Development and Evaluation of UTDoA as a Positioning Method in LTE

(1)

Development and Evaluation of UTDoA as a Positioning Method in LTE

THOMAS ARTHUR HERBERT BRESSNER

Master’s Degree Project Stockholm, Sweden 2015

XR-EE-SB 2015:004

(2)

(3)

Degree Project in Signal Processing (EQ272X)

Development and Evaluation of UTDoA as a Positioning Method in LTE

Thomas Arthur Herbert Bressner 890419-2872

Supervisors: Dr. Sara M. Razavi (ER), Henrik Rydén (ER), Per Zetterberg (KTH) Examiner: Prof. Dr. Mats Bengtsson (KTH)

Period: Spring Semester – 2015

Linköping, 2015

Postaddress: Ericsson AB Tel.: 010-7190000

Datalinjen 4 Fax:

SE-583 30 Linköping E-Mail:

Web: www.ericsson.se

(4)

(5)

Declaration

I hereby declare that I wrote my Degree Project in Signal Processing (EQ272X) on my own and that I have followed the regulations relating to good scientific practice of the Kungliga Tekniska Högskolan (KTH) in its latest form. I did not use any unacknowledged sources or means and I marked all references I used literally or by content.

Linköping, 2015

Thomas Arthur Herbert Bressner 890419-2872

iii

(6)

iv

(7)

Abstract

Although positioning has been one of the main target study areas in mobile communication in the last decade, it still receives strong attention in recent years focusing more on the indoor users.

Nowadays, a wide range of different methods are available to estimate the position of the target user under certain circumstances. One of these methods is Uplink Time Difference of Arrival (UTDoA), which has been defined in 3GPP Release 11 for Long Term Evolution (LTE) networks, and is the focus of this master thesis. In Uplink based positioning, to estimate the position of a User Equipment (UE), the UE only needs to generate and transmit the reference signal and the main computational effort of time estimation, is moved from the UE towards the network side.

This might be one advantage compared to Observed Time Difference of Arrival (OTDoA), while further performance properties of UTDoA in LTE are investigated in the course of this master thesis.

In parallel with the 3GPP Study Item on Indoor Positioning which mainly was based on downlink OTDoA, this thesis studies on the potential use of UTDoA in LTE under the same type of agreed deployment scenarios and simulation parameters. For time estimation based on the Sounding Reference Signals (SRSs), the uplink channel has been modeled and simulated. Finally, the position estimation of the UE is derived by multilateration techniques using the time/distance estimations of the received SRS at each eNodeB.

The metrics of positioning results are based on Cumulative Distribution Functions (CDF) of horizontal and vertical positioning error. The study shows that reasonable horizontal position accuracy can be achieved, while a number of pico cells are added to the network to enhance the macro-only scenario. However, this positive effect could not be observed in vertical position estimation. A further investigated aspect is the influence by other active UEs considered as interference. The outcome shows, that the accuracy is strongly and negatively affected by introducing interference. A final observation focuses on the SRS bandwidth and that for bandwidths below 10 MHz additional degradations in performance are seen.

v

(8)

vi

(9)

Sammanfattning

Trots att positionering har varit ett av huvudmålen för forskning inom mobil kommunikation under det senaste decenniet, får det fortfarande mycket uppmärksamhet och under de senaste åren har forskningen fokuserats mer på inomhusanvändare. Idag finns en mängd olika metoder för att estimera positionen för en specifik användare under vissa omständigheter. En av dessa metoder är Uplink Time Difference of Arrival (UTDoA) som har definierats i 3GPP Release 11 för Long Term Evolution (LTE)-nätverk och är fokus för detta examensarbete. I positionering baserad på upplänken, vilken används för att skatta positionen för en User Equipment (UE), behöver UE:n bara generera och sända en referenssignal och den huvudsakliga beräkningskraften för tidsestimeringen flyttas från UE:n till nätsidan. Detta kan vara en fördel jämfört med Observed Time Difference of Arrival (OTDoA), detta examensarbete undersöker ytterligare prestandaegenskaper hos UTDoA i LTE.

Parallellt med 3GPP:s studie för inomhuspositionering, som huvudsakligen baseras på ner- länk OTDoA, studerar denna avhandling den potentiella användningen av UTDoA i LTE med samma typ av överenskomna scenarier och simuleringsparametrar. För tidsuppskattning baserad på Sounding Reference Signals (SRSs) har kanalen upplänkmodellerats och simulerats. Slutligen är positionsestimeringen av UE:n härledd genom multilaterationstekniker med hjälp av tids- och distansestimeringar av de mottagna SRS vid varje eNodeB.

De mått som används för positioneringsresultat baseras på kumulativ fördelningsfunktion av det horisontella och vertikala positioneringsfelet. Studien visar att en rimlig noggrannhet kan uppnås i den horisontella dimensionen då ett antal pico-celler adderas till nätverket för att förbättra makroscenariot. Denna positiva effekt kunde emellertid inte observeras i den vertikala positionsestimeringen. En ytterligare undersökt aspekt är påverkan av andra aktiva UEs, som betraktas som interferens. Resultaten visar att noggrannheten är starkt negativt påverkad då störningar i form av interferens införs. En slutlig observation fokuserar på bandbredden av SRS och det visar på försämringar i prestanda för bandbredder under 10 MHz.

vii

(10)

viii

(11)

ix

Acknowledgement

This study was performed at Ericsson Research in Linköping Sweden during the spring of 2015.

I would like to express my gratitude to my supervisors at Ericsson, Dr. Sara M. Razavi and Henrik Rydén. Together they introduced me to the research problem of UTDoA in LTE and supported me with useful comments, advices and explained the problems I have encountered during my work. Thank you both for spending so much time, and especially for reading my report multiple times.

I would also like to thank my supervisor at KTH, Per Zetterberg, for his comments during my master thesis and my examiner/MERIT coordinator, Prof. Dr. Mats Bengtsson, for his support during my entire time at KTH.

Furthermore I would like to express my sincere thanks to my family, who contributed by helping me with my application, removals and being available when I needed support.

Finally, I want to thank the European Union for the financial support and making this great double master degree ’MERIT’ possible.

Linköping, June 2015 Thomas Arthur Herbert Bressner

(12)

x

(13)

List of Figures

2.1 An indoor scenario. . . 6

2.2 Requesting SRS transmission by a serving cell. . . 7

2.3 E-UTRA supported positioning network architecture [1]. . . 8

2.4 Transmission and reception scheme of an OFDM signal [2]. . . 9

2.5 An uplink LTE signal structure. . . 12

2.6 Frequency and non-frequency hopping transmitting schemes. . . 12

2.7 Available groups and sequences [3]. . . 14

2.8 The Gold sequence scheme [4]. . . 15

3.1 Scenario deployment with 49 UEs: (a) Scenario 1; (b) Scenario 2. . . 18

4.1 Structure of the implementation. . . 22

4.2 Obtaining time of arrival by using a cross-correlation. . . 23

4.3 Autocorrelation of SRS with narrow and wide bandwidths. . . 23

4.4 Cross-correlation between RX signal and SRS with different bandwidths. . . 24

4.5 Tapped Delay Line (TDL) model. . . 24

4.6 Cross-correlation properties of an RX signal at different time instances. . . 26

4.7 Positioning estimation with TDoA. . . 29

4.8 Gauss-Newton algorithm fed with perfect estimated distances. . . 32

5.1 Histogram plot of estimations with different lower filtering boundaries (ΩPAR). . . . 34

5.2 The PDF of time estimation accuracy with different intervals of ΩPAR. . . 35

5.3 Horizontal position accuracy for weighted and non-weighted SRS with 10 MHz bandwidth of Scenario 2. . . 37

5.4 Horizontal position accuracy comparison between the use of zero and four pico cells per macro cell. . . 37

5.5 Vertical position accuracy comparison between the use of zero and four pico cells per macro cell. . . 38

5.6 Horizontal position accuracy comparison of UTDoA and OTDoA. . . 40

5.7 Horizontal position accuracy dependency on the number of active UEs. . . 42

5.8 Horizontal position accuracy comparison between different bandwidth sizes. . . 44

xiii

(16)

xiv List of Figures

(17)

List of Tables

2.1 (Non-)Frequency hopping - pros and cons [5]. . . 11

2.2 SRS interference management. . . 16

3.1 Scenario deployment . . . 18

3.2 UTDoA assumptions and parameters . . . 19

4.1 SRS Verification . . . 25

5.1 Parameter setup for investigating on Q1. . . 34

5.2 Percentage of available estimates depending on the chosen ΩPAR. . . 36

5.3 Obtained position accuracy. . . 39

5.4 Assumptions for OTDoA and PRS characterization [6]. . . 39

5.5 Horizontal position accuracy comparison between OTDoA and UTDoA. . . 41

5.7 Interference impact on the horizontal position accuracy. . . 42

5.9 Horizontal position accuracy for different SRS bandwidths. . . 43

A.1 Summary of factors impacting position accuracy (GSM/GPRS) based on Table 5-3 of [7] . . . 49

B.1 QPSK 12 sequence based on Table 5.5.1.2-1 [8]. . . 51

B.2 QPSK 24 sequence based on Table 5.5.1.2-2 [8]. . . 52

xv

(18)

xvi List of Tables

(19)

List of Nomenclature

Abbreviations

Symbol Description

3GPP 3rd Generation Partnership Project A-GPS Assisted-Global Positioning System

AoA Angle of Arrival

AWGN Additive White Gaussian Noise CDF Cumulative Distribution Function

CID Cell IDentification

CMRS Commercial Mobile Radio Service

CP Cyclic Prefix

DM-RS DeModulation-Reference Signal E-OTD Enhanced-Observed Time Difference E-SMLC Evolved-Serving Mobile Location Centre E-UTRA Evolved-UMTS Terrestrial Radio Access ECID Enhanced Cell IDentification

eNodeBs evolved Node Bs

FCC Federal Communications Commission GN-Algorithm Gauss Newton-Algorithm

GPS Global Positioning System

GSM Global System for Mobile Communications IDFT Inverse Discrete Fourier Transform

ISD Inter-Site Distance

ISI InterSymbol Interference

xvii

(20)

xviii List of Nomenclature

LMU Location Measurement Unit

LOS Line Of Sight

LPP LTE Positioning Protocol

LTE Long Term Evolution

NLOS Non Line Of Sight

OFDM Orthogonal Frequency Division Multiplexing OFDMA Orthogonal Frequency Division Multiple Access OTDoA Observed Time Difference of Arrival

PAPR Peak to Average Power Ratio

PAR Peak to Average Ratio

PDF Probability Density Function

PoA Phase of Arrival

QPSK Quadrature Phase Shift Keying

RB Resource Block

RE Resource Element

RSS Received Signal Strength

RSTD Reference Signal Time Difference

SC-FDMA Single Carrier-Frequency Division Multiple Access SRS Sounding Reference Signal

TDL Tapped Delay Line

TDoA Time Difference of Arrival

UE User Equipment

UMTS Universal Mobile Telecommunications System UTDoA Uplink Time Difference of Arrival

Notations

Symbol Description

αk Step lengths

αp_˜ Cyclic shift

(21)

List of Nomenclature xix

¯

γ^u,v[n] Base sequence

δaccuracy Distance corresponding to T_accuracy

ǫ Minimum allowed change after an iteration γu^(α),v[n] Sounding reference signal

Rˆsi[t],rj[t][τ ] Cross-correlation between s_i[t] of desired UE i and r[t]

Tˆ_delay Time delay

ˆ

x Estimated position vector

ΩPAR Peak to Average Ratio of the correlated signal

τ Correlation lag

d_i Distance from UE to eNodeB i

H Hessian matrix

R Weighting matrix

V[x] General loss function

u Sequence group number

v Base sequence index

aq[m] qth root Zadoff-Chu sequence

c[i] Pseudo-random sequence

chi,j[t] Channel between UE i and eNodeB j at time t d[n] Output signal of the IDFT

fgh[ns] Group hopping pattern fss Sequence-shift pattern

L^max_seq Maximum length of an SRS sequence n[t] Additive white Gaussian noise

n_s Slot index

Nap Number of antenna ports N_ID^Cell Total number of cell IDs

N_SC^RB Quantity of subcarriers for resource blocks n^cs,˜_SRS^p Cyclic shift value, depending on the antenna port

(22)

xx List of Nomenclature

n^cs_SRS Cyclic shift value, given by higher-layers NZC Length of Zadoff-Chu sequence

p Antenna port number

rj[t] Received signal at eNodeB j at time t

r_j,t_max Received signal at eNodeB j accumulated over the duration of tmax

si[t] Transmitted SRS signal from UE i at time t

T_1.tap Exact time delay

Taccuracy Time difference of the first tap and the estimated tap T_DoA_ij Time Difference of Arrival between eNodeB i and j TCP Duration of a cyclic prefix

xi,y_i,z_i Known coordinates of eNodeB i

c Speed of light in vacuum

q Sequence index

x,y,z Unknown coordinates of UE

(23)

1 Introduction

The mobile communication market has been a growing market in the previous decades and the development process in many areas continues to improve the user experience. One of these areas is Location-Based Services (LBS) which gained strong attention in recent years. Future trends show the need of enhancing existing methods to provide a quicker and a more accurate estimation of a user position [9].

Depending on the type of LBS, requirements and potentials of the used positioning method can differ. To set an example, for mobile phone-based vehicle positioning and tracking, Assisted- Global Positioning System (A-GPS) and Observed Time Difference of Arrival (OTDoA) are the commonly used methods to estimate the location [10]. Further, to provide an even more precise location estimation, additional information as traffic maps can be included.

As in the previous mentioned example, accumulation of several positioning information sources could lead to a more accurate estimation. Nevertheless, there are LBS which are time and accuracy urgent and due to their environmental circumstances they are not able to draw on other sources such as using maps or Global Positioning System (GPS) [11]. Such constraints can be found in an emergency call scenario with an indoor located user. Due to the signal attenuation caused by construction materials and the scattered signals by roofs, walls and other obstacles, the weak GPS signal is not suitable to be used indoors [12]. In addition, indoor maps of buildings, who can help to determine the position, are still seldom available. This motivates the estimation of position by using mobile communication signals.

The recent mobile communication generation, Long-Term Evolution (LTE), has several position methods defined, one of them is OTDoA. This method uses the measured timing of downlink signals received from multiple evolved Node Bs (eNodeBs) to locate the User Equipment (UE), which can be any device used directly by an end-user to communicate, in relation to neighbouring eNodeBs. The use of TDoA as a location metric in comparison with Angle of Arrival (AoA), Phase of Arrival (PoA) and Received Signal Strength (RSS), has the advantage of being less likely to suffer under multipath propagation and heavy shadow fading in indoor locations [13]. Beside OTDoA, TDoA can be used with the uplink signal in LTE and hence is called UTDoA. In the uplink case the Sounding Reference Signal (SRS) is transmitted by the UE, while the eNodeBs receive the SRS and the processing is done by the Location Measurement Unit (LMU). This leads to better performances while at the same time does not have an impact on UE implementation, no impact on device battery and a good outdoor and indoor position estimation result [14].

Nevertheless, UTDoA is not available at this time, which might be caused by the additional

1

(24)

2 1 Introduction

hardware needed, such as the Location Measurement Unit (LMU), to enable this method. Up to now, UTDoA has been just defined in 3rd Generation Partnership Project (3GPP) LTE Release 11, but not yet implemented in any products, since further investigation needs to be done. This master thesis particularly focuses on the achievable estimation accuracy by applying UTDoA in LTE and its accuracy potentials in comparison with the established OTDoA in LTE. Furthermore, this thesis includes the investigation on the impact of interference and the use of different bandwidths for SRS.

1.1 Problem Formulation

As the introduction already pointed out, the investigation on UTDoA in LTE is one of the recent methods to provide a sufficient accuracy of indoor positioning. This leads to the need of creating a reasonable scenario and modifying the company’s internal simulator in order to evaluate the use of UTDoA in LTE. Through the simulation and performance evaluation, the following objectives are supposed to be achieved.

Q1 How is the precision of the time estimations?

Q2 How is the position estimated accuracy?

Q3 How is the position accuracy of UTDoA in comparison with OTDoA?

Q4 How is the UTDoA position accuracy affected by interference from other users?

Q5 How is the positioning accuracy affected by using different SRS bandwidths?

1.2 Methodology

Provided by Ericsson, an internal wireless communication simulator is used in the course of this master thesis. First, to approach the formulated problems the simulator needs to be modified to feature UTDoA. Afterwards, a dispatcher with information about the scenario needs to be created in Matlab. The dispatcher is used to feed the simulator with all required parameters and to process and save the obtained simulation results.

Key elements of the implementation are generating the SRS, defining the scenario, time estimation and positioning estimation. All parts are presented in detail in this report and an overview of the relationship between all parts is given in Figure 4.1.

1.3 Ethical and societal consideration

Thousands of emergency calls are made worldwide everyday. To set an example, according to the Federal Communications Commission (FCC), an estimated 240 million calls are made to 9-1-1 in the U.S. and about 33 % are wireless calls [15]. To provide a quick and uncomplicated help, it is

(25)

1.4 Outline 3

important for the rescuers to know the location of the emergency. This can be problematic when the call is made from a mobile device and the client cannot give information about the current position. Therefore, using the Commercial Mobile Radio Service (CMRS) is a convenient method to obtain a position estimate of a client.

On that purpose, the U.S. Federal Communications Commission (FCC) adopted the Fourth Report and Order in February 2015 [16]. In this report, the issue of positioning 9-1-1 calls originating from indoor locations was addressed and defined new requirements on CMRS indoor localization. According to the report, all CMRS providers must provide horizontal location within 50 meters [16]:

• in 40 % of all wireless 911 calls within 2 years.

To fulfill these standards, different methods can be used in CMRS, such as OTDoA and UTDoA.

Because of that in Section 5.3 the performance of the established OTDoA and the investigated UTDoA is additionally compared with the required FCC standards.

Another societal aspect is the lower impact on the UE battery service life compared to OT- DoA. This is achieved, because the main computational effort, as time and position estimation, is done by the network, through which UTDoA becomes interesting for other LBSs, since one of the main limitations of a mobile device is its short battery service life. Nevertheless, UTDoA is more likely to be compromised by interference and might be not able to provide simultaneously its service to the same amount of UEs as OTDoA. Therefore, the research questions Q4 asks about the influence of other UEs.

1.4 Outline

As a next step, the UTDoA theory is explained in Chapter 2. This includes background information about other available methods and an explanation of the notion of UTDoA. Further, the SRS, which is the centerpiece of UTDoA, is described in detail. Thereafter, Chapter 3 introduces the applied scenarios and the assumptions, which are required to approach the formulated problems. After- wards, Chapter 4 deals with the implementation and explains the idea and problems of time and positioning estimation. Additionally, some verification tests are performed to support the reliability of the implementation. Moreover, all required knowledge about UTDoA and the implementation are provided, whereupon the research outcome is presented in Chapter 5. Finally, in Chapter 6 and 7 the conclusions are drawn and the future perspectives are given respectively.

(26)

4 1 Introduction

(27)

2 Uplink TDOA Positioning Theory

The purpose of the second chapter is to provide a view into the theory of positioning methods with a focus on the concept of UTDoA and the generation of its centrepiece signal, SRS.

2.1 Positioning Methods

There are different strategies for the objective of estimating the UE position. Those can be distinguished based upon the used infrastructure as Global Positioning System (GPS), wireless communication systems as Global System for Mobile Communications (GSM), Long-Term Evolution (LTE) or Wireless Local Area Network (W-LAN), and furthermore. All different infrastructures have one thing in common, they provide a signal to be used for estimating the actual position. This includes, the need of a proper method to acquire a precise result. Similar to the infrastructure, there are more than one possible way which can lead to the target. Depending on the circumstances and chosen infrastructure, methods such as Angle of Arrival (AoA), Phase of Arrival (PoA), Received Signal Strength (RSS), Cell ID (CID) and Time Difference of Arrival (TDoA) may be applied. These mentioned methods have their advantages and disadvantages, which need to be considered for choosing the best approach. To set an example, the use of AoA for an indoor positioning estimation, Figure 2.1, might be difficult since a high number of reflections, which leads to changing angles, can be expected. In contrast, TDoA has certain strengths with respect to AoA, particularly in ability to process close range multipath propagation in urban environments and geolocation of wideband signals [17]. Further, advantages are simpler antenna requirements and amenability to low cost sensor network deployments.

As it has been already mentioned in Chapter 1, the focus of this work is on the Uplink TDoA in LTE. Indeed, UTDoA is not a recently newborn idea but the use of this method in LTE just gained importance during the last years. One of the reasons is the good performance which UTDoA showed in GSM and GPRS, published in the technical report [7] from 3GPP and described as "The ability to achieve high levels of location accuracy and availability in urban and indoor environments allows UTDOA to mitigate the performance limitations of A-GPS and E-OTD in these same environments.". This great performance can be achieved in urban and suburban areas with UTDoA, where A-GPS just shows in suburban and rural areas a great performance. A summary of factors impacting location accuracy in GSM and GPRS is given in Appendix A. For a better understanding on how UTDoA can be implemented into an LTE system, and to see the advantages and disadvantages, the following subsections can be followed.

5

(28)

6 2 Uplink TDOA Positioning Theory

Fig. 2.1: An indoor scenario.

2.2 Concept of UTDoA

The basic concept of UTDoA, shown in Figure 2.2, is to request the position of a UE which replies with transmitting a reference signal, called Sounding Reference Signal (SRS). With the use of this reference signal the eNodeB can estimate the time of arrival. In relation with the estimated SRS time of arrival at other eNodeBs a time difference can be computed, while the serving cell can be used as a reference. Finally the position can be estimated by solving the resulting non-linear equations.

An in-depth look is given in Figure 2.3, based on [1]. It depicts the architecture of Evolved-UMTS Terrestrial Radio Access (E-UTRA) for uplink use with UTDoA support. The entities at the two ends are the UE and the location server, called Evolved Serving Mobile Location Centre (E-SMLC), which are communicating with each other via an eNodeB by using LTE Positioning Protocol (LPP). If there is a need to estimate the position of a UE, the E-SMLC can indicate the serving cell of the targeted UE to direct the target UE to transmit SRS signals. Furthermore, the E-SMLC provides the configuration of the required SRS signals, as periodic/non-periodic, bandwidth and frequency/group/sequence hopping. These configuration options are explained in Section 2.3.3 in more detail. After receiving the transmitted SRS signals, the location needs to be determined. For that purpose the Location Measurement Units (LMUs) are used which are attached to each eNodeB. To determine the location, the LMUs needs information about the target UE configuration data and the position measurements, which the LMU gets via the interconnection between the LMU and E-SMLC, called SLm [18]. Both, LMU and SLm, are the

(29)

2.3 Uplink Processing 7

Fig. 2.2: Requesting SRS transmission by a serving cell.

two new components needed to provide UTDoA [1].

As seen in Figure 2.3, the new required parts are located at the network side. Since there is no impact on the UE implementation, a roll-out becomes easier. In addition, the only task on the UE side is to generate and transmit the SRS which needs less computational effort than calculating a time estimate [1]. This means, in comparison with OTDoA, where the reference signal is generated by the network and the time estimation is done by the UE, the computational effort on the UE is reduced. A further improvement can be obtained due to better receiving properties of an eNodeB system and it enables more advanced signal detection algorithms with its higher computational abilities.

On the other hand, the UE is limited by its transmitting power, which makes it tougher to detect its signal for adjacent cells. Moreover, as a UE has an omnidirectional antenna with a gain of 0 dBi, UTDoA suffers more under interference and noise compared to a downlink signal approach. An additional drawback of UTDoA is the need of supplementary equipment, such as the LMU, which causes additional costs on network providers.

2.3 Uplink Processing

The previous section already presented an overview about how UTDoA works. Now, to be able to implement UTDoA, a deeper knowledge about how the uplink in LTE operates is needed. For that purpose, the following sections do not maintain to cover the entire uplink theory in LTE, but to provide the required knowledge from generating the reference signal to modulating the desired

(30)

LTE

User Equipment eNodeB

SLP E-SMLC P-GW

MME

LCS Sever SUPL/LPP

SLm LPP

S1-MME

LPPa

SLs

S1-U S5 Lup

SLP – SUPL Locaon Plaorm SUPL – Secure User Plane Locaon

LMU

S-GW

MME– Mobility Management Enty

LPP – LTE Posioning Protocol [3 GPP TS 36.355]

P-GW – Packet Data Network Gateway

S-GW – Serving Gateway

E-SMLC – Evolved Serving Mobile Locaon Center

LPPa – LTE Posioning Protocol Annex [3 GPP TS 36.455]

LMU – Locaon Measurment Unit LCS – Locaon Service

Fig. 2.3: E-UTRA supported positioning network architecture [1].

signal onto a carrier.

2.3.1 Single-Carrier Frequency Division Multiple Access

Depending on the used link direction, downlink or uplink, two multiple-access schemes are defined in LTE. In case of downlink Orthogonal Frequency Division Multiple Access (OFDMA) is used, while for the uplink, Single-Carrier Frequency Division Multiple Access (SC-FDMA) is applied. The reason for these two schemes are caused by different circumstances at the UE and the eNodeB transmitters. While in downlink the transmitter has plenty of resources available, the uplink transmitter is much more restricted. Therefore, in uplink SC-FDMA is selected since it provides a low Peak to Average Power Ratio (PAPR) which results in reduced power consumption in comparison with OFDMA [19].

As Figure 2.4 depicts, SC-FDMA uses basically the same operations as OFDMA, due to that both divide the transmission bandwidth into multiple parallel subcarriers and maintain the orthogonality by adding Cyclic Prefix (CP) as a guard interval. The difference between both techniques is the added Inverse Discrete Fourier Transform (IDFT). Hereby, the data symbols are not assigned to each subcarrier independently, but instead the symbols are modulated such that they are transmitted at the same time instant [19].

In case of the simulator implementation, there is no need to make use of a discrete Fourier transform, since the SRS is generated in frequency domain. Therefore, to receive the desired SC-FDMA signal, the OFDM modulation can be directly applied onto the generated reference sequence. Thus, the incoming parallel reference bit-stream ˜d₀ to ˜dLmax

seq is transformed to the time

(31)

IDFT DFT

Cyclic

preﬁx Channel Remove

preﬁx

d₀ d[0]

dL_seq-1 d[Nc-1]

x[1]=d[Nc-L+1]

x[L-1]=d[N_c-1]

x[L]=d[0]

x[Nc+L-1]=d[Nc-1] y[Nc+L-1]

y[1]

y[L-1]

y[L]

y[Nc+L-1]

y[L] y₀

yN -1 SOFDM(f)

Frequency

~

N-point DFT

N-point IDFT

SC- FDMA OFDMA

max

sc RS

Fig. 2.4: Transmission and reception scheme of an OFDM signal [2].

domain, by the inverse Fourier transform with a sampling rate fs = _T¹

s = N · ∆f , where N is the FFT length and ∆f is the subcarrier spacing. The nominal bandwidth of the OFDM signal is defined by L^max_seq · ∆f and implies that Nc, which needs to be chosen such that the sampling theorem is sufficiently fulfilled, should exceed L^max_seq with a sufficient margin. The output signal of the IDFT is given by [3]

d[n] =

L^max_seq−1

X

k=0

ake^j2πknT^s =

L^max_seq−1

X

k=0

ake^j2πkn^N =

Nc−1

X

k=0

a^′_ke^j2πkn^N , (2.1)

where

a^′_k =

(a_k, 0 ≤ k < L^max_seq,

0, L^max_seq ≤ k < N. (2.2)

As Equation 2.2 shows, the input values are extended with zeros to reach the required length. After the IDFT, the signal is spread over the used frequency spectrum. As the next step, the CP is added to the signal. The idea of using CP is to add redundant information onto the signal by prefixing of the OFDM symbol with a repetition of the end. This has the advantage that the CP works as a guard interval and eliminates the Inter-Symbol Interference (ISI) from the previous symbol. For the uplink case, there are two CP modes, where the normal CP mode has a duration of T_CP = 5.2µs for the first symbol and T_CP = 4.7µs for the remaining six symbols. In case of using the extended CP mode, the duration of the CP is always T_CP = 16.7µs while there are just 6 symbols instead of 7 in one slot. A more detailed description on symbol and slot, is given in Section 2.3.3.

2.3.2 Reference Signals: SRS and DM-RS

There are two types of reference signals defined in the uplink frequency of an LTE system. Both signals are used to sound a certain frequency spectrum based on Zadhoff Chu sequence, but they

(32)

differentiate between their applications [3].

The intention of the DeModulation-Reference Signal (DM-RS) is to handle the impact of the radio channel of the received data at the eNodeB by estimating the channel. This requires sending the signal together with data and a sufficient high repetition frequency. Based on the 3GPP definition the DM-RS signal is scheduled every fourth Single Carrier-Frequency Division Multiple Access (SC-FDMA) symbol of a slot.

In contrast to the DM-RS, the SRS is not transmitted simultaneously with data. This reduces the interference and provides the reference signal with a better detectability. Because of that the SRS signal is more suitable in terms of location determination.

On the other hand, since during an SRS transmission no data is allowed to be transmitted, the use of SRS reduces the possible throughput of a system. Because of that, the SRS can be just transmitted on the last SC-FDMA symbol of a subframe. A more detailed description is given in the next section and its subsections.

2.3.3 Sounding Reference Signal

The Sounding Reference Signal is designated to be used as a sounding signal in the uplink. Because of its properties, it is suitable for estimating the position of a UE. For a better understanding, why an SRS is robust against interference, a closer look on its properties is given. These information are important for the implementation of the SRS in the simulator, as it was needed for this work.

The following subsections are mainly based on the information collected from the 3GPP definition of SRS given in [8] and [20], all further used sources are mentioned directly in the text.

2.3.3.1 Scheduling

For a better understanding on how an SRS signal is generated, the structure of an LTE signal needs to be known. Figure 2.5 depicts the basic structure used for scheduling signals in LTE. The biggest segment is called frame which can be divided into 10 subframes with an equidistant of 1 ms. An additional subdivision is slots which have a length of 0.5 ms and consist of 7 symbols. Up to now, only time domain is considered. By taking the frequency into account, a symbol consists of 12 subcarriers and forms a Resource Block (RB) with a size of 7 symbols times 12 subcarriers.

Each element of a RB is called Resource Element (RE), which is the smallest entity in LTE. An SRS can be allocated only to the last symbol of a subframe. Depending on the SRS configuration, defined by the network, all RE of a symbol are used or just every second RE. This odd and even pattern is called Comb-Spectrum and has the advantages that the same SRS can be used by two UEs during one symbol. This means the number of available orthogonal signal doubles from 8 to 16. Depending on the configuration an SRS can be retransmitted. In case of an aperiodic configuration the SRS can be transmitted just once triggered by the network. Another case is the periodic configuration, which allows the transmission of SRS once every 2 ms up to once every

(33)

Option Advantage Drawback

Non-freq. hop. - Entire freq. band can be sounded - Lower received power density.

with an SRS transmission. - When used in aperiodic mode, - Longer SRS length the result can be compromised

better detectability. by temporary channel distortion.

Freq. hop. - Higher received power density. - Needs more OFDM symbols to - Robust against temporary sound the entire freq. band.

channel distortion. - Smaller SRS length, worse detectability.

Table 2.1: (Non-)Frequency hopping - pros and cons [5].

160 ms, due to that frequency hopping can be used [21].

The concept of frequency hopping is to generate smaller SRS signals and spreading the transmission over the entire frequency during different time slots. This is shown in Figure 2.6, where possible schemes of frequency and non-frequency hopping is plotted. The possible positive and negative effects of both options are listed in Table 2.3.3.1 [5].

The potential advantage of non-frequency hopping is that the entire frequency band can be sounded at once and the detectability is higher for wide band signals as for narrow band. On the other hand, the use of the entire frequency band will result in a lower received power density which is correlated with a lower hearability. In case of using frequency hopping the SRS is transmitted on smaller frequency bands, and hence has a higher received power density but at the same time the use of narrow band signals are more difficult to detect. This difficulty of narrow/wide band detection is discussed more precisely in Section 4.1. Another drawback is that more Orthogonal Frequency-Division Multiplexing (OFDM) symbols are used to sound the entire frequency band and during each SRS transmission the entire band is reserved for SRS transmission.

2.3.3.2 Sequence Generation

The previous section showed the allocation of the SRS in the time domain. In terms of the frequency domain the SRS length, L^max_seq, depends on the system bandwidth and on the chosen option of frequency hopping which is decided by the network. The possible length are all units of 4 RB in the frequency domain and are defined in [8] Section 5.5.3.2. For instance, in case of a system bandwidth of 10 MHz there are 50 RB available, based on [8] Section 5.5.3.2 Table 5.5.3.2-2, the SRS length, L^max_seq , without frequency hopping is 48 RB long. When frequency hopping is applied, L^max_seq would be a fraction of 48 RB, depending on the defined frequency hopping size by the network. The generic description of an SRS formula in the frequency domain with a maximum length of L^max_seq is given by

γu^(a,v^p^˜⁾[n] = e^iα^p^˜ⁿ¯γ^u,v[n] , 0 ≤ n < L^max_seq, (2.3)

(34)

0 ms 5 ms

Slot

Subframe

Frame

12 Subcarriers

Slot 1: 7 Symbols Slot 2: 7 Symbols Resource Block

Resource Element 10 ms

Fig. 2.5: An uplink LTE signal structure.

Time

Non-frequency hopping SRS

SRS realizaons

Frequency hopping SRS or

Subframe

Fig. 2.6: Frequency and non-frequency hopping transmitting schemes.

(35)

and consists of two parts influenced by the parameters u, v, and αp_˜which are the group and sequence numbers and the cyclic shift. The cyclic shift affects the exponential part of the function and is described by

αp_˜= 2πn^cs,˜_SRS^p

8 , (2.4)

n^cs,˜_SRS^p = (n^cs_SRS + 8¯p Nap

) mod 8, (2.5)

with

˜

p∈ {0, 1, . . . , Nap− 1}. (2.6)

As Equations 2.4 and 2.5 show, the cyclic shift depends on the number of mobile antenna ports, N_ap, and the cyclic shift value n^cs_SRS which is configured separately for periodic and each configuration of aperiodic sounding by the higher-layer parameters. The exponential function is called cyclic shift since a cyclic shift in time domain is equivalent to a phase rotation in frequency domain. By applying a cyclic shift onto a base sequence, ¯γ^u,v[n] , up to 8 orthogonal signals can be generated. Which has the advantage of reusing a base sequence while effecting a low interference between each other.

The second part is the actual base sequence, ¯γ^u,v[n], which can be generated with two different methods. Depending on the maximum length of the reference signal sequence, L^max_seq, the sequence is generated based on Zadoff-Chu or Quadrature phase-shift keying (QPSK) sequence.

The reason for using two different sequences generation methods, is to be able to generate at least 30 different sequences which allows to separate between 30 different groups of sequences. This has the advantage that neighbour cells are using a different group of sequences, which is important to provide UTDoA simultaneously in all cells. This requirement is not fulfilled when Zadoff-Chu is applied and the sequence length is smaller than 3 times the resource block size in the frequency domain, N_SC^RB, which is expressed as a number of subcarriers. This means in case of L^max_seq ≥ 3N_SC^RB the base sequence is

¯

γ^u,v[n] = aq[n mod NZC], (2.7)

where a_q[m] is the qth root Zadoff-Chu sequence and expressed by

aq[m] = e⁻ⁱ^πqm(m+1)^NZC with 0 ≤ m ≤ NZC− 1. (2.8) The symbol q, in Equation 2.8, is the Zadoff-Chu index. Depending on this index, a different sequence is generated. As seen in the formula

q=

¯ q+ 1

2

+ v(−1)^⌊2¯^q⌋ (2.9)

(36)

Group 0

Group 1

Group 29

1 2 3 4 7 6 7 8 96 100 108

u

v

Fig. 2.7: Available groups and sequences [3].

and

¯

q= NZC

(u + 1)

31 , (2.10)

the index depends on the Zadoff-Chu length, N_ZC, and the group number and sequence index u and v, which are depending on the group and sequence hopping options. All together, there can be 30 groups in a system, where

u∈ {0, 1, . . . , 29}, (2.11)

is the group index and two sequences when the sequence length is above L^max_seq >6 · N_SC^RB as shown in Figure 2.7.

Both, group and sequence hopping, can be used in a hopping or non-hopping mode, with the restriction that just one can be used for hopping at each time. The hopping modes are used to randomize the reuse of a sequence in a system. Therefore, both hopping values, u and v, are based on a pseudo-random sequence which is a length-31 Gold sequence, Figure 2.8. Before u and v can be calculated by Equations 2.17 and 2.19 the Gold sequence c[n] needs to be generated. Generically, the output sequence c[n] of length MP N is defined by [4]

c[n] = (x1[n + Nc] + x2[n + Nc]) mod 2, (2.12) x1[n + 31] = (x1[n + 3] + x1[n]) mod 2, (2.13) x₂[n + 31] = (x2[n + 3] + x2[n + 2] + x2[n + 1] + x2[n]) mod 2, (2.14) where n = 0, 1, . . . , MP N − 1 and to ensure a pseudo-random sequence the Gold sequence algorithm must be in a steady state, to fulfill this requirement the first N_c = 1600 symbols are removed.

x1[n + 31] starts with a sequence of a one followed by zeros, which as shown in Figure 2.8. The second part, x₂[n + 31], to calculate c[n] is initialized by cinitand calculated for group hopping by

cinit= n^RS_ID 30

. (2.15)

(37)

0

+ +

0 0 0 0 0 0 0 0 0 0 1

X

X X X X X X X X X X X

+

x¹

x2

C(N)

Fig. 2.8: The Gold sequence scheme [4].

In case of sequence hopping by

cinit = n^RS_ID 30

· 2⁵+ (n^RS_ID + ∆ss) mod 30, (2.16) where ∆ss ∈ 0, 1, . . . , 29 is configured by higher layers and n^RS_ID = N_ID^cell = 0, 1, 2, . . . , 503 is based on the cell IDs. These pseudo-random sequences can be then used to calculate the hopping values. In terms of the group index, the value is defined by

u= (fgh[ns] + fss) mod 30, (2.17) where f_ss = n^RS_ID mod 30 is the sequence-shift pattern, and fgh[ns] defines the group hopping option by

f_gh[ns] =

(0, group hopping disabled

P7

i=0(c[8ns+ i] · 2ⁱ), group hopping enabled (2.18) where c(i) is the pseudo-random sequence. In terms of sequence index, the length of the sequence is important. As seen in Figure 2.7, there are just more than one sequences available for L^max_seq >

6 · N_SC^RB, this means

v=

(c[ns], L^max_seq >6N_SC^RB and sequence hopping enabled

0, otherwise. (2.19)

On the other hand, if

L^max_seq = 2N_SC^RB or L^max_seq = N_SC^RB, (2.20) a QPSK sequence is used to generate the base sequence. In this case

¯

γu,v[n] = e^iϕ[n]^π⁴ with 0 ≤ n ≤ L^max_seq − 1 (2.21)

(38)

Option Advantage

Base sequences Relatively low but still non-zero mutual correlation.

Group/sequence hopping Base sequence separation.

Cyclic shift 8 orthogonal signals in one cell.

Comb-spectrum Each Cyclic Shift can be used twice per cell.

No data transmission No interference by data transmission.

Table 2.2: SRS interference management.

where ϕ[n] is given in the Appendix B, and is based on Table 5.5.1.2-1 and 5.5.1.2-2 of the technical specification [8].

2.3.3.3 Interference Management

As described, a number of methods are used to lower the influence of interference by other UEs.

A complete list of all options and their advantages is given in Table 2.2.

The used base sequence, based on Zadoff-Chu or QPSK, enables a relatively low but still non-zero mutual correlation. Since it can happen that several UEs have the need to send an SRS at the same time, the base sequences are multiplied with a cyclic shift as it is shown in Equation 2.3.

Hence, eight orthogonal signals can be generated. Further through the use of a comb-spectrum, the number of possible orthogonal signals can be doubled. Thereby, more orthogonal signals can be generated which causes a low intra-cell interference, nevertheless the impact of inter-cell interference needs to be considered, because of the limited number of groups. On this purpose group and sequence hopping are introduced. It is based on a Gold sequence, and leads to a better distribution of the reused base sequence in the network. Last but no least, beside the already mentioned techniques, is the feature of forbidding data transmission during SRS transmission, an effective technique to decrease the amount of interference.

(39)

3 Scenario, System Models and Assumptions

To obtain adequate and comparable results, proper scenarios, system models and assumptions are required. In the following sections the applied specifications are listed.

3.1 Scenarios

To achieve a result which is reliable and comparable to results from other research groups, a standardized scenario needs to be chosen. Therefore, the scenarios considered in this thesis are according to 3GPP agreements in indoor positioning study of LTE Release 13 [6].

This master thesis is limited to the first case scenario described in [6], which is called ’Out- door macro + outdoor small cell deployment scenarios and outdoor macro-only deployment scenario (outdoor small cells = 0)’. In this study the number of outdoor pico cells equals to zero, Scenario 1, and four, Scenario 2, are considered. This means the case of having zero or four pico cells in each cell is investigated. The deployment consist of a hexagonal grid with seven macro sites where each has 3 sectors and the Inter-Site Distance (ISD) is 500 meters. Further, to provide a more realistic approach, a wraparound option of the deployment is integrated by which the simulation becomes equivalent to simulating the network as if all sites were surrounded by other sites. A detailed description of the radio distance wrapping technique can be found in [22].

A graphical presentation of Scenario 1 and Scenario 2 is given in Figure 3.1, and more details about the applied parameters is described in the outline of the scenario, given in Table 3.1. Beside these basic definitions, there are UTDoA specific parameters given in Table 3.2. Depending on the investigated problem formulation the applied UTDoA specific parameters can change, which are provided in the result chapter for each formulated problem.

3.1.1 Standard and Emergency Case

In case of an emergency, precise position estimation is required. For that purpose the accuracy can be increased by allowing just one UE to transmit its signal whilst the others are muted. At the same time, this case gives information about the upper bound of the applied UTDoA method under the given scenario constraints.

The standard case describes the scenario as it usually is. This means there is no data communication allowed during an SRS transmission, although there can be other SRS transmissions.

17

(40)

18 3 Scenario, System Models and Assumptions

X-Direction [m]

-800 -600 -400 -200 0 200 400 600 800

Y-Direction [m]

-600 -400 -200 0 200 400 600

8

12 15

6 4

Deployment - 49 UE, 0 Picos

21 19 10

11

1 16 20

3 5

13 14

18

2

7 17 9

(a) (b)

X-Direction [m]

-800 -600 -400 -200 0 200 400 600 800

Y-Direction [m]

-600 -400 -200 0 200 400 600

1 2

3

4

5

6

7

8 9

10

11 12

13 14

15

16 17

18 19

20

21

Deployment - 49 UE, 4 Picos

Fig. 3.1: Scenario deployment with 49 UEs: (a) Scenario 1; (b) Scenario 2.

Parameter Outdoor macro cell Outdoor pico cell

Layout Hexagonal grid, 3 sectors per site, Distributed in each macro cell

7 Macro sites, ISD = 500m

Carrier frequency 2.0 GHz 2.0 GHz

Carrier number 1 1

Total PBS,TX(Ptotal per carrier) 46dBm 30dBm

Distance-dependent path loss 3D-UMa (Table 7.2-1 in TR36.873 [23]) 3D-UMi (Table 7.2-1 in TR36.873 [23])

Penetration outdoor 0dB 0dB

Penetration indoor 20dB+0.5din; dinuniform[0, min(25, 20dB+0.5din; dinuniform[0, min(25, UE-to-eNodeB distance)] UE-to-eNodeB distance)]

Shadowing 3D-UMa (Table 7.3-6 in TR36.873 [23]) 3D-UMi (Table 7.3-6 in TR36.873 [23])

Antenna pattern 3D, referring to TR36.819 [24] 3D, referring to TR36.819 [24]

Antenna height 25m + α, where α uniform[-5, 25] 10m + β, where β uniform[-5, 10]

UE height hUE=3(nfl- 1) + 1.5 m

where, nfl uniform[1,nfl]

Antenna gain + connector loss 17 dBi 5 dBi

Antenna gain of UE 0 dBi

Fast fading channel between eNB and UE 3D-UMa (Table 7.3-6 in TR36.873 [23]) 3D-UMi (Table 7.3-6 in TR36.873 [23])

Antenna configuration 1Tx2Rx in UL

Nr. of buildings/clusters per macro cell 1 Nr. of floors per buildings nfl= 8

UE dropping 2/3 UEs randomly and uniformly

dropped within the buildings/clusters, 1/3 UEs randomly and uniformly dropped throughout the macro geographical area. 20% UEs are outdoor and 80% UEs are indoor.

Radius for UE dropping in a building/cluster 70m

Minimum distance (2D distance) Macro - UE : 35m

cluster center-cluster center: 2·50m

UE noise figure 9dB

UE speed 3km/h

Table 3.1: Scenario deployment

(41)

3.2 System Model 19

Parameter Value

UE power class 23dBm

RB allocation for UL-RTOA measurement Periodic SRS, 100 SRS transmissions where these SRS transmissions follow a periodic configuration

of 10ms periodicity [25].

System Bandwidth 100RBs (20 MHz)

75RBs (15 MHz) 50RBs (10 MHz) 25RBs (5 MHz) 12RBs (3 MHz) 6RBs (1.4 MHz)

Frequency hopping On/off

Pico cells 0/4

Muting Emergency/Standard

Cyclic Prefix Normal

SRS power control P srs_offset= 0dB

Quantization FFT size mode of 2048

UL loading and interference Existing methodology in R1-103410 [26].

Table 3.2: UTDoA assumptions and parameters

3.2 System Model

To provide a realistic scenario proper system models are required. For that reason all applied system models are three dimensional models based on 3GPP agreements in indoor positioning study of LTE Release 13 [6]. Based on the 3GPP recommendation, the used model to describe shadowing from obstacles affecting the wave propagation is taken from Table 7.3-6 in TR36.873 [23]. The same table includes also the fast fading channel model between eNodeB and UE. The distance-dependent path loss model is taken from Table 7.3-1 in TR36.873 [23]. Another system component, is the 3D antenna pattern which is described in TR36819 [24]. Further applied models and parameters are listed in Table 3.1 in Section 3.1.

3.3 Assumptions

Some of the applied approaches, as the deployment and the different applied models, are already listed in Tables 3.1 and 3.2. Beside these limitations, further restrictions are applied.

One restriction is the total number of allowed SRS in a cell at the same instant of time.

According to the 3GPP definition of UTDoA and the use of SRS, there are a maximum number of 16 orthogonal signals of a base sequences. Because of that, there can not be more than 16 UEs

(42)

20 3 Scenario, System Models and Assumptions

transmitting their SRS in one cell at the same time.

By definition, no data transmission are allowed during an SRS transmission. This means if there is at least one UE which transmits its SRS, no other UE can transmit data in the same cell. In terms of the simulation, there will be no data transmission in the entire system. This can be realistic in the standard case where many UEs have to transmit their SRS, but also for the emergency case where a high accuracy is required.

Another restriction belongs to the use of group and sequence hopping, explained in section 2.3.3.2. These techniques are used to randomize the distribution of the SRS sequences in the system and to lower the interference. In case of no pico cells, the use of these hopping techniques is not required since there are just 21 cells in the simulation. This means there are just 21 different SRS base sequences needed. In the second case, where four pico cells per cell are available, the number of needed base sequences increases to 105 and outnumbers the available base sequences.

By the use of group or sequence hopping the shortage of base sequences can be solved. Neverthe- less, the group and sequence hopping techniques are not investigated in this thesis. Instead, both hopping modes are disabled and the distribution is based on the cell identification number, which is distributed by the internal simulator to each cell. How the group and sequence hopping can affect the outcome, in case of using pico cells, might be important for future studies.

(43)

4 Implementation, Metrics and Verification

The preceding chapters already discussed essential parts as the theory of the SRS and the applied scenario. For a better understanding, Figure 4.1 shows the relationship of all components. As the sketch shows, there are implementation required in Matlab and C++. The Matlab code includes information about the scenario, as well as the final position estimation, while in C++ the signal generation and processing including the time estimation is done.

Beside the applied metric to determine the time and positioning estimate, the impacts of the considered circumstances onto the accuracy are explained. Finally, the achieved properties and convergence are shown to verify the functionality. Altogether, the purpose of this chapter is to support the reliability of the results presented in Chapter 5.

4.1 Time Estimation

The applied time estimation is a key part of an accurate positioning estimation and can be the bottleneck of the final positioning estimation. The challenging part is to approximate the correct received time, of a desired signal, out of an interference and noise corrupted received signal.

The received signal at eNodeB j is described by

rj[t] =

n

X

i=0

(si∗ chi,j)[t] + nj[t], (4.1) where s_i[t] is the single SRS transmitted by UE i of n active UEs, which is influenced by the channel ch_i,j[t] between UE i and eNodeB j and nj[t] is the Additive White Gaussian Noise (AWGN). To obtain the time delay of the desired reference signal of UE i, which is si[t], needs to be cross- correlated with the desired reference signal. With the correlation function, given by [27]

Rˆsi[t],rj[t][τ ] =

T 2−1

X

t=−^T₂

rj[t]s^∗_i[t − τ ], (4.2)

which is a convolution between the received signal at eNodeB j and the reference signal. The time- delay ˆTdelayis ascertained by taking the value of τ at which the cross-correlation function is for the first time above half of maximum [27]:

21

(44)

22 4 Implementation, Metrics and Verification

Scenario

Dispatcher: Matlab

CDF Plots

GN- Algorithm

Simulator : C++

Generang

SRS OFDM

Channel Determine

RX signal

Correlaon

Obtain Time/distance

2 RX Antennas average

Obtain PAR

Fig. 4.1: Structure of the implementation.

Tˆdelay = argmin

τ

( Rˆ_s_i_[t],r_j_[t][τ ]

max{ ˆRsi[t],rj[t]} >3dB )

. (4.3)

Depending on the used bandwidth of the reference signal, multipath and on influences as interference and noise the decision of the correct peak can be challenging. This means not always the highest peak is the peak of interest. A description of the challenge with narrow and wide bandwidth signals is given in Section 4.1.1 and the impact of multipath on the correlation in Section 4.1.2.

4.1.1 The Impact of the Bandwidth

As the scheme in Figure 4.2 shows, by doing a cross-correlation the correlation degree at each lag can be obtained. Caused by the interference of other signals and additional noise, the output of the cross-correlation will not look so ideal as it is given in Figure 4.2. An additional significant factor is the applied SRS bandwidth. For a better understanding the autocorrelation is plotted in Figure 4.3. It depicts the different resulting autocorrelation for SRS with narrow and wide bandwidths, while narrow signals tending to have slow changing slope and wide signals have a sharp peak.

Because of the sharp peak a more distinct result can be obtained in case of using a wide band signal, while for a narrow band signal more equal peaks appear. This relation between bandwidth and detectability is shown in Figure 4.4, which depicts an example of a cross correlation scheme with a narrow and a wide band signal. As this example shows, in the wide band case a clear peak can be detected, while the narrow band case shows many equal peaks and the direct consequence result in ambiguity. Furthermore, since the detected peaks are almost leveled, influences from interference and noise can lead to detecting a wrong peak as the desired one.

Development and Evaluation of UTDoA as a Positioning Method in LTE

Development and Evaluation of UTDoA as a Positioning Method in LTE

THOMAS ARTHUR HERBERT BRESSNER

Master’s Degree Project Stockholm, Sweden 2015

Degree Project in Signal Processing (EQ272X)

Development and Evaluation of UTDoA as a Positioning Method in LTE

Thomas Arthur Herbert Bressner 890419-2872

Declaration

Abstract

Sammanfattning

Acknowledgement

Contents

List of Figures

List of Tables

List of Nomenclature

Abbreviations

Notations

1 Introduction

1.1 Problem Formulation

1.2 Methodology

1.3 Ethical and societal consideration

1.4 Outline

2 Uplink TDOA Positioning Theory

2.1 Positioning Methods

2.2 Concept of UTDoA

2.3 Uplink Processing

2.3.1 Single-Carrier Frequency Division Multiple Access

2.3.2 Reference Signals: SRS and DM-RS

2.3.3 Sounding Reference Signal

+ +

+

3 Scenario, System Models and Assumptions

3.1 Scenarios

3.1.1 Standard and Emergency Case

3.2 System Model

3.3 Assumptions

4 Implementation, Metrics and Verification

4.1 Time Estimation

4.1.1 The Impact of the Bandwidth