I-III].
Chapter 4 covers stochastic channel modeling, reviewing the WINNER,
COST 2100, IEEE 802.11ad and METIS models and their applicability
in the mm-wave bands. Based on channel sounding campaigns in large
indoor environments such as a shopping mall and a train station, a sim-
plified spatio-temporal channel model is proposed and parametrized at 60
and 70 GHz in [V]. Directional 60-GHz measurements in a shopping mall
are used to parametrize the WINNER II channel model, which is reported
in [IV]. In [VI], cross-polarization characteristics of indoor environments
26
Introduction
are studied. The results show that in the studied shopping mall the polar-
ization is preserved better at mm-waves than at microwave frequencies,
allowing more efficient polarization multiplexing. A method to evaluate
line-of-sight (LOS) probability using point clouds is presented in [VII],
where the LOS probability is derived in two environments which until
now have been lacking LOS probability models, an open square and a
shopping mall. A proposed generic exponential model is found to perform
excellently in the studied environments, and a study on frequency depen-
dency shows that the LOS probability at mm-waves differs from that at
lower frequencies due to the narrower Fresnel zone in the mm-wave re-
gion.
In Chapter 5, channel sounding techniques are reviewed with a focus on
mm-wave frequencies. It is noted that channel measurements are time-
consuming and that one sounder can usually not account for all the chan-
nel properties of interest. For instance, wideband, directional sounders
are generally unable to capture the time variance of dynamic channels.
Channel measurements used for validating the point cloud-based chan-
nel prediction tools and parametrizing stochastic channel models are por-
trayed.
The last topic of the thesis is the use of channel models for evaluation
of radio link performance, which is discussed in Chapter 6. The primary
use cases for site-specific and stochastic channel models are depicted in
order to provide the “big picture” of wireless communications and to jus-
tify the need for propagation channel modeling. The point cloud-based
propagation prediction tool is used to study the mutual orthogonality of
mm-wave multi-user channels using large antenna arrays in [VIII], where
it is found that the number of active users should be smaller at mm-waves
compared to microwave frequencies to guarantee effective spatial multi-
plexing.
Apart from publications [I-VIII], the author has authored or co-authored
several other publications related to mm-wave channel modeling [25, 26,
29–39].
27
2. The Wireless Propagation Channel
In order to model propagation channels for evaluating network perfor-
mance, channels must first be studied and characterized. Let us start
with the definition: A wireless propagation channel is the medium through
which a transmitted radio wave reaches the receiver. In urban and in-
door environments, the medium includes besides the air both natural and
man-made structures such as vegetation, building walls, pavement and
furniture. If the first Fresnel zone between the transmitter and receiver
is free of obstacles, the channel is denoted as a LOS channel, whereas
blockage leads to a non-line-of-sight (NLOS) condition. In LOS channels,
the direct path through the air is the shortest and strongest one, and its
amplitude can be calculated with the Friis’ law. As real environments
are not in vacuum, the radio wave will interact with objects in the envi-
ronment causing multipath propagation, that is, weaker, delayed copies
of the original signal, to arrive at the receiver. The mechanisms leading
to multipaths are generally divided into reflection, scattering and diffrac-
tion. Furthermore, all these paths can be attenuated due to shadowing
objects. Next, the propagation mechanisms at mm-wave frequencies are
discussed.
2.1 Propagation mechanisms at mm-waves
2.1.1 Reflection and transmission
A specular reflection occurs when an electromagnetic field interacts with
an electrically large, smooth and distant surface. Snell’s law, stating that
the incident and outgoing angles are identical, can be applied and the re-
flection loss for perpendicular and parallel polarization can be calculated
28
The Wireless Propagation Channel
Table 2.1. Relative permittivities and penetration losses of materials around 60 GHz.
Material |εr| Att [dB/cm] Ref.
Brick 2.6–4.4 1.5–14.7 [17,40–42]Concrete 3.1–6.5 3.7–17 [17,41–43]Plasterboard 2.3–2.8 0.09–2.4 [13,41–43]Glass 6.2–8.9 2.8–18.8 [41–45]Plywood/wood 1.6–2.8 0.8–12 [41–43,46–48]Vinyl floor 6.5–7.8 5.5–6.9 [46]
from the Fresnel reflection coefficients
Rperp =cos θi −
√εr − sin2 θi
cos θi +√εr − sin2 θi
,
Rpara =−εr cos θi +
√εr − sin2 θi
εr cos θi +√εr − sin2 θi
, (2.1)
where θi is the angle of incidence and εr is the complex relative per-
mittivity describing the dielectric properties of the material. A specular
reflection is illustrated by Figure 2.1. Due to the importance of permittiv-
ity in determining the strength of a reflected wave, many measurement
campaigns have been dedicated to dielectric properties of typical urban
and indoor materials. Typical values for the absolute value of εr in the
60-GHz band are listed in Table 2.1.
At mm-wave the penetration losses are usually very large and therefore
accounting for penetration can be simplified to only inducing extra atten-
uation similarly to the Motley–Keenan model [15, 49]. Typical building
material losses at 60 GHz are shown in Table 2.1, where Att is the pen-
etration loss in dB/cm. It can be seen that Att varies wildly even for a
single material because the material samples used in measurements are
not identical and their composition depends on, e.g., the moisture and
possible metallic reinforcements. Also the permittivity is seen to vary de-
pending on the specific material sample.
θi θ
i
εr
n
Figure 2.1. Illustration of specular reflection. The vector n denotes the surface normal.
29
The Wireless Propagation Channel
σh [mm]
0 0.2 0.4 0.6 0.8 1
Rro
ugh/R
smoo
th
0
0.2
0.4
0.6
0.8
1
θi = 75°
θi = 45°
θi = 15°
Figure 2.2. Effective reflection coefficient.
2.1.2 Scattering
When an electromagnetic wave impinges on a surface which is rough or
small compared to the wavelength, the equations (2.1) are no longer valid
and the propagation phenomenon is referred to as scattering. Many stud-
ies have been made to study the effect of rough surfaces, such as introduc-
ing an effective reflection coefficient which models the power scattered to
the specular direction [15]
Rrough = Rsmooth exp[2(kσh cos θi)2], (2.2)
where Rsmooth is the Fresnel reflection coefficient calculated with (2.1),
k = 2π/λ is the wavenumber for the wavelength λ and σh is the standard
deviation of the height distribution. To illustrate the effect of the surface
roughness at 60 GHz, the ratio Rrough/Rsmooth, which ideally should be
close to 1, is plotted as a function of σh for different incident angles in
Figure 2.2. Other well-known surface roughness criteria are the Rayleigh
criterion, requiring σh < λ/(8 cos θi) for a smooth surface, or the stricter
Fraunhofer criterion which presumes σh < λ/(32 cos θi) for a smooth sur-
face [50,51]. For θi = {15◦, 45◦, 75◦}, the Rayleigh criterion gives σh values
ranging from 0.65 to 2.4 mm at 60 GHz, while the Fraunhofer criterion
gives values between 0.16 and 0.6 mm. As a reference, Table 2.2 shows
the surface roughness, i.e., the standard deviation in height σh, for com-
mon indoor and urban materials. Most indoor materials have σh < 0.2 and
can thus be considered smooth, implying that surface roughness does not
require particular consideration. On the other hand, it can be observed
30
The Wireless Propagation Channel
Table 2.2. Surface roughness for indoor and urban materials.
Material σh [mm] Ref.
Acrylic 0.001 [52]Glass 0.01 [40,53]Vinyl floor 0.028 [52]Plaster 0.05–0.15 [54]Wooden panel < 0.1 [40]Wallpaper 0.13 [54]Hardwood floor 0.14 [52]Smooth concrete 0.2 [55]Brick 0.3–2 [40]Rough asphalt 0.9 [55]
that outdoor materials such as brick and asphalt are clearly rougher.
A more holistic method for dealing with scattering mechanisms, includ-
ing both real surface roughness and the effect of small objects, is the
so-called effective roughness (ER) approach [56]. The ER assumes that
all plane surfaces contribute to diffuse scattering because the building
databases do not include details about small irregularities. Two models
have been proposed to model the scattering, a non-directional model with
a Lambertian scattering pattern and a single-lobe directive model, which
focuses the scattered signal in the direction of the specular reflection.
Among these, the single-lobe directive model has been found to perform
better than the Lambertian model for many typical wall materials [57].
In the directive single-lobe model, presented in Figure 2.3, the scattered
field is written as
|Es|2 =(SEi
rirs· λ
4π
)2
· dS cos θiFαR
·(1 + cosψR
2
)αR
, (2.3)
where Ei is the incident field, S = |Es|/|Ei| is the scattering coefficient in
the range [0, 1], ri and rs are the distances from the scatterer to the trans-
mitter and receiver, respectively, dS is the area of the surface element, ψR
is the angle between the direction of the reflected wave and the scattering
direction, αR determines the width of the scattering lobe so that a higher
value gives more directive scattering, and FαR is a scaling coefficient de-
fined as
FαR =1
2αR·αR∑j=0
(αR
j
)· Ij , (2.4)
where
Ij =2π
j + 1·[cos θi ·
j−12∑
ω=0
(2ω
ω
)· sin
2ω θi22ω
]( 1−(−1)j
2)
. (2.5)
31
The Wireless Propagation Channel
ri
rs
θi
θr
rr
ψR
dS
Figure 2.3. Single-lobe directive scattering model where ri and rs are the directions ofthe incident and scattering waves and rr is the direction of the specular re-flection with the outgoing angle θr = θi [57].
2.1.3 Diffraction
At microwave frequencies, diffraction has been known as a major contrib-
utor especially in urban NLOS scenarios [58, 59]. At mm-waves, diffrac-
tion is negligible in most cases [60] with the exception of weak NLOS
links and when the receiver is in the transition region close to the shadow
boundary [17,61].
When considering diffraction in deterministic field prediction, Huygens’
principle is commonly used [62]. In short, it states that an object being
radiated by a field can be seen as a secondary source which re-radiates
a field to the receiver. In wireless communications especially the wedge
diffraction is of interest because building corners, both indoors and out-
doors, are the main objects causing diffraction [15]. The amplitude of a
diffracted path can be calculated with the uniform theory of diffraction
(UTD) for an imperfectly conducting right-angle wedge [63]:
D = D1 +R0RnD2 +R0D3 +RnD4, (2.6)
where R0 and Rn are the reflection coefficients for the illuminated and
shadow faces, respectively, and Dg(g = 1, 2, 3, 4) depend on the frequency,
distances and angles [64]. The diffracted field is calculated from the
diffraction coefficient as
EUTD =E0 exp(−jk(s′ + s))
s′ + sD
√s′ + s
s′s, (2.7)
where E0 is the field at the transmitting (Tx) antenna and s′ and s are the
32
The Wireless Propagation Channel
τ
| h(τ
)|
Figure 2.4. Example of a channel impulse response.
distances from the edge to the transmitting and reveiving (Rx) antennas,
respectively.
2.2 Characterization of wireless channels
2.2.1 The double-directional channel
The environment and propagation mechanisms described above give rise
to multipath components, which cause the received signal to be spread
out in both delay and angle. Looking only at the delay τ , the received
signal can be depicted by a channel impulse response (CIR) h(τ) shown
in Figure 2.4. Taking into account the angle for each multipath at both
the Tx and Rx sides, as well as the delay, we end up with the so-called
double-directional impulse response consisting of L multipaths [65]
h(τ,ΩRx,ΩTx) =L∑l=1
αl exp (jϕl)δ(τ − τl)δ(ΩRx −ΩRx
l )δ(ΩTx −ΩTxl ), (2.8)
where ΩTx = [φTx ϑTx] and ΩRx = [φRx ϑRx] are vectors composed of az-
imuth and elevation angles on the Tx and Rx sides, respectively, αl, ϕl
and τl are the amplitude, phase and delay of the lth multipath compo-
nent, and δ(·) stands for the Dirac delta function.
Furthermore, due to the movement of the environment, the transmitter
or the receiver, the channel can also be time-variant. This may cause indi-
vidual Doppler shifts in the different multipath components, which causes
the original signal to be spread out also in frequency. The distribution of
frequency shifts for the multipaths can be modeled by the Doppler spec-
trum. The effect of the Doppler spread can be significant in narrowband
systems, but is usually small in wideband systems.
33
The Wireless Propagation Channel
2.2.2 Channel characterization metrics
A single CIR experiences small scale fading due to constructive and de-
structive interference of multipaths, which can be seen as a variation in
signal amplitude or phase over a short time. To filter out the small scale
fading, a number of CIRs taken over a small area can be averaged to form
a power delay profile (PDP):
PDP (τ) =1
M
M∑m=1
|h(τ)|2, (2.9)
where M is the number of CIRs.
To simplify the characterization of the channel power and delay disper-
sion, a set of parameters can be extracted from the PDP [15]. Path loss
(PL) describes how much the power is attenuated in the channel and can
be calculated simply as the integral over the PDP for those delay bins
lying over the noise floor. The first delay parameter is the mean delay,
which is derived as
τm =
∫∞−∞ PDP (τ)τdτ∫∞−∞ PDP (τ)dτ
. (2.10)
The second, and more important parameter describing the delay disper-
sion, is the root mean square (rms) delay spread:
τrms =
√∫∞−∞ PDP (τ)τ2dτ∫∞−∞ PDP (τ)dτ
− τm. (2.11)
The importance of the delay spread is due to its proportionality with bit
error rate (BER) and inverse correlation with the coherence bandwidth
[15].
Analogous to the delay dispersion, also the angular dispersion can be ex-
pressed with the angular spread from the power angular spectrum (PAS),
which characterizes how the signal power varies over an angle. The az-
imuth spread is computed with [15,66]
Sφ =
√∫ | exp(jφ)− μφ|2PAS(φ)dφ∫PAS(φ)dφ
, (2.12)
where φ is the azimuth angle and
μφ =
∫exp(jφ)PAS(φ)dφ∫
PAS(φ)dφ. (2.13)
The elevation spread Sϑ can be determined similarly from the elevation
34
The Wireless Propagation Channel
angle ϑ. Also other definitions of the angular spreads exist [67]. The de-
lay and angular dispersion parameters are commonly used for describing
channels, and hence they can also be used for comparing the agreement
between measured and predicted channels.
2.2.3 Depolarization
Due to edges and rough surfaces, the polarization of a transmitted wave
is not perfectly preserved, but a part of the power is coupled to the or-
thogonal polarization [62]. This phenomenon, called depolarization, can
be characterized by the cross-polarization ratio (XPR), which is the ratio
between the co-polarized and cross-polarized powers.
35
3. Site-Specific Channel Modeling
For modeling a radio channel, channel measurements always provide the
most accurate option, or put in another way, the “truth”. However, mea-
surements, as will be discussed in Chapter 5, can sometimes be very ex-
hausting, especially when a large amount of directional data is required
for statistical purposes, e.g., [68]. The long measurement durations can
cause limitations as all measurement sites are not accessible for long pe-
riods of time. Furthermore, the placement of antennas may be infeasible
due to non-stationary objects such as vehicles and people. It must also be
kept in mind that the antenna radiation pattern affect the measurements.
In contrast to measurements, using simulation tools to estimate the
channel does not suffer from all these limitations. Simulation, often de-
noted as deterministic field prediction, relies on electromagnetic theory to
determine the electromagnetic fields in a given environment as detailed
in Chapter 2. To obtain accurate channel estimation results, the detailed
geometry and material properties of the entire environment should be
known [15]. The highest precision is given by so-called full-wave ap-
proaches such as the method of moments (MoM) or the finite-difference
time-domain (FDTD) method, which solve the Maxwell’s equations for a
discretized environmental model given that the most accurate represen-
tation of the environment is available.. As the resolution of the geome-
try should be less than one wavelength [69, 70], full-wave simulations of
large environments are computationally prohibitive, particularly at mm-
wave frequencies. At microwave frequencies, simple indoor geometries
have been characterized using both MoM and FDTD [69,71–73]. Another
full-wave method for repetitive structures has been reported in [74].
The growing need for site-specific network planning in the early nineties
necessitated field prediction tools with reduced computational complex-
ity [75, 76]. This initiated the development of ray-based methods, relying
36
Site-Specific Channel Modeling
on high frequency approximations of Maxwell’s equations. The approx-
imations are based on the geometric optics assumption that the wave-
length of the propagating radio wave is much smaller than the objects
in the surrounding environment and that radio waves can be treated as
infinitely narrow rays, which reflect and penetrate according to Snell’s
law. Among ray-based methods, ray tracing is the most established one
and is used in many commercially available software such as the Wireless
InSite [77], WaveSight [78], Winprop [79] and Volcano [80]. Moreover,
many universities and research institutes have developed their own ray
tracing engines [81–84]. Some works use a hybrid method combining ray
tracing and FDTD [85,86]. Another prediction method relying on the ray
assumption was introduced in [30], where the 3D model of the environ-
ment is obtained through laser scanning in the form of point clouds. Point
cloud-based methods eliminate the need for manually building detailed
environmental models, but require different computation techniques to
model the propagation mechanisms as will be discussed in Section 3.2.
3.1 Ray tracing
Ray tracing rests on the so-called image principle, where the image of the
transmitter is determined with respect to each planar surface (first order
reflection) and combination of surfaces (second and higher order reflec-
tions) [87]. The environment is represented by surfaces, which for urban
locations might be available from online databases such as Google Earth
[88], but can also be built using computer-aided design software [89]. All
possible rays are then formed by tracing the ray from the transmitter im-
ages to the receiver, along with diffracted rays (discussed in Section 2.1.3).
The power of the paths are computed from free-space path loss, reflection
loss (Eq. 2.1) and penetration loss. For improved prediction accuracy,
also scattering has to be included. For this purpose, scattering models
presented in Section 2.1.2 are good candidates.
To validate ray tracing tools, measured channels are commonly used as
a reference. The simplest metric for validation is, as presented in Section
2.2.2, the path loss. Delay and angular spreads are also popular compar-
ison metrics. Table 3.1 presents the prediction accuracy obtained from
literature in terms of power, delay and angular metrics. The comparison
shows that path loss can be accurately predicted in both LOS and NLOS
scenarios. The predicted delay spread is also in good agreement with
37
Site-Specific Channel Modeling
Table 3.1. Prediction accuracy for path loss, delay spread and angular spread.
Environment Freq. Meas. Pred. Error Ref.P
ath
loss
Indoor (NLOS) 28 GHz 4.5 dB [90]Indoor (LOS) 58 GHz 67.1 dB 67.2 dB [91]Indoor (LOS) 58 GHz 1.2 dB [92]Indoor (NLOS) 58 GHz 0.5 dB [92]Indoor (LOS) 58 GHz 48 dB 46 dB [93]
Del
aysp
read
Office (LOS) 61 GHz 6.3 ns 6.6 ns 1.2 ns [82]Office (LOS) 61 GHz 6.3 ns 6.3 ns 1.1 ns [94]Urban (NLOS) 28 GHz 58.6 ns 50.7 ns [95]Indoor (LOS) 58 GHz 4.1 ns 4.1 ns [91]Indoor (LOS) 58 GHz 1.4 ns [92]Indoor (NLOS) 58 GHz 2.6 ns [92]Indoor (LOS) 60 GHz 13.3 ns 14.3 ns [93]Office (LOS) 60 GHz 10.8 ns 18.6 ns [96]Office (LOS) 60 GHz 11.8 ns 14.1 ns [96]Urban 60 GHz 63.5 ns 63.3 ns [97]Office (LOS) 73 GHz 9 ns 8 ns [89]Office (NLOS) 73 GHz 13 ns 12 ns [89]Urban 73 GHz 6 ns 5 ns [68]Office (LOS) 60 GHz 12.3 ns 13.1 ns [98]
Azi
mut
hsp
read
Urban (NLOS) 28 GHz 8.5° 4.1° [95]Urban (NLOS) 28 GHz 33.5° 17.3° [95]Urban (NLOS) 28 GHz 52.9° 47.9° [95]Indoor (LOS) 58 GHz 41.5° 55.1° [91]Indoor (LOS) 58 GHz 18.4° 16.5° [91]Urban 73 GHz 35° 27° [68]
Ele
vati
onsp
read
Urban (NLOS) 28 GHz 4.7° 1.1° [95]Urban (NLOS) 28 GHz 8.8° 7.2° [95]Indoor (LOS) 58 GHz 7.3° 3.3° [91]Urban 73 GHz 3° 6° [68]
measurements when looking at mean values. The azimuth and elevation
spreads clearly suffer from large errors, and in NLOS links the predicted
spreads are heavily underestimated. As many works present the compar-
ison only in terms of mean values, the reasons for the discrepancy is not
always evident. However, papers which present the agreement in the form
of a PDP or a PAS, such as [82,91,97,99,100], highlight the fact that the
3D models are usually quite simple and therefore ray tracing is not able to
predict even the specular peaks correctly. Ray tracing also fails in predict-
ing a large number of weaker paths that might originate from electrically
small objects or rough surfaces which are not included in the database,
that is, scattering. For example in [101], lamps and bookshelves were
found to cause strong scattering. The effect of furniture is investigated in,
e.g., [102], which shows that the difference between PASs in an empty and
38
Site-Specific Channel Modeling
a furnished room can be significant. The importance of considering the de-
tails of the environment is discussed in, e.g., [97, 101, 103–107]. Further
ray tracing results, whose agreement with measurements have not been
quantified but presented visually, can be found in [89,99–101,107–113].
3.2 Point cloud-based propagation prediction
Due to the fact that simplified geometrical descriptions lead to the afore-
mentioned inaccuracies, a field prediction method relying on accurate en-
vironmental data in the form of point clouds has been developed [I]. The
point clouds are obtained with laser scanners, which use moveable mir-
rors to steer a laser beam in different directions. An object will cause the
beam to reflect back to the scanner, where the distance to the object is de-
termined, for instance, by comparing the phase of the transmitted and re-
ceived beams [114]. For each reflecting object, the x-, y- and z-coordinates
are determined from the distance and the direction of the laser beam. The
position accuracy of laser scanners is in the order of 5 mm [115]. To cover
an environment without shadowed areas, scans are made in multiple lo-
cations and the data is combined to form one complete point cloud. For
example, the small office in [I] and the open square in [VIII] were scanned
with two and 20 locations, respectively, where each scan lasted less than 4
minutes. To decrease the computational load of the simulation, the point
cloud density can be reduced with different sampling methods [116].
Ray-tracing software require a surface representation of the environ-
ment, but finding a surface description from the point cloud, commonly
known as meshing, is not a straight-forward task [116]. The single-lobe
scattering model presented in (2.3) is on the other hand not restricted by
the format of the environment model and can be used directly with the
point cloud. Using a scattering model can also be justified based on the
importance of scattering in improving the accuracy of deterministic field
prediction, as was discussed in Section 2.1.2.
3.2.1 Point cloud-based prediction of diffusive propagationchannels in small rooms
Assuming scattering to be the dominant propagation mechanism, the to-
tal field can be estimated as the sum of scattering from all points in the
point cloud, where scattering is calculated with the single-lobe directive
model (2.3). To determine the angles θi and ψR, the normal vector for the
39
Site-Specific Channel Modeling
Figure 3.1. Point cloud of a small office room [I].
local surface dS of each point is required (see Figure 2.3). A simple ap-
proach to find the angles is to find the normal from the three neighboring
points [30], or for a more stable solution the normal can be acquired from
fitting a plane to the Nn neighboring points (In [III], Nn = 8). The area
dS of each point is similarly calculated from the average distance to the
4 neighboring points dave as dS = d2ave. A scattered path is defined as the
propagation path from the Tx antenna, scattered from a point in the point
cloud and received by the Rx antenna. Each path can be described as
P ={τl, αl,Ω
Txl ,ΩRx
l
}L
l=1, (3.1)
where the delay τ is determined by the propagation distance of each path,
α is the amplitude computed with (2.3), ΩTx = [φTx ϑTx] and ΩRx =
[φRx ϑRx] are vectors composed of azimuth and elevation angles on the
Tx and Rx sides, respectively and L is the number of paths. Using the
discrete Fourier transform we can calculate the channel transfer function
(CTF) as a sum of the paths with
Hi =L∑l=1
GRx(ΩRx)αlGTx(ΩTx) exp [−j(2πτlfi − ξ)], (3.2)
where GTx and GRx are the gains on the Tx and Rx sides, respectively, f is
the frequency with I frequency steps, 1 ≤ i ≤ I and ξ is a phase uniformly
distributed over [0 2π). Using the inverse discrete Fourier transform, the
transfer function H can be converted back to a CIR:
hn =1
I
I∑i=1
Hi ej2πfiτn , (3.3)
40
Site-Specific Channel Modeling
where n corresponds to the nth delay bin, 1 ≤ n ≤ I. To account for the
effects of small-scale fading in deriving the PDP, the phase of the scattered
paths ξ in (3.2) can be varied randomly. The PDP can be calculated as an
ensemble average of M small-scale realizations of the CIRs with (2.9).
As it is difficult to know the material parameters for a propagation envi-
ronment exactly [41, 42] as highlighted by Table 2.1, all the points in the
point cloud are assumed to have the same scattering model parameters αR
and S in (2.3). Using channel measurements, the scattering model param-
eters are then tuned so that the measured and predicted PDPs agree as
well as possible. It should be noted that because the diffuse scattering is
observed both due to surface roughness and diffraction from small objects,
it is difficult to treat objects separately. Recent results also support the
assumption that different objects and materials can share similar scatter-
ing model parameters when they have similar surface roughness [117]. To
find the optimum scattering model parameter values for a single link, the
parameter values are tuned by a brute force search to minimize the cost
function
ε =1
J
J∑j=1
∣∣∣PDPmeas(j)− PDPpred(j)∣∣∣, (3.4)
where PDPmeas and PDPpred are the measured and predicted PDPs, re-
spectively, and 1 ≤ j ≤ J are the delay indices for which the corre-
sponding power levels are within a 30 dB dynamic range seen from the
maximum signal amplitude of the PDP.
3.2.2 Shadowing detection
At mm-waves, shadowing due to walls, furniture and people [118, 119]
causes severe signal attenuation as shown in, e.g., Table 2.1. Therefore
the detection of shadowing is a necessary feature in any deterministic
field prediction tool. As a surface representation is not available in point
cloud environments, shadowing is detected by searching for points within
the first Fresnel zone between the Tx and Rx antennas, as proposed in
[33]. In other words, a path is free of blockage if
lp ≤ d3D +λ
2, (3.5)
where lp is the length of the path from the Tx antenna via a point in the
point cloud to the Rx antenna and d3D is the distance from the Tx to the Rx
antenna [VII]. Based on the material of the shadowing object, penetration
41
Site-Specific Channel Modeling
Tx n Rx
ndavg
Figure 3.2. Point cloud densification for improved shadowing detection. Blue points de-note the original point cloud, red squares denote dense points and the grayellipse marks the Fresnel zone [III].
loss is added to the amplitude of shadowed paths.
The point cloud density is a crucial factor determining a successful de-
tection of shadowing objects. In [33], it is found that with low point den-
sities, the second Fresnel zone should be used instead of the first order
zone. To assure that shadowing is detected also close to the link ends, an
improved shadowing detection method was proposed in [III]. The point
cloud is densified, that is, new points are interpolated around objects close
to the Fresnel zone ends as shown in Figure 3.2.
3.2.3 Prediction of overall channel using point cloud data
In small environments where there are a lot of small objects, scattering
might be a dominant propagation mechanism, as seen in [I], [II]. How-
ever, in larger spaces with smooth walls, also specular reflections have
to be modeled. Moreover, even physically small objects such as computer
screens might act as specular reflectors at mm-waves [120]. A method for
simulating specular reflections is presented in [121], where points lying in
the first Fresnel zone between the image source and the receiver are iden-
tified as specular reflection points. The method sums up the contributions
from points within the first Fresnel zone, and thus requires a high point
density to accurately predict the amplitude of a reflected path. This be-
comes especially challenging in mm-wave bands where the Fresnel zones
have small dimensions, since a dense point cloud increases the compu-
tational burden significantly. To calculate specular reflections with point
clouds having lower density, an approach similar to [121] is developed
in [III]. In this new method, it is assumed that the surface is larger than
the first Fresnel zone and that the amplitude thus can be calculated di-
rectly with the Fresnel equations shown in (2.1). To assure that specular
reflections are calculated from flat surfaces, the plane depth δplane of the
local surface around the reflection point is checked. For a particular point,
this is accomplished by computing δplane as the maximum variation in the
direction of the normal among the eight neighboring points (see Figure
42
Site-Specific Channel Modeling
n
δplane
Figure 3.3. Determination of plane depth δplane. Blue points denote neigboring points.
3.3), and checking whether this value is smaller than a predefined maxi-
mum plane depth δmax. The value δmax is dependent on the environment
and the point cloud density, and is usually determined heuristically. All
points which are found within the nth Fresnel zone and which satisfy the
smoothness criterion δplane < δmax are said to be specular reflection points
p0,spec. However, in practical implementations there are cases where mul-
tiple nearby specular reflection points are found within the same Fresnel
zone. These points must be grouped such that a specular reflection from
the same object is not calculated multiple times, thus overestimating the
amplitude of the reflected path. In particular, specular reflection points
lying inside the same Fresnel zone are grouped, and in each group the
point closest to the Fresnel zone center forms a specular reflector (SR).
Every point p0,spec then either belongs to an SR or forms an SR of its own.
Finally, the specular paths are calculated from the SRs only.
In severe NLOS scenarios also diffraction around corners is taken into
account. The relevant wedges where diffraction occurs are modeled by
rows of points, and the diffracted field is calculated with the UTD diffrac-
tion formulas presented in Section 2.1.3.
3.3 Contribution of the thesis
3.3.1 Validation of propagation prediction based on diffusescattering
The validity of the diffusive propagation prediction method has been in-
vestigated in [I] in two indoor environments:, an ultrasonic inspection
room (UIR) with approximate dimensions of 7.2×6.1×2.6 m3, and a small
office with dimensions of 4.5×4.3×2.9 m3. The results show that the scat-
tering lobe width αR has a much smaller effect on the predicted channel
compared to the scattering coefficient S, and that a single value for S can
be used within one scenario. In the ultrasonic inspection room and the
office, the best values for S are found to be 0.9 and 0.5, respectively. Ex-
emplary PDPs and PASs from the small office are presented in Figures
3.4 and 3.5, which show that the predicted and measured channels agree
43
Site-Specific Channel Modeling
Table 3.2. Prediction accuracy for the point cloud-based diffuse prediction method [I].
Env. Metric Meas. Pred. Error
UIR (LOS) τm 6.8 ns 7.3 ns 0.6 nsSmall office (LOS) τm 5.3 ns 5.3 ns 0.1 nsUIR (LOS) τrms 1.9 ns 1.9 ns 0.2 nsSmall office (LOS) τrms 3.1 ns 2.6 ns 0.5 nsSmall office (LOS) Sφ 23.0° 22.7° 2.6°Small office (LOS) Sϑ 16.7° 16.4° 0.6°
excellently both in delay and azimuth domains. The prediction accuracy
is also presented in Table 3.2, from which it can be observed that errors
in both delay and angular spreads are smaller compared to the results in
Table 3.1. It should be noted that due to the small size of the rooms and
the absence of objects leading to high shadowing losses, the prediction is
carried out with only first-order scattering and by neglecting the effects
of shadowing.
It is difficult to assess the general applicability of the optimized scat-
tering model parameters as the availability of those is still very scarce.
Moreover, the optimum values αR and S do not depend on the material
alone, but also on the amount of fixtures and furniture in the environ-
ment.
3.3.2 Influence of point cloud density
As the computational burden is proportional to the number of points in a
point cloud, the effect of point cloud density has been studied at 60 GHz
in [II]. The study is performed in the same small office described earlier
with point cloud sizes ranging from 1000 points to 676 000 points, cor-
responding to point separations of roughly 30 and 1 cm, or 60λ and 2λ,
Delay [ns]0 10 20 30 40 50
Am
plitu
de [
dB]
-110
-100
-90
-80
-70
-60 MeasuredPredicted
Figure 3.4. Measured and predicted PDPs in small office [I].
44
Site-Specific Channel Modeling
Azimuth angle at Tx [°]0 60 120 180 240 300 360
Azi
mut
h an
gle
at R
x [ °]
45
90
135
[dB]
-110
-100
-90
-80
-70
(a)
Azimuth angle at Tx [°]0 60 120 180 240 300 360
Azi
mut
h an
gle
at R
x [ °]
45
90
135
[dB]
-110
-100
-90
-80
-70
(b)
Figure 3.5. a) Measured and b) predicted power angular spectrum for small office [I].
respectively. A comparison of optimum scattering model parameters and
prediction accuracy for delay and angular spreads reveals virtually no dif-
ferences between different densities, which can be explained by the fact
that the strongest paths can be well predicted with all densities. Yet, the
PDPs for the low point cloud densities can be seen to fluctuate more and
have a higher number of peaks, as seen from the PDPs in Figure 3.6. This
results from the increased relative power of a single scatterer and the un-
certainty in determining the surface normals due to the small number of
points. A further decrease in the number of points is believed to high-
light this phenomenon even more. The predicted PDPs also emphasize
the effect of omitting double-order scattering, as the prediction fails in
representing the tail of the measured PDP, i.e., delays larger than 30 ns.
However, this tail is more than 40 dB weaker compared to the maximum
amplitude and will thus have a negligible effect on the channel.
3.3.3 Shadowing detection
The method to detect shadowing, which was proposed in [33], is used
in [III] to validate the channel prediction method in NLOS conditions.
Also an improved shadowing detection method which is described in Sec-
tion 3.2.2 is presented to account for shadowing objects close to the link
ends. In [VII], the shadowing detection method is applied for deriving
45
Site-Specific Channel Modeling
Delay [ns]0 10 20 30 40 50
Am
plitu
de [
dB]
-130
-120
-110
-100
-90
-80
-70 Measured1k5k20k49k676k
Figure 3.6. Measured and predicted PDPs with different point cloud densities, where “k”in the legend denotes 1000 [II].
LOS probability models, which is discussed in Section 4.6.4.
3.3.4 Validation of overall channel prediction tool
In [III], the overall prediction tool is studied in a cafeteria with approx-
imate dimensions of 14×13.5×2.8 m3, which contains smooth walls and
windows as well as scattering objects including tables, chairs, computer
screens and lamps. A solid wall is separating the main cafeteria from an
adjacent smaller space, allowing measurements with the direct path com-
pletely blocked. Channel sounding in the band 61–65 GHz is performed
in 3 LOS and 3 NLOS locations, including a 0–360° azimuth sweep on the
Tx side to study the directional characteristics of the channel. Similar
to the scattering model parameters, the relative permittivity is assumed
to be equal for all materials and optimized based on a brute force search
minimizing the delay spread error compared to measured channels. The
optimization yields relative permittivity values between 3.5 and 6 and S
of around 0.8 in LOS links and 0.6 in NLOS links, while the scattering
lobe width αR is fixed to 1 because it is found to be of small influence com-
pared to S. A comparison between measured and predicted PDPs, PASs
and channel metrics are presented in Figures 3.7 and 3.8 and Table 3.3.
The result shows that the prediction accuracy in LOS channels is excel-
lent in terms of path loss, mean delay and delay spread, and very good for
angular spread. The agreement between measured and predicted links
is very good also for NLOS links. Furthermore, it is shown that spec-
ular reflections can be modeled by a single relative permittivity for all
surfaces as a result of common indoor materials having similar reflection
coefficients. The only exception is metal, which causes underestimation
46
Site-Specific Channel Modeling
Delay [ns]0 50 100 150 200
Am
plitu
de [
dB]
-140
-120
-100
-80 MeasuredPredicted
(a)
Delay [ns]0 50 100 150 200
Am
plitu
de [
dB]
-140
-120
-100
-80 MeasuredPredicted
(b)
Figure 3.7. Measured and predicted PDPs for a) LOS location Tx1, b) NLOS location Tx4in [III].
Azimuth angle [°]0 90 180 270 360
Am
plitu
de [
dB]
-140
-120
-100
-80 MeasuredPredicted
(a)
Azimuth angle [°]0 90 180 270 360
Am
plitu
de [
dB]
-140
-120
-100
-80 MeasuredPredicted
(b)
Figure 3.8. Measured and predicted PASs for a) LOS location Tx1, b) NLOS location Tx4in [III].
of the path amplitude and should be modeled with a separate permittiv-
ity value. In general, the more different materials are considered, i.e., the
higher number of relative permittivities that are optimized, the better the
agreement between measurements and prediction should be.
Table 3.3. Comparison of measured (m.) and predicted (p.) large scale parameters for theoverall channel prediction [III].
Link Tx PL [dB] τm [ns] τrms [ns] Sφ [°]type m. p. m. p. m. p. m. p.
LO
S 1 78.0 78.4 11.9 12.1 5.1 5.6 16.8 19.52 77.4 78.7 11.1 11.1 6.9 6.9 14.0 19.53 79.1 80.4 14.3 14.1 7.7 7.3 20.2 22.0
NL
OS 4 98.8 97.3 50.3 48.2 18.4 18.1 44.5 41.7
5 98.9 96.6 52.2 46.6 14.4 10.3 32.3 27.06 97.1 98.5 53.6 59.1 11.3 22.7 38.7 44.0
47
4. Stochastic Channel Modeling
In contrast to site-specific models, stochastic channel models aim at repro-
ducing the statistical properties of a propagation channel in terms of, for
instance, received power, delay or angular dispersion. They can be used
for transmission technique design or performance comparison. Stochas-
tic models consider balance between accuracy and simplicity, and their
requirements depend on the system for which the models are built. For
instance, wireless systems in hospitals require a much higher reliability
compared to conventional indoor networks [122]. The importance of accu-
rate models can also be seen in network planning, where overestimating
the path loss leads to increased costs due to, e.g., redundancy in base
station deployment, but underestimation leads to unsatisfactory coverage
and thus decreased QoE.
Stochastic channel models are usually formulated as a set of mathemat-
ical equations, including parameters describing the characteristics of the
environment and the deployment such as the antenna height, the street
width and the path loss. The parametrization of channel models is done
either with channel measurements [123,124] or by utilizing deterministic
field prediction such as ray tracing [125–128]. The most well-known chan-
nel models are path loss models such as the Okumura–Hata model [129]
and the COST 231–Walfisch-Ikegami model [130]. As the wireless sys-
tems have become more and more complex, the channel models have de-
veloped to include more features of the channel, e.g., directional proper-
ties. Next, the most common modern channel models and their applica-
bility at mm-waves is reviewed.
48
Stochastic Channel Modeling
Table 4.1. WINNER II scenarios.
Scenario Definition
A1 Indoor office/residentialB1 Urban micro-cell hotspotB2 Bad urban micro-cell
(same as B1 + long delays)B3 Large indoor hall hotspot
(train station, airport)
4.1 WINNER
The 3rd generation partnership project (3GPP) released the spatial chan-
nel model (SCM) in 2002, which was developed for cellular systems with
multiple antennas in the frequency range 2–5 GHz [131]. The original
model was designed for outdoor links and specified in three scenarios:
Suburban macro, urban macro and urban micro. It was parametrized in
two dimensions, considering only azimuth angles and neglecting the ele-
vation domain. Later the development of the SCM lead to the WINNER
(wireless world initiative new radio) [132], SCME (SCM Extension) [133],
WINNER II [134], WINNER+ [135], IMT-Advanced [136] and QuaDRiGa
[137] models, covering a multitude of outdoor and indoor scenarios and ex-
tensions such as elevation angles, continuous time evolution, frequencies
up to 6 GHz and bandwidths up to 100 MHz. The WINNER-based models
are some of the most widely used channel models and have been validated
by several measurement campaigns [138, 139]. Among the many scenar-
ios for which WINNER is parametrized, the scenarios most relevant to
mm-wave systems are listed in Table 4.1.
The WINNER model is a geometry-based stochastic channel model
(GSCM), where the large-scale propagation parameters (LSPs), such as
delay spread or shadow fading, are determined randomly based on sta-
tistical distributions extracted from comprehensive measurements [134].
The model is antenna independent, which means that it can be used
with different antenna configurations. For most scenarios, a distance-
dependent LOS probability function is used to determine the channel con-
dition, and the parameters are defined separately for LOS and NLOS. A
dependency between the different LSPs was observed in many channel
measurements, and are hence taken into account using their correlation.
Based on the LSPs and tabulated distribution functions, small-scale pa-
rameters (SSPs), i.e parameters taking into account the physical prop-
erties of rays including delays, powers and directions of departure and
49
Stochastic Channel Modeling
arrival, are formed, thus fully describing the propagation channel. Chan-
nel coefficients are then generated by applying random initial phases to
the rays. Each scenario has a predefined number of clusters ranging from
8 to 15 in LOS cases and 10 to 20 in NLOS cases, and the number of rays
per cluster is fixed to 20.
WINNER models apply the so-called “drop” concept, meaning that the
LSPs are assumed constant in a single channel segment, but have no cor-
relation with LSPs in adjacent channel segments, i.e., other drops. Spa-
tial consistency, meaning that two closely located mobile stations (MSs)
experience similar power, delay and angular dispersion, is thus not sup-
ported. Therefore the WINNER model is incapable of modeling, for in-
stance, device-to-device (D2D) links, where both link ends are moving.
Furthermore, the WINNER model assume that antenna arrays are elec-
trically small, which may not be a valid assumption for very large antenna
arrays [25].
4.2 COST 2100
The COST 2100 channel model [140, 141], developed within the COST
(European cooperation in science and technology) framework, resembles
the WINNER model in being a GSCM and having similar LSPs, SSPs
and clusters. In contrast to WINNER, where the clusters are drawn ran-
domly in each drop, the COST 2100 model has fixed cluster positions. The
position-defined clusters allow smooth user movement and close-by MSs
are able to share the same scatterers. Three type of clusters are defined,
namely local clusters, single-bounce clusters and twin clusters. The local
clusters are always visible to the MS, while single-bounce and twin clus-
ters are associated with a visibility region, which is a circular region in-
side which a cluster is visible. Signal paths propagating through the twin
clusters are reflected several times between the transmitter and receiver.
A drawback of the COST 2100 channel model is that parametrization of
the cluster parameters such as the radii of the visibility regions requires
identification of wave scatters from the channel sounding, which is not
always straightforward [25, 142]. Moreover, alike the WINNER model,
also the COST 2100 model does not support D2D links because the COST
model is designed for cellular channels where the base station is always
fixed, while the D2D link may have mobility on both link ends. In an
attempt to improve the COST 2100 model, a modification of the original
50
Stochastic Channel Modeling
model framework for covering D2D links has been proposed in, e.g., [143].
4.3 IEEE 802.11ad
The IEEE 802.11ad channel model was developed by the IEEE 802.11
task group AD for 60 GHz wireless local area networks (WLANs) [144,
145]. The model aims at taking into account all the relevant character-
istics of 60 GHz propagation channels, and supports beamforming and
non-stationary channels due to moving people. The model is based on
clustered rays and provides accurate space-time and polarization charac-
teristics of each ray, which consist of the LOS path and first and second
order reflections. Channel sounding and ray tracing has been used to
parametrize the model for three indoor scenarios, namely a conference
room, a cubicle and a living room. As the layouts for the three environ-
ments are specified very precisely, the parameters may not be valid in
other similar environments [25]. Moreover, no diffuse scattering is in-
cluded in the model.
4.4 METIS channel model
The European 7th framework project METIS (Mobile and wireless com-
munications Enablers for the Twenty-twenty Information Society) was
founded to lay a foundation for 5G [26, 146]. To establish the first 5G
channel model, the following requirements were identified:
• Support of a wide range of network topologies, such as D2D
• Frequency bands up to 86 GHz and bandwidths up to 500 MHz
• Support for very large antenna arrays
• Spatial consistency and mobility
• Realistic modeling of specular reflections
The final model consists of a deterministic map-based model, a stochastic
model model, and a hybrid model [146]. The map-based model is meant
for use cases where realistic spatial channel characteristics are needed,
such as for large antenna arrays. The model is based on simple 3D geome-
tries and ray tracing, including the propagation mechanisms discussed in
Section 2.1. Shadowing and scattering objects are placed randomly in the
51
Stochastic Channel Modeling
environment. If it is possible to compromise the details of the generated
radio channels with reduced computational load, different propagation
mechanisms may be turned off. Validation work for the map-based model
has been presented in, e.g., [147]. On the other hand, the METIS stochas-
tic model follows the WINNER framework, and is specified separately for
a number of scenarios and frequency bands. The METIS hybrid channel
model takes advantage of the map-based and stochastic models for vary-
ing levels of demands in accuracy and complexity. It obtains the path loss
and shadowing from the map-based model, and other parameters from the
stochastic model. The usability of the METIS model at mm-wave frequen-
cies is still unknown as very little validation work has been conducted.
4.5 Other mm-wave channel modeling works
The project MiWEBA (Millimetre-Wave Evolution for Backhaul and Ac-
cess) has also contributed to mm-wave channel modeling [148]. The chan-
nel modeling approach follows the same general structure as the IEEE
802.11ad model, but where the 802.11ad model takes into acount only
deterministic rays, the MiWEBA model is quasi-deterministic, combining
deterministic rays, rays from random objects and rays from moving ob-
jects [149].
4.6 Contribution of the thesis
4.6.1 Spatio-temporal channel model for large indoorenvironments
In [V], a simple stochastic channel model structure is proposed for the 60-
and 70-GHz bands based on channel sounding in large indoor environ-
ments. The measurements reveals that clustering is not apparent in the
studied environments, in contrast to the channel modeling frameworks of
WINNER, COST 2100 and IEEE 802.11ad. Moreover, the results show
that specular paths dominate over diffuse paths, and that propagation at
60 and 70 GHz is very similar with slightly faster power decay at 70 GHz
compared to 60 GHz. The channel model is defined for LOS channels and
takes into account both specular and diffuse paths. Based on detailed
instructions, CIRs can be generated for a given link distance, frequency
52
Stochastic Channel Modeling
Azimuth angle [°]90 180 270 360
Del
ay [
ns]
150
100
50
0 -120
-100
-80
-60
(a)
Azimuth angle [°]90 180 270 360
Del
ay [
ns]
150
100
50
0 -120
-100
-80
-60
(b)
Figure 4.1. PADPs from a) the measurement and b) the proposed channel model [V].
and bandwidth. An example of a measured power angular delay profile
(PADP) and a PADP generated with the channel model are presented in
Figure 4.1. The validity of the channel model is shown in terms of path
loss and delay spread.
4.6.2 Parametrization of WINNER channel model in shoppingmall at 60 GHz
A first parametrization of the WINNER at 60 GHz is presented in [IV],
in which both LSPs and SSPs are derived for a shopping mall. Both LOS
and obstructed LOS (OLOS) links are measured, where the shadowing for
OLOS links occurs due to pillars. The parameters are diplayed in Table
4.2, where PL stands for the path loss, μ is the mean value and σ is the
standard deviation. An initial validation of the delay spread and K-factor
is performed, showing good agreement between measured channels and
channels produced by the WINNER implementation described in [150].
Furthermore, parametrizations based on simulations for a cafeteria and
an open square are reported in [26].
4.6.3 Characterization of cross-polarization at 70 GHz
In [VI], wideband channel sounding at 70 GHz is conducted in four in-
door sites, namely an empty office, a furnished office, a shopping mall
and a railway station. By rotating a horn antenna at the Tx side, both
co-polarized (co-pol) and cross-polarized (x-pol) channels are measured.
Using a peak detection algorithm to find paths in the PDPs, as shown in
Figure 4.2(a), the path-wise cross polarization ratio (XPR) is calculated as
the ratio between the co- and x-pol path amplitudes. The cumulative dis-
tribution function (CDF) is shown in Figure 4.2(b), in which also the an-
53
Stochastic Channel Modeling
Delay [ns]0 50 100 150
Am
plitu
de [
dB]
-130
-120
-110
-100
-90
-80
-70Co-pol PDPX-pol PDPCo-pol peaksX-pol peaks
(a)
XPR, XPD [dB]10 20 30 40
CD
F
0
0.2
0.4
0.6
0.8
1XPRXPD
(b)
Figure 4.2. a) Co- and x-pol PDPs and peaks, b) CDF of XPR and XPD [VI].
tenna cross-polarization discrimination (XPD) is portrayed. The results
show that the mean XPR value in the studied environments is around
23 dB, which is close to the XPR value of 20 dB specified in the IEEE
802.11ad channel model for 60 GHz [144], and clearly higher than XPRs
of 4–12 dB specified in WINNER II for frequencies below 6 GHz [134].
However, it must be kept in mind that the obtained XPR values might be
underestimated as many of the high XPR values might not be detected
due to the poor dynamic range in the x-pol measurement and the antenna
XPD of roughly 34 dB.
4.6.4 Line-of-sight probability at millimeter-wave frequencies
Applying point cloud data and the shadowing detection method described
in Section 3.2.2, a novel method to evaluate LOS probability is proposed
in [VII]. The LOS probability is calculated in two new scenarios, an open
square and a shopping mall, as well as in an office environment, as de-
picted in Figure 4.3. Base stations (BSs) are deployed on typical loca-
tions, such as by the ceiling, and a very high number of MSs are placed
in locations where users can go. For each BS-MS link, the shadowing is
checked with (3.5), and finally the LOS probability as a function of link
distance is calculated for the three scenarios separately. Existing LOS
probability models including the ITU-R and a linear model as well as our
proposed generic exponential model are parametrized. The LOS probabil-
ity for the three scenarios differed notably due to the size and structure of
the environment, but the exponential model shows excellent performance
in all scenarios. A study on the impact of frequency shows that due to
the increasing Fresnel zone radius with decreasing frequency, the LOS
probability at 2.4 GHz is clearly lower than at 63 GHz. This implies that
the ray assumption is not valid at microwave frequencies in the studied
54
Stochastic Channel Modeling
Link distance [m]0 20 40 60 80 100 120
LO
S pr
obab
ility
0
0.2
0.4
0.6
0.8
1CalculatedITU-R UMiExponentialLinear
(a)
Link distance [m]0 20 40 60 80 100
LO
S pr
obab
ility
0
0.2
0.4
0.6
0.8
1CalculatedITU-R UMiExponentialLinear
(b)
Link distance [m]0 5 10 15 20 25 30
LO
S pr
obab
ility
0
0.2
0.4
0.6
0.8
1CalculatedITU-R InHExponentialLinear
(c)
Figure 4.3. LOS probability at 63 GHz in a) an open square, b) a shopping mall and c) anoffice [VII].
scenarios.
55
Stochastic Channel Modeling
Table 4.2. WINNER II parameters at 70 GHz in a shopping mall.
LOS OLOS
PL = A log10(d) +B [dB] A 18.4 3.6B 68.8 94.3
Delay spread (DS) μ −8.28 −7.78log10([s]) σ 0.32 0.10
Azimuth spread of departure (ASD) μ 1.09 1.61log10([°]) σ 0.43 0.11
Azimuth spread of arrival (ASA) μ 1.19 1.62log10([°]) σ 0.47 0.14
Shadow fading (SF) [dB] σ 1.2 2.1
K-factor [dB] μ 7.9 N/Aσ 5.8 N/A
Cross-correlations
ASD [°] vs DS [s] 0.4 0.5ASA [°] vs DS [s] 0.2 0.3
ASA [°] vs SF [dB] 0.0 0.1ASD [°] vs SF [dB] 0.0 −0.1
DS [s] vs SF [dB] 0.2 −0.4ASD [°] vs ASA [°] 0.0 0.0ASD [°] vs K [dB] −0.4 N/AASA [°] vs K [dB] −0.3 N/A
DS [s] vs K [dB] −0.2 N/ASF [dB] vs K [dB] 0.2 N/A
Delay scaling parameter rτ 2.5 2.0
XPR [dB] μ 20 2σ 20 2
Per cluster shadowing [dB] σ 2.5 5.3
Number of clusters 4 10Number of rays per cluster 20 20Cluster ASD [°] 1.5 1.5Cluster ASA [°] 1.5 1.5
Correlationdistance [m]
DS [s] 1 0.5ASD [°] 2 0.5ASA [°] 1 1SF [dB] 0.5 0.5K [dB] 1 N/A
56
5. Millimeter-Wave Channel Sounding
Measurements have always been the foundation for understanding the
propagation behavior of radio signals. Since Marconi’s first long-range
experiments in 1895, devices for detecting the radio signals have become
increasingly sophisticated and complex, allowing a more detailed charac-
terization of different radio wave propagation phenomena. Field strength
measurements as a function of link distance were conducted already be-
fore the Second World War [151], and still in the 1960s, the field strength
was the principal area of interest within the wireless communications
community. The next step was to study the delay of multipath compo-
nents [152], and finally, in the 1990s, focus was laid also on directional
properties of the channel [15, 153]. Simultaneously, the environments
have changed from large, outdoor spaces to smaller, indoor scenarios.
Also the frequency of interest has shifted from tens of megahertz to even
terahertz frequencies. Although the first mm-wave experiments were con-
ducted already in the 19th century [154], it wasn’t until the 1990s when
mm-wave channel measurements were being performed from a wireless
communications’ perspective [155]. Present mm-wave equipment includ-
ing wideband systems and huge antenna arrays enables a very detailed
study on the effect of small objects in the environment, e.g., [106].
Preferably, measurements should be able to characterize every dimen-
sion of a propagation channel, including the power, delay, azimuth and
elevation angles on both Tx and Rx sides, Doppler spread, polarization
and time variance. However, present channel sounders have certain lim-
itations and, e.g., mm-wave sounders are not able to capture all of the
aforementioned aspects simultaneously, as will be discussed next.
57
Millimeter-Wave Channel Sounding
5.1 Narrowband measurements
When one is interested only in the field strength, i.e., the path loss from
the transmitter to the receiver, narrowband measurements are the easi-
est option. A narrowband signal can be measured for instance by trans-
mitting a continuous wave generated by a Gunn oscillator and observ-
ing the received signal with a spectrum analyzer [60, 156]. Due to the
simple channel sounder structure, the instrumentation costs are low and
the measurement duration is very short, allowing measurements of time-
variant channels. To produce the baseband signal for mm-wave chan-
nels, the oscillator does not have to work at mm-wave frequencies, but
the oscillator frequency can be multiplied to the desired frequency. The
radio signal can be downconverted to lower frequencies also at the detec-
tion [15]. Examples of narrowband measurements at mm-waves can be
found in [48,157,158].
5.2 Wideband channel measurements
To estimate the delays and amplitudes of individual multipaths, that is,
to measure a CIR, wideband measurements are required. These can be
conducted either in the delay or frequency domains.
5.2.1 Measurements in the delay domain
Wideband channel sounding in the delay domain is usually performed
by transmitting short pulses or continuous wave modulated by a pseu-
dorandom sequence, and sampling the received signal in the delay do-
main using a sliding correlator or an A/D sampling card [159–161]. The
CIR is obtained from a convolution of the transmitted and received sig-
nals. Sounding in the delay domain is very fast and can thus be used
to measure time-variant channels. The drawbacks include a quite com-
plex instrumentation and a limited delay resolution caused by the limited
speed of the sampling unit. Results from wideband mm-wave channel
sounding campaigns using delay domain techniques have been shown in,
e.g., [82,162–166].
58
Millimeter-Wave Channel Sounding
5.2.2 Measurements in the frequency domain
In frequency domain measurements, the frequency is swept either con-
tinuously or with discrete frequency steps referred to as the frequency-
stepping method [167]. Frequency-stepping is slower compared to a con-
tinuous sweep, but it provides higher measurement accuracy. Such mea-
surement systems are usually realized with a vector network analyzer
(VNA), and the time domain response, i.e., the CIR, is obtained by an in-
verse Fourier transform of the frequency transfer function. As the band-
width is inversely proportional to the delay resolution, a large bandwidth
is essential in resolving multipath components from each other. One
drawback of VNA-based systems is that a cable connection is needed be-
tween the transmitter and the receiver, which limits the range of the link
distance [168]. By using optical fibre cables, the limit can nonetheless
be extended to cover hundreds of meters [36]. Compared to delay do-
main sounders, VNA-based systems are also slower and cannot be used
for time-variant channels [159]. The main advantage of using a VNA is,
beside the simple sounder structure, the ability for very wideband chan-
nel measurements. VNA-based channel sounding at mm-wave frequen-
cies has been conducted in [I-VI], [36–38,60,159,169–176].
5.3 Directional channel measurements
From the antenna design point of view, the directional properties of the
channel at both the Tx and Rx sides are highly important. To determine
the direction of the multipath components there are two distinct methods:
rotating a directional antenna or using an antenna array [15].
5.3.1 Rotation of directional antenna
A very directive antenna, typically with a half-power beamwidth less than
10°, is installed on the link end where directional characteristics are of
interest. The antenna is then rotated so that the antenna is pointing in
different directions, and at each rotation angle the radio channel is mea-
sured. The angular resolution can be improved by having a narrower an-
tenna beamwidth, but at the same time the number of pointing directions
is increased. The mechanical rotation of the antenna is a time-consuming
method, especially when using a VNA to measure the CIRs. A full 3D scan
at one link end, including antenna pointing directions in both azimuth
59
Millimeter-Wave Channel Sounding
and elevation domains, can easily take tens of minutes, and for such mea-
surements at both link ends the total measurement duration can be sev-
eral hours. Thus, only very static directional channels can be measured
accurately. However, the use of directional antennas decreases the need
for post-processing as the CIRs are measured individually for each direc-
tion. To remove the effect of the antennas when deriving channel metrics
such as path loss, the antenna gains have to be subtracted from the mea-
sured channel data, e.g., [36]. Directional channel measurements with
rotating antennas are presented in [III–VI], [36–38,162–164,177,178].
5.3.2 Antenna array measurements
When using an antenna array, the antenna elements should be as omni-
directional as possible and the directional information is determined by
array signal processing. The relative locations of the antennas should be
well defined and the separation between antennas should be in the order
of one wavelength [15]. The array can either be a real array, where the
CIRs for all antennas can be measured simultaneously when the sounder
is equipped with multiple radio frequency (RF) chains, or a multiplexed
array in which case a switch is used to measure the antenna elements
one by one with a single RF chain. The array can be modeled virtually
by moving a single antenna between predefined positions. With the vir-
tual antenna technique the measurement complexity and hardware costs
are low, but they come at a price of increased measurement time. For
example, in [179], the duration of measuring a single link is more than
20 minutes. Directional characterization using antenna arrays has been
done in [I],[II], [65,179–181].
5.4 Polarization measurements
To measure the depolarization due to the environment, orthogonal polar-
izations are required at the Tx or Rx antenna. A common solution at mm-
waves is to use a horn antenna, for which two orthogonal polarizations
are obtained by rotating the antenna by 90°. Polarimetric measurements
at mm-waves have been reported in for example [VI], [102,120,182,183].
60
Millimeter-Wave Channel Sounding
5.5 Millimeter-wave channel sounding campaigns for 5G scenarios
So far, a number of mm-wave channel measurements have been performed
in a large variety of environments. The first campaigns were dedicated
to point-to-point links [155, 184–187], and in the nineties the research
turned to indoor environments [167,188–191]. As the use of wireless com-
munication systems has become more widespread, new scenarios for net-
work deployment have been defined. Nowadays, mm-waves are foreseen
to be used in point-to-point links, i.e., backhaul links, and in short-range
access point and cellular scenarios, so-called hot spots and small cells.
Considering 5G scenarios, a need for mm-wave communications are seen
in, e.g., urban micro scenarios such as street canyons and open squares,
stadiums, indoor environments such as offices, shopping malls, airports
and train stations, and outdoor-to-indoor scenarios [27, 28, 192–194]. Al-
though mm-wave indoor measurement results are widely available [195],
the number of works in these new environments, especially in the large in-
door spaces, is small. Results found in the literature include, e.g., 28 GHz
measurements in a train station and an airport terminal [196]. Multi-
frequency channel measurements in an airport are reported in [38].
5.6 Contribution of the thesis
In this work, mm-wave channel sounding has been performed mainly for
the following purposes:
1. To acquire general knowledge of mm-wave propagation in different
scenarios.
2. For material parameter tuning and validation of the point cloud-
based simulation tool (Section 3.2) in [I-III,VIII].
3. For parametrization of stochastic channel models in [IV–VI].
For the work, 60- and 70-GHz measurements have been conducted in an
ultrasonic inspection room, small and large offices, a shopping mall, a
train station and an open square. A few insights acquired from the mea-
surements are listed below
• Small spaces containing plenty of fixtures can be dominated by dif-
fuse scattering [I],[II].
61
Millimeter-Wave Channel Sounding
• In large open environments, specular paths are dominant and can
carry over 90% of the power in LOS links [V].
• In NLOS links, the second and third order reflections can account
for the majority of the power [III].
• Multipath clustering is not evident in large indoor environments
[IV],[III].
• XPR is higher at mm-waves compared to microwave frequencies [VI].
62
6. Applications for Channel Models
The performance of wireless systems can be evaluated from many differ-
ent perspectives, but they all require some kind of a channel model. Link
level design, such as comparing transmission techniques, needs stochas-
tic channel models which are independent of specific locations, while base
station deployment requires site-specific models. Next, a short overview
is given as to typical usage of both stochastic and site-specific channel
models.
6.1 The use of stochastic channel models
6.1.1 Coding and modulation
In wireless communications, coding and modulation are used to convert
the data into a form that can be transmitted over the wireless channel and
received by the receiver in an efficient and reliable manner. In [197,198],
various coding schemes are compared in terms of bit error rate (BER) and
throughput. The work in [199] compares different modulations and coding
rates to evaluate the performance of wireless systems in a hospital.
6.1.2 Mobile terminal antenna design
Mobile phones and others mobile terminals need to function regardless
of the orientation of the device, or with the influence of the user. Thus,
the polarization plays an important role and should be considered by
the channel model used for evaluating mobile antenna performance. The
work presented in [200] investigates the influence of antenna placement
in a handset on, e.g., the capacity. The handset antenna placement is also
studied in [201], which also looks at the effect of the user on the antenna
63
Applications for Channel Models
efficiency.
6.1.3 Network design
Network design involves for instance planning the cell layout, determin-
ing the cell radius and designing BS antennas. In [24], the influence of
the antenna and cell radius is investigated with regard to the signal-to-
interference ratio (SIR) and signal-to-interference-and-noise ratio (SINR).
It is observed that beamforming is needed to achieve sufficient network
coverage. In [202], the SINR is studied for different BS antenna down
tilt angles, concluding that tilt angles between 4° and 8° are good for
cells with a radius of 300 m. The work reported in [203] evaluates the
throughput of different multiple-input multiple-output (MIMO) antenna
array configurations.
6.1.4 Capacity and throughput evaluation
For the end user in a wireless system, the throughput, i.e., the data rate,
is one of the most relevant factors affecting the quality of service and
has thus been investigated widely. In [204], a statistical mm-wave chan-
nel model is used for system simulation, showing that mm-wave systems
can offer an order of magnitude increase in capacity compared to 4G net-
works without increasing the cell density. In [205], the performance in
high speed trains is studied in terms of BER and throughput. The differ-
ence between a 2D and a 3D channel model is presented in [206], which
shows that the 2D model underestimates the channel throughput by 20%
compared to the 3D model.
6.2 The use of site-specific channel models
6.2.1 Coverage analysis
Coverage analysis is one of the most typical simulations in network de-
ployment, including tasks such as finding the optimal locations of base
stations. Ray tracing tools have popularly been used for this purpose, as
presented in, e.g., [207,208]. The coverage along with the effects of coding
and antenna sectorization is studied in [209]. In [210], the influence of the
room geometry, wall material and antenna locations are studied. A study
of an urban small cell mm-wave backhaul network, including for instance
64
Applications for Channel Models
0 10 20 300
2
4
6
8
Inter−user distance (m)A
vera
ge n
o. o
fco
mm
on s
catte
rers
(a)
0 10 20 300
0.5
1
Inter−user distance (m)
Cor
rela
tion
coef
fici
ent
6x6 array10x10 array16x16 array20x20 array
(b)
Figure 6.1. Average number of a) common scatterers and b) pairwise orthogonality be-tween two channel vectors for different array sizes [VIII].
outage probabilities and coverage, is presented in [211].
6.2.2 Base station antenna design
In [212], different BS antenna arrays are investigated in an office envi-
ronment in terms of the throughput. It is shown that placing the anten-
nas only in the azimuth domain gives higher throughput that distributing
them in elevation or both azimuth and elevation. A similar investigation
in outdoor macro and pico cells is conducted in [213], where the influence
of BS antenna configurations on the outage and throughput is presented.
6.3 Contribution of thesis
In [VIII], the mutual orthogonality of mm-wave massive multiuser (MU)-
MIMO channels is studied in an open square. The channel data is gener-
ated with the point cloud-based simulation tool described in Section 3.2,
resulting in both LOS and OLOS links. Shadowing is caused by lamp-
posts, trees and people. It is seen that the inter-user distance has a
clear correlation with the number of common scatterers, as seen in Figure
6.1(a). Moreover, the influence of the antenna array size on the pairwise
orthogonality between two channel vector is shown, as depicted by Fig-
ure 6.1(b). Furthermore, the result suggests that the number of active
users should be smaller than at microwave frequencies because mm-wave
channels are sparser in terms of multipaths compared to lower frequen-
cies [182]. The capacity analysis suggests that the separation between
users should be twice the correlation distance of shadowing, 16 m, to take
advantage of spatial multiplexing.
65
7. Summary of Publications
[I] Sixty Gigahertz Indoor Radio Wave Propagation PredictionMethod Based on Full Scattering Model
A propagation prediction method relying on accurate point cloud data,
obtained by laser scanning, and a single-lobe directive scattering model, is
presented. The prediction method is applied in two indoor environments,
an ultrasonic inspection room and a small office, and the scattering model
parameters are tuned based on measurements at 60 GHz. The agreement
between measured and predicted PDPs, mean delay, rms delay spread,
PASs, azimuth spread and elevation spread is found to be very good.
[II] Impacts of Room Structure Models on the Accuracy of 60 GHzIndoor Radio Propagation Prediction
The impact of the point density on the prediction accuracy of the point
cloud-based propagation prediction method is analyzed in a small office
with five densities ranging from an average point separation of 1 to 30
cm. The different densities yield similar estimates of the rms delay spread
and azimuth and elevation spreads compared to 60 GHz measurements,
showing that the simulation speed can be enhanced by lowering the point
cloud density without compromising on the prediction accuracy. The PDPs
show more fluctuation with lower density.
66
Summary of Publications
[III] Indoor Propagation Channel Simulations at 60 GHz Using PointCloud Data
The point cloud-based propagation prediction method is extended by in-
tegrating all relevant propagation mechanisms including specular reflec-
tion, scattering, diffraction and shadowing. The material parameters are
tuned based on 60 GHz measurements in a cafeteria, showing that typ-
ical indoor materials can be modeled with a single relative permittivity.
The method is validated in both LOS and NLOS scenarios, in terms of
path loss, mean delay, rms delay spread and azimuth spread, as well as
by PDPs and PASs. Results show very small prediction errors of 0.5 dB,
0.3 ns and 3° for power, delay and angular domains in LOS links, and
relative errors of only 10% in NLOS links.
[IV] Radio Propagation Measurements and WINNER IIParametrization for a Shopping Mall at 60 GHz
Directional wideband channel sounding in the 60 GHz band is conducted
in a shopping mall to acquire PADPs for both LOS and OLOS links. In
OLOS cases, shadowing is caused by pillars. WINNER II parameters for
power, delay and angles are derived along with those for correlation and
clustering. An initial validation shows good agreement between measured
channels and channels reproduced with the WINNER II implementation
for the rms delay spread and K-factor.
[V] A Statistical Spatio-Temporal Radio Channel Model for LargeIndoor Environments at 60 and 70 GHz
Channel sounding at 60 and 70 GHz in four indoor spaces is used to de-
velop a novel stochastic channel model framework. Specular and diffuse
components are modeled separately, and in contrast to common channel
models, clustering is not included because the measurements do not show
apparent clustering effects of multipaths. Parameters are determined
separately for the two frequencies and the four environments, and in-
structions on implementing the channel model are given. The validity of
the model is demonstrated in terms of path loss and rms delay spread.
67
Summary of Publications
[VI] Polarisation Characteristics of Propagation Paths in Indoor 70GHz Channels
Polarimetric directional wideband channel sounding is conducted in in-
door environments at 70 GHz in order to study the effect of depolariza-
tion. Peak detection is used to find specular propagation paths in PDPs
of co-pol and x-pol links, and the XPR is calculated as the ratio between
the co-pol and x-pol path powers. The XPR is seen to vary between 10 and
30 dB with a mean value of 23 dB. The result shows that in the studied
environment the polarization is better preserved at mm-waves compared
to microwaves, implying that polarization diversity can be used more ef-
fectively.
[VII] Evaluation of Millimeter-Wave Line-of-Sight Probability WithPoint Cloud Data
A novel method to evaluate LOS probability based on point cloud data
and Fresnel zones is presented. The LOS probability is calculated in two
new scenarios, i.e., an open square and a shopping mall, as well as an
office environment. The ITU-R, a linear model and a proposed generic
exponential model are parametrized for all three scenarios. The exponen-
tial model performs excellently in all scenarios despite large differences
among them. The dependency of the frequency is also portrayed.
[VIII] On the Mutual Orthogonality of Millimeter-Wave Massive MIMOChannels
The mutual orthogonality of mm-wave massive MU-MIMO channels are
studied in an open square scenario. The dependency of mutual orthog-
onality on the inter-user distance, number of active users and antenna
array size is characterized. The results show that the number of active
users should be small, and that a separation distance of at least twice the
correlation distance of shadowing (16 m) is required to assure efficient
spatial multiplexing.
68
8. Conclusions
This thesis focuses on mm-wave channel modeling for future 5G wireless
communication systems. The main scientific contributions of the work
include tools and insights for both site-specific and stochastic channel
modeling, and emphasize the need for more detailed descriptions of the
model frameworks and environment descriptions at mm-waves compared
to lower frequencies.
The first part of the thesis describes deterministic field prediction, point-
ing out that even if good agreement between measured and predicted
power and delay metrics can be achieved in terms of mean values, ac-
curate databases are required to obtain good prediction accuracy for the
angular domain and in terms of, e.g., power delay profiles. Moreover, the
effect of scattering is emphasized. This work focuses on developing a novel
field prediction tool relying on accurate environment data in the form of
point clouds. Methods to account for relevant propagation mechanisms,
including specular reflections, scattering, diffraction and shadowing, are
detailed. The point cloud-based prediction method is validated in both
diffuse- and specular-dominant scenarios, showing excellent agreement
in power, delay and angular domains for both LOS and NLOS links.
The second subject of the thesis deals with stochastic channel modeling.
A review of the common channel modeling frameworks stresses that the
use of mm-waves introduces new challenges related to, e.g., massive an-
tenna arrays and D2D links. A novel stochastic spatio-temporal channel
model structure is proposed, which in contrast to common models does
not consider clustering as it is not found apparent in the channel mea-
surements. A detailed channel model implementation recipe is given and
the validity at 60 and 70 GHz is demonstrated by studying path loss and
delay spread. Measurements are also used for deriving parameters for
the WINNER II channel model at 60 GHz in a shopping mall and for an
69
Conclusions
initial validation of the parameters. Moreover, 70-GHz channel sounding
results are used to characterize depolarization in indoor environments,
showing that the XPR at mm-waves is usually over 20 dB and can thus
offer improved polarization diversity compared to lower frequency bands.
Last, a novel method to evaluate LOS probability based on point clouds is
presented.
In the third part of the thesis, mm-wave channel sounding is discussed
in short, comparing pros and cons of typical sounding equipment. It is
noted that no mm-wave sounder can measure everything at once, and
that compromises have to made between speed and accuracy. A litera-
ture review affirms that many of the large indoor spaces envisioned for
mm-wave deployment are lacking channel sounding results. Lastly, a few
insights acquired through mm-wave channel sounding are provided. For
example, large indoor spaces are dominated by specular paths and do not
show clear signs of multipath clustering.
The last part of the thesis is dedicated to applications for channel mod-
els, and provides exemplary use cases for both stochastic and site-specific
channel models, such as BS antenna design and throughput evaluations.
A study on mutual orthogonality for massive mm-wave MU-MIMO chan-
nels is performed with the aid of point cloud-based propagation prediction.
The results show that compared to microwave frequencies, mm-wave sys-
tems must have fewer active users and larger inter-user distances to allow
efficient spatial multiplexing.
Despite the many valuable contributions presented in this work, there
is still a great deal of efforts required to ensure the successful deploy-
ment of mm-wave networks. Among these, the most significant task is
to validate the existing channel model frameworks by means of channel
sounding and accurate field prediction tools in various mm-wave bands
and environments. The results from these actions can be used to derive
models which are valid in an extremely wide range of frequencies, or to
point out possible model deficiencies and propose solutions. Last, as no
thesis devoted to wireless communications is complete without a refer-
ence to higher frequencies, it must be mentioned that the future will most
likely bring even higher frequencies into the midst of our super-connected
society [214].
70
References
[1] Ericsson, “Ericsson mobility report,” Ericsson, Tech. Rep., Nov. 2015.[Online]. Available: www.ericsson.com/res/docs/2015/mobility-report/ericsson-mobility-report-nov-2015.pdf
[2] W. Xie and H. Liang, “A case study: Innovation strategy assessment ofthe leading smartphone companies,” in 2013 Suzhou-Silicon Valley-BeijingInt. Innovation Conf. (SIIC), Suzhou, China, July 8-9, 2013, pp. 121–126.
[3] Akamai, “Akamai’s state of the internet: Q1 2015 report,” Akamai,Tech. Rep., June 2015. [Online]. Available: www.stateoftheinternet.com/resources-connectivity-2015-q1-state-of-the-internet-report.html
[4] Cisco, “Cisco visual networking index: Global mobile data traf-fic forecast update, 2015-2020,” white paper, Feb. 2016. [Online].Available: www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/mobile-white-paper-c11-520862.html
[5] J. Pullen, “Livestreaming apps will totally crush your dataplan,” Apr. 2015. [Online]. Available: http://time.com/3768107/meerkat-periscope-data-plan
[6] T. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, G. Wong,J. Schulz, M. Samimi, and F. Gutierrez, “Millimeter wave mobile commu-nications for 5G cellular: It will work!” IEEE Access, vol. 1, pp. 335–349,2013.
[7] Z. Pi and F. Khan, “An introduction to millimeter-wave mobile broadbandsystems,” IEEE Commun. Mag., vol. 49, no. 6, pp. 101–107, June 2011.
[8] Nokia, “looking ahead to 5G,” white paper, 2014.
[9] Samsung Electronics, “5G Vision,” white paper, Aug. 2015. [On-line]. Available: www.samsung.com/global/business-images/insights/2015/Samsung-5G-Vision-2.pdf
[10] O. Sevimli, V. Dyadyuk, D. Abbott, L. Stokes, S. Smith, J. Archer, M. Shen,R. Kendall, and J. Tello, “Multi-gigabit wireless test bed at millimetrewaves,” in IEEE/ACES Int. Conf. Wireless Commun. Appl. Comput. Elec-tromagn., Honolulu, HI, Apr. 3-7, 2005, pp. 434–437.
[11] H. Singh, J. Oh, C. Kweon, X. Qin, H.-R. Shao, and C. Ngo, “A 60 GHzwireless network for enabling uncompressed video communication,” IEEECommun. Mag., vol. 46, no. 12, pp. 71–78, Dec. 2008.
71
References
[12] FCC, “Use of spectrum bands above 24 GHz for mobile radio services,”FCC-15-138, Oct. 2015.
[13] C. Anderson and T. Rappaport, “In-building wideband partition loss mea-surements at 2.5 and 60 GHz,” IEEE Trans. Wireless Commun., vol. 3,no. 3, pp. 922–928, May 2004.
[14] S. Rangan, T. Rappaport, and E. Erkip, “Millimeter-wave cellular wirelessnetworks: Potentials and challenges,” Proc. IEEE, vol. 102, no. 3, pp. 366–385, Mar. 2014.
[15] A. F. Molisch, Wireless Communications, 2nd ed. New York: IEEE Press-Wiley, 2011.
[16] S. Geng, “Millimeter wave and UWB propagation for high throughput in-door communications,” Ph.D. dissertation, Aalto Univ., Helsinki, Finland,2011.
[17] P. Vainikainen et al., “Propagation,” in EURASIP Book Series on SignalProcessing and Communications, Volume 3: Smart Antennas – State of theArt, T. Kaiser et al., Ed. Hindawi, 2005, pp. 211–240.
[18] J. McDonnell, “Characteristics of the indoor wireless propagation environ-ment at microwave and millimetre frequencies,” in IEE Colloq. Radio Com-mun. Microwave Millimetre Wave Freq. (Dig. No. 1996/239), London, UK,Dec. 16, 1996, pp. 1–6.
[19] M. Peter and W. Keusgen, “Analysis and comparison of indoor widebandradio channels at 5 and 60 GHz,” in Proc. 3rd European Conf. AntennasPropag. (EuCAP), Berlin, Germany, Mar. 23-27, 2009, pp. 3830–3834.
[20] M.-T. Martínez-Inglés, J.-M. Molina-García-Pardo, J.-V. Rodríguez,J. Pascual-García, and L. Juan-Llácer, “Experimental comparison betweencentimeter- and millimeter-wave ultrawideband radio channels,” RadioSci., vol. 49, pp. 450–458, June 2014.
[21] H. Yang, F. Smulders, and M. Herben, “Indoor channel measurementsand analysis in the frequency bands 2 GHz and 60 GHz,” in Proc. 16thIEEE Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC), Berlin,Germany, Sept. 11-14, 2005, pp. 579–583.
[22] W. Hong, K. H. Baek, Y. Lee, Y. Kim, and S. T. Ko, “Study and prototyp-ing of practically large-scale mmWave antenna systems for 5G cellular de-vices,” IEEE Commun. Mag., vol. 52, no. 9, pp. 63–69, Sept. 2014.
[23] K. Haneda, “Channel models and beamforming at millimeter-wave fre-quency bands,” IEICE Trans. Commun., vol. E98-B, no. 5, pp. 755–772,May 2015.
[24] Q. Li, H. Shirani-Mehr, T. Balercia, H. Niu, A. Papathanassiou, and G. Wu,“Millimeter wave channel model and system design considerations,” inProc. 2015 IEEE Int. Conf. Commun. Workshop (ICCW), London, UK, June8-12, 2015, pp. 1214–1219.
[25] J. Medbo, K. Börner, K. Haneda, V. Hovinen, T. Imai, J. Järveläinen,T. Jämsä, A. Karttunen, K. Kusume, J. Kyröläinen, P. Kyösti, J. Meinilä,
72
References
V. Nurmela, L. Raschkowski, A. Roivainen, and J. Ylitalo, “Channel mod-elling for the fifth generation mobile communications,” in Proc. 8th Euro-pean Conf. Antennas Propag. (EuCAP), The Hague, Netherlands, Apr. 6-11,2014.
[26] METIS, “Deliverable D1.4 METIS channel models,” METIS, DeliverableICT-317669-METIS/D1.4, Feb. 2015. [Online]. Available: www.metis2020.com/documents/deliverables/
[27] mmMagic, “6-100 GHz channel modelling for 5G: Measurementand modelling plans in mmMAGIC,” white paper, Feb. 2016. [On-line]. Available: bscw.5g-mmmagic.eu/pub/bscw.cgi/d76988/mmMAGIC_WhitePaper-W2.1.pdf
[28] MiWEBA, “D1.1: Definition of scenarios and use cases,” MiWEBA,Deliverable FP7-ICT 608636/D1.1, Dec. 2013. [Online]. Available:www.miweba.eu/#Publications
[29] Aalto University et al., “5G channel model for bands up to 100 GHz,” whitepaper, May 2016. [Online]. Available: www.5gworkshops.com/5GCM.html
[30] J. Järveläinen, K. Haneda, M. Kyrö, V.-M. Kolmonen, J. Takada, andH. Hagiwara, “60 GHz radio wave propagation prediction in a hospitalenvironment using an accurate room structural model,” in Proc. Lough-borough Antennas Propag. Conf. (LAPC), Loughborough, UK, Nov. 12-13,2012.
[31] A. V. Räisänen, J. Ala-Laurinaho, K. Haneda, J. Järveläinen, A. Kart-tunen, M. Kyrö, V. Semkin, A. Lamminen, and J. Säily, “Studies on e-bandantennas and propagation,” in Proc. Loughborough Antennas Propag. Conf.(LAPC), Loughborough, UK, Nov. 11-12, 2013, pp. 176–180.
[32] K. Haneda, J. Järveläinen, A. Karttunen, M. Kyrö, and J. Putkonen, “In-door short-range radio propagation measurements at 60 and 70 GHz,” inProc. 8th European Conf. Antennas Propag. (EuCAP), The Hague, Nether-lands, Apr. 6-11, 2014.
[33] J. Järveläinen, M. Kurkela, A. Karttunen, K. Haneda, and J. Putkonen,“70 GHz radio wave propagation prediction in a large office,” in Proc.Loughborough Antennas Propag. Conf. (LAPC), Loughborough, UK, Nov.10-11, 2014, pp. 420–424.
[34] K. Haneda, J. Järveläinen, A. Khatun, and K. i. Takizawa, “Spatial coexis-tence of millimeter-wave distributed indoor channels,” in Proc. IEEE 81stVeh. Technol. Conf. (VTC-Spring), Glasgow, Scotland, May 11-14, 2015, pp.1–5.
[35] S. Sun et al., “Propagation path loss models for 5G urban micro- andmacro-cellular scenarios,” in Proc. IEEE 83rd Veh. Technol. Conf. (VTC-Spring), Nanjing, China, May 15-18, 2016.
[36] K. Haneda, S. L. H. Nguyen, J. Järveläinen, and J. Putkonen, “Estimatingthe omni-directional pathloss from directional channel sounding,” in Proc.10th European Conf. Antennas Propag. (EuCAP), Davos, Switzerland, Apr.10-15, 2016.
73
References
[37] R. Naderpour, J. Vehmas, S. L. H. Nguyen, J. Järveläinen, and K. Haneda,“Spatio-temporal channel sounding in a street canyon at 15, 28 and 60GHz,” in 27th Ann. IEEE Int. Symp. Pers. Indoor Mobile Radio Commun.(PIMRC), 2016, to be published.
[38] J. Vehmas, J. Järveläinen, S. L. H. Nguyen, R. Naderpour, and K. Haneda,“Millimeter-wave channel characterization at Helsinki airport in the 15,28, and 60 GHz bands,” in Proc. IEEE 84th Veh. Technol. Conf. (VTC-Fall),2016, to be published.
[39] A. Karttunen, C. Gustafson, A. Molisch, J. Järveläinen, and K. Haneda,“Censored multipath component cross-polarization ratio modeling,” IEEETrans. Antennas Propag., 2016, submitted for publication.
[40] B. Langen, G. Lober, and W. Herzig, “Reflection and transmission be-haviour of building materials at 60 GHz,” in Proc. 5th IEEE Int. Symp.Pers. Indoor Mobile Radio Commun. (PIMRC), The Hague, the Nether-lands, Sept. 18-23, 1994, pp. 505–509 vol.2.
[41] J. Lu, D. Steinbach, P. Cabrol, P. Pietraski, and R. Pragada, “Propagationcharacterization of an office building in the 60 GHz band,” in Proc. 8thEuropean Conf. Antennas Propag. (EuCAP), The Hague, Netherlands, Apr.6-11, 2014, pp. 809–813.
[42] D. Ferreira, I. Cuinas, R. Caldeirinha, and T. Fernandes, “A review on theelectromagnetic characterisation of building materials at micro- and mil-limetre wave frequencies,” in Proc. 8th European Conf. Antennas Propag.(EuCAP), The Hague, Netherlands, Apr. 6-11, 2014, pp. 145–149.
[43] L. Correia and P. Frances, “Transmission and isolation of signals in build-ings at 60 GHz,” in Proc. 6th IEEE Int. Symp. Pers. Indoor Mobile RadioCommun. (PIMRC), vol. 3, Toronto, ON, Sept. 27-29, 1995, pp. 1031–1034.
[44] R. Felbecker, W. Keusgen, A. Kortke, and M. Peter, “Estimation of effectivepermittivity and effective thickness of inhomogeneous materials at 52 -70 GHz,” in Proc. 3rd European Conf. Antennas Propag. (EuCAP), Berlin,Germany, Mar. 23-27, 2009, pp. 3881–3885.
[45] H. Zhao, R. Mayzus, S. Sun, M. Samimi, J. K. Schulz, Y. Azar, K. Wang,G. N. Wong, F. Gutierrez, and T. S. Rappaport, “28 GHz millimeter wavecellular communication measurements for reflection and penetration lossin and around buildings in New York City,” in Proc. IEEE Int. Conf. Com-mun. (ICC), Budapest, Hungary, June 9-13, 2013, pp. 5163–5167.
[46] J. Ahmadi-Shokouh, S. Noghanian, E. Hossain, M. Ostadrahimi, and J. Di-etrich, “Reflection coefficient measurement for house flooring materials at57-64 GHz,” in Proc. IEEE Global Telecommun. Conf. (GLOBECOM), Hon-olulu, HI, Nov 30-Dec. 4, 2009, pp. 1–6.
[47] K. Sato, T. Manabe, T. Ihara, H. Saito, S. Ito, T. Tanaka, K. Sugai, N. Ohmi,Y. Murakami, M. Shibayama, Y. Konishi, and T. Kimura, “Measurementsof reflection and transmission characteristics of interior structures of of-fice building in the 60-GHz band,” IEEE Trans. Antennas Propag., vol. 45,no. 12, pp. 1783–1792, Dec. 1997.
74
References
[48] N. Moraitis and P. Constantinou, “Indoor channel measurements and char-acterization at 60 GHz for wireless local area network applications,” IEEETrans. Antennas Propag., vol. 52, no. 12, pp. 3180–3189, Dec. 2004.
[49] A. J. Motley and J. M. P. Keenan, “Personal communication radio coveragein buildings at 900 MHz and 1700 MHz,” Electronics Lett., vol. 24, no. 12,pp. 763–764, June 1988.
[50] P. Beckmann and A. Spizzichino, The Scattering of Electromagnetic WavesFrom Rough Surfaces. Oxford: Pergamon Press, 1963.
[51] D. Didascalou, M. Dottling, N. Geng, and W. Wiesbeck, “An approach toinclude stochastic rough surface scattering into deterministic ray-opticalwave propagation modeling,” IEEE Trans. Antennas Propa., vol. 51, no. 7,pp. 1508–1515, July 2003.
[52] S. Wang, “Effects of surface roughness on the indoor particle depositionrate constant,” Ph.D. dissertation, Syracuse Univ., Syracuse, NY, May2012.
[53] H. Morita, T. Yokosuka, A. Hatanaka, R. Takeuchi, Y. Dan, and M. Doi,“Electrostatic potential distribution characteristics of glass surfaces invacuums,” in Proc. 23rd Int. Sympo. Discharges Electr. Insulation Vacuum(ISDEIV), vol. 1, Bucharest, Romania, Sept. 15-19, 2008, pp. 54–57.
[54] R. Piesiewicz, C. Jansen, D. Mittleman, T. Kleine-Ostmann, M. Koch, andT. Kürner, “Scattering analysis for the modeling of THz communicationsystems,” IEEE Trans. Antennas Propag., vol. 55, no. 11, pp. 3002–3009,Nov. 2007.
[55] E. S. Li and K. Sarabandi, “Low grazing incidence millimeter-wave scat-tering models and measurements for various road surfaces,” IEEE Trans.Antennas Propag., vol. 47, no. 5, pp. 851–861, May 1999.
[56] V. Degli-Esposti, “A diffuse scattering model for urban propagation predic-tion,” IEEE Trans. Antennas Propag., vol. 49, no. 7, pp. 1111–1113, July2001.
[57] V. Degli-Esposti, F. Fuschini, E. M. Vitucci, and G. Falciasecca, “Measure-ment and modelling of scattering from buildings,” IEEE Trans. AntennasPropag., vol. 55, no. 1, pp. 143–153, Jan. 2007.
[58] T. Iwama, E. Moriyama, H. Ryuko, S. Sekizawa, and T. Sarauwatari, “In-vestigation of propagation characteristics above 1 GHz for microcellularland mobile radio,” in Proc. IEEE 40th Veh. Technol. Conf. (VTC), Orlando,FL, May 6-9, 1990, pp. 396–400.
[59] F. Niu and H. L. Bertoni, “Path loss and cell coverage of urban microcells inhigh-rise building environments,” in Proc. IEEE Globecom, Huoston, TX,Nov. 29-Dec. 2, 1993, pp. 266–270 vol.1.
[60] P. Smulders and L. Correia, “Characterisation of propagation in 60 GHzradio channels,” Electron. Commun. Eng. J., vol. 9, no. 2, pp. 73–80, Apr.1997.
75
References
[61] M. Jacob, S. Priebe, R. Dickhoff, T. Kleine-Ostmann, T. Schrader, andT. Kürner, “Diffraction in mm and sub-mm wave indoor propagation chan-nels,” IEEE Trans. Microw. Theory Techn., vol. 60, no. 3, pp. 833–844, Mar.2012.
[62] H. F. Bertoni, Radio Propagation for Modern Wireless Systems. UpperSaddle River, NJ: Prentice Hall, 2000.
[63] P. Holm, “A new heuristic UTD diffraction coefficient for nonperfectly con-ducting wedges,” IEEE Trans. Antennas Propag., vol. 48, no. 8, pp. 1211–1219, Aug. 2000.
[64] R. Kouyoumjian and P. Pathak, “A uniform geometrical theory of diffrac-tion for an edge in a perfectly conducting surface,” Proc. IEEE, vol. 62,no. 11, pp. 1448–1461, Nov. 1974.
[65] M. Steinbauer, A. Molisch, and E. Bonek, “The double-directional radiochannel,” IEEE Antennas Propag. Mag., vol. 43, no. 4, pp. 51–63, Aug.2001.
[66] B. H. Fleury, “First- and second-order characterization of direction disper-sion and space selectivity in the radio channel,” IEEE Trans. Inf. Theory,vol. 46, no. 6, pp. 2027–2044, Sept. 2000.
[67] G. D. Durgin, Space-time wireless channels. Upper Saddle River, NJ:Prentice Hall, 2003.
[68] H. C. Nguyen, G. R. MacCartney, T. Thomas, T. S. Rappaport, B. Vejlgaard,and P. Mogensen, “Evaluation of empirical ray-tracing model for an urbanoutdoor scenario at 73 GHz E-Band,” in Proc. IEEE 80th Veh. Technol.Conf. (VTC-Fall), Vancouver, BC, Sept. 14-17, 2014, pp. 1–6.
[69] A. Alighanbari and C. Sarris, “Rigorous and efficient time-domain mod-eling of electromagnetic wave propagation and fading statistics in indoorwireless channels,” IEEE Trans. Antennas Propag., vol. 55, no. 8, pp. 2373–2381, Aug. 2007.
[70] R. Burkholder and T.-H. Lee, “Adaptive sampling for fast physical opticsnumerical integration,” IEEE Trans. Antennas Propag., vol. 53, no. 5, pp.1843–1845, May 2005.
[71] I. Kavanagh, “Developing a method of moments based indoor propagationmodel,” in IEEE Int. Conf. Comp. Tool (EUROCON), Salamanca, Spain,Sept. 8-11, 2015, pp. 1–6.
[72] J.-H. Dou, X.-S. Yang, J. Chen, and B.-Z. Wang, “Performance analysis oftime-reversed ultra-wideband indoor channel by simulation,” in Int. Conf.Microw. Millimeter Wave Technol. (ICMMT), Chengdu, China, May 8-112010, pp. 1424–1427.
[73] U. Virk, K. Haneda, V.-M. Kolmonen, J.-F. Wagen, and P. Vainikainen,“Full-wave characterization of indoor office environment for accurate cov-erage analysis,” in Proc. Int. Conf. Electromagn. Adv. Appl. (ICEAA),Torino, Italy, Sept. 9-13, 2013, pp. 1197–1200.
76
References
[74] C.-P. Lim, J. Volakis, K. Sertel, R. Kindt, and A. Anastasopoulos, “Indoorpropagation models based on rigorous methods for site-specific multipathenvironments,” IEEE Trans. Antennas Propag., vol. 54, no. 6, pp. 1718–1725, June 2006.
[75] J. McKown and J. Hamilton, R.L., “Ray tracing as a design tool for radionetworks,” IEEE Netw., vol. 5, no. 6, pp. 27–30, Nov. 1991.
[76] R. Valenzuela, “A ray tracing approach to predicting indoor wireless trans-mission,” in Proc. IEEE 43rd Veh. Technol. Conf. (VTC), Secaucus, NJ, May18-20, 1993, pp. 214–218.
[77] Remcom (2016, Feb. 22). Wireless insight wireless EM propagationsoftware. [Online]. Available: www.remcom.com/wireless-insite
[78] Wavecall (2016, Feb. 22). Wavesight. [Online]. Available: wavecall.com/downloads/WaveSight_060330.pdf
[79] AWE Communications (2016, Feb. 22). Proman wave propagation andradio network planning. [Online]. Available: www.awe-communications.com/Products/ProMan
[80] Siradel (2016, Feb. 22). Volcano propagation model. [Online]. Available:www.siradel.com/wireless-3d-design/software/volcano-software/
[81] V. Degli-Esposti, D. Guiducci, A. de’Marsi, P. Azzi, and F. Fuschini, “Anadvanced field prediction model including diffuse scattering,” IEEE Trans.Antennas Propag., vol. 52, no. 7, pp. 1717–1728, July 2004.
[82] W. Peter, W. Keusgen, and R. Felbecker, “Measurement and ray-tracingsimulation of the 60 GHz indoor broadband channel: Model accuracy andparameterization,” in Proc. 2nd European Conf. Antennas Propag. (Eu-CAP), Edinburgh, UK, Nov. 11-16, 2007, pp. 1–8.
[83] J. Nuckelt, M. Schack, and T. Kürner, “Deterministic and stochastic chan-nel models implemented in a physical layer simulator for Car-to-X commu-nications,” Adv. Radio Sci., vol. 9, pp. 165–171, Aug. 2004.
[84] N. Zhang, J. Dou, L. Tian, X. Yuan, X. Yang, S. Mei, and H. Wang, “Dy-namic channel modeling for an indoor scenario at 23.5 GHz,” IEEE Access,vol. 3, pp. 2950–2958, Dec. 2015.
[85] Y. Wang, S. Safavi-Naeini, and S. Chaudhuri, “A hybrid technique basedon combining ray tracing and FDTD methods for site-specific modeling ofindoor radio wave propagation,” IEEE Trans. Antennas Propag., vol. 48,no. 5, pp. 743–754, May 2000.
[86] M. Thiel and K. Sarabandi, “3D-wave propagation analysis of indoor wire-less channels utilizing hybrid methods,” IEEE Trans. Antennas Propag.,vol. 57, no. 5, pp. 1539–1546, May 2009.
[87] Z. Yun and M. Iskander, “Ray tracing for radio propagation modeling: Prin-ciples and applications,” IEEE Access, vol. 3, pp. 1089–1100, 2015.
[88] Z. Yun, M. F. Iskander, S. Y. Lim, D. He, and R. Martinez, “Radio wavepropagation prediction based on 3-D building structures extracted from2-D images,” IEEE Antennas Wireless Propag. Lett., vol. 6, pp. 557–559,2007.
77
References
[89] S. Sun, T. S. Rappaport, T. A. Thomas, and A. Ghosh, “A preliminary 3Dmm wave indoor office channel model,” in Proc. Int. Conf. Comp., Netw.Commun. (ICNC), Garden Grove, CA, Feb. 16-19, 2015, pp. 26–31.
[90] B. Kim, H. Kim, D. Choi, Y. Lee, W. Hong, and J. Park, “28 GHz propa-gation analysis for passive repeaters in NLOS channel environment,” inProc. 9th European Conf. Antennas Propag. (EuCAP), Lisbon, Portugal,May 13-17, 2015, pp. 1–4.
[91] M.-T. Martínez-Inglés, D. Gaillot, J. Pascual-García, J.-M. Molina-García-Pardo, M. Lienard, and J.-V. Rodríguez, “Deterministic and experimentalindoor mmW channel modeling,” IEEE Antennas Wireless Propag. Lett.,vol. 13, pp. 1047–1050, May 2014.
[92] H. Yang, M. Herben, and P. Smulders, “Indoor radio channel fading analy-sis via deterministic simulations at 60 GHz,” in Proc. 3rd Int. Symp. Wire-less Commun. Syst. (ISWCS), Valencia, Spain, Sept. 6-8, 2006, pp. 144–148.
[93] M. Kazemi, A. Abdipur, and A. Mohammadi, “Indoor propagation MIMOchannel modeling in 60 GHz using SBR based 3D ray tracing technique,”in Proc. 2nd Conf. Millimeter-Wave and THz Technol. (MMWaTT) 2012,Tehran, Iran, Dec. 24-25, 2012, pp. 25–28.
[94] R. Felbecker, L. Raschkowski, W. Keusgen, and M. Peter, “Electromagneticwave propagation in the millimeter wave band using the NVIDIA OptiXGPU ray tracing engine,” in Proc. 6th European Conf. Antennas Propag.(EuCAP), Prague, Czech Republic, Mar. 26-30, 2012, pp. 488–492.
[95] S. Hur, S. Baek, B. Kim, Y. Chang, A. Molisch, T. Rappaport, K. Haneda,and J. Park, “Proposal on millimeter-wave channel modeling for 5G cellu-lar system,” IEEE J. Sel. Topics Signal Process., vol. PP, no. 99, pp. 1–1,2016.
[96] G. Rougeron, F. Gaudaire, Y. Gabillet, and K. Bouatouch, “Simulation ofthe indoor propagation of a 60 GHz electromagnetic wave with a time-dependent radiosity algorithm,” Comp. Graph., vol. 26, no. 1, pp. 125–141,Feb. 2002.
[97] B. Goktepe, M. Peter, R. Weiler, and W. Keusgen, “The influence of streetfurniture and tree trunks in urban scenarios on ray tracing simulationsin the millimeter wave band,” in Proc. European Microw. Conf. (EuMC),Paris, France, Sept. 7-10, 2015, pp. 195–198.
[98] Y. K. Yoon, M. w. Jung, and J. Kim, “Extraction of channel parameter onthe office in LOS indoor building environment,” in Proc. 12th Int. Conf.Adv. Commun. Technol. (ICACT), vol. 1, Phoenix Park, Korea, Feb. 7-10,2010, pp. 570–573.
[99] V. Degli-Esposti et al., “Ray-tracing-based mm-wave beamforming assess-ment,” IEEE Access, vol. 2, pp. 1314–1325, Oct. 2014.
[100] J. Pascual-García, J. M. Molina-García-Pardo, M. T. Martínez-Inglés, J. V.Rodríguez, and N. Saurin-Serrano, “On the importance of diffuse scatter-ing model parameterization in indoor wireless channels at mm-wave fre-quencies,” IEEE Access, vol. PP, no. 99, pp. 1–1, 2016.
78
References
[101] C. Gustafson, K. Haneda, S. Wyne, and F. Tufvesson, “On mm-wave mul-tipath clustering and channel modeling,” IEEE Trans. Antennas Propag.,vol. 62, no. 3, pp. 1445–1455, Mar. 2014.
[102] S. Guerin, “Indoor wideband and narrowband propagation measurementsaround 60.5 GHz in an empty and furnished room,” in Proc. IEEE 46thVeh. Technol. Conf. (VTC), vol. 1, Atlanta, GA, Apr. 28-May 1, 1996, pp.160–164 vol.1.
[103] S. Collonge, G. Zaharia, and G. El Zein, “Influence of the furniturere on60 GHz radio propagation in a residential environment,” in Proc. Int.Symp. Signals, Circuits and Syst., vol. 2, Iasi, Romania, July 10-11, 2003,pp. 413–416.
[104] J. S. Lu, P. Cabrol, D. Steinbach, and R. V. Pragada, “Measurement andcharacterization of various outdoor 60 GHz diffracted and scattered paths,”in Proc. IEEE Military Commun. Conf. (MILCOM), San Diego, CA, Nov.18-20, 2013, pp. 1238–1243.
[105] J. S. Lu, H. L. Bertoni, and V. Degli-Esposti, “Scale model investigation ofmechanisms for scattering from office buildings at 2 GHz,” IEEE Trans.Antennas Propag., vol. 62, no. 12, pp. 6435–6442, Dec. 2014.
[106] J. Medbo, H. Asplund, and J.-E. Berg, “60 GHz channel directional charac-terization using extreme size virtual antenna array,” in Proc. IEEE 26thAnnu. Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC), HongKong, Aug. 30-Sept. 2, 2015, pp. 176–180.
[107] M. Ghaddar, L. Talbi, G. Y. Delisle, and J. LeBel, “Deflecting-obstacle ef-fects on signal propagation in the 60 GHz band,” IEEE Trans. AntennasPropag., vol. 61, no. 1, pp. 403–414, Jan. 2013.
[108] P. F. M. Smulders, “Geometrical optics model for millimetre-wave indoorradio propagation,” Electron. Lett., vol. 29, no. 13, pp. 1174–1176, June1993.
[109] C.-P. Lim, M. Lee, R. J. Burkholder, J. L. Volakis, and R. J. Marhefka, “60GHz indoor propagation studies for wireless communications based on aray-tracing method,” EURASIP J. Wirel. Commun. Netw., vol. 2007, no. 1,pp. 49–49, Jan. 2007.
[110] B. Neekzad, K. Sayrafian-Pour, J. Perez, and J. Baras, “Comparison ofray tracing simulations and millimeter wave channel sounding measure-ments,” in Proc. IEEE 18th Int. Symp. Pers. Indoor Mobile Radio Commun.(PIMRC), Athens, Greece, Sept. 3-7, 2007, pp. 1–5.
[111] V. Mei, Y. Zhang, and H. Lin, “A novel 3D ray-tracing acceleration tech-nique based on Kd-tree algorithm for radio propagation prediction in com-plex indoor environment,” in Proc. Prog. Electromagn. Res. Symp. (PIERS),Guangzhou, China, Aug. 25-28, 2014, pp. 2164–2168.
[112] W. Keusgen, R. J. Weiler, M. Peter, M. Wisotzki, and B. Göktepe, “Propaga-tion measurements and simulations for millimeter-wave mobile access ina busy urban environment,” in Proc. 39th Int. Conf. Infrared, Millimeter,Terahertz Waves (IRMMW-THz), Tucson, AZ, Sept. 14-17, 2014, pp. 1–3.
79
References
[113] Z. Zhang, J. Ryu, S. Subramanian, and A. Sampath, “Coverage and chan-nel characteristics of millimeter wave band using ray tracing,” in IEEE Int.Conf. Commun. 2015 (ICC), London, UK, June 8-12, 2015, pp. 1380–1385.
[114] K.-H. Thiel and A. Wehr, “Performance capabilities of laser-scanners -an overview and measurement principle analysis,” in Proc. ISPRS work-ing group VIII/2 "Laser-scanners for Forest and Landscape Assesment",Freiburg, Germany, Oct. 3-6, 2004, pp. 14–18.
[115] Leica Geosystems, “HDS Laser Scanners & SW,” accessedJune 16th, 2016. [Online]. Available: hds.leica-geosystems.com/en/HDS-Laser-Scanners-SW_5570.htm
[116] J. Bohm and M. Pateraki, “From point samples to surfaces – on meshingand alternatives,” in Proc. ISPRS Commission V Symp. "Image Engineer-ing and Vision Metrology", Dresden, Germany, Sept. 25-27, 2012.
[117] F. Fuschini, S. Häfner, M. Zoli, R. Müller, E. M. Vitucci, D. Dupleich,M. Barbiroli, J. Luo, E. Schulz, V. Degli-Esposti, and R. S. Thomä, “Itemlevel characterization of mm-wave indoor propagation,” EURASIP J. Wire-less Commun. Netw., vol. 2016, no. 1, pp. 1–12, 2016.
[118] S. Collonge, G. Zaharia, and G. E. Zein, “Influence of the human activityon wide-band characteristics of the 60 GHz indoor radio channel,” IEEETrans. Wireless Commun., vol. 3, no. 6, pp. 2396–2406, Nov. 2004.
[119] M. Jacob, C. Mbianke, and T. Kürner, “A dynamic 60 GHz radio chan-nel model for system level simulations with MAC protocols for IEEE802.11ad,” in Proc. IEEE Int. Symp. Consumer Electron. (ISCE 2010),Braunschweig, Germany, June 7-10, 2010, pp. 1–5.
[120] D. Dupleich, F. Fuschini, R. Müller, E. Vitucci, C. Schneider, V. Degli-Esposti, and R. Thomä, “Directional characterization of the 60 GHz indoor-office channel,” in Proc. XXXIth URSI General Assembly Scientific Symp.(URSI GASS), Beijing, China, Aug. 16-23, 2014, pp. 1–4.
[121] U. Virk, J.-F. Wagen, and K. Haneda, “Simulating specular reflections forpoint cloud geometrical database of the environment,” in Proc. Loughbor-ough Antennas Propag. Conf. (LAPC), Loughborough, UK, Nov. 2-3, 2015.
[122] M. Kyrö, K. Haneda, J. Simola, K. Takizawa, H. Hagiwara, and P. Vainikai-nen, “Statistical channel models for 60 GHz radio propagation in hospitalenvironments,” IEEE Trans. Antennas Propag., vol. 60, no. 3, pp. 1569–1577, Mar. 2012.
[123] S. Y. Seidel, T. S. Rappaport, S. Jain, M. L. Lord, and R. Singh, “Path loss,scattering and multipath delay statistics in four European cities for digi-tal cellular and microcellular radiotelephone,” IEEE Trans. Veh. Technol.,vol. 40, no. 4, pp. 721–730, Nov. 1991.
[124] M. Zhu, G. Eriksson, and F. Tufvesson, “The COST 2100 channel model:Parameterization and validation based on outdoor MIMO measurementsat 300 MHz,” IEEE Trans. Wireless Commun., vol. 12, no. 2, pp. 888–897,Feb. 2013.
80
References
[125] S. Hur, Y.-J. Cho, T. Kim, J. Park, A. Molisch, K. Haneda, and M. Peter,“28 GHz channel modeling using 3D ray-tracing in urban environments,”in Proc. 9th European Conf. Antennas Propag. (EuCAP), Lisbon, Portugal,Apr. 13-17, 2015, pp. 1–5.
[126] T. Thomas, H. C. Nguyen, G. MacCartney, and T. Rappaport, “3D mmWavechannel model proposal,” in Proc. IEEE 80th Veh. Technol. Conf. (VTC-Fall), Vancouver, Canada, Sept. 14-17, 2014, pp. 1–6.
[127] T. Kürner and M. Jacob, “Application of ray tracing to derive channel mod-els for future multi-gigabit systems,” in Proc. Int. Conf. ElectromagneticsAdv. App. (ICEAA ’09), Torino, Italy, Sept. 14-18, 2009, pp. 517–520.
[128] Y. Chang, S. Baek, S. Hur, Y. Mok, and Y. Lee, “A novel dual-slope mm-wavechannel model based on 3D ray-tracing in urban environments,” in Proc.IEEE 25th Ann. Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC),Washington DC, Sept. 2-5, 2014, pp. 222–226.
[129] M. Hata, “Empirical formula for propagation loss in land mobile radio ser-vices,” IEEE Trans. Veh. Technol., vol. 29, no. 3, pp. 317–325, Aug. 1980.
[130] COST Action 231, “Digital mobile radio towards future generation sys-tems,” European Commission, Final report, 1999.
[131] 3GPP, “Spacial channel model for MIMO simulations,” ETSI, Tech.Rep. TR 25.996 v6.1.0, Sept. 2003. [Online]. Available: www.3gpp.org/DynaReport/25996.htm
[132] D. Baum et al., “IST-2003-507581 WINNER D5.4 v.1.4, Final Reporton Link Level and System Level Channel Models,” IST-WINNER, Tech.Rep., Nov. 2005. [Online]. Available: www.ist-winner.org/deliverables_older.html
[133] D. S. Baum, J. Hansen, J. Salo, G. Del Galdo, M. Milojevic, and P. Kyösti,“An interim channel model for beyond-3G systems: extending the 3GPPspatial channel model (SCM),” in Proc. IEEE 61st Veh. Technol. Conf.(VTC-Spring), vol. 5, Stockholm, Sweden, May 30-June 1, 2005, pp. 3132–3136 Vol. 5.
[134] P. Kyösti et al., “IST-4-027756 WINNER II deliverable 1.1.2. v.1.2,WINNER II channel models,” IST-WINNER2, Tech. Rep., July 2007.[Online]. Available: www.ist-winner.org/deliverable.html
[135] P. Heino et al., “D5.3: WINNER+ Final Channel Models,” CELTIC, Tech.Rep., June 2010. [Online]. Available: projects.celtic-initiative.org/winner+/deliverables_winnerplus.html
[136] ITU-R, “Guidelines for evaluation of radio interface technologies for IMT-Advanced,” ITU, Rep. ITU-R M.2135-1, Dec. 2009. [Online]. Available:www.itu.int/pub/R-REP-M.2135-1-2009
[137] S. Jaeckel, L. Raschkowski, K. Borner, and L. Thiele, “QuaDRiGa: A 3-D multi-cell channel model with time evolution for enabling virtual fieldtrials,” IEEE Trans. Antennas Propag., vol. 62, no. 6, pp. 3242–3256, June2014.
81
References
[138] J. Medbo, M. Riback, and J. E. Berg, “Validation of 3GPP spatial channelmodel including WINNER wideband extension using measurements,” inProc. IEEE 64th Veh. Technol. Conf. (VTC-Fall), Montreal, QC, Sept. 25-28, 2006, pp. 1–5.
[139] C. Schneider, M. Narandzic, M. Kaske, G. Sommerkorn, and R. S. Thoma,“Large scale parameter for the WINNER II channel model at 2.53 GHz inurban macro cell,” in Proc. IEEE 71st Veh. Technol. Conf. (VTC-Spring),Taipei, Taiwan, May 16-19, 2010, pp. 1–5.
[140] L. Liu, C. Oestges, J. Poutanen, K. Haneda, P. Vainikainen, F. Quitin,F. Tufvesson, and P. D. Doncker, “The COST 2100 MIMO channel model,”IEEE Wireless Commun., vol. 19, no. 6, pp. 92–99, Dec. 2012.
[141] R. Verdone and A. Zanella, Eds., Pervasive mobile and ambient wirelesscommunications, COST Action 2100. London, UK: Springer, 2012.
[142] J. Poutanen, K. Haneda, L. Liu, C. Oestges, F. Tufvesson, and P. Vainikai-nen, “Parameterization of the COST 2100 MIMO channel model in indoorscenarios,” in Proc. 5th European Conf. Antennas Propag. (EuCAP), Rome,Italy, Apr. 11-15, 2011, pp. 3606–3610.
[143] V. Nurmela and P. Kyösti, “A spatially consistent radio channel model en-abling dual mobility,” in Proc. IEEE 80th Veh. Technol. Conf. (VTC-Fall),Vancouver, BC, Sept. 14-17, 2014, pp. 1–5.
[144] A. Maltsev et al., “Channel models for 60 GHz WLAN systems,” IEEE802.11ad, doc. IEEE 802.11-09/0334r8, May 2010. [Online]. Available:www.ieee802.org/11/Reports/tgad_update.htm
[145] A. Maltsev, R. Maslennikov, A. Sevastyanov, A. Lomayev, and A. Khoryaev,“Statistical channel model for 60 GHz WLAN systems in conference roomenvironment,” in Proc. 4th European Conf. Antennas Propag. (EuCAP),Barcelona, Spain, Apr. 12-16, 2010, pp. 1–5.
[146] J. Medbo, P. Kyösti, K. Kusume, L. Raschkowski, K. Haneda, T. Jämsä,V. Nurmela, A. Roivainen, and J. Meinilä, “Radio propagation modeling for5G mobile and wireless communications,” IEEE Commun. Mag., vol. 54,no. 6, pp. 144–151, June 2016.
[147] I. Carton, W. Fan, P. Kyösti, and G. F. Pedersen, “Validation of 5G METISmap-based channel model at mmwave bands in indoor scenarios,” in Proc.10th European Conf. Antennas Propag. (EuCAP), Davos, Switzerland, Apr.10-15, 2016.
[148] MiWEBA, “D5.1: Channel modeling and characterization,” MiWEBA,Deliverable FP7-ICT 608637/D5.1, June 2014. [Online]. Available:www.miweba.eu/#Publications
[149] A. Maltsev, A. Sadri, A. Pudeyev, I. Bolotin, A. Davydov, G. Morozov, andR. Weiler, “Partially adaptive arrays application for MU-MIMO mode in aMmWave small cells,” in Proc. IEEE 26th Annu. Int. Symp. Pers. IndoorMobile Radio Commun. (PIMRC), Hong Kong, Aug. 30-Sept. 2, 2015, pp.315–319.
82
References
[150] L. Hentilä, P. Kyösti, M. Käske, M. Narandzic, and M. Alatossava (2007,Dec.). MATLAB implementation of the WINNER phase II channel modelver1.1. [Online]. Available: www.ist-winner.org/phase_2_model.html
[151] G. Wickizer, “Field-strength survey, 52.75 megacycles from Empire StateBuilding,” Proc. IRE, vol. 28, no. 7, pp. 291–296, July 1940.
[152] D. Cox, “910 MHz urban mobile radio propagation: Multipath characteris-tics in New York City,” IEEE Trans. Commun., vol. 21, no. 11, pp. 1188–1194, Nov. 1973.
[153] D. Kitchener, J. Dalley, and M. Smith, “Antenna diversity in outdoor mi-crocells,” in 9th Int. Conf. Antennas Propag., vol. 2, Eindhoven, the Nether-lands, Apr. 4-7, 1995, pp. 1–4 vol.2.
[154] J. C. Wiltse, “History of millimeter and submillimeter waves,” IEEE Trans.Microw. Theory Techn., vol. 32, no. 9, pp. 1118–1127, Sept. 1984.
[155] B. Rembold, “Communication systems at millimeter waves,” in Proc. 22ndEur. Microw. Conf., vol. 1, Helsinki, Finland, Sept. 5-7, 1992, pp. 53–72.
[156] J. Lähteenmäki and T. Karttaavi, “Measurement of dielectric parametersof wall materials at 60 GHz band,” Electronics Lett., vol. 32, no. 16, pp.1442–1444, Aug. 1996.
[157] M. Bensebti, J. McGeehan, and M. Beach, “Indoor multipath radio propa-gation measurements and characterisation at 60 GHz,” in Proc. 21st Eu-ropean Microw. Conf., vol. 2, Stuttgart, Germany, Sept. 9-12, 1991, pp.1217–1222.
[158] M. Sanchéz, A. Hammoudeh, E. Grindrod, and J. Kermoal, “Coherencebandwidth characterization in an urban microcell at 62.4 GHz,” IEEETrans. Veh. Technol., vol. 49, no. 2, pp. 607–613, Mar. 2000.
[159] M. Kyrö, “Radio wave propagation and antennas for millimeter-wave com-munications,” Ph.D. dissertation, Aalto Univ., Helsinki, Finland, 2012.
[160] W. Newhall, “Wideband propagation measurement results, simulationmodels, and processing techniques for a sliding correlator measurementsystem,” Master’s thesis, Virginia Polytech. Univ., Blacksburg, VA, 1997.
[161] T. Zwick, T. Beukema, and H. Nam, “Wideband channel sounder with mea-surements and model for the 60 GHz indoor radio channel,” IEEE Trans.Veh. Technol., vol. 54, no. 4, pp. 1266–1277, July 2005.
[162] H. Xu, V. Kukshya, and T. Rappaport, “Spatial and temporal character-istics of 60-GHz indoor channels,” IEEE J. Sel. Areas Commun., vol. 20,no. 3, pp. 620–630, Apr. 2002.
[163] S. Collogne, G. Zaharia, and G. El Zein, “Experimental investigation ofthe spatial and temporal characteristics of the 60 GHz radio propagationwithin residential environments,” in Proc. Int. Symp. Signals, Circuits andSyst., vol. 2, Iasi, Romania, July 10-11, 2003, pp. 417–420 vol.2.
[164] X. Yin, C. Ling, and M.-D. Kim, “Experimental multipath-cluster charac-teristics of 28-GHz propagation channel,” IEEE Access, vol. 3, pp. 3138–3150, Dec. 2015.
83
References
[165] E. Lemos Cid, M. Garcia Sanchez, and A. Vazquez Alejos, “High speedtransmission at 60 GHz for 5G communications,” in IEEE Int. Symp. An-tennas Propag. USNC/URSI Nat. Radio Sci. Meeting, Vancouver, BC, July19-24, 2015, pp. 1007–1008.
[166] T. Rappaport, F. Gutierrez, E. Ben-Dor, J. Murdock, Y. Qiao, and J. Tamir,“Broadband millimeter-wave propagation measurements and models us-ing adaptive-beam antennas for outdoor urban cellular communications,”IEEE Trans. Antennas Propag., vol. 61, no. 4, pp. 1850–1859, Apr. 2013.
[167] P. Smulders and A. Wagemans, “Frequency-domain measurement of themillimeter wave indoor radio channel,” IEEE Trans. Instrum. Meas.,vol. 44, no. 6, pp. 1017–1022, Dec. 1995.
[168] J. Kivinen, “Development of wideband radio channel measurementand modeling techniques for future radio systems,” Ph.D. dissertation,Helsinki Univ. Technol., Helsinki, Finland, 2001.
[169] V. Guillet, “Narrowband and wideband characteristics of 60 GHz radio pro-pagation in residential environment,” Electron. Lett., vol. 37, no. 21, pp.1310–1311, Oct. 2001.
[170] A. Siamarou, “Broadband wireless local-area networks at millimeterwaves around 60 GHz,” IEEE Antennas Propag. Mag., vol. 45, no. 1, pp.177–181, Feb. 2003.
[171] C. Liu, E. Skafidas, and R. Evans, “Characterization of the 60 GHz wirelessdesktop channel,” IEEE Trans. Antennas Propag., vol. 55, no. 7, pp. 2129–2133, July 2007.
[172] S. Ranvier, J. Kivinen, and P. Vainikainen, “Millimeter-wave MIMO radiochannel sounder,” IEEE Trans. Instrum. Meas., vol. 56, no. 3, pp. 1018–1024, June 2007.
[173] J. Ahmadi-Shokouh, R. Rafi, A. Taeb, and S. Safavi-Naeini, “Real-time millimeter-wave MIMO channel measurements,” in IEEE AntennasPropag. Soc. Int. Symp. (APSURSI), Chicago, IL, July 8-14, 2012, pp. 1–2.
[174] M.-T. Martínez-Inglés, C. Sanchis-Borras, J.-M. Molina-García-Pardo, J.-V. Rodríguez, and L. Juan-Llácer, “Experimental evaluation of an indoorMIMO-OFDM system at 60 GHz based on the IEEE802.15.3c standard,”IEEE Antennas Wireless Propag. Lett., vol. 12, pp. 1562–1565, Nov. 2013.
[175] M. Lei, J. Zhang, T. Lei, and D. Du, “28-GHz indoor channel measure-ments and analysis of propagation characteristics,” in Proc. IEEE 25thAnn. Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC), Washing-ton DC, Sept. 2-5, 2014, pp. 208–212.
[176] J. H. Kim, M.-W. Jung, Y. K. Yoon, Y. J. Chong, and M. S. Song, “60 and28 GHz delay spread measurements and simulation at indoor,” in Proc.Int. Conf. Inf. Commun. Technol. Convergence (ICTC), Busan, Korea, Oct.22-24, 2014, pp. 148–150.
[177] N. Zhang, X. Yin, S. Lu, M. Du, and X. Cai, “Measurement-based angu-lar characterization for 72 GHz propagation channels in indoor environ-ments,” in Proc. Globecom Workshops, Austin, TX, Dec. 8-12, 2014, pp.370–376.
84
References
[178] R. Müller, R. Herrmann, D. Dupleich, C. Schneider, and R. Thomä, “Ul-trawideband multichannel sounding for mm-wave,” in Proc. 8th EuropeanConf. Antennas Propag. (EuCAP), The Hague, the Netherlands, Apr. 6-11,2014, pp. 817–821.
[179] S. Ranvier, M. Kyrö, K. Haneda, T. Mustonen, C. Icheln, and P. Vaini-kainen, “VNA-based wideband 60 GHz MIMO channel sounder with 3-Darrays,” in Proc. IEEE Radio Wireless Symp. (RWS ’09), San Diego, CA,Jan. 18-22, 2009, pp. 308–311.
[180] N. Moraitis and P. Constantinou, “Millimeter wave propagation measure-ments and characterization in an indoor environment for wireless 4G sys-tems,” in Proc. IEEE 16th Int. Symp. Pers. Indoor Mobile Radio Commun.(PIMRC), vol. 1, Berlin, Germany, Sept. 11-14, 2005, pp. 594–598.
[181] M.-S. Choi, G. Grosskopf, and D. Rohde, “A novel wideband space-timechannel measurement method at 60 GHz,” in Proc. IEEE 16th Int. Symp.Pers. Indoor Mobile Radio Commun. (PIMRC), vol. 1, Berlin, Germany,Sept. 11-14, 2005, pp. 584–588.
[182] A. Maltsev, R. Maslennikov, A. Sevastyanov, A. Khoryaev, and A. Lomayev,“Experimental investigations of 60 GHz WLAN systems in office environ-ment,” IEEE J. Sel. Areas Commun., vol. 27, no. 8, pp. 1488–1499, 2009.
[183] M. Kim, K. Umeki, K. Wangchuk, J. i. Takada, and S. Sasaki, “Polarimetricmm-wave channel measurement and characterization in a small office,” inProc. IEEE 26th Annu. Int. Symp. Pers. Indoor Mobile Radio Commun.(PIMRC), Hong Kong, Aug. 30-Sept. 2, 2015, pp. 764–768.
[184] G. Weibel and H. Dressel, “Propagation studies in millimeter-wave linksystems,” IEEE Proc., vol. 55, no. 4, pp. 497–513, Apr. 1967.
[185] E. Violette, R. Espeland, R. DeBolt, and F. Schwering, “Millimeter-wavepropagation at street level in an urban environment,” IEEE Trans. Geosci.Remote Sens., vol. 26, no. 3, pp. 368–380, May 1988.
[186] G. Lovnes, J. Reis, and R. Raekken, “Channel sounding measurements at59 GHz in city streets,” in Proc. 5th IEEE Int. Symp. Pers. Indoor MobileRadio Commun. (PIMRC), The Hague, the Netherlands, Sept. 18-23, 1994,pp. 496–500 vol.2.
[187] H. Xu, T. Rappaport, R. Boyle, and J. Schaffner, “38 GHz wideband point-to-multipoint radio wave propagation study for a campus environment,” inProc. IEEE 49th Veh. Technol. Conf. (VTC), vol. 2, Houston, TX, May 16-20,1999, pp. 1575–1579 vol.2.
[188] L. Talbi and G. Delisle, “Wideband propagation measurements and mod-eling at millimeter wave frequencies,” in Proc. IEEE Globecom, San Fran-sisco, CA, Nov. 28-Dec. 2, 1994, pp. 47–51 vol.1.
[189] A. Hammoudeh and G. Allen, “Millimetric wavelengths radiowave propa-gation for line-of-sight indoor microcellular mobile communications,” IEEETrans. Veh. Technol., vol. 44, no. 3, pp. 449–460, Aug. 1995.
[190] G. Kalivas, M. El-Tanany, and S. Mahmoud, “Millimeter-wave channelmeasurements with space diversity for indoor wireless communications,”IEEE Trans. Veh. Technol., vol. 44, no. 3, pp. 494–505, Aug. 1995.
85
References
[191] A. Kato, T. Manabe, Y. Miura, K. Sato, and T. Ihara, “Measurements of mil-limeter wave indoor propagation and high-speed digital transmission char-acteristics at 60 GHz,” in Proc. 8th IEEE Int. Symp. Personal, Indoor Mo-bile Radio Commun. (PIMRC), vol. 1, Helsinki, Finland, Sept. 1-4, 1997,pp. 149–154.
[192] METIS, “Deliverable D1.1 scenarios, requirements and KPIs for 5G mobileand wireless system,” METIS, Deliverable ICT-317669-METIS/D1.1, Apr.2013. [Online]. Available: www.metis2020.com/documents/deliverables/
[193] MiWaveS, “Heterogeneous wireless network with mmWave small cellaccess and backhauling,” white paper, Jan. 2015. [Online]. Available:www.miwaves.eu/publications.html
[194] Millimetre Wave Transmission (mWT); Applications and use cases of mil-limetre wave transmission, ETSI Std. GS mWT 002, 2015.
[195] G. R. MacCartney, T. S. Rappaport, S. Sun, and S. Deng, “Indoor officewideband millimeter-wave propagation measurements and channel mod-els at 28 and 73 GHz for ultra-dense 5G wireless networks,” IEEE Access,vol. 3, pp. 2388–2424, 2015.
[196] M.-D. Kim, J. Liang, H.-K. Kwon, and J. Lee, “Directional delay spreadcharacteristics based on indoor channel measurements at 28 GHz,” in Proc.26th IEEE Ann. Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC),Hong Kong, Aug. 30-Sept. 2, 2015, pp. 403–407.
[197] A. R. M. Sidek and A. Z. Sha’ameri, “Markov model for performance es-timation of adaptive channel coding schemes with various acknowledg-ment strategies,” in Proc. Int. Conf. Comp. Commun. Eng. (ICCCE), KualaLumpur, Malaysia, May 11-12, 2010, pp. 1–5.
[198] Siemens, “Simulation results of TFCI coding performance for TDD,” 3GPP,Tech. Rep. TSGR1#6(99)865, 1999.
[199] M. Kyrö, K. Takizawa, K. Haneda, and P. Vainikainen, “Validation ofstatistical channel models for 60 GHz radio systems in hospital environ-ments,” IEEE Trans. Biomedical Eng., vol. 60, no. 5, pp. 1458–1462, May2013.
[200] L. R. Ximenes and A. L. F. de Almeida, “Performance evaluation of mobilephone antennas in physical MIMO channels using polarized spherical har-monic decomposition,” in Proc. IEEE 72nd Veh. Technol. Conf. (VTC-Fall),San Francisco, CA, Sept. 5-8, 2011, pp. 1–5.
[201] F. Harrysson, J. Medbo, A. F. Molisch, A. J. Johansson, and F. Tufves-son, “Efficient experimental evaluation of a MIMO handset with user in-fluence,” IEEE Trans. Wireless Commun., vol. 9, no. 2, pp. 853–863, Feb.2010.
[202] X. Lu, E. Kunnari, J. Leinonen, O. Piirainen, M. Vainikka, and M. Juntti,“LTE uplink power control and base station antenna down tilt in a 3Dchannel model,” in Proc. European Wireless Conf. (EW), Lucca, Italy, Apr.12-15, 2010, pp. 377–381.
86
References
[203] S. Suo, Y. Yang, Y. Tang, and Y. Wang, “Evaluation of MIMO structure andchannel model in LTE system,” in Proc. Fourth Int. Conf. Commun. Netw.China (ChinaCOM), Xian, China, Aug. 26-28, 2009, pp. 1–5.
[204] M. R. Akdeniz, Y. Liu, M. K. Samimi, S. Sun, S. Rangan, T. S. Rappaport,and E. Erkip, “Millimeter wave channel modeling and cellular capacityevaluation,” IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1164–1179,June 2014.
[205] J. Rodríguez-Pineiro, P. Suárez-Casal, M. Lerch, S. Caban, J. A. García-Naya, L. Castedo, and M. Rupp, “LTE downlink performance in high speedtrains,” in Proc. IEEE 81st Veh. Technol. Conf. (VTC-Spring), Glasgow,Scotland, May 11-14, 2015, pp. 1–5.
[206] L. Hentilä, P. Kyösti, and P. Heino, “Evaluation of beam forming and multiantenna techniques in non-stationary propagation scenarios with HW em-ulator,” in Proc. Int. ITG Workshop Smart Antennas (WSA), Dresden, Ger-many, Mar. 7-8, 2012, pp. 347–351.
[207] Z. Al-Daher, L. P. Ivrissimtzis, and A. Hammoudeh, “Outdoor coverage &capacity estimation in small cells with a 3D ray tracing model,” in Proc.Loughborough Antennas Propag. Conf. (LAPC), Loughborough, UK, Nov.11-23, 2013, pp. 31–36.
[208] N. F. Abdullah, A. A. Goulianos, T. H. Barratt, A. G. L. Freire, D. E.Berraki, S. M. D. Armour, A. R. Nix, and M. A. Beach, “Path-loss andthroughput prediction of IEEE 802.11ad systems,” in Proc. IEEE 81st Veh.Technol. Conf. (VTC-Spring), Glasgow, Scotland, May 11-14, 2015, pp. 1–5.
[209] D. Dardari and V. Tralli, “High-speed indoor wireless communications at60 GHz with coded OFDM,” IEEE Trans. Commun., vol. 47, no. 11, pp.1709–1721, Nov. 1999.
[210] J. Hansen, “A novel stochastic millimeter-wave indoor radio channelmodel,” IEEE J. Sel. Areas Commun., vol. 20, no. 6, pp. 1240–1246, Aug.2002.
[211] J. Góra and K. Safjan, “Design and optimization aspects of wireless back-haul operated in the millimeter wave spectrum,” in Proc. IEEE 25th Ann.Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC), Washington DC,Sept. 2-5, 2014, pp. 1575–1579.
[212] T. A. Thomas, F. W. Vook, E. Visotsky, and S. Sun, “System-level perfor-mance of different array types for an indoor mmWave system,” in Proc.IEEE 82nd Veh. Technol. Conf. (VTC-Fall), Boston, MA, Sept. 6-9, 2015,pp. 1–5.
[213] S. Zhang, D. Kong, E. Mellios, G. Hilton, A. Nix, T. A. Thomas, andA. Ghosh, “Impact of BS antenna number and array geometry on single-user LTE-A data throughputs in realistic macro and pico cellular envi-ronments,” in Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), NewOrleans, LA, Mar. 9-12, 2015, pp. 1464–1469.
[214] T. Yilmaz and O. B. Akan, “On the use of low terahertz band for 5G indoormobile networks,” Comput. Electr. Eng., vol. 48, pp. 164 – 173, 2015.
87
Errata
Publication VII
In Table II, the parameters for Eqs. (3) and (4) should be swapped. In Fig.
3, the exponential model should refer to (4), and the linear model should
refer to (3).
88
This thesis focuses on mm-wave channel modeling for future 5G wireless communication systems. The main contributions of the work include simulations tools and insights acquired through channel measurements. The work emphasize the need for more detailed descriptions of the model frameworks and environment descriptions at mm-waves compared to lower frequencies.
Aalto-D
D 16
4/2
016
9HSTFMG*agjhbf+
ISBN 978-952-60-6971-5 (printed) ISBN 978-952-60-6972-2 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Electrical Engineering Department of Radio Science and Engineering www.aalto.fi
BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS
Jan Järveläinen M
easurement-Based M
illimeter-W
ave Radio C
hannel Simulations and M
odeling A
alto U
nive
rsity
2016
Department of Radio Science and Engineering
Measurement-Based Millimeter-Wave Radio Channel Simulations and Modeling
Jan Järveläinen
DOCTORAL DISSERTATIONS