BRNO UNIVERSITY OF TECHNOLOGYVYSOKÉ UČENÍ TECHNICKÉ V BRNĚ

FACULTY OF ELECTRICAL ENGINEERING ANDCOMMUNICATIONFAKULTA ELEKTROTECHNIKYA KOMUNIKAČNÍCH TECHNOLOGIÍ

DEPARTMENT OF RADIO ELECTRONICSÚSTAV RADIOELEKTRONIKY

INTRA- AND OUT-OF-VEHICLE CHANNELMEASUREMENTS AND MODELINGMĚŘENÍ A MODELOVÁNÍ KANÁLŮ UVNITŘ A VNĚ VOZIDEL

DOCTORAL THESISDIZERTAČNÍ PRÁCE

AUTHORAUTOR PRÁCE

Pavel Kukolev

SUPERVISORŠKOLITEL

prof. Ing. Aleš Prokeš, Ph.D.

BRNO 2016

ABSTRACTThe dissertation is focused on channel measurements and modeling for vehicle-to-Xcommunication and on localization. In order to realize an integrated intelligent trans-portation system (ITS), it is important to estimate channel features for intra-vehicle andout-of-vehicle scenarios. For this propose the following activities are carried out: simula-tion of the 802.11p PHY; comparison with 802.11a; channel measurements for differentscenarios based on the 802.11p and ultra-wideband (UWB); creating channel models for802.11p and UWB; UWB measurements to assess performance of localization.

• The vehicle-to-X communication is supposed on the IEEE 802.11p standard. Thedissertation presents the differences between IEEE 802.11a and IEEE 802.11pphysical layer standards through the simulation results of the transmission overa HIPERPLAN/2 channel. Further, the simulation of the 802.11p signal transmis-sion over ITU-R M.1225 channel, which includes pedestrian and vehicle modelswith different relative delays and average power, is presented. The influence of thechannel on the signal is analyzed using MATLAB simulation in terms of bit errorrate (BER).

• The dissertation reports vehicular channel measurements in the frequency band of5.8 GHz for IEEE 802.11p standard and for UWB (3-11 GHz). Experiments forboth intra-vehicle and out-of-vehicle environments are carried out. It was observedthat the large-scale variations (LSVs) of the power delay profiles (PDPs) can bebest approximated through a two-term exponential decay model for the 802.11pprotocol, in contrast to the Saleh-Valenzuela (S-V) model which is suitable forUWB systems.

• For each measurement, the LSV trend was used to construct the respective channelimpulse response (CIR). Next, the CIR is used in 802.11p simulation to evaluatethe BER performance, following a Rician model. The results of the BER simula-tion shows the suitability of the protocol for in-car as well as out-of-car wirelessapplications. The simulation for out-of-car parameters indicate that the error per-formances do not vary much and it is possible to determine an average BER curvefor the whole set of data.

• The randomness in UWB channel for small positional variations around a car,parked in an underground garage, is reported. The path loss (PL) is found to bemonotonically increasing with distance but varies randomly with angle and heightand thereby renders signal strength based ranging inaccurate for such scenarios.On the other hand, arrival time of the first ray can be used for reliable estimation ofdistance, independent on transmitter angle or height. The number of clusters in thePDP is reduced with distance but the nature of the profile remains fairly consistentwith angle. The S-V model parameters also vary with distance and height but theiraverage values are close to the IEEE 802.15.3 recommended channel model.

• For localization applications the distance between the antennas is calculated ex-ploiting the linear dependence of distance on delay from PDP. The coordinates ofa transmitting antenna are found with the help of two receiving antennas follow-ing a two-dimensional (2-D) time-of-arrival (TOA) based localization technique.A comparison of the calculated coordinates with the original ones exhibits an er-ror of less than 6% which supports the suitability of the proposed approach forlocalization of the cars.

KEYWORDSIntra-vehicle channel; Out-of-vehicle channel; IEEE 802.11p; Channel measurement;Ultra-wideband; Power delay profile; Bit error rate; Saleh-Valenzuela model; Localiza-tion; Time of arrival.

ABSTRAKTDisertační práce je zaměřena na měření a modelování kanálu uvnitř a vně vozidla prokomunikaci a lokalizaci. Pro účely vytvoření integrovaného inteligentního dopravníhosystému ITS (Intelligent transportation system) je důležitý odhad vlastnosti kanálů provnitřní a venkovní scénáře. Za tímto účelem je vhodné provést řadu činností, které jsouobsahem disertační práce: Simulace fyzické vrstvy 802.11p, její srovnávání s 802.11a,měření kanálu pro různé scénáře pro 802.11p a pro širokopásmový systém (UWB),vytvoření modelů kanálů pro 802.11p a UWB a výzkum vlastností lokalizace založené naměření v pásmu UWB.

• Výzkum komunikace vozidla s okolím založená na IEEE 802.11p standardu. Jed-ním z cílů disertační práce je ukázat rozdíly mezi standardy fyzické vrstvy IEEE802.11a a IEEE 802.11p prostřednictvím simulace s použitím modelu kanáluHIPERPLAN/2. V práci je uvedena simulace přenosu signálu 802.11p kanálemITU-R M.1225 s odlišným zpožděním a středním výkonem (pro chodce a vozidla).Vliv kanálu na signál je analyzován za použití simulace v prostředí MATLABupomocí vyhodnocení chybovosti.

• Určení vlastností kanálů v kmitočtovém pásmu 5,8 GHz pro standard IEEE 802.11pa UWB. Experimenty byly prováděny pro vnitřní a vnější prostředí vozidla. Bylozjištěno, že pro protokol 802.11p může být trend (dlouhodobý vývoj) profilu PDP(power delay profile) nejlépe aproximován pomocí modelu obsahujícího dvě klesajícíexponenciální funkce, na rozdíl od Saleh-Valenzuelova (S-V) modelu, který je vícevhodný pro UWB systémy pracující v pásmu 3 až 11 GHz.

• Vytvoření odpovídající impulzní odezvy (CIR) s využitím trendu PDP. Informaceo CIR byla použita pro simulaci 802.11p za účelem vyhodnocení chybovosti připoužití Ricianova modelu. Výsledky odhadu BER ukazují vhodnost protokolu provnitřní a vnější prostředí bezdrátových aplikací. Výsledky simulací dále ukazují, žese chybovost zásadně nemění a proto je možné určit střední křivku BER pro celousadu změřených dat.

• Určení vlivu malé změny polohy antény na vlastnosti kanálu. Práce ukazuje náhod-nost parametrů UWB kanálu pro malé změny polohy antény okolo vozidla, zaparko-vaného v podzemní garáži. Ztráty šířením jsou monotónně rostoucí se vzdáleností,avšak náhodně se mění v závislosti na úhlu a výšce antén, a proto je vyhodno-cení vzdálenosti pomocí síly signálu pro tyto scénáře nevhodné. Na druhé straněmůže být pro spolehlivé určení vzdálenosti bez ohledu na úhel nebo výšku anténypoužita doba příchodu prvního svazku. Ověření vlivu změn konfigurace kanálu naparametry S-V modelu. Práce demonstruje závislost parametrů Saleh-Valenzuelamodelu v na vzdálenosti a výšce antén, avšak ukazuje, že jejich průměrné hodnotyjsou blízké IEEE 802.15.3 standardu.

• Ověření možnosti lokalizace pomocí metody TOA (time of arrival). Vzdálenostmezi anténami byla určena z profilu PDP s využitím lineární závislosti vzdálenostina zpoždění. Souřadnice vysílací antény byly nalezeny pomocí dvou přijímacíchantén pomocí 2-D lokalizační techniky TOA. Porovnání vypočtených souřadnics původními vykazuje chybu menší než 6%, což ukazuje vhodnost navrženéhopřístupu pro lokalizaci vozidel.

KLÍČOVÁ SLOVAKanal uvnitř vozidel; Kanal ve vozidel; IEEE 802.11p; Měření kanalu; UWB; PDP; BER;Saleh-Valenzuelova model; Localizace; TOA.

KUKOLEV, Pavel Intra- and out-of-vehicle channel measurements and modeling: doc-toral thesis. BRNO: Brno University of Technology, Faculty of Electrical Engineeringand Communication, Department of Radio Electronics, 2016. 80 p. Supervised by prof.Ing. Aleš Prokeš, Ph.D.

DECLARATION

I declare that I have written my doctoral thesis on the theme of “Intra- and out-of-vehiclechannel measurements and modeling” independently, under the guidance of the doctoralthesis supervisor and using the technical literature and other sources of information whichare all quoted in the thesis and detailed in the list of literature at the end of the thesis.

As the author of the doctoral thesis I furthermore declare that, as regards the creationof this doctoral thesis, I have not infringed any copyright. In particular, I have notunlawfully encroached on anyone’s personal and/or ownership rights and I am fully awareof the consequences in the case of breaking Regulation S 11 and the following of theCopyright Act No 121/2000 Sb., and of the rights related to intellectual property rightand changes in some Acts (Intellectual Property Act) and formulated in later regulations,inclusive of the possible consequences resulting from the provisions of Criminal ActNo 40/2009 Sb., Section 2, Head VI, Part 4.

BRNO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .author’s signature

ACKNOWLEDGEMENT

I would like to express great thanks to my supervisor Prof. Ing. Aleš Prokeš, Ph.D. forhis patient guidance on my study, for his precious advice, and for all his devoted time,that made this research feasible.The research is not the work of one person. I need to express my gratitude to Dr.Aniruddha Chandra, Ph.D. and Dr. Ing. Christoph Mecklenbräuker (Vienna Universityof Technology, Institute of Telecommunications, Austria).I thank my parents and my friends for their emotional support.This work was supported by the Czech Science Foundation, Project No. 13-38735S"Research into wireless channels for intra-vehicle communication and positioning", andby National Sustainability Program under grant LO1401. These supports are gratefullyacknowledged.

BRNO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .author’s signature

ACKNOWLEDGEMENT

Research described in this doctoral thesis has been implemented in the laboratoriessupported byt the SIX project; reg. no. CZ.1.05/2.1.00/03.0072, operational programVýzkum a vývoj pro inovace.

BRNO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .author’s signature

Faculty of Electrical Engineeringand CommunicationBrno University of TechnologyPurkynova 118, CZ-61200 BrnoCzech Republic

http://www.six.feec.vutbr.cz

CONTENTS

1 Introduction 13

2 State of Art 162.1 Channel measurements and modeling . . . . . . . . . . . . . . . . . . 162.2 Localization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3 Objectives of the thesis 20

4 Channel Measurements and Modeling 214.1 PHY simulation results of the IEEE 802.11p . . . . . . . . . . . . . . 21

4.1.1 Overview of PHY of the 802.11p . . . . . . . . . . . . . . . . . 214.1.2 Simulation model of 802.11p . . . . . . . . . . . . . . . . . . . 234.1.3 Comparison of the 802.11a and 802.11p . . . . . . . . . . . . . 244.1.4 BER performance of 802.11p standard over an ITU-R Multi-

path channel . . . . . . . . . . . . . . . . . . . . . . . . . . . 274.2 Intra-vehicle channel measurement based on the IEEE 802.11p . . . . 33

4.2.1 Measurement setup . . . . . . . . . . . . . . . . . . . . . . . . 334.3 Intra-vehicle channel description based on the IEEE 802.11p . . . . . 36

4.3.1 Large-scale variations for Intra-Vehicle channel . . . . . . . . . 374.3.2 Small-scale variations for Intra-Vehicle channel . . . . . . . . . 384.3.3 BER-performance of the 802.11p over Intra-Vehicle Channel . 404.3.4 Comparison with UWB . . . . . . . . . . . . . . . . . . . . . . 43

4.4 Around of vehicle channel measurement based on the IEEE 802.11pand UWB systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444.4.1 Out-of-vehicle channel measurement setup . . . . . . . . . . . 45

4.5 Out-of-vehicle channel description in 5.8 GHz band . . . . . . . . . . 474.5.1 BER-performance of the 802.11p over Out-of-vehicle Channel 47

4.6 Out-of-vehicle channel description in UWB . . . . . . . . . . . . . . . 504.6.1 Path Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504.6.2 PDP Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5 Localization using out-of-vehicle measurements in UWB 555.1 Distance estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5.1.1 Delay distance dependence . . . . . . . . . . . . . . . . . . . . 575.2 Two dimensional localization . . . . . . . . . . . . . . . . . . . . . . . 58

6 Conclusion 63

List of abbreviations 65

References 68

A List of publication 79

LIST OF FIGURES1.1 The DSRC band allocations in different regions and countries. . . . . 144.1 The 802.11p OFDM system model. . . . . . . . . . . . . . . . . . . . 234.2 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over

the Rayleigh channel Model A . . . . . . . . . . . . . . . . . . . . . . 264.3 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over

the Rayleigh channel Model B . . . . . . . . . . . . . . . . . . . . . . 264.4 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over

the Rayleigh channel Model C . . . . . . . . . . . . . . . . . . . . . . 274.5 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over

the Rayleigh channel Model E . . . . . . . . . . . . . . . . . . . . . . 274.6 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11p with coding rate

1/2 over the Rayleigh channels with different Models . . . . . . . . . 284.7 BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11p with coding rate

3/4 over the Rayleigh channels with different Models . . . . . . . . . 284.8 BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of

802.11p with coding rate 1/2 over Pedestrian Channel A . . . . . . . 304.9 BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of

802.11p with coding rate 3/4 over Pedestrian Channel A . . . . . . . 314.10 BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of

802.11p with coding rate 1/2 over Vehicular Channel B . . . . . . . . 314.11 BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of

802.11p with coding rate 3/4 over Vehicular Channel B . . . . . . . . 324.12 Simulation results of coded transmission with coding rate 1/2 (3

Mbps) and 3/4 (4.5 Mbps) over Pedestrian Channel A and Vehic-ular Channel B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.13 The underground parking lot with the vehicle under test. The re-ceive/transmit antennas inside the car. . . . . . . . . . . . . . . . . . 33

4.14 The antennas patterns for different bands. . . . . . . . . . . . . . . . 344.15 The schematic diagram of the intra-vehicle measurement setup. . . . 354.16 The transmitter and receiver antenna positions for the intra-vehicle

channel measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . 364.17 The PDPs for in/out-side measurements for different Tx-Rx distances. 384.18 Normalized LSV (𝛾) of the PDP for LOS/ NLOS measurements car-

ried out for different Tx-Rx separations for in/out-side measurements. 394.19 SSV (𝜉) of the PDP for LOS/ NLOS measurements carried out for

different Tx-Rx separations for in/out-side measurements. . . . . . . . 40

4.20 Empirical CDF of SSV fitted with three statistical distributions (lo-gistic, GEV, and normal) for different Tx-Rx separations, from leftto right, top: d = 0.53 m, d = 0.94 m, d = 1.28 m, bottom: d = 2.03m, d = 3.38 m. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.21 BER performance of IEEE 802.11p standard over the measured chan-nels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.22 The UWB and the 802.11p normalized PDP vs normalized time sam-ples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.23 The vehicle under test and measuring equipments. . . . . . . . . . . . 454.24 Placement of the receiving antennas for around-vehicle measurements. 464.25 Placement of the transmitting antenna for around-vehicle measure-

ments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.26 The PDPs for measurements points in 5.8 GHz band. . . . . . . . . . 494.27 LSV of PDP for measurements points in 5.8 GHz band for out-of-

vehicle measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . 494.28 BER results based on the 802.11p over measured channels. . . . . . . 514.29 PDPs for Tx-Rx separation (𝑑) = 3m. . . . . . . . . . . . . . . . . . . 534.30 PDPs for Tx-Rx separation (𝑑) = 9m. . . . . . . . . . . . . . . . . . . 545.1 Superposition of PDPs for Rx1 for all Tx positions. . . . . . . . . . . 565.2 Delay versus distance. . . . . . . . . . . . . . . . . . . . . . . . . . . 605.3 The principle of 2D localization. . . . . . . . . . . . . . . . . . . . . . 605.4 Actual and estimated position of the Tx antenna. . . . . . . . . . . . 61

LIST OF TABLES4.1 The main differences between IEEE 802.11p and 802.11a . . . . . . . 224.2 Modulation schemes and mappings . . . . . . . . . . . . . . . . . . . 224.3 Simulation parameters to compare 802.11p and 802.11a standards . . 244.4 Description of the HIPERLAN/2 . . . . . . . . . . . . . . . . . . . . 254.5 The HIPERLAN/2 channel models . . . . . . . . . . . . . . . . . . . 254.6 The ITU-R channel models for Pedestrian and Vehicular test envi-

ronments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.7 Simulation parameters over ITU-R channels with different speed . . . 294.8 Detailed settings for intra-vehicle measurements . . . . . . . . . . . . 374.9 Parameter values for LSV of in/out-side setups. . . . . . . . . . . . . 394.10 Two sample K-S test 𝑝 values for fitting SSV with different continuous

distributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414.11 Characteristics of the channel models for in/out-side measurements. . 424.12 Distances between transmitting and receiving antennas . . . . . . . . 484.13 Parameters of LSV for measurements points in 5.8 GHz band for

out-of vehicle measurements. . . . . . . . . . . . . . . . . . . . . . . . 504.14 Parameters for Out-of-vehicle channel. . . . . . . . . . . . . . . . . . 504.15 PL values (in dB) for all Tx locations. . . . . . . . . . . . . . . . . . 524.16 S-V parameter values . . . . . . . . . . . . . . . . . . . . . . . . . . . 545.1 Distance of Rx1 from Tx antenna positions . . . . . . . . . . . . . . . 575.2 Distance of Rx2 from Tx antenna positions . . . . . . . . . . . . . . . 585.3 Distance of Rx3 from Tx antenna positions . . . . . . . . . . . . . . . 595.4 Parameters of polynomial fitting . . . . . . . . . . . . . . . . . . . . . 595.5 Comparison between measured and calculated coordinates of Tx . . . 62

1 INTRODUCTIONWireless communication has been developed and used in many areas of our life.The level of wireless communications has been grown in the latest years. Initiallythis growth was in the cellular mobile’s area from analog systems to 2G (as GSM,CDMA), 3G (GPRS, WCDMA), 4G (LTE). On other side, progress in digital elec-tronics has given cheaper smart devices and wireless sensors, which have become partof our life. The IEEE standards for wireless area network (WLAN), which includesmachine-to-machine (M-2-M), vehicle-to-X (V-2-X) communications, together withcloud systems have changed conception of communication networks. In our daysmore wireless communication are introduced in real life. The applications of newtechnologies related with wireless networks are being developed in transport areatoo.

Modern vehicles include more and more digital and smart technologies: for nav-igation, location information, safety and intra-vehicle communication. This appli-cations are directed on reducing of car accidents, traffic jams and help on the roads.

Vehicular Ad Hoc Network (VANET) is a application for V-2-X communica-tion, where different technologies are applied. Well known the mobile technologies,such as 2G and 3G, are out to be unsuitable to VANET, because latency and pri-ority service are used, which contrary requirements for safety-critical applications.The development of decentralized communication technology is chosen for V-2-Xcommunication. Wireless communication channels named Dedicated Short RangeCommunications (DSRC) are designed for automotive use and content set of proto-cols and standards involving everything from physical layer (PHY) for VANET [1].The American Society for Testing and Materials (ASTM) started standardization ofDSRC for vehicle roadside communications [2]. The standard is called 802.11p de-scribed in wireless access in vehicular environments (WAVE) [3,4], and governed bya non-profit organization, the car-2-car communication consortium (C2C-CC) [5].

The Federal Communications Commission (FCC) allowed a 75 MHz band from5.850-5.925 GHz for DSRC in the USA. The DSRC spectrum in the EU is a 30 MHzband from 5.875-5.905 GHz. Spectrum allocated to DSRC in other countries canbe with status "in use", "allocated" or "potential". The DSRC band allocations indifferent regions and countries are illustrated in Figure 1.1.

Vehicle area networks (VAN) and VANETs generally refer to networks betweencars and infrastructure points located along the road side [6,7]. These networks areaimed to deliver information about the traffic, to ensure safety of the passengers,and to provide driver assistance and passenger entertainment [8].

Presently there had been an upsurge in the interest of the automobile industriesregarding wider applications of intra-vehicular wireless transmission. The use of

13

Fig. 1.1: The DSRC band allocations in different regions and countries.

wireless communication in the vehicle cuts down the wiring harness, which willreduce the weight and cost of the car, eliminate the need of drilling to pass the cables;and will simplify the design as well as production processes. The IEEE 802.11pprotocol was specified primarily for V-2-V/V-2-I scenarios where both transmitterand receiver are placed outside the vehicle [9]. In order to realize fully integratedintelligent transportation systems (ITS), it is important to evaluate the efficacy ofthe protocol for wireless connections in in-car (both transmitter and receiver areinside the vehicle) and out-of-car (transmitter is inside and receiver is outside, orvice-versa) scenarios as well.

Another application of V-2-X communication is locating a personal vehicle, beit safety-critical, e.g. railway crossing surveillance [10], or non safety-critical, e.g.finding a vehicle in parking lot [11]. There exists different global navigation satellitesystems (GNSS) such as GPS in USA, GLONASS in Russia, and Galileo in Europe,to determine the location of a car in outdoor scenarios. Car localization is alsopossible to some extent with the help of terrestrial cellular mobile networks [12],another vehicles and Roadside Units (RSUs) [13]. However, finding exact locations ofvehicles in an enclosed space represents a major challenge. The positioning systemsmentioned before cease to work in confined and underground areas and it is necessaryto find a suitable method for such scenarios.

A short introduction and overview were given above about the current progressin C-2-C technology and applications of communications and localizations. Thesecond chapter briefly describes measurements provided for C-2-X communicationfor different scenarios for channel modeling and calculating of channel parameters.

14

This chapter also presents results of UWB localization in buildings. The definitionof the dissertation objectives follows in chapter 3.

Chapters 4 and 5 make the core of this thesis and describe simulations and theexperiments. Chapter 4 discusses PHY simulation model of the IEEE 802.11p andcomparison with 802.11a based on simulation results over fading channels. Chapter5 demonstrates measurements for intra-vehicle and out-of-vehicle scenarios in UWBand for 802.11p standard. This chapter also includes two-term exponential channelmodeling for 802.11p and Saleh-Valenzuela (S-V) model for UWB. Section 5.6 addi-tionally represents localization of vehicle in the underground garage utilizing UWB.Finally, the main results are summarized in conclusions.

15

2 STATE OF ART

2.1 Channel measurements and modelingChannel modeling in most cases is based on real-world measurements of the chan-nel. The measurements are used to understand the nature of the channel, extractmodel parameters and validate of existing channel models. Design and implemen-tation of measurement campaigns should be carried out in a manner such that allthe major characteristics can be obtained. This task is difficult in reality, becauseeach measurement is limited in measuring equipment, measurement conditions andscenarios.

Real-world measurements have been prepared and realized for vehicular com-munication systems in the past under different scenarios and measurement setups.Generally, vehicular channel measurements fall under two broad categories: vehicleto vehicle (V-2-V) and vehicle to infrastructure (V-2-I) channels.

Channel measurements and modeling based on 802.11p

A general review on advances and challenges in V-2-V channel measurements isprovided in [14]. For both V-2-V and V-2-I scenarios, the propagation conditionscan be grouped into highway, rural, urban and suburban, based on the classificationscheme of cellular channels which are well studied. This classification was used formeasurements in [15–19]. In [15], speed-separation diagrams were prepared to ana-lyze channel propagation, channel Doppler spread and coherence time based on themeasurements of V-2-V propagation at 5.9 GHz channel in suburban scenario. Themeasuring system for on-road experimental test to characterize V-2-C wireless com-munication channels are described in [16]. In [17], the power delay profiles (PDPs),RMS delay spread, approximate amplitude fading distributions and frequency cor-relation function were provided using a spread-stepped correlator technique at 5GHz in three areas: large cities, open highway areas, and small cities. The PDPswere characterized using an exponential decay cluster model, path loss (PL) models.Further, Doppler spread statistics were derived from channel measurements at 5.9GHz for the car-to-car channel in rural, highway, and urban environments describedin [18]. Paper [19] presents measuring of the sensitivity to the position of the an-tennas based on radio reception performance the IEEE 802.11a in the 5 GHz bandfor roof- and in-vehicle scenario.

There are some other measurements attempted to imitate the real-world scenarioin a better way but were more complex and involved more parameters. The scenarioswere designed based on the previous classification but includes novel parameters such

16

as traffic analysis and driving directions [20–23]. In [20], the paper presents resultsfor PL, PDPs, and delay-Doppler spectra at the 5.2 GHz band on the highway whencars move in opposite directions on high speed. Results of the V-2-V measurement inthe 5 GHz band for delay spread, amplitude statistics, and correlations are presentedin [21]. In [22], time-variant channel impulse response (CIR) is extracted frommeasured data by the multiple input multiple output (MIMO) channel sounder at5.7 GHz in typical urban traffic scenarios. In [23], delay spread and propagationaspects were discussed for various vehicular environment parameters, i.e. speed,roads and building construction, vehicle population, and traffic statistics.

More measuring campaigns were directed to scenarios based on specific situ-ations and locations. Measurements in tunnels, on bridges, slopes, intersectionsand in direct line-of-sight (LOS) with buildings and other vehicles are presentedin [24–33]. Reference [24] presents the PDPs for collision avoidance application.Parametrization of PL categorized in LOS and non line-of-sight (NLOS) paths forV-2-V MIMO channel at 5.7 GHz are discussed in [25]. Authors of [27] analyze theNLOS reception of the signals according to the IEEE 802.11p in a typical urbanintersection. Channel model using real-world measurements in a city scenario basedon IEEE 802.11p is shown in [28]. The relationship between types of vehicles andthe average power of the received signal and the signal distribution, as well as theeffects of vehicle shadowing on IEEE 802.11p based communication are investigatedin [29]. The LOS propagation results for road intersections are presented in [30],in terms of PDPs, PL and delay spreads. Shadow fading channel models catego-rized into LOS and obstructed LOS, based on measurements in urban and highwayscenarios are presented in [34]. Signal propagation in urban area prediction tech-nique for V-2-V situations are presented and compared with real measurements andsimulations in [32]. Various methods for channel modeling in tunnels are reviewedin [33], including numerical methods solving Maxwell equations, waveguide or modalapproach, ray tracing based methods and two-slope PL modeling.

In [35–39] channel measurements for signal propagation modeling were consid-ered for inter-vehicle scenarios. The inter-vehicle channel at 5.2 GHz were measuredin realistic road traffic scenario and statistical functions were derived [35]. In [36],NLOS, PL and fading models were presented based on measurements at intersec-tions for inter-vehicular 5.9 GHz band. Geometry-based stochastic channel modelsare used for describing the fading of V-2-V radio propagation channel at 5.3 GHz oncampus, highway, suburban and urban areas in [37]. In [38], the propagation modelwas obtained with the “Wiedemann-Model” tool to generate realistic scenarios forindividual driving actions. Channel measurement results based on the IEEE 802.11afor road traffic scenario are shown in [39].

The articles [40,41] present the distribution of channel parameters based on non-

17

stationary vehicular channel measurements following a bimodal Gaussian mixturedistribution. In [40], the time-frequency-dependent local scattering function wasestimated from measurements. Moreover, the PDP, the Doppler power spectraldensity, the RMS delay spread and the RMS Doppler spread are analyzed. Article[41] shows that the small-scale fading of the envelope of the first delay bin of Riciandistribution varies with K-factor.

It must be noted here that every measuring campaign is unique and is differentfrom others, even the same scenarios or the same frequency band. The main dif-ferences are vehicle types, locations of the antennas and configuration of roads andbuildings. A roof-mounted antenna is popular in measurement configurations. Thisconfiguration is presented in [17,18,42–46] for single input single output (SISO) andin [20, 22, 24, 47] for MIMO. Places of antenna location inside the vehicle on theroof and on the windshield are shown in [21]. The various positions on the roofwere used in [19, 27]. Authors in [48] describe the impact of antenna placement onchannel parameters based on channel measurements performed with four omnidi-rectional antennas mounted on the roof, bumper, windscreen, and left-side mirrorof the transmitter and receiver cars.

Initially, on the first step, measurements for V-2-V and V-2-I were carried outin the cellular mobile frequency band of 900 MHZ used in road payment systems[42, 43, 49, 50]. Vehicle channel measurements were performed at frequencies of 5GHz [17,19–21,35,47,51], 5.9 GHz [15,16,18,25–29,52], and 2.4 GHz [44,53].

The reference [54] describes measurement based narrow-band channel models atfrequency of 5.2 GHz. Reports from some inter-vehicular measurement campaignsin the 5.9 GHz band are also available in [55]. A wide set of PDP measurements andproposed tapped delay line models based on Markov chains were provided in [55].

The MIMO measurements for V-2-V at 5.2 GHz were performed in highwayand rural environments in [56]. Measurements at 5.6 GHz band performed withMIMO for urban intersection scenario, LOS and NLOS situations are discussedin [57]. In [58], V-2-V measurements are performed for single input multiple output(SIMO) to explain PL, delay spread, maximum excess delay, the standard deviationof slow fading and K-factor. Rician K-factor from time-frequency variability basedon vehicular channel measurements at 5.6 GHz are analyzed in [59]. The urban com-munication scenario at intersection in 6 GHz band is analyzed in [60]. Reference [61]describes the geometry-based channel models for existing measurement campaigns,the most important environments, and the delay spread and Doppler spreads. Sim-ulation of V-2-V propagation model and comparison with cellular channels are pre-sented in [62].

The characterization of the 5 GHz intra-vehicle communication channel is onlyattempted recently in [63], where the authors present PDPs, delay spread, and

18

statistical channel models for a minivan and a bus.

Channel measurements and modeling based on UWB

As far as intra-vehicular channel measurements are concerned, most of the arti-cles [64,65] are focused on UWB as UWB is a short-range transmission technology.Theoretical limits on UWB range estimation and practical algorithms in terms ofCramer-Rao and Ziv-Zakai are discussed in [66]. The frequency domain measure-ments of the intra-vehicle UWB channel were prepared in [67]. Authors in [68]present clustering channel models based on the channel measurement in frequencydomain from 2 to 16 GHz in different intra-vehicle scenarios and developed by mod-ifying the IEEE 802.15.3a [69]. The performance of multi-band OFDM in intra-vehicle communication is studied in [70]. The authors in [71] report about a com-parison of broadband channel sounding experiments over a different range of cars.References [72, 73] describe a detailed UWB channel modeling campaign for intra-vehicular environments. Time-domain channel sounding and modeling the UWBpropagation channel with modified S-V PDP model are presented in [74]. The sta-tionary regions based on the correlation between consecutive PDPs are extractedfrom UWB channel measurements results in [75].

2.2 LocalizationThe UWB network may be integrated for intra-vehicle and around vehicles local-ization. Measurement studies of communication and positioning inside vehicle areavailable in [76–78]. The authors of [76] describe the localization and tracking tech-nology based on the UWB. In [77, 78], the CIR measured for the frequency range3-11 GHz are presented via a two-part exponentially decaying envelope-delay profilefor the intra-vehicle localization system based on the time of arrival (TOA). TheUWB sensor networks were used for localization through coherent and non-coherentapproaches of data fusion for imaging and object localization in [79]. Vehicle pene-tration loss and PDPs depending on angle around of car were calculated from UWBmeasurement in [80].

19

3 OBJECTIVES OF THE THESISThe objectives of the dissertation thesis are formulated as follows:

• Design and implementation of simulation model based on the IEEE 802.11p.Performance comparison of the IEEE 802.11p and the IEEE 802.11a overfading channels. The BER performance of the IEEE 802.11p in dependencewith a vehicular speed.

• Measuring real-world stationary channels in frequency domain for the intra-vehicle scenario. Channel modeling based on the 802.11p measurement results.Separation of large-scale variation (LSV) and small-scale variation (SSV) fromPDP, and extracting CIR for using it in simulation model. The BER perfor-mance of the IEEE 802.11p of intra-vehicle channels.

• Measuring the IEEE 802.11p (5.8 GHz band) and UWB (3-11 GHz band) chan-nels in frequency domain for the out-of-vehicle scenario. The BER performanceof the IEEE 802.11p over the out-of-vehicle channels. Channel modeling forUWB using S-V model.

• Analysis of a localization technique designed for UWB system from measure-ment results for out-of-vehicle scenario.

20

4 CHANNEL MEASUREMENTS AND MOD-ELING

4.1 PHY simulation results of the IEEE 802.11pThis chapter provides a description of the simulation model recommended for theIEEE 802.11p protocol. The main task of chapter is to determine the differencesbetween standards 802.11p and 802.11a. For comparison, bit error rate (BER) wasselected. For simulation model channels appropriate for both standards were used.The IEEE 802.11p protocol inherits the principle of the 802.11a, but has otheroptions in connection with another bandwidth. Thus, the concept model for bothstandards will be used.

Transmitting raw bits in the network medium is function of PHY, which is thelowest layer of the Open Systems Interconnection (OSI). Wireless communicationrequires effort on this layer, especially in channel modeling and high reliability [81].

4.1.1 Overview of PHY of the 802.11p

PHY of the V-2-X communication system is based on the IEEE 802.11p standard.The IEEE 802.11p is modified version of the IEEE 802.11a standard [82]. Original802.11a was optimized for work in local area networks with low mobility, it couldn’tprocess information in system with high delays. Amendments defined in the IEEE802.11p aims to ensure interoperability between wireless devices that interact inrapidly changing communication environments and handle situations where trans-actions must be completed in a time-frame, much shorter than that of the IEEE802.11a [83].

PHY of the IEEE 802.11p and the IEEE 802.11a are based on OFDM. Themain differences between the IEEE 802.11a and the IEEE 802.11p are presented inTable 4.1 [84]. An OFDM with 64 sub-carriers is used for transmission, but only52 sub-carriers are utilized: 48 – for the actual data, 4 – pilot sub-carriers. Pilotsub-carriers transmit a fixed pattern for disregard frequency and phase offset at thereceiver side. IEEE 802.11a uses modulation schemes BPSK, QPSK, 16-QAM or64-QAM in combination with different coding rates. A nominal data rate could befrom 6 to 54 Mbit/s for mode with 20 MHz bandwidth. The IEEE 802.11p utilizesthe half with 10 MHz bandwidth, it is used to make signal robust against fading,resulting in corresponding data rate reduction.

The rate-dependent parameters of the IEEE 802.11p standard PHY, which uti-lizes BPSK, QPSK, 16-QAM, and 64-QAM as sub-carrier modulation schemes in

21

combination with rates 1/2, 2/3, and 3/4 convolutional codes to obtain a variabledata bit rate from 3 up to 27 Mbit/s which are listed in Table 4.2 [85, 86].

Tab. 4.1: The main differences between IEEE 802.11p and 802.11a

Parameters IEEE 802.11p IEEE 802.11a ChangesBit rate (Mbit/s) 3, 4.5, 6, 9 6, 9, 12, 18 Half

12, 18, 24, 27 24, 36, 48, 54Modulation BPSK, QPSK, 16-QAM, 64-QAM No changesCoding rate 1/2, 2/3, 3/4 No changes

Number of sub-carriers 52 No changesSymbol duration (𝜇s) 8 4 Double

Guard time (𝜇s) 1.6 0.8 DoubleFFT period (𝜇s) 6.4 3.2 Double

Preamble duration (𝜇s) 3.2 1.6 DoubleSub-carrier spacing (MHz) 0.15625 0.3125 Half

Tab. 4.2: Modulation schemes and mappings

Data Coding Coded bits Coded bits Data bitsrate Modulation rate, per per OFDM per OFDM

(Mbit/s) 𝑅 sub-carriers, symbol, symbol,𝑁𝐵𝑃 𝑆𝐶 𝑁𝐶𝐵𝑃 𝑆 𝑁𝐷𝐵𝑃 𝑆

3 BPSK 1/2 1 48 244.5 BPSK 3/4 1 48 366 QPSK 1/2 2 96 489 QPSK 3/4 2 96 7212 16-QAM 1/2 4 192 9618 16-QAM 3/4 4 192 14424 64-QAM 2/3 6 288 19227 64-QAM 3/4 6 288 216

22

4.1.2 Simulation model of 802.11p

The latest technologies, such as 802.11p, are based on OFDM PHY, because theOFDM-based transmission systems can provide high-rate transmission and highspectrum efficiency in channel environments.

Fig. 4.1: The 802.11p OFDM system model.

The configuration of the model prepared in accordance with the recommenda-tions for IEEE 802.11a standard [87]. A simulation model was implemented usingMATLAB and SIMULINK. The structures of transmitters and receivers for the802.11p and 802.11a correspond to blocks in Figure 4.1.

The simulation model includes following signal processing stages: transmitter,channel, receiver.

The transmitter:• Convolutional coding with a coding rate, 𝑅 = 1/2 and 𝑅 = 3/4;• BPSK modulation scheme;• Inverse fast Fourier transform (IFFT) for 64 samples;• Adding cyclic prefix (CP) with length of one-fourth of the symbol duration;

The channel:

23

• Transmitting the signal with 8 𝜇s (for 802.11p) and 4 𝜇s (for 802.11a) durationover the fading channel HIPERPLAN/2;

The receiver:• Removing CP;• Fast Fourier transform (FFT) for 64 sub-carriers;• Least Squares (LS) channel estimation;• BPSK demodulation;• Decoding error codes by using the algorithm Viterbi hard decision.

For getting accurate results, ten thousand bits are transmitted. More parameters ofsimulation are presented in Table 4.3.

4.1.3 Comparison of the 802.11a and 802.11p

To determine the differences between BER performances of the standards 802.11aand 802.11p channel models HIPERLAN/2 are used as fading channel models.Channel characteristics of HIPERLAN/2 channels are used for a both standardsbecause they have the same structure published in [88]. Table 4.4 presents descrip-tion of the channel. Parameters of channel are described in Table 4.5.

Tab. 4.3: Simulation parameters to compare 802.11p and 802.11a standards

Parameters Mode 1 Mode 2802.11p 802.11a 802.11p 802.11a

Bandwidth (MHz) 20 10 20 10Inner modulation BPSK

Coding rate 1/2 3/4Location of pilots (№) 7 21 35 49

Transmission HIPERLAN/2:channel models A, B, C, E

The transmission of data was simulated for varying 𝐸𝑏/𝑁0 over fading Rayleighchannels: Model A, Model B, Model C, and Model E that are shown in Figures 4.2,4.3, 4.4 and 4.5, two BER curves were obtained: for standards IEEE 802.11a andIEEE 802.11p.

For low values of average 𝐸𝑏/𝑁0, the errors are same. The average 𝐸𝑏/𝑁0 requiredfor BER of 10−4 is found at 2 dB for mode 1 and 3 dB for mode 2 over ChannelA, for BER of 10−4 – 3 dB for mode 1 and 0 dB for mode 2 over Channel B, andfor BER of 10−4 – 3 dB for mode 1 and 5 dB for mode 2 over Channel C. Figure

24

Tab. 4.4: Description of the HIPERLAN/2

Environment DescriptionA Typical office environment for NLOS

condition with 50 ns RMS delay spreadB Typical large open space and office environment for NLOS

condition and 100 ns RMS delay spreadC Typical large open space environment for NLOS

condition and 150 ns RMS delay spreadE Typical large open space environment for NLOS

condition and 250 ns RMS delay spread

Tab. 4.5: The HIPERLAN/2 channel models

Model A Model B Model C Model ETap Delay Average Delay Average Delay Average Delay Average

Number (ns) power (ns) power (ns) power (ns) power(dB) (dB) (dB) (dB)

1 0 0.0 0 -2.6 0 -3.3 0 -4.92 10 -0.9 10 -3.0 10 -3.6 10 -5.13 20 -1.7 20 -3.5 20 -3.9 20 -5.24 30 -2.6 30 -3.9 30 -4.2 40 -0.85 40 -3.5 50 0.0 50 0.0 70 -1.36 50 -4.3 80 -1.3 80 -0.9 100 -1.97 60 -5.2 110 -2.6 110 -1.7 140 -0.38 70 -6.1 140 -3.9 140 -2.6 190 -1.29 80 -6.9 180 -3.4 180 -1.5 240 -2.110 90 -7.8 230 -5.6 230 -3.0 320 0.0

4.5 shows the errors at standards and different modes, results for the IEEE 802.11pindicate better performance even for the IEEE 802.11a.

The BER performance of IEEE 802.11p is almost the same for the channel ModelsA, B, C, and E (Figures 4.6 and 4.7). For IEEE 802.11a the BER are slightly greaterfor channel Model E than for channel Model A, B, and C, as the mean delay exceedsthe cyclic prefix.

The IEEE 802.11p has different BER performance than the IEEE 802.11a. Thisis based on the difference of bandwidth. BER performance of IEEE 802.11p is found

25

Fig. 4.2: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over theRayleigh channel Model A

Fig. 4.3: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over theRayleigh channel Model B

to be superior when compared to that for IEEE 802.11a owing to a longer CP timeand increases with growing path gain and delay channel.

26

Fig. 4.4: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over theRayleigh channel Model C

Fig. 4.5: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11a and 802.11p over theRayleigh channel Model E

4.1.4 BER performance of 802.11p standard over an ITU-RMultipath channel

Channel model ITU-R is an empirical model and developed in accordance with theITU-R M.1225 Recommendation. This type of modeling was used to simulate the

27

Fig. 4.6: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11p with coding rate 1/2over the Rayleigh channels with different Models

Fig. 4.7: BER as a function of 𝐸𝑏/𝑁0 for signals of 802.11p with coding rate 3/4over the Rayleigh channels with different Models

IEEE 802.11p signal. Among the recommendations for channel ITU-R, which aredescribed in [89], there are two models of channels, characterizing different propaga-tion environments. For each model, pedestrians and motor vehicles, different delayspread profiles are set. Low delay spread profile A, describes the rms delay spread370 ns for Pedestrian. Profile B is 4000 ns for Vehicular. The model for Pedestrian

28

Channel A has 4 taps and Vehicular Channel B has 6 taps which are presented inTable 4.6.

Tab. 4.6: The ITU-R channel models for Pedestrian and Vehicular test environments

Pedestrian Channel A Vehicular Channel BTaps Relative Average Relative Average

delay (ns) power (dB) delay (ns) power (dB)1 0 0 0 -2.52 110 -9.7 300 03 190 -19.2 8900 -12.84 410 -22.8 12900 -10.05 - - 17100 -25.26 - - 20000 -16.0

Simulation model is used from 4.1.2. Table 4.7 presents modes for simulation.The Pedestrian Channel A and Vehicular Channel B in simulation model of theIEEE 802.11p standard are the components of MATLAB SIMULINK.

Tab. 4.7: Simulation parameters over ITU-R channels with different speed

Parameters Mode 1 Mode 2802.11p 802.11p

Bandwidth (MHz) 20 20Inner modulation BPSK

Coding rate 1/2 and 3/4 1/2 and 3/4Location of pilots (№) 7 21 35 49

Transmission Pedestrian Vehicularchannel Channel Channelmodels A B

Speed of 5, 30 , 60 30, 60 , 90transmitter (km/h) 130, 260

One of the key factors affecting the performance of the IEEE 802.11p standardmodel ITU-R M.1225 channels is the effect of the Doppler shift. Figures 4.8 and 4.9show the errors at different speeds of the transmitter relative to the receiver overthe ITU-R Pedestrian Channel A, specified in Table 4.6. The simulation was carriedout by BPSK OFDM modulation and coding rates 1/2 and 3/4, which corresponds

29

to the transmission rates 3 Mbps and 4.5 Mbps respectively. At low speed, about 5km/h, better performance is achieved, with an increase in speed of the transmitter,the probability of error increases. Results of 3 Mbps rate indicate better performanceeven at a speed of 30 km/h, when the rate of 4.5 Mbps BER is not reduced below0.007 units.

Figures 4.10 and 4.11 show the simulation results over ITU-R Vehicular ChannelB. The model also includes a BPSK OFDM modulation and coding rates 1/2 and3/4. With the growth of the Doppler shift and intersymbol interference (ISI) for avehicle speed of 260 km/h the BER performance does not show favorable results.

Figure 4.12 presents side-by-side comparison of PHY model performance in var-ious propagation test environments. All curves are plotted for coded BPSK trans-mission without being punctured. Pedestrian model Channel A assumes a vehiclevelocity of 5 km/h. For Vehicular Models, the speed of the transmitter is 90 km/h.

ITU-R Pedestrian Channel A has a maximum relative delay of 410 ns and RMS45 ns, so the impulse response of the channel is shorter than the guard interval.Thus, when the speed of movement of the transmitter is 5 km/h, BER correspondsto 10−3 at 𝐸𝑏/𝑁0 of 20 and 26 dB, depending on the data rate. The ITU-R Thereforethe efficiency of the model is reduced to a level of 10−1 at the speed of the transmitter90 km/h and data rate is 3 Mbps, and efficiency drops with increasing data ratesup to 4.5 Mbps, while maintaining the speed of the transmitter.

Fig. 4.8: BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of802.11p with coding rate 1/2 over Pedestrian Channel A

30

Fig. 4.9: BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of802.11p with coding rate 3/4 over Pedestrian Channel A

Fig. 4.10: BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of802.11p with coding rate 1/2 over Vehicular Channel B

31

Fig. 4.11: BER as a function of 𝐸𝑏/𝑁0 on relative vehicle velocity for signals of802.11p with coding rate 3/4 over Vehicular Channel B

Fig. 4.12: Simulation results of coded transmission with coding rate 1/2 (3 Mbps)and 3/4 (4.5 Mbps) over Pedestrian Channel A and Vehicular Channel B.

32

4.2 Intra-vehicle channel measurement based onthe IEEE 802.11p

4.2.1 Measurement setup

Measurements were carried out on the basis of the car Skoda Octavia Liftback. Thiscar was parked on the lower (sixth) floor of the underground parking garage of theFaculty of Electrical Engineering (FEEC), Brno University of Technology (BUT).The car is a typical mid-size vehicle with dimensions outside: 4.659 × 1.814 ×1.462 m. Due to the car location of the underground and walls and floors are fromreinforced concrete, the narrowband interference is suppressed. There were no othervehicle around the measured car. Figure 4.13 presents the parking lot with the carunder test and used equipments.

Tx

Rx 2 Rx 1

Rx 3

Rx 2

Rx 1

Tx

(a) (b)

Fig. 4.13: The underground parking lot with the vehicle under test. The receive/-transmit antennas inside the car.

The measurements were performed using one transmitter antenna (Tx) and threereceiver antennas (Rx), which are shown in Figure 4.13. The four identical omnidi-rectional conical monopole antenna were used for the Tx and Rx. The the circularazimuth H-plane pattern of the antennas can be confirmed from the radiation pat-terns shown in Figure 4.14.

33

-20dBi

-10dBi

0dBi

10dBi

90o

60o

30o0o

-30o

-60o

-90o

-120o

-150o

180o150o

120o

E-plane f = 3.0 GHzf = 6.5 GHzf = 10.0 GHz

-20dBi

-10dBi

0dBi

10dBi

90o

60o

30o0o

-30o

-60o

-90o

-120o

-150o

180o150o

120o

H-plane

Fig. 4.14: The antennas patterns for different bands.

Measuring equipment was installed at the outside, near to the car. All doorsand windows were closed, except for the driver’s window, which was slightly opendue to the cables connecting the antennas to measuring and recording equipment.

The four-port vector network analyzer (VNA) E5071C (Agilent Technologies)[90] was used to measure and record the scattering parameters. To connect the ana-lyzer and the Tx and Rx antennas phase stable coaxial cables were used. Calibrationof VNA was performed only with cables, without connected antennas. The transmitpower was chosen at 5 dBm.

The measurement results are the complex valued transfer function in the fre-quency domain. The VNA presents CTF as s-parameter 𝑆𝑥1, where x ∈ {2, 3, 4} isnumber of Rx port and 1 is Tx port of VNA.

Measurement parameters are set in accordance with recommendations of theIEEE 802.11p except frequency range of 100 MHz (5.775 to 5.875 GHz). Frequencystep of 𝑓𝑠 = 0.125 MHz was chosen for 801 carriers. The bandwidth, 𝐵𝑤, for 1sub-channel is 10 MHz. The time range was calculated as 𝑡𝑑 = 1/𝑓𝑠 = 8 𝜇s. Theinformation converted into the spatial domain presents a maximum propagationdistance is 𝐿𝑑𝑚𝑎𝑥 = 𝑐/𝑓𝑠 = 2.4 km and a propagation space resolution 𝑃𝑠𝑟 = 𝑐/𝐵𝑤

= 3 m, where 𝑐 is the light speed.Figure 4.15 shows the schematic diagram of the measurements setup using VNA

for frequency domain channel measurement. After recording the CTF complex 𝑆𝑥1

was uploaded to MATLAB. Since the IEEE 802.11p standard has bandwidth of 10MHz, the data in frequency domain with bandwidth of 100 MHz were partitionedinto 10 MHz segments, response of each of them are close. All results were translatedfrom frequency to time domain using the complex IFFT with rectangular window[91]: (ℎ(𝜏) = ℱ−1𝐻(𝑓)), where 𝑆𝑥1 used as 𝐻(𝑓). Results in the time domain werepresented as averaging over ten sub-channels.

The Tx antenna was placed at three different locations inside and at two locations

34

outside the car. Positions inside the vehicle were chosen at the right rear set, in themiddle of the at armrest, and at the driver’s seat. Outside positions were set infront of the car and near the right headlamp, the Tx antenna was installed on atripod with height of 1.09 m. The Rx antennas positioned inside of the car: one - onthe roof in the rear part of the vehicle and two - on the windshield, in the left andright upper corners. A total of 15 different combinations between the Tx and Rxantennas were performed with distances from 0.53 to 3.38 m. All transmitter andreceiver antenna positions are shown in Figure 4.16. Table 4.8 presents parametersof all combinations for measuring. Both the LOS and NLOS scenarios were realized.

Tx

Antenna

Rx

Antenna

Vector

network

analyzer

Transmission

S21 = Y(ω) / X(ω)= H(ω)

X(ω) Y(ω)

IFFT h(t)

Port 2Port 1

Fig. 4.15: The schematic diagram of the intra-vehicle measurement setup.

35

Fig. 4.16: The transmitter and receiver antenna positions for the intra-vehicle chan-nel measurements.

4.3 Intra-vehicle channel description based on theIEEE 802.11p

The generic multipath actual CIR is described as:

ℎ𝑐ℎ(𝜏) =𝑁−1∑︁𝑖=0

𝑎𝑖 exp (𝑗𝜃𝑖)𝛿(𝜏 − 𝜏𝑖). (4.1)

where 𝜏𝑖 is the propagation delay of the 𝑖-th multipath component (MPC), 𝑎𝑖 exp (𝑗𝜃𝑖)is the complex amplitude coefficient of the 𝑖-th MPC, and 𝛿(.) is the Dirac deltafunction, and N is the total number of MPCs.

As we have one realization, the PDP can be presented through the squared CIR,𝑃 (𝜏) = |ℎ𝑐ℎ(𝜏)|2, and mathematically decomposed into a sum of large scale variation(LSV) and small scale variation (SSV): 𝑃 (𝜏) = 𝛾(𝜏) + 𝜉(𝜏), where 𝛾(𝜏) denotes theLSV and 𝜉(𝜏) is the SSV.

The measured PDP for different Tx-Rx combinations are shown in Figure 4.17.As expected, power levels for LOS are significantly higher compared to the NLOS

36

Tab. 4.8: Detailed settings for intra-vehicle measurements

Measurement Tx-Rx Tx Rx RemarksNo. separation (m) position position1 0.53 2R 2R LOS2 0.69 4R 4M LOS3 0.74 2M 2L LOS4 0.85 2M 2R LOS5 0.87 2M 4M LOS6 0.94 2R 2L LOS7 1.26 4R 2R NLOS8 1.28 2R 4M NLOS9 1.41 4R 2L NLOS10 1.86 0R 2L Tx at an angle11 2.03 0M 2L Tx in front12 2.08 0M 2R Tx in front13 2.74 0R 2R Tx at an angle14 3.15 0R 4M Tx at an angle15 3.38 0M 4M Tx in front

measurements. The power levels for NLOS conditions for both inside and outsideof the vehicle present show rapid variations near the noise floor (-110dB).

4.3.1 Large-scale variations for Intra-Vehicle channel

Analytical performance calculated for closed and small areas presented an exponen-tial decay of the PDP [92,93]. The intra-vehicular signal propagation environmentsare similar to small spaces, the LSV can be demonstrated a two-term exponentialmodel:

𝛾(𝜏) = 𝐴 exp (𝐵𝜏) + 𝐶 exp (𝐷𝜏) ; 0 < 𝜏 < 𝑡𝑑/2 (4.2)

The first term describes power from direct and major reflected rays. The secondterm is close to linear with a very low slope includes the power from diffused MPCs.The two-term exponential model 4.2 gives more flexible fitting than single-term ex-ponential or linear models. This model avoids discontinuities in the separated LSV,which appear when PDP divided into several parts, and the individual descriptionfor each of these parts [93].

To evaluate the quality of the definition of LSV model (4.2), according to the

37

0 0.5 1 1.5 2 2.5 3 3.5 4

x 10-6

-130

-120

-110

-100

-90

-80

-70

-60

-50

-40

-30

Time Delay [s]

PD

P [d

B]

0.53m LOS Tx:2R Rx:2R0.94m LOS Tx:2R Rx:2R1.28m NLOS Tx:2R Rx:4M2.03m NLOS Tx:0M Rx:2L3.38m NLOS Tx:0M Rx:4M

In-car

Out-of-car

Fig. 4.17: The PDPs for in/out-side measurements for different Tx-Rx distances.

PDP, the mean square error (MSE) between the PDP and the LSV was used:

MSE = 1𝑀

𝑀∑︁𝑘=1

[𝑃𝑘(𝜏) − 𝛾𝑘(𝜏)]2 ; 0 < 𝜏 < 𝑡𝑑/2 (4.3)

The minimum MSE is 6.59 × 10−5, averaged over all the 𝑀 = 15 measurements.Figure 4.18 presents the LSV for the two-term exponential model (4.2) for dif-

ferent Tx-Rx distances of 0.53 m, 0.87 m, 1.28m, 2.74 m, and 3.15 m. It is observedthat LSV for the NLOS case drops more slowly than for the LOS scenario as inearlier observations [94].

Table 4.9 describes the parameters of the two-term exponential model for fivedistances from Figure 4.18. The coefficients A and C show similar values for groupsin-car and out-of-car. The coefficient B decreases steadily with increasing Tx-Rxseparation whereas D increases.

4.3.2 Small-scale variations for Intra-Vehicle channel

As stated earlier, the measured PDP is the mathematical sum of LSV and SSV,consequently SSV can be separated from the PDP, 𝜉(𝜏) = 𝑃 (𝜏) − 𝛾(𝜏). SSVs for 5different distances were calculated, and the results are shown in Figure 4.19.

The characterization of the SSV is focused on finding the appropriate randomprocess using the normal, the logistic, and the generalized extreme value (GEV)distributions. For both UWB [77] and milimeter wave [95] SSVs were based on the

38

0 0.5 1 1.5 2 2.5 3 3.5 4

x 10-6

-70

-60

-50

-40

-30

-20

-10

0

Time Delay [s]

Nor

mal

ized

LS

V (

γ) o

f PD

P [d

B]

0.53m LOS Tx: 2R Rx:2R0.94m LOS Tx: 2R Rx: 2L1.28m NLOS Tx: 2R Rx: 4M2.03m NLOS Tx: 0M Rx: 2L3.38m NLOS Tx: 0M Rx: 4M

Out-of-car

In-car

Fig. 4.18: Normalized LSV (𝛾) of the PDP for LOS/ NLOS measurements carriedout for different Tx-Rx separations for in/out-side measurements.

Tab. 4.9: Parameter values for LSV of in/out-side setups.

Tx-Rx Parameter valuesseparation (m) 𝐴 𝐵 (×104) 𝐶 𝐷 (×105)

0.53 -49.81 0.63 45.98 -7.050.94 -48.56 0.64 43.89 -6.471.28 -50.66 0.46 45.37 -5.202.03 -47.36 0.34 38.82 -3.683.38 -45.32 0.14 38.25 -4.28

GEV distribution [96]. The logistic family of distributions [97] appeared to be a goodfit in ray-cluster based modeling at 5 GHz [98] and modeling the probability densityfunctions (PDF) of the RMS delay spread at 2.35 GHz [99].The normal distributionwas also added as a reference. The PDF of these three distributions: the normal,the logistic, and the GEV are given in (4.4), (4.5), and (4.6), respectively.

𝑓(𝑥|𝜇, 𝜎) = 1𝜇

√2𝜋

exp[︃

−(𝑥 − 𝜇)2

2𝜎2

]︃(4.4)

𝑓(𝑥|𝜇, 𝜎) =exp

(︁𝑥−𝜇

𝜎

)︁𝜎[︁1 + exp

(︁𝑥−𝜇

𝜎

)︁]︁2 (4.5)

39

0 0.5 1 1.5 2 2.5 3 3.5 4

x 10-6

-15

-10

-5

0

5

10

Time Delay [s]

SS

V (

ξ) o

f PD

P [d

B]

0.53m LOS Tx: 2R Rx:2R0.94m LOS Tx: 2R Rx: 2L1.28m NLOS Tx: 2R Rx: 4M2.03m NLOS Tx: 0M Rx: 2L3.38m NLOS Tx: 0M Rx: 4M

Fig. 4.19: SSV (𝜉) of the PDP for LOS/ NLOS measurements carried out for differ-ent Tx-Rx separations for in/out-side measurements.

𝑓(𝑥|𝑘, 𝜇, 𝜎) = 1𝜎

exp(︁−𝛽−𝑘−1)︁

𝛽−1−𝑘−1 (4.6)

where 𝛽 = 1 + 𝑘 𝑥−𝜇𝜎

. The parameters, 𝜇, 𝜎, and 𝑘 are the location, scale, and shapeparameters, respectively.

The two-sample Kolmogorov-Smirnov (K-S) test [100] was performed to compareSSV and distributions. From Table 4.10 𝑝-values for the logistic distribution insidethe car and for the GEV distribution outside the car are highest. Figure 4.20shows the cumulative distribution function (CDF) of measured SSV fitted withthree distributions for results measured at different distances. The logistic and theGEV are the best candidates among the three for describing the SSV in in-car andout-of-car situations, respectively.

4.3.3 BER-performance of the 802.11p over Intra-VehicleChannel

For BER simulation MATLAB SIMULINK was used. A detailed description of thesimulation methodology and parameters for generating the signal can be found inSection 4.1.2 and in [84]. A block-scheme diagram that corresponds to the SISOtransmission model is presented in Figure 4.1.

40

Tab. 4.10: Two sample K-S test 𝑝 values for fitting SSV with different continuousdistributions.

Tx Tx-Rx GEV Logistic Normalposition separation (m)

0.53 0.3456 0.8838 0.7085In-car 0.94 0.8838 0.8838 0.7085

1.28 0.7085 0.7085 0.8838Out-of-car 2.03 0.9997 0.6297 0.5142

3.38 0.8838 0.7085 0.8838

-6 -4 -2 0 2 40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cu

mul

ativ

e P

rob

abili

ty

DataLogisticGEVNormal

-6 -4 -2 0 2 4 60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cu

mul

ativ

e P

rob

abili

ty

DataLogisticGEVNormal

-10 -5 0 5 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)C

umu

lativ

e P

rob

abi

lity

DataLogisticGEVNormal

-15 -10 -5 0 5 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cum

ula

tive

Pro

ba

bilit

y

DataLogisticGEVNormal

-10 -8 -6 -4 -2 0 2 4 6 80

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cum

ula

tive

Pro

ba

bilit

y

DataLogisticGEVNormal

(a) (b) (c) (d) (e)

-6 -4 -2 0 2 40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cu

mul

ativ

e P

rob

abili

ty

DataLogisticGEVNormal

-6 -4 -2 0 2 4 60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cu

mul

ativ

e P

rob

abili

ty

DataLogisticGEVNormal

-10 -5 0 5 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cum

ula

tive

Pro

ba

bilit

y

DataLogisticGEVNormal

-15 -10 -5 0 5 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cum

ula

tive

Pro

ba

bilit

y

DataLogisticGEVNormal

-10 -8 -6 -4 -2 0 2 4 6 80

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SSV (ξ)

Cum

ula

tive

Pro

ba

bilit

y

DataLogisticGEVNormal

(a) (b) (c) (d) (e)

Fig. 4.20: Empirical CDF of SSV fitted with three statistical distributions (logistic,GEV, and normal) for different Tx-Rx separations, from left to right, top: d = 0.53m, d = 0.94 m, d = 1.28 m, bottom: d = 2.03 m, d = 3.38 m.

For the channel block in Figure 4.1, the tap gains were calculated by samplingthe LSV, after every 10 ns. The respective CIRs are shown in Table 4.11. It is

41

interesting to find that real-life measurements exhibit a very fast decrease of the tapgains, similar to some earlier reports [101]. Thus the derived tap-delay line modelexhibits power-delay characteristics found in LOS multipath propagation dominatedby a strong first peak and followed by several other peaks. This implies that we canuse a Rician model. Rician distribution had been successfully utilized for non-geometrical stochastic modeling of vehicular channels [53]. A Rician model alsoencompasses the Rayleigh NLOS model as a special case and this can be seen inTable 4.11 that the for NLOS situations the tap gains do not decrease very fast.

We have not used the SSV (and the corresponding fitted distributions) for BERcalculation as BER is an average measure. In fact the idea to separate SSV wasto develop a model free from such local measurement variations. However, thevariations may be added by simulating an appropriate fitted distributions in order toaccurately reconstruct the PDP. This may be useful in studying other time domainchannel parameters such as RMS delay spread etc which is out of scope for thecurrent thesis.

Tab. 4.11: Characteristics of the channel models for in/out-side measurements.

0.53m LOS 0.94m LOS 1.28m NLOS 2.03m NLOS 3.38m NLOSK-factor (dB) 21.30 20.07 19.61 15.15 15.40Tap delay (ns) Tap gain (dB)

0 0 0 0 0 010 -11.78 -10.61 -9.61 -6.06 -6.6920 -17.69 -16.24 -14.90 -10.28 -11.0630 -20.69 -19.27 -18.27 -13.22 -13.9340 -22.26 -20.93 -20.32 -15.28 -15.8050 -23.11 -21.87 -21.58 -16.74 -17.0460 -23.62 -22.45 -22.39 -17.76 -17.8570 -23.95 -22.83 -22.91 -18.50 -18.4080 -24.20 -23.10 -23.27 -19.04 -18.7690 -21.41 -23.33 -23.53 -19.43 -19.01

The BER results for different distances between Tx and Rx, for LOS and NLOScases, and for intra-vehicle and out-of-vehicle situations are shown in Figure 4.21.Results can be classified quite unambiguously into two groups, one for intra-vehicleand the other for out-of-vehicle measurements. The BER worsens for higher Tx-Rx distance inside the car, but outside the vehicle the BER values almost remainunaffected when the Tx-Rx separation is altered. The BER results for intra-vehiclesimulations are slightly better than the results for the out-of-vehicle channel. For a

42

target BER of 10−5 the required (𝐸𝑏/𝑁0) differs by 6 dB.

-10 -5 0 5 10 15 20 25

10-6

10-5

10-4

10-3

10-2

10-1

100

BE

R

Eb/N

0 [dB]

0.53 m LOS Tx: 2R Rx: 2R0.94 m LOS Tx: 2R Rx: 2L1.28 m NLOS Tx: 2R Rx: 4M2.03 m NLOS Tx: 0M Rx: 2L 3.38 m NLOS Tx: 0M Rx: 4MAWGN

In-car

Out-of-car

Fig. 4.21: BER performance of IEEE 802.11p standard over the measured channels.

4.3.4 Comparison with UWB

UWB possesses a great potential for high-speed data communication and preciselocalization operations in cluttered closed-spaces hostile towards radio frequency(RF) signal propagation. Traditionally the PDP for UWB channels are describedwith a modified S-V model [74]. However, in a recent work by Demir et al. [102],the authors found that the PDP for UWB transmission in vehicular networks maybe characterized by segmenting the PDP (in dB scale) into several linear slopes.Inspired by these facts, we tried to compare our results for 802.11p protocol withmeasurements for UWB (3 GHz - 11 GHz) performed using the same VNA basedsetup. The goal was to find if one can attain similar LSV trends in wider bandwidthsas well.

The number of measured points were the same, however, due to the larger BW(8 GHz) and a frequency step size of 𝑓𝑠 = 100 MHz, we have a smaller time range,𝑡𝑑 = 1/𝑓𝑠 =10 ns. Thus we cannot compare the PDPs in an one-to-one basis.For analyzing LSV models we use a scaled comparison instead. For the UWBmeasurement, the propagation space resolution is 𝑃𝑠𝑟 = 𝑐/𝐵𝑤 = 3 cm, where 𝑐 isthe light speed, and the maximum propagation distance is 𝐿𝑑𝑚𝑎𝑥 = 𝑐/𝑓𝑠 = 30 m.

43

0 5 10 15 20 25 30 35 40-100

-90

-80

-70

-60

-50

-40

-30

-20

-10

0

Normalized Time Samples

Nor

mal

ized

PD

P [d

B]

1.28 m (802.11p)1.28 m (UWB)0.53 m (802.11p)0.53 m (UWB)

Fig. 4.22: The UWB and the 802.11p normalized PDP vs normalized time samples.

Figure 4.22 shows the UWB and the 802.11p normalized PDP vs normalizedtime samples, which used as basis instead of delay of MPCs. For the UWB, it wasfound that the PDP constitutes one (or a few) major peaks followed by somewhatlinear decreasing slope. The reader may also note that there are the delay of thefirst peak and lower power of PDP for the UWB. In addition, the delay and themaximum value of the peak are more for larger distances between the transmittingand receiving antennas. However, this phenomena is not observed in narrowband5.8 GHz measurements for 802.11p.

4.4 Around of vehicle channel measurement basedon the IEEE 802.11p and UWB systems

The Section presents out-of-vehicle channel measurement results in the 5.8 GHzfrequency band and UWB. Experiments for a total of 23 points for different distancesand different angles around the parked car are carried out. The LSV separated by thetwo-term exponential decay model is used to calculate the corresponding CIR andto construct a Rician tap-delay multipath channel model for 5.8 GHz band. BERperformance for the IEEE 802.11p for some measured locations is then evaluatedthrough MATLAB based simulation.

44

4.4.1 Out-of-vehicle channel measurement setup

Measurements are performed with one transmitter and three receiver antennasaround the car Skoda Octavia. The car is parked in the same underground garage asin Section 4.3. Experiments are realized with VNA following the basic principle asoutlined in [103]. For the measurements car is parked at the middle of the six floor,away from walls and other cars. Figure 4.23 presents equipments for measuring anddata recording and the car with closed doors and windows. Equipments are installedoutside the car on the opposite side of the measurement area.

VNATx Rx3

Rx1Rx2

Fig. 4.23: The vehicle under test and measuring equipments.

The measurements were performed in the ultra-wide band (3 GHz to 11 GHz)and for the 802.11p standard (5.8 GHz) in the same points for Tx and Rx, for thatreason the VNA Agilent Technologies E5071C was used. The Tx and Rx antennasare connected to the four ports of the VNA through phase stable coaxial cables. Theprinciples of measuring of the 𝑠-parameter values, 𝑠𝑥1; 𝑥 ∈ {2, 3, 4}, are describedalso in Section 4.2.1. The frequency domain data are converted to time domain CIRusing IFFT.

For measurements omnidirectional identical conical antennas are used. Figure4.14 shows the measured radiation patterns of antennas for 5.8 GHz band and forUWB. For both bands the H-planes are circular and there is no effect of antenna

45

orientation when antennas are placed at same height.The positions of the receiving antennas are displayed in Figure 4.24 The first

receiving antenna, Rx1, is installed on the roof of the vehicle near car radio antenna.The second receiver, Rx2, is mounted also on the roof at a distance of 1.13 m fromRx1. The third antenna, Rx3, is located on the car bonnet near windshield. Allthree antennas create one axis, Rx1 and Rx2 are installed at a height of 1.5 m andRx3 – 1 m.

1,5m

1,0m

1,04m 1,13mRx1Rx2

Rx3

Rx3

Rx1-Rx2

Fig. 4.24: Placement of the receiving antennas for around-vehicle measurements.

The transmitting antenna is set at a height of 1.5 m above ground (at the samelevel of Rx1 and Rx2) with the help of a tripod stand and its position is changed to23 different locations as depicted with blue solid circles in Fig. 4.25. The distancesof Tx from all three Rx antennas are listed in Table 4.12. Tx installation pointsfollow concentric semi-circles around Rx1 with an increasing radius of 3 m, 4.5 m, 6m, 7.5 m, and 9 m at an uniform angular separation of 45∘, i.e. at angles of 0∘, 45∘,90∘, 135∘, and 180∘ relative to the roll axis which connects Rx1 with Rx3 via Rx2.This arrangement ensures that there always exists a direct LOS path between Tx

46

1.5m

3m

4.5m

6m

7.5m

9m

RX1RX2RX3

Fig. 4.25: Placement of the transmitting antenna for around-vehicle measurements.

and Rx1 (as well as between Tx and Rx2) whereas the path between Tx and Rx3 ismostly NLOS with LOS path being available at only certain angles and distances.

4.5 Out-of-vehicle channel description in 5.8 GHzband

The PDP is calculated from CIR and described in previous Section 4.3. Figure 4.26presents the measured normalized PDP for measurement points for LOS conditionoutside of the car for 𝑇𝑥1. The power reduces to about 45 dB in the considereddelay range, from -8 to -53 dB. The noise for non-normalized PDP is -110 dB leveland corresponds to previous out-of-vehicle measurements.

Following previous Section 4.3, the measured PDP at each location can be ex-pressed as a sum of LSV and SSV. The LSV are separated from PDP by two-term-exponential model (4.2) and presented in Figure 4.27. It can be seen that the LSVcurves form a distinct group. Table 4.13 shows the parameters of the model fordifferent measurement points.

4.5.1 BER-performance of the 802.11p over Out-of-vehicleChannel

For constructing of BER as a function of (𝐸𝑏/𝑁0) the simulation model was usedfrom Section 4.1.2. Simulation was prepared in MATLAB utilizing system parame-

47

Tab. 4.12: Distances between transmitting and receiving antennas

Measurement Angle Distance Distance DistanceNo. (degrees) for Rx1 (m) for Rx2 (m) for Rx3 (m)1 4.5 3.37 2.332 0 6.0 4.87 3.823 7.5 6.37 5.324 9.0 7.87 6.825 3.0 2.34 2.146 4.5 3.79 3.357 45 6.0 5.26 4.728 7.5 6.75 6.169 9.0 8.24 7.6210 3.0 3.21 3.7311 90 4.5 4.64 5.0112 6.0 6.11 6.3913 7.5 7.58 7.8214 3.0 3.88 4.8115 4.5 5.36 6.2516 135 6.0 6.85 7.7117 7.5 8.34 9.1818 9.0 9.83 10.6619 3.0 4.13 5.220 4.5 5.63 6.6921 180 6.0 7.13 8.1922 7.5 8.63 9.6923 9.0 10.13 11.19

ters based on the recommendations for the IEEE 802.11p protocol and described indetail in [84] and Section 4.1.2.

Following Section 4.3 LOS cases are described by the Rician distributions [53].To construct a tap-delay multipath Rician channel the LSV was used. Table 4.14presents the calculated 10 tap gains and 𝐾 factors for different positions.

The BER performance using the Rician channel based on the data in Table 4.14for six locations are presented in Figure 4.28. The BER curves can be groupedeasily as the maximum difference is less than 4 dB at a BER = 10−4. An averagechannel BER performance was calculated over channel based on average coefficientsof the LSV. The average channel error result has non-significant deviation from BERcurves for six locations and can effectively serve as a design basis for intra-vehicle

48

Fig. 4.26: The PDPs for measurements points in 5.8 GHz band.

Fig. 4.27: LSV of PDP for measurements points in 5.8 GHz band for out-of-vehiclemeasurements.

link budget calculations.

49

Tab. 4.13: Parameters of LSV for measurements points in 5.8 GHz band for out-ofvehicle measurements.

No. of ParametersMeasurement 𝐴 𝐵 𝐶 𝐷

1 -109.3 0.002080 34.01 -0.28722 -114.9 0.001692 35.29 -0.25953 -112.0 0.002545 35.72 -0.43454 -114.6 0.001976 29.04 -0.24925 -111.7 0.001641 40.39 -0.27806 -121.6 0.003020 38.98 -0.4942

Tab. 4.14: Parameters for Out-of-vehicle channel.Tap Tap gain (dB)

delay (ns) 1 2 3 4 5 60 0 0 0 0 0 010 -4.36 -4.34 -6.73 -3.31 -4.99 -7.7820 -7.65 -7.65 -11.01 -5.92 -8.79 -12.6030 -10.14 -10.17 -13.75 -7.98 -11.70 -15.6140 -12.03 -12.10 -15.53 -9.61 -13.92 -17,5250 -13.48 -13.59 -16.70 -10.91 -15.62 -18.7660 -14.59 -14.74 -17.48 -11.94 -16.93 -19.5970 -15.46 -15.63 -18.03 -12.78 -17.95 -20.1780 -16.13 -16.34 -18.43 -13.45 -18.74 -20.6090 -16.67 -16.91 -18.74 -14.00 -19.37 -20.93

𝐾 factor (dB) 17.93 18.24 19.86 15.43 20.36 22.40

4.6 Out-of-vehicle channel description in UWB

4.6.1 Path Loss

The PL (𝛼) data is computed from VNA record through averaging the inverse ofsquared CTF over the whole frequency range

𝛼 = 10 log10

⎡⎣ 1𝑁VNA

𝑁VNA∑︁𝑛=1

1|𝐻(𝑓𝑛)|2

⎤⎦ (4.7)

where the index 𝑛 is used to signify discrete tones generated by VNA.The PL values calculated from (4.7) are listed in Table 4.15. Unavailability

of data is occurred at (9m, 90∘) due to walls and at (3m, 0∘) the car bonnet ob-

50

Fig. 4.28: BER results based on the 802.11p over measured channels.

structed the measurement location. As expected, the PL value increase monoton-ically with Tx-Rx separation following a log-distance model for a given angle andantenna height. However, for a particular distance, random PL variation aroundthe average loss value is observable when the angle is changed. This indicates thatreceived signal strength indication (RSSI) based ranging and localization algorithmswould perform poorly around a car. Another interesting aspect is, although the el-evation and Tx-Rx separation do not change much with antenna height, there isa large increase in the PL when the Tx antenna is brought down from 1.5m (Rxantenna level) to 1m above the ground.

4.6.2 PDP Clusters

In this subsection, we would analyze the PDP variability, when the PDP can becharacterized with clustered propagation as described by the S-V model. Accordingto the S-V model, the discrete impulse response may be expressed as

ℎ(𝑡) =𝑁𝑐∑︁

𝑛=1

𝑁𝑟,𝑛∑︁𝑚=1

𝛽𝑚,𝑛 exp(𝑗𝜃𝑚,𝑛)𝛿(𝑡 − 𝑇𝑛 − 𝜏𝑚,𝑛) (4.8)

where 𝑁𝑐 is the number of clusters, 𝑁𝑟,𝑛 is the number of rays in the 𝑛th cluster,and 𝑇𝑛 is the arrival time of the 𝑛th cluster. The magnitude, phase, and addi-tional delay of the 𝑚th ray within the 𝑛th cluster are given by 𝛽𝑚,𝑛, 𝜃𝑚,𝑛, and 𝜏𝑚,𝑛,respectively. The inter- and intra-cluster exponential decay rates, namely Γ and𝛾, define the magnitude of individual rays according to 𝛽2

𝑚,𝑛 = 𝛽21,1 exp[−(𝑇𝑛 −

51

Tab. 4.15: PL values (in dB) for all Tx locations.

Tx height (ℎ) = 1.5mTx-Rx Tx angles

distance (m) 0∘ 45∘ 90∘ 135∘ 180∘ Average3.0 - 58.57 58.99 60.99 60.82 59.844.5 61.82 62.65 62.58 64.47 64.13 63.136.0 64.30 65.16 64.91 66.10 66.24 65.347.5 65.94 67.23 66.65 68.41 68.03 67.259.0 67.42 68.31 - 70.02 70.01 68.94

Tx height (ℎ) = 1mTx-Rx Tx angles

distance (m) 0∘ 45∘ 90∘ 135∘ 180∘ Average3.0 - 64.33 61.73 62.10 61.38 62.394.5 66.50 66.34 64.32 64.45 64.13 65.156.0 67.49 67.83 66.27 66.54 66.45 66.927.5 68.65 69.51 67.93 68.19 68.46 68.559.0 69.67 69.91 - 69.69 69.53 69.70

𝑇1)/Γ] exp(−𝜏𝑚,𝑛/𝛾). The arrival time of clusters and rays within clusters fol-low independent Poisson processes, i.e. Pr(𝑇𝑛|𝑇𝑛−1) = Λ exp[−Λ(𝑇𝑛 − 𝑇𝑛−1)] andPr(𝜏𝑚,𝑛|𝜏𝑚−1,𝑛) = 𝜆 exp[−𝜆(𝜏𝑚,𝑛 − 𝜏𝑚−1,𝑛)], where Pr(·) denotes probability, and Λand 𝜆 represent cluster and ray arrival rates, respectively.

Figures 4.29 and 4.30 depict the PDPs at two indicative distances, namely at3 m and 9 m, for the three different Tx angles, 45∘, 135∘, and 180∘. The MPCsin the PDPs are grouped into clusters, and the overall PDPs are described withS-V channel model. The individual clusters in each PDP are calculated using theautomatic clustering algorithm described in [104] and validating the correctness ofthe algorithm by visual inspection. The automatic clustering algorithm is based onidentifying the changes in the slope of the impulse response based on the assumptionthat all changes in the slope correspond to the start of a new cluster.

From Figure 4.29 one can find that the cluster arrival times do not alter signif-icantly with Tx angle. However, the number of resolvable clusters reduces from 4to 3, when antenna height is changed from 1 m to 1.5 m. For longer distances, asseen in Figure 4.30, the signal reaches the noise level (-110 dB) within a short delayspan and only two distinct clusters are observable. The effect of antenna height isalso no longer visible when Tx-Rx separation is large.

In Table 4.16, we put the S-V parameter values computed using the mathematical

52

0 5 10 15-120

-100

-80

-60

Delay [ns]

PD

P [

dB

] h = 1m, d = 3m, angle 450

0 5 10 15-120

-100

-80

-60

Delay [ns]

PD

P [

dB

]

h = 1m, d = 3m, angle 1350

0 5 10 15-120

-100

-80

-60

Delay [ns]

PD

P [

dB

]

h = 1m, d = 3m, angle 1800

0 5 10 15-120

-100

-80

-60

Delay [ns]

PD

P [

dB

]

h = 1.5m, d = 3m, angle 450

PDPLocal peaksCluster

0 5 10 15-120

-100

-80

-60

Delay [ns]P

DP

[d

B]

h = 1.5m, d = 3m, angle 1350

0 5 10 15-120

-100

-80

-60

Delay [ns]

PD

P [

dB

]

h = 1.5m, d = 3m, angle 1800

Fig. 4.29: PDPs for Tx-Rx separation (𝑑) = 3m.

expressions given in [105]. The variations with angles, as described before, are notlarge and is not mentioned in the table. The ray decay, 𝛾, for 1 m height is lowerthan 1.5 m, as there are more clusters with shorter durations, i.e. having a larger1/Λ value. However, 𝛾 increases with distance for 1 m as number of clusters reduce,which is not necessarily the case for 1.5 m. The cluster decay, Γ, decreases withdistance for both antenna heights, as for larger distances there are only two clusterheads and the decay slope is high. The ray arrival time is almost constant for all thecases. The values are also quite close to the values for UWB channel in the range of 4-10 m (CM3), Γ = 14, 𝛾 = 7.9, 1/Λ = 15, 1/𝜆 = 0.48, all in nanoseconds, as reportedin the IEEE 802.15.3 standard [106]. A direct comparison is not straightforwardas the measurements for the standard are done for NLOS whereas most of ourmeasurements had a clear LOS path.

53

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

] h = 1m, d = 9m, angle 450

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

]

h = 1m, d = 9m, angle 1350

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

]

h = 1m, d = 9m, angle 1800

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

]

h = 1.5m, d = 9m, angle 450

PDPLocal peaksCluster

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

]

h = 1.5m, d = 9m, angle 1350

0 5 10-120

-100

-80

Delay [ns]

PD

P [

dB

] h = 1.5m, d = 9m, angle 1800

Fig. 4.30: PDPs for Tx-Rx separation (𝑑) = 9m.

Tab. 4.16: S-V parameter values

Tx height (ℎ) = 1.5mS-V Tx-Rx distance (𝑑) in m

parameters (ns) 3.0 4.5 6.0 7.5 9.0Γ 18.80 19.54 25.46 4.60 2.46𝛾 10.43 8.81 5.31 7.57 4.55

1/Λ 4.99 4.93 6.67 5.60 5.881/𝜆 0.40 0.47 0.52 0.45 0.36

Tx height (ℎ) = 1.5mS-V Tx-Rx distance (𝑑) in m

parameters (ns) 3.0 4.5 6.0 7.5 9.0Γ 16.19 16.70 11.56 12.52 3.18𝛾 4.99 6.31 6.16 6.32 5.67

1/Λ 3.98 5.01 5.34 4.48 5.881/𝜆 0.42 0.48 0.43 0.43 0.39

54

5 LOCALIZATION USING OUT-OF-VEHICLEMEASUREMENTS IN UWB

5.1 Distance estimationThe frequency domain data obtained through the experiments consists of discretechannel frequency response recorded at equally spaced 𝑁(= 801) points spanningthe entire ultra-wide bandwidth of 𝐵𝑤 = 8 GHz. Post-processing of this frequencydomain data involves IFFT operation in combination with a rectangular window.This window is generally recommended for applications where good separation be-tween the MPCs in the resultant PDP is required [91]. The corresponding distanceresolution is given by the ratio

𝑑𝑟 = 𝑐

𝐵(5.1)

where 𝑐 = 2.998 × 108 m/s and inverse of the bandwidth is equal to the timeresolution

𝑇𝑟 = 1𝐵

= 1𝑓max − 𝑓min

(5.2)

where 𝑓min(= 3GHz) and 𝑓max(= 11GHz) are the lowest and highest frequency pointsof the data set. The maximum propagation distance measured

𝑑max = 𝑐

𝑓𝑠

(5.3)

depends on the frequency step 𝑓𝑠; 𝑓𝑠 = 𝐵/(𝑁 − 1).For calculating the distances between Tx and Rx antennas TOA technique is

used. The TOA is based on the detection of the first ray received by a particularRx antenna. This is achieved by a search algorithm that compares individual signalsamples of the PDP with a certain threshold to avoid noise and identify the firstamplitude peak that corresponds to the LOS time delay, 𝑇𝑑. The distance betweenTx and Rx antennas is calculated by

𝑑 = 𝑇𝑑 × 𝑐 (5.4)

In Figure 5.1, we superimpose all PDPs corresponding to Rx1 obtained for dif-ferent Tx antenna positions. The PDPs are plotted in the distance domain insteadof the delay domain for better understanding. The PDPs for same Tx-Rx separationbut with different angles with Rx1-Rx2-Rx3 axis form distinct groups. Thus we mayconclude that there is almost no effect of angular variation in distance estimation.Further, the peaks in the same group overlaps with each other and is very close tothe actual distance shown with red dotted lines.

55

0 1 2 3 4 4.5 5 6 7 7.5 8 9 10-70

-65

-60

-55

-50

-45

-40

-35

-30

-25

-20

-15

Distance [m]

PDP

[dB

]

Fig. 5.1: Superposition of PDPs for Rx1 for all Tx positions.

Table 5.1 gives the real and calculated distance between Tx and Rx1 for all the 23different measurement points. In most of the cases the deviation is about 2.5% withthe maximum deviation being 5% which translates to an error of 15 cm at a distanceof 3 m. The same information for Rx2 and Rx3 are given in Tables 5.2 and 5.3 wherea maximum of 6% deviation is noticed. The maximum deviation is for measurementno. 10 for all three Rx antennas, and the deviation for all other measurements arevery low indicating some measurement error that might have occurred during thatparticular measurement. Even though the error is 15 cm, 16 cm, and 17 cm for Rx1,Rx2, and Rx3 respectively, which is small compared to the dimension of the vehicle.

While analyzing Tables 5.1, 5.2 and 5.3, one may find a positive bias in thecalculated distance, i.e. the calculated distance is always longer than the actualdistance. It is interesting to see some correlation among the error statics for allthe three Rx antennas, both on on the roof and on the windshield. This may berelated with the calibration plane and phase center of the antenna. The VNA wascalibrated only for the coaxial cables without considering coaxial adapters of theantennas. Also due to the discrete nature of the PDP time axis, the first peakcan not be detected accurately. The distances between Rx and Tx antennas weremeasured by a ruler, and there is also a possibility of error in the measurement.

56

Tab. 5.1: Distance of Rx1 from Tx antenna positions

Measurement Position Real Measured Deviation DeviationNo. (degrees) distance (m) distance (m) (m) (%)1 4.5 4.6125 0.11 2.52 0 6.0 6.1125 0.11 1.883 7.5 7.65 0.15 24 9.0 9.15 0.15 1.675 3.0 3.0375 0.04 1.256 4.5 4.6125 0.11 2.57 45 6.0 6.15 0.15 2.58 7.5 7.6125 0.11 1.59 9.0 9.225 0.23 2.510 3.0 3.15 0.15 511 90 4.5 4.6125 0.11 2.512 6.0 6.15 0.15 2.513 7.5 7.6125 0.11 1.514 3.0 3.1125 0.11 3.7515 4.5 4.575 0.08 1.6716 135 6.0 6.1125 0.11 1.8817 7.5 7.6125 0.11 1.518 9.0 9.225 0.23 2.519 3.0 3.0375 0.04 1.2520 4.5 4.5 0.0 021 180 6.0 6.0375 0.04 0.6222 7.5 7.575 0.08 123 9.0 9.075 0.07 0.83

5.1.1 Delay distance dependence

Figure 5.2 shows the linear relationship between the Tx-Rx distance (𝑑) and theassociated delay (𝑇𝑑). The polynomial fitting of measurement results allows us toconstruct substantially similar linear functions for all receivers in the form:

𝑑 = 𝑎 × 𝑇𝑑 + 𝑏 (5.5)

The results of the fitting are shown in Table 5.4. The calculated coefficients, 𝑎 and𝑏, are fairly constant for all three Rx while sum of squares due to error (SSE) aswell as and root mean square error (RMSE) are close to zero and the coefficientof determination (𝑅-square) is close to unity. This indicates the goodness of fit

57

Tab. 5.2: Distance of Rx2 from Tx antenna positions

for Rx2

Measurement Real Measured DeviationNo. distance (m) distance (m) (%)1 3.37 3.45 2.372 4.87 4.95 1.643 6.37 6.49 1.844 7.87 8.02 1.915 2.34 2.36 0.96 3.79 3.86 2.017 5.26 5.4 2.628 6.75 6.86 1.699 8.24 8.5 3.1610 3.21 3.37 5.2511 4.64 4.76 2.6512 6.11 6.22 1.9613 7.58 7.69 1.3614 3.88 4.01 3.3615 5.36 5.44 1.4716 6.85 6.94 1.3417 8.34 8.44 1.218 9.83 10.05 2.2219 4.13 4.2 1.6920 5.63 5.62 0.0921 7.13 7.2 0.9822 8.63 8.74 1.2523 10.13 10.24 0.06

and based on the coefficients for the linear function, it is possible to interpolate thedistance for other values of LOS path delay.

5.2 Two dimensional localizationNext, we describe the simple two dimensional (2-D) localization performed fromthe distance estimates as obtained from the first peak detection of PDPs in thedistance domain. Two receivers on the car roof, Rx1 and Rx2, are used to locatethe Tx antenna as the Tx antenna is placed at same height, and it is sufficient to

58

Tab. 5.3: Distance of Rx3 from Tx antenna positions

for Rx3

Measurement Real Measured DeviationNo. distance (m) distance (m) (%)1 2.33 2.47 6.222 3.82 3.94 3.083 5.32 5.47 2.914 6.82 7.01 2.825 2.14 2.21 3.226 3.35 3.45 3.077 4.72 4.91 48 6.16 6.3 2.319 7.62 7.87 3.3810 3.73 3.9 4.6911 5.01 5.1 1.7412 6.39 6.49 1.4713 7.82 7.91 1.214 4.81 4.95 2.8615 6.25 6.34 1.4116 7.71 7.8 1.1717 9.18 9.26 0.8618 10.66 10.87 1.9819 5.2 5.25 1.0320 6.69 6.71 0.2721 8.19 8.25 0.6922 9.69 9.79 0.9823 11.19 11.29 0.85

Tab. 5.4: Parameters of polynomial fitting

Receiver a b SSE R-square RMSERx1 3.38 ×10−9 6.81 ×10−11 7.96 ×10−19 0.9993 1.86 ×10−10

Rx2 3.38 ×10−9 9.65 ×10−11 9.36 ×10−19 0.9993 2.02 ×10−10

Rx3 3.34 ×10−9 3.27 ×10−10 1.062 ×10−18 0.9993 2.15 ×10−10

Average 3.36 ×10−9 1.75 ×10−10 2.92 ×10−18 0.9993 2 ×10−10

find only 2-D coordinates. Figure 5.3 summarizes the geometrical topology wherethe co-ordinate system has the unknown Tx antenna node 𝐵(𝑥𝑇 𝑥, 𝑦𝑇 𝑥) at the origin

59

2 3 4 5 6 7 8 9 10 110.5

1

1.5

2

2.5

3

3.5

4x 10-8

Distance [m]

Del

ay [s

]

Rx1 calculated

Rx1 measured

Rx2 measured

Rx2 calculated

Rx3 measured

Rx3 calculated

Fig. 5.2: Delay versus distance.

and node 𝐴(𝑥𝑅𝑥1 , 𝑦𝑅𝑥1) and node 𝐵(𝑥𝑅𝑥2 , 𝑦𝑅𝑥2) denote the location of Rx1 and Rx2,respectively. It may be noted that 𝑦𝑅𝑥1 = 𝑦𝑅𝑥2 , because Rx1 and Rx2 are on thesame axis, and 𝑥𝑅𝑥1 − 𝑥𝑅𝑥2 = 1.13m, as evident from the placement in Figure 4.24.

y

xB(xTx, yTx) xRx1

yRx1

xRx2

C(xRx2, yRx2

) A(xRx1, yRx1

)

α

β

1

Fig. 5.3: The principle of 2D localization.

As the actual co-ordinates of Rx1 are known (relative to the parking lot floorplan), it is enough to find the relative co-ordinates (𝑥𝑅𝑥1 , 𝑦𝑅𝑥1) in order to locate

60

the Tx antenna. Let 𝛼 be the angle between 𝐴𝐵 and 𝑋−axis, then

𝑥𝑅𝑥1 = 𝐵𝐴 cos 𝛼 (5.6)

𝑦𝑅𝑥1 = 𝐵𝐴 sin 𝛼 (5.7)

From the theory of parallel line angles, it is easy to show that the angle 𝛼 is equalto an alternate angle 𝛽 between 𝐶𝐴 and 𝐵𝐴. Also it is possible to find the angle 𝛽

from the sides of Δ𝐴𝐵𝐶

𝛽 = cos−1(︃

𝐶𝐴2 + 𝐵𝐴2 − 𝐵𝐶2

2 × 𝐶𝐴 × 𝐵𝐴

)︃(5.8)

The side 𝐶𝐴 is fixed and equal to 𝑥𝑅𝑥1 − 𝑥𝑅𝑥2 , whereas 𝐵𝐴 and 𝐵𝐶 are distancescalculated using LOS delays of PDP of 𝑅𝑥1 and 𝑅𝑥2, respectively. Thus, combining(5.6, 5.7) and (5.8) through the relation 𝛼 = 𝛽, we obtain

𝑥𝑅𝑥1 = 𝐵𝐴 × cos[︃cos−1

(︃𝐶𝐴2 + 𝐵𝐴2 − 𝐵𝐶2

2 × 𝐶𝐴 × 𝐵𝐴

)︃]︃

= 𝐵𝐴 × 𝐶𝐴2 + 𝐵𝐴2 − 𝐵𝐶2

2 × 𝐶𝐴 × 𝐵𝐴= 𝐶𝐴2 + 𝐵𝐴2 − 𝐵𝐶2

2 × 𝐶𝐴

(5.9)

and𝑦𝑅𝑥1 = 𝐵𝐴 × sin

[︃cos−1

(︃𝐶𝐴2 + 𝐵𝐴2 − 𝐵𝐶2

2 × 𝐶𝐴 × 𝐵𝐴

)︃]︃(5.10)

Fig. 5.4: Actual and estimated position of the Tx antenna.

In the second step, a detailed map of calculated Tx coordinates are constructedin Fig. 5.4 and compared with the measured coordinates. The goal of constructing

61

this 2-D localization map is to assess whether the simple TOA algorithm resultscorresponds at least roughly to reality, i. e. whether it possible to reliably detect avehicle on a particular parking lot. In our situation it is not necessary to have anaccurate coordinates (in mm) but only rough estimation of the car position (in cm).Table 5.5 lists the measured and calculated coordinates of Tx antenna along withthe deviation for each measured location. The deviation is few centimeters at mostand indicates that UWB is robust in locating other cars in the vicinity.

The main problem associated with the localization using two receivers is theinability to determine the zone location of the transmitter relative to the receiveraxis. However, in this particular case, it was not a serious issue as we consideredonly one side of the vehicle with LOS scenario for roof-antennas.

Tab. 5.5: Comparison between measured and calculated coordinates of Tx

Measurement 𝑥𝑎 𝑦𝑎 𝑥𝑎 𝑦𝑎 𝑥𝑎 𝑦𝑎

number measured measured calculated calculated deviation deviation1 4.50 0.00 4.71 0.00 0.21 0.002 6.00 0.00 6.26 0.00 0.26 0.003 7.50 0.00 7.84 0.00 0.34 0.004 9.00 0.00 9.11 0.00 0.11 0.005 2.12 2.12 2.18 2.12 0.05 0.006 3.17 3.19 3.38 3.14 0.21 -0.057 4.25 4.23 4.40 4.30 0.15 0.078 5.29 5.31 5.37 5.40 0.07 0.089 6.36 6.37 5.87 6.78 -0.49 0.4210 -0.01 3.00 -0.08 3.15 -0.07 0.1511 0.00 4.50 -0.06 4.61 -0.06 0.1112 -0.02 6.00 0.15 6.15 0.18 0.1513 0.03 7.50 0.06 7.61 0.03 0.1114 -2.11 2.13 -2.27 2.13 -0.16 0.0015 -3.19 3.18 -3.26 3.21 -0.07 0.0416 -4.27 4.22 -4.20 4.44 0.07 0.2217 -5.32 5.28 -5.30 5.47 0.03 0.1918 -6.35 6.38 -6.47 6.57 -0.12 0.2019 -3.00 0.00 -3.16 0.00 -0.16 0.0020 -4.50 0.00 -4.48 0.47 0.02 0.4721 -6.00 0.00 -6.24 0.00 -0.24 0.0022 -7.50 0.00 -7.83 0.00 -0.33 0.0023 -9.00 0.00 -9.37 0.00 -0.37 0.00

62

6 CONCLUSIONThis dissertation thesis is focused on the wireless systems aimed at intra-vehicle andout-of-vehicle communications operating in narrow- and ultra-wide bands.

A description of measurements results as well as the description of a channelmeasurement setups realized for the 5 GHz band and UWB systems are given in thestate-of-the-art part. To provide a more specific insight on vehicle communicationand localization systems, the IEEE 802.11p standard and 3-11 GHz ultra wide bandsystems have been chosen as indicative examples which currently happens to be themajor and the most promising vehicular communication systems, respectively.

When summarized the main outcomes of the dissertation are:• Simulation models describing IEEE 802.11p and 802.11a standards are created

and the simulation results are compared. The BER performance gives the bestresults for the model 802.11p for channels with large delay. The 802.11p proto-col is more suitable for vehicle communications, where a signal is transmittedover long distances. This is confirmed during the calculation of the BER fordifferent speeds of movement of the receiver experiencing the Doppler effect.

• Extensive intra-vehicle channel measurements in the 5.8 GHz band were car-ried out. A double exponential decay model was used for describing the basictrend of the measured PDPs and we utilized it for IEEE 802.11p BER sim-ulation. The BER achieves the recommended values for all variants of thechannel, and it can be concluded that 802.11p standard (in particular thePHY protocol stack) can be adopted for in-vehicle communication systems.

• Channel measurements outside car were carried out in the 5.8-GHz and 3-11GHz bands in an underground garage with different Tx-Rx locations.For the IEEE 802.11p channel model, a double exponential decay model wasused to separate LSV from measurement PDP. The LSVs were utilized to simu-late BER and the BER results satisfy the recommended values for all locations.An average channel model is proposed that can be used for communicationlink design outside the car.Results for a comprehensive measurement campaign to study the variations ofthe channel characteristics are reported for UWB propagation around a parkedcar. It was found that MPCs can be roughly grouped into clusters followingthe well-known S-V model. The cluster profile do not alter significantly withangle but changes with antenna height. Also, the number of clusters reducewith distance as it becomes difficult to segregate clusters. PL values and delayof the first arriving MPC are also reported which serve as guidelines for rangingalgorithms.

• Based on the measurements results in 3-11 GHz band it is possible to uti-

63

lize a distance-delay linear relationship which is exploited to find distance ofthe transmitting antenna from receiver antennas. The estimated distances areused for TOA based 2-D localization with only two receivers and the calcu-lated coordinates exhibits small error. The proposed solution is sufficient todetermine the positions of nearby vehicles in an underground garage. Obvi-ously the simple mechanism is unable to provide the information whether thecar is on left/right side of the roll axis. However, an extension with a thirdreceiver which is not on the same axis can solve the problem.

64

LIST OF ABBREVIATIONS2G Second Generation

3G Third Generation

4G Fourth Generation

ASTM American Society for Testing and Materials

BER Bit Error Rate

BPSK Binary Phase Shift Keying

C2C-CC Car 2 Car Communication Consortium

CDF Cumulative Distribution Function

CDMA Code Division Multiple Access

CIR Channel Impulse Response

CTF Channel Transfer Function

DSRC Dedicated Short Range Communications

FCC Federal Communications Commission

FEEC Faculty of Electrical Engineering

FFT/IFFT fast Fourier transform / inverse fast Fourier transform

GEV Generalized Extreme Value

GLONASS GLObal NAvigation Satellite System

GNSS Global Navigation Satellite System

GPRS General Packet Radio Service

GSM Global System for Mobile Communications

ISI Intersymbol interference

FFT/IFFT fast Fourier Transform / inverse

ITS Intelligent Transportation System

65

ITU–R International Telecommunication UnionRadiocommunication Sector

K-S test Kolmogorov-Smirnov test

LOS Line-of-Sight

LS Least Squares

LSV Large-Scale Variation

LTE Long-Term Evolution

M-2-M Machine to Machine

MIMO Multiple Input Multiple Output

MPC Multipath Components

MSE Mean Square Error

NLOS non Line-of-Sight

OFDM Orthogonal Frequency-Division Multiplexing

OSI Open Systems Interconnection

PDF Probability Density Function

PDP Power Delay Profile

PHY Physical Layer

PL Path loss

QAM Quadrature Amplitude Modulation

QPSK Quadrature Phase Shift Keying

RF Radio frequency

RMS Root Mean Square

RMSE Root Mean Square Error

RSSI Received Signal Strength Indication

Rx Receiver

66

SIMO Single Input Multiple Output

SISO Single Input Single Output

SSE Squares Due to Error

SSV Small-Scale Variation

TOA Time of Arrival

Tx Transmitter

UWB Ultra-Wide Band

V-2-I Vehicle to Infrastructure

V-2-V Vehicle to Vehicle

V-2-X Vehicle to X

VAN Vehicle Area Network

VANET Vehicular Ad Hoc Network

VNA Vector Network Analyzer

WAVE Wireless Access in Vehicular Environments

WCDMA Wideband Code Division Multiple Access

WLAN Wireless Area Network

67

REFERENCES[1] H. Hartenstein and K. Laberteaux, VANET: vehicular applications and inter-

networking technologies. Wiley Online Library, 2010, vol. 1.

[2] H. J. Miller and S.-L. Shaw, Geographic information systems for transporta-tion: principles and applications. Oxford University Press on Demand, 2001.

[3] I. V. T. Society, “Ieee guide for wireless access in vehicular environments(wave) - architecture,” IEEE Std 1069 P1609.0/D7.0, Mar. 2014.

[4] Y. J. Li, “An overview of the dsrc/wave technology,” in Quality, Reliability,Security and Robustness in Heterogeneous Networks. Springer, 2010, pp.544–558.

[5] C. . C. C. Consortium, “Overview of the c2c-cc system,” Manifesto, Aug. 2007.

[6] J. Gozalvez, M. Sepulcre, and R. Bauza, “IEEE 802.11p vehicle to infrastruc-ture communications in urban environments,” IEEE Commun. Magz., vol. 50,no. 5, pp. 176–183, May 2012.

[7] M. Faezipour, M. Nourani, A. Saeed, and S. Addepalli, “Progress and chal-lenges in intelligent vehicle area networks,” Commun. ACM, vol. 55, no. 2, pp.90–100, Feb. 2012.

[8] H. Hartenstein and K. P. Laberteaux, “A tutorial survey on vehicular ad hocnetworks,” IEEE Commun. Magz., vol. 46, no. 6, pp. 164–171, Jun. 2008.

[9] C. F. Mecklenbrauker, A. F. Molisch, J. Karedal, F. Tufvesson, A. Paier,L. Bernado, T. Zemen, O. Klemp, and N. Czink, “Vehicular channel charac-terization and its implications for wireless system design and performance,”Proc. IEEE, vol. 99, no. 7, pp. 1189–1212, Jul. 2011.

[10] M. Govoni, F. Guidi, E. M. Vitucci, V. Degli Esposti, G. Tartarini, and D. Dar-dari, “Ultra-wide bandwidth systems for the surveillance of railway crossingareas,” IEEE Communications Magazine, vol. 53, no. 10, pp. 117–123, October2015.

[11] S. Baghaee, S. Zubeyde Gurbuz, and E. Uysal-Biyikoglu, “Implementationof an enhanced target localization and identification algorithm on a magneticWSN,” IEICE Transactions on Communications, vol. E98-B, no. 10, pp. 2022–2032, October 2015.

68

[12] G. Mao, B. Fidan, and B. D. Anderson, “Wireless sensor network localizationtechniques,” Computer networks, vol. 51, no. 10, pp. 2529–2553, 2007.

[13] C.-H. Ou, “A roadside unit-based localization scheme for vehicular ad hocnetworks,” International Journal of Communication Systems, vol. 27, no. 1,pp. 135–150, 2014.

[14] C.-X. Wang, X. Cheng, and D. I. Laurenson, “Vehicle-to-vehicle channel mod-eling and measurements: recent advances and future challenges,” Communi-cations Magazine, IEEE, vol. 47, no. 11, pp. 96–103, 2009.

[15] L. Cheng, B. E. Henty, D. D. Stancil, F. Bai, and P. Mudalige, “Mobile vehicle-to-vehicle narrow-band channel measurement and characterization of the 5.9ghz dedicated short range communication (dsrc) frequency band,” SelectedAreas in Communications, IEEE Journal on, vol. 25, no. 8, pp. 1501–1516,2007.

[16] L. Cheng, B. Henty, D. D. Stancil, F. Bai, and P. Mudalige, “A fully mobile,gps enabled, vehicle-to-vehicle measurement platform for characterization ofthe 5.9 ghz dsrc channel,” in Antennas and Propagation Society InternationalSymposium, 2007 IEEE. IEEE, 2007, pp. 2005–2008.

[17] D. W. Matolak, I. Sen, W. Xiong, and N. T. Yaskoff, “5 ghz wireless channelcharacterization for vehicle to vehicle communications,” in Military Communi-cations Conference, 2005. MILCOM 2005. IEEE. IEEE, 2005, pp. 3016–3022.

[18] J. Kunisch and J. Pamp, “Wideband car-to-car radio channel measurementsand model at 5.9 ghz,” in Vehicular Technology Conference, 2008. VTC 2008-Fall. IEEE 68th. IEEE, 2008, pp. 1–5.

[19] S. Kaul, K. Ramachandran, P. Shankar, S. Oh, M. Gruteser, I. Seskar, andT. Nadeem, “Effect of antenna placement and diversity on vehicular networkcommunications,” in Sensor, Mesh and Ad Hoc Communications and Net-works, 2007. SECON’07. 4th Annual IEEE Communications Society Confer-ence on. IEEE, 2007, pp. 112–121.

[20] A. Paier, J. Karedal, N. Czink, C. Dumard, T. Zemen, F. Tufvesson, A. F.Molisch, and C. F. Mecklenbräuker, “Characterization of vehicle-to-vehicleradio channels from measurements at 5.2 ghz,” Wireless personal communica-tions, vol. 50, no. 1, pp. 19–32, 2009.

69

[21] I. Sen and D. W. Matolak, “Vehicle–vehicle channel models for the 5-ghzband,” Intelligent Transportation Systems, IEEE Transactions on, vol. 9,no. 2, pp. 235–245, 2008.

[22] P. Paschalidis, M. Wisotzki, A. Kortke, W. Keusgen, and M. Peter, “A wide-band channel sounder for car-to-car radio channel measurements at 5.7 ghzand results for an urban scenario,” in Vehicular Technology Conference, 2008.VTC 2008-Fall. IEEE 68th. IEEE, 2008, pp. 1–5.

[23] A. Molisch, F. Tufvesson, J. Kåredal, and C. F. Mecklenbräuker, “A surveyon vehicle-to-vehicle propagation channels,” IEEE Wireless Communications,vol. 16, no. 6, pp. 12–22, 2009.

[24] A. Paier, L. Bernadó, J. Karedal, O. Klemp, and A. Kwoczek, “Overview ofvehicle-to-vehicle radio channel measurements for collision avoidance applica-tions,” in Vehicular Technology Conference (VTC 2010-Spring), 2010 IEEE71st. IEEE, 2010, pp. 1–5.

[25] R. Meireles, M. Boban, P. Steenkiste, O. Tonguz, and J. Barros, “Experi-mental study on the impact of vehicular obstructions in vanets,” in VehicularNetworking Conference (VNC), 2010 IEEE. IEEE, 2010, pp. 338–345.

[26] P. Paschalidis, K. Mahler, A. Kortke, M. Peter, and W. Keusgen, “Pathlossand multipath power decay of the wideband car-to-car channel at 5.7 ghz,”in Vehicular Technology Conference (VTC Spring), 2011 IEEE 73rd. IEEE,2011, pp. 1–5.

[27] T. Mangel, M. Michl, O. Klemp, and H. Hartenstein, “Real-world measure-ments of non-line-of-sight reception quality for 5.9 ghz ieee 802.11 p at inter-sections,” in Communication Technologies for Vehicles. Springer, 2011, pp.189–202.

[28] C. Sommer, D. Eckhoff, R. German, and F. Dressler, “A computationallyinexpensive empirical model of ieee 802.11 p radio shadowing in urban envi-ronments,” in Wireless On-Demand Network Systems and Services (WONS),2011 Eighth International Conference on. IEEE, 2011, pp. 84–90.

[29] M. Segata, B. Bloessl, S. Joerer, C. Sommer, R. Lo Cigno, and F. Dressler,“Short paper: Vehicle shadowing distribution depends on vehicle type: Resultsof an experimental study,” in Vehicular Networking Conference (VNC), 2013IEEE. IEEE, 2013, pp. 242–245.

70

[30] J. Karedal, F. Tufvesson, T. Abbas, O. Klemp, A. Paier, L. Bernadó, and A. F.Molisch, “Radio channel measurements at street intersections for vehicle-to-vehicle safety applications,” in Vehicular Technology Conference (VTC 2010-Spring), 2010 IEEE 71st. IEEE, 2010, pp. 1–5.

[31] T. Abbas, L. Bernadó, A. Thiel, C. F. Mecklenbrauker, and F. Tufvesson,“Measurements based channel characterization for vehicle-to-vehicle commu-nications at merging lanes on highway,” in Wireless Vehicular Communications(WiVeC), 2013 IEEE 5th International Symposium on. IEEE, 2013, pp. 1–5.

[32] E. Giordano, R. Frank, G. Pau, and M. Gerla, “Corner: a realistic urbanpropagation model for vanet,” in Wireless On-demand Network Systems andServices (WONS), 2010 Seventh International Conference on. IEEE, 2010,pp. 57–60.

[33] A. Hrovat, G. Kandus, and T. Javornik, “A survey of radio propagation mod-eling for tunnels,” Communications Surveys & Tutorials, IEEE, vol. 16, no. 2,pp. 658–669, 2014.

[34] T. Abbas, K. Sjöberg, J. Karedal, and F. Tufvesson, “A measurement basedshadow fading model for vehicle-to-vehicle network simulations,” InternationalJournal of Antennas and Propagation, vol. 2015, 2015.

[35] J. Maurer, T. Fugen, and W. Wiesbeck, “Narrow-band measurement and anal-ysis of the inter-vehicle transmission channel at 5.2 ghz,” in Vehicular Tech-nology Conference, 2002. VTC Spring 2002. IEEE 55th, vol. 3. IEEE, 2002,pp. 1274–1278.

[36] T. Mangel, O. Klemp, and H. Hartenstein, “5.9 ghz inter-vehicle communica-tion at intersections: a validated non-line-of-sight path-loss and fading model,”EURASIP Journal on Wireless Communications and Networking, vol. 2011,no. 1, pp. 1–11, 2011.

[37] O. Renaudin, V.-M. Kolmonen, P. Vainikainen, and C. Oestges, “Widebandmeasurement-based modeling of inter-vehicle channels in the 5-ghz band,”Vehicular Technology, IEEE Transactions on, vol. 62, no. 8, pp. 3531–3540,2013.

[38] J. Maurer, T. Schäfer, and W. Wiesbeck, “A realistic description of the envi-ronment for inter-vehicle wave propagation modelling,” in Vehicular Technol-ogy Conference, 2001. VTC 2001 Fall. IEEE VTS 54th, vol. 3. IEEE, 2001,pp. 1437–1441.

71

[39] J. Maurer, T. Fugen, T. Schafer, and W. Wiesbeck, “A new inter-vehiclecommunications (ivc) channel model,” in Vehicular Technology Conference,2004. VTC2004-Fall. 2004 IEEE 60th, vol. 1. IEEE, 2004, pp. 9–13.

[40] L. Bernadó, T. Zemen, F. Tufvesson, A. F. Molisch, and C. F. Mecklenbrauker,“Delay and doppler spreads of nonstationary vehicular channels for safety-relevant scenarios,” Vehicular Technology, IEEE Transactions on, vol. 63,no. 1, pp. 82–93, 2014.

[41] L. Bernado, T. Zemen, F. Tufvesson, A. F. Molisch, and C. F. Mecklenbrauker,“Time-and frequency-varying k-factor of non-stationary vehicular channels forsafety-relevant scenarios,” Intelligent Transportation Systems, IEEE Transac-tions on, vol. 16, no. 2, pp. 1007–1017, 2015.

[42] J. S. Davis, J.-P. P. Linnartz et al., “Vehicle to vehicle rf propagation measure-ments,” in Signals, Systems and Computers, 1994. 1994 Conference Record ofthe Twenty-Eighth Asilomar Conference on, vol. 1. IEEE, 1994, pp. 470–474.

[43] J. Turkka and M. Renfors, “Path loss measurements for a non-line-of-sightmobile-to-mobile environment,” in ITS Telecommunications, 2008. ITST2008. 8th International Conference on. IEEE, 2008, pp. 274–278.

[44] G. Acosta and M. A. Ingram, “Model development for the wideband express-way vehicle-to-vehicle 2.4 ghz channel,” in Wireless Communications and Net-working Conference, 2006. WCNC 2006. IEEE, vol. 3. IEEE, 2006, pp. 1283–1288.

[45] J. S. Otto, F. E. Bustamante, and R. A. Berry, “Down the block and aroundthe corner the impact of radio propagation on inter-vehicle wireless commu-nication,” in Distributed Computing Systems, 2009. ICDCS’09. 29th IEEEInternational Conference on. IEEE, 2009, pp. 605–614.

[46] S. Biswas, R. Tatchikou, and F. Dion, “Vehicle-to-vehicle wireless communi-cation protocols for enhancing highway traffic safety,” IEEE Commun. Magz.,vol. 44, pp. 74–82, Jan. 2006.

[47] O. Renaudin, V.-M. Kolmonen, P. Vainikainen, and C. Oestges, “Widebandmimo car-to-car radio channel measurements at 5.3 ghz,” in Vehicular Tech-nology Conference, 2008. VTC 2008-Fall. IEEE 68th. IEEE, 2008, pp. 1–5.

72

[48] T. Abbas, J. Karedal, and F. Tufvesson, “Measurement-based analysis: theeffect of complementary antennas and diversity on vehicle-to-vehicle commu-nication,” Antennas and Wireless Propagation Letters, IEEE, vol. 12, pp. 309–312, 2013.

[49] F. M. Pallarés, F. J. P. Juan, and L. Juan-Llácer, “Analysis of path lossand delay spread at 900 mhz and 2.1 ghz while entering tunnels,” VehicularTechnology, IEEE Transactions on, vol. 50, no. 3, pp. 767–776, 2001.

[50] D. G. Dudley, M. Lienard, S. F. Mahmoud, and P. Degauque, “Wireless propa-gation in tunnels,” Antennas and Propagation Magazine, IEEE, vol. 49, no. 2,pp. 11–26, 2007.

[51] Q. Wu, D. W. Matolak, and I. Sen, “5-ghz-band vehicle-to-vehicle channels:models for multiple values of channel bandwidth,” Vehicular Technology, IEEETransactions on, vol. 59, no. 5, pp. 2620–2625, 2010.

[52] P. Alexander, D. Haley, and A. Grant, “Cooperative intelligent transport sys-tems: 5.9-ghz field trials,” Proceedings of the IEEE, vol. 99, no. 7, pp. 1213–1235, 2011.

[53] G. Acosta-Marum and M. A. Ingram, “Six time- and frequency- selective em-pirical channel models for vehicular wireless LANs,” IEEE Vehicular Technol-ogy Magazine, vol. 2, no. 4, pp. 4–11, December 2007.

[54] A. Paier, J. Karedal, N. Czink, H. Hofstetter, C. Dumard, T. Zemen,F. Tufvesson, C. F. Mecklenbrauker, and A. F. Molisch, “First results fromcar-to-car and car-to-infrastructure radio channel measurements at 5.2 ghz,”in Personal, Indoor and Mobile Radio Communications, 2007. PIMRC 2007.IEEE 18th International Symposium on. IEEE, 2007, pp. 1–5.

[55] L. Cheng, B. E. Henty, R. Cooper, D. D. Stancil, and F. Bai, “A measurementstudy of time-scaled 802.11 a waveforms over the mobile-to-mobile vehicularchannel at 5.9 ghz,” Communications Magazine, IEEE, vol. 46, no. 5, pp.84–91, 2008.

[56] J. Karedal, F. Tufvesson, N. Czink, A. Paier, C. Dumard, T. Zemen,C. F. Mecklenbräuker, and A. F. Molisch, “A geometry-based stochasticmimo model for vehicle-to-vehicle communications,” Wireless Communica-tions, IEEE Transactions on, vol. 8, no. 7, pp. 3646–3657, 2009.

[57] T. Abbas, J. Nuckelt, T. Kurner, T. Zemen, C. F. Mecklenbrauker, andF. Tufvesson, “Simulation and measurement-based vehicle-to-vehicle channel

73

characterization: Accuracy and constraint analysis,” Antennas and Propaga-tion, IEEE Transactions on, vol. 63, no. 7, pp. 3208–3218, 2015.

[58] A. Roivainen, P. Jayasinghe, J. Meinilau, V. Hovinen, and M. Latva-aho,“Vehicle-to-vehicle radio channel characterization in urban environment at2.3 ghz and 5.25 ghz,” in Personal, Indoor, and Mobile Radio Communication(PIMRC), 2014 IEEE 25th Annual International Symposium on. IEEE, 2014,pp. 63–67.

[59] L. Bernado, T. Zemen, F. Tufvesson, A. F. Molisch, and C. F. Mecklenbrauker,“Time-and frequency-varying-factor of non-stationary vehicular channels forsafety-relevant scenarios,” Intelligent Transportation Systems, IEEE Transac-tions on, vol. 16, no. 2, pp. 1007–1017, 2015.

[60] K. Mahler, W. Keusgen, F. Tufvesson, T. Zemen, and G. Caire, “Propagationof multipath components at an urban intersection,” in Vehicular TechnologyConference (VTC Fall), 2015 IEEE 82nd. IEEE, 2015, pp. 1–5.

[61] A. F. Molisch, F. Tufvesson, J. Karedal, and C. Mecklenbrauker, “Propaga-tion aspects of vehicle-to-vehicle communications-an overview,” in Radio andWireless Symposium, 2009. RWS’09. IEEE. IEEE, 2009, pp. 179–182.

[62] S. Dhar, R. Bera, R. Giri, S. Anand, D. Nath, and S. Kumar, “An overviewof v2v communication channel modeling,” the proceedings of ISDMISC, pp.12–14, 2011.

[63] D. W. Matolak and A. Chandrasekaran, “5 ghz intra-vehicle channel char-acterization,” in Vehicular Technology Conference (VTC Fall), 2012 IEEE.IEEE, 2012, pp. 1–5.

[64] W. Xiang, “A vehicular ultra-wideband channel model for future wireless intra-vehicle communications (ivc) systems,” in Vehicular Technology Conference,2007. VTC-2007 Fall. 2007 IEEE 66th. IEEE, 2007, pp. 2159–2163.

[65] T. Tsuboi, J. Yamada, N. Yamauchi, M. Nakagawa, and T. Maruyam, “Uwbradio propagation for intra vehicle communications,” in Ultra Modern Telecom-munications & Workshops, 2009. ICUMT’09. International Conference on.IEEE, 2009, pp. 1–5.

[66] I. Guvenc, S. Gezici, and Z. Sahinoglu, “Ultra-wideband range estima-tion: Theoretical limits and practical algorithms,” in Ultra-Wideband, 2008.ICUWB 2008. IEEE International Conference on, vol. 3. IEEE, 2008, pp.93–96.

74

[67] M. Schack, J. Jemai, R. Piesiewicz, R. Geise, I. Schmidt, and T. Kürner,“Measurements and analysis of an in-car uwb channel,” in Vehicular Technol-ogy Conference, 2008. VTC Spring 2008. IEEE. IEEE, 2008, pp. 459–463.

[68] L. Liu, Y. Wang, N. Zhang, and Y. Zhang, “Uwb channel measurement andmodeling for the intra-vehicle environments,” in Communication Technology(ICCT), 2010 12th IEEE International Conference on. IEEE, 2010, pp. 381–384.

[69] R. Fisher, R. Kohno, M. McLaughlin, and M. Welborn, “Ds-uwb physical layersubmission to ieee 802.15 task group 3a,” IEEE P802, pp. 15–04, 2005.

[70] F. Maehara, R. Kaneko, A. Yamakita, and S. Goto, “Coverage performanceof mb-ofdm uwb in-car wireless communication,” in Communication SystemsNetworks and Digital Signal Processing (CSNDSP), 2010 7th InternationalSymposium on. IEEE, 2010, pp. 417–421.

[71] M. Schack, R. Geise, I. Schmidt, R. Piesiewicz, and T. Kürner, “Uwb channelmeasurements inside different car types,” EuCAP 2009, 2009.

[72] T. Kobayashi, “Measurements and characterization of ultra wideband prop-agation channels in a passenger-car compartment,” IEICE transactions onfundamentals of electronics, communications and computer sciences, vol. 89,no. 11, pp. 3089–3094, 2006.

[73] W. Niu, J. Li, and T. Talty, “Intra-vehicle uwb channels in moving andstaionary scenarios,” in Military Communications Conference, 2009. MIL-COM 2009. IEEE. IEEE, 2009, pp. 1–6.

[74] ——, “Ultra-wideband channel modeling for intravehicle environment,”EURASIP Journal on Wireless Communications and Networking, vol. 2009,no. 1, pp. 1–12, 2009.

[75] Y. Zahedi, R. Ngah, M. Mokayef, and K. Zahedi, “Stationarity regions forultra-wideband channels,” 2015.

[76] C. Kocks, E. Scheiber, D. Xu, A. Viessmann, S. Wang, G. H. Bruck, andP. Jung, “A localization and tracking application for uwb,” in Applied Sciencesin Biomedical and Communication Technologies, 2009. ISABEL 2009. 2ndInternational Symposium on. IEEE, 2009, pp. 1–5.

[77] J. Blumenstein, T. Mikulasek, R. Marsalek, A. Chandra, A. Prokes, T. Zemen,and C. Mecklenbrauker, “In-vehicle uwb channel measurement, model and

75

spatial stationarity,” in Vehicular Networking Conference (VNC), 2014 IEEE.IEEE, 2014, pp. 77–80.

[78] J. Blumenstein, A. Prokes, T. Mikulasek, R. Marsalek, T. Zemen, andC. Mecklenbräuker, “Measurements of ultra wide band in-vehicle channel-statistical description and toa positioning feasibility study,” EURASIP Jour-nal on Wireless Communications and Networking, vol. 2015, no. 1, pp. 1–10,2015.

[79] R. Thomä, O. Hirsch, J. Sachs, and R. Zetik, “Uwb sensor networks for posi-tion location and imaging of objects and environments,” Short-Range WirelessCommunications: Emerging Technologies and Applications, pp. 97–112, 2009.

[80] U. T. Virk, K. Haneda, V.-M. Kolmonen, P. Vainikainen, and Y. Kaipainen,“Characterization of vehicle penetration loss at wireless communication fre-quencies,” in Antennas and Propagation (EuCAP), 2014 8th European Con-ference on. IEEE, 2014, pp. 234–238.

[81] R. Kraemer and M. Katz, Short-range wireless communications: emergingtechnologies and applications. John Wiley & Sons, 2009.

[82] I. 802.11p, “Draft standard for information technology - telecommunicationsand information exchange between systems - local and metropolitan area net-works - specific requirements: Wireless access in vehicular environments,”Tech. Rep., 2009.

[83] V. Shivaldova, “Implementation of ieee 802.11 p physical layer model insimulink,” 2010.

[84] P. Kukolev, “Comparison of 802.11 a and 802. 11p over fading channels,”Elektrorevue, vol. 4, no. 1, pp. 7–11, 2013.

[85] ——, “Ber performance of 802.11 p standard over an itu-r multipath channel,”Proc. Radioelektronika, Pardubice, Czech Republic, pp. 391–396, 2013.

[86] I. C. Society, “Ieee standard for information technology–telecommunicationsand information exchange between systems–local and metropolitan area net-works–specific requirements–part 11: Wireless lan medium access control(mac) and physical layer (phy) specifications amendment 6: Wireless accessin vehicular environments,” IEEE Std 802, Jul. 2010.

[87] “802.11a–1999 – ieee standard for telecommunications and information ex-change between systems - lan/man specific requirements - part 11: Wireless

76

medium access control (mac) and physical layer (phy) specifications: Highspeed physical layer in the 5 GHz band,” pp. 1–102, 1999.

[88] J. Medbo, “Channel models for hiperlan/2,” ETSI/BRAN document no.3ERI085B,’98, 1998.

[89] R. Jain, “Channel models a tutorial1,” 2007.

[90] M. Müller, “Wlan 802.11 p measurements for vehicle to vehicle (v2v) dsrcapplication note,” Rohde & Schwarz, pp. 7–8, 2009.

[91] F. J. Harris, “On the use of windows for harmonic analysis with the discretefourier transform,” Proceedings of the IEEE, vol. 66, no. 1, pp. 51–83, 1978.

[92] J. Hansen, “An analytical calculation of power delay profile and delay spreadwith experimental verification,” Communications Letters, IEEE, vol. 7, no. 6,pp. 257–259, 2003.

[93] H. Yang, P. F. Smulders, and M. H. Herben, “Channel characteristics andtransmission performance for various channel configurations at 60 ghz,”EURASIP Journal on Wireless Communications and Networking, vol. 2007,no. 1, pp. 43–43, 2007.

[94] S. Guerin, Y. J. GUO, and S. K. BARTON, “Indoor propagation measure-ments at 5ghz for hiperlan,” in IEE conference publication. Institution ofElectrical Engineers, 1997, pp. 2–306.

[95] J. Blumenstein, T. Mikulasek, R. Marsalek, A. Prokes, T. Zemen, andC. Mecklenbrauker, “In-vehicle mm-wave channel model and measurement,”in Vehicular Technology Conference (VTC Fall), 2014 IEEE 80th. IEEE,2014, pp. 1–5.

[96] S. Nadarajah and S. Kotz, “Extreme value distributions: Theory and applica-tions,” 2000.

[97] N. Balakrishnan and V. B. Nevzorov, A primer on statistical distributions.John Wiley & Sons, 2004.

[98] H. El-Sallabi, D. S. Baum, P. Zetterberg, P. Kyosti, T. Rautiainen, andC. Schneider, “Wideband spatial channel model for mimo systems at 5 ghz inindoor and outdoor environments,” in Vehicular Technology Conference, 2006.VTC 2006-Spring. IEEE 63rd, vol. 6. IEEE, 2006, pp. 2916–2921.

77

[99] Y. Guo, J. Zhang, C. Tao, L. Liu, and L. Tian, “Propagation characteristics ofwideband high-speed railway channel in viaduct scenario at 2.35 ghz,” Journalof Modern Transportation, vol. 20, no. 4, pp. 206–212, 2012.

[100] F. J. Massey Jr, “The kolmogorov-smirnov test for goodness of fit,” Journalof the American statistical Association, vol. 46, no. 253, pp. 68–78, 1951.

[101] A. Paier, “The vehicular radio channel in the 5 GHz band,” PhD Thesis,Technischen Universitat Wien, Vienna, Austria, Oct. 2010.

[102] U. Demir, C. U. Bas, and S. C. Ergen, “Engine compartment uwb chan-nel model for intravehicular wireless sensor networks,” Vehicular Technology,IEEE Transactions on, vol. 63, no. 6, pp. 2497–2505, 2014.

[103] T. Dammes, W. Endemann, and R. Kays, “Frequency domain channel mea-surements for wireless localization - practical considerations and effects of themeasurement,” in 18th European Wireless Conference (EW), Poznan, Poland,April 2012, pp. 1–8.

[104] M. Corrigan, A. Walton, W. Niu, J. Li, and T. Talty, “Automatic uwb clus-ters identification,” in Radio and Wireless Symposium, 2009. RWS’09. IEEE.IEEE, 2009, pp. 376–379.

[105] F. Grejták and A. Prokes, “Uwb-ultra wideband characteristics and the saleh-valenzuela modeling,” Acta Electrotechnica et Informatica, vol. 13, no. 2, p. 32,2013.

[106] A. F. Molisch, J. R. Foerster, and M. Pendergrass, “Channel models for ultra-wideband personal area networks,” Wireless Communications, IEEE, vol. 10,no. 6, pp. 14–21, 2003.

78

A LIST OF PUBLICATION1. Kukolev, P.; Chandra, A.; Mikulasek, T.; Prokes, A.; Zemen, T.; Mecklen-

brauker, C. In-vehicle channel sounding in the 5.8 GHz band. EURASIPJournal on Wireless Communications and Networking, 2015, roč. 2015, č. 1,s. 1-9. ISSN: 1687-1499.

2. Kukolev, P.; Chandra, A.; Mikulasek, T.; Prokes, A. Out of vehicle chan-nel sounding in 5.8 GHz band. In 7th International Workshop on ReliableNetworks Design and Modeling. 2015. s. 341-344. ISBN: 978-1-4673-8049- 2.

3. Kukolev, P. BER performance of 802.11p standard over an ITU-R Multi-path channel. In Proceedings of 23th International Conference RADIOELEK-TRONIKA 2013. 2013. s. 391-396. ISBN: 978-1-4673-5517- 9.

4. Kukolev, P. Comparison of 802.11a and 802. 11p over fading channels. Elek-trorevue - Internetový časopis (http://www.elektrorevue.cz), 2013, roč. 2013,č. 2, s. 7-11. ISSN: 1213- 1539.

5. Chandra, A.; Prokes, A.; Mikulasek, T.; Blumenstein, J.; Kukolev, P.; Zemen,T.; Mecklenbrauker, C. Frequency-Domain In-Vehicle UWB Channel Model-ing. IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2016, roč.PP, č. 99, s. 1-12. ISSN: 0018-9545.

6. Chandra, A.; Prokes, A.; Blumenstein, J.; Kukolev, P.; Vychodil, J.; Miku-lasek, T.; Zemen, T.; Mecklenbrauker, C. Intra Vehicular Wireless ChannelMeasurements. In European Project Space on Information and Communica-tion Systems. Portugal: SCITEPRESS – Science and Technology Publica-tions, Lda., 2016. s. 3-28. ISBN: 978-989-758-155- 7.

7. Chandra, A.; Kukolev, P.; Mikulasek, T.; Prokes, A. Frequency-Domain In-Vehicle Channel Modelling in mmW Band. In In 1st International Forum onResearch and Technologies for Society and Industry (RTSI 2015). 2015. s.1-5. ISBN: 978-1-4673-8166- 6.

8. Chandra, A.; Kukolev, P.; Mikulasek, T.; Prokes, A. Autoregressive Model ofChannel Transfer Function for UWB Link inside a Passenger Car. In Proceed-ings of the 19th International Conference on Communications (part of CSCC’15). WSEAS Applied Informatics and Communications. 2015. s. 238-241.ISBN: 978-1-61804-318- 4. ISSN: 1790-5117.

9. Kukolev, P.; Chandra, A.; Prokes, A. BER Performance of 802.11p in SISO,MISO, and MIMO Fading Channels. In Proceedings of the 19th InternationalConference on Communications (part of CSCC ’15). WSEAS Applied Infor-matics and Communications. 2015. p. 89-93. ISBN: 978-1-61804-318- 4.ISSN: 1790- 5117.

10. Kukolev, P.; Chandra, A.; Prokes, A.; Mikulasek, T. UWB channel variability

79

around a car in an underground parking lot. IEEE Antennas and WirelessPropagation Letters. VOL. XX, 2016. Submitted.

11. Kukolev, P.; Chandra, A.; Mikulasek, T.; Prokes, A. Out-of-Vehicle TOAbased Localization in Ultra-Wide Band. Advances in Vehicular Sensor Net-works and Urban Computing. International Journal of Distributed SensorNetworks. Hindawi Publishing Corporation. 2016. Submitted.

80