多輸入多輸出正交分頻多工行動網路宏觀天線分集合併機制之研究

(1)

!

୯!ҥ!Ҭ!೯!ε!Ꮲ!

!

ႝᐒᏢଣ!!!ႝߞᏢำ!

!

ᅺ!γ!ፕ!Ў!!

!

ӭᒡΕӭᒡр҅ҬϩᓎӭπՉ୏ᆛၡֻᢀ

Ϻጕϩ໣ӝٳᐒڋϐࣴز!

!

! !

Macro-Diversity Antenna Combining for

MIMO OFDM Cellular Mobile Networks

!

ࣴزғǺڬণᐇ!

!

!!!ࡰᏤ௲௤ǺЦᆿ։!௲௤!

!

(2)

ӭᒡΕӭᒡр҅ҬϩᓎӭπՉ୏ᆛၡֻᢀϺጕϩ໣ӝٳᐒڋϐࣴز

!

Macro-Diversity Antenna Combining for MIMO OFDM Cellular Mobile

Networks

ࣴ ز ғǺڬণᐇ StudentǺZhe-Hua Chou

ࡰᏤ௲௤ǺЦᆿ։ AdvisorǺLi-Chun Wang

୯ ҥ Ҭ ೯ ε Ꮲ

ႝᐒᏢଣ!!ႝߞᏢำ

ᅺ γ ፕ Ў

A Thesis

Submitted to College of Electrical and Computer Engineering National Chiao Tung University

in partial Fulfillment of the Requirements for the Degree of

Master of Science in

Communication Engineering June 2010

Hsinchu, Taiwan, Republic of China

ύ๮҇୯ΐΜΐԃΎД

(3)

ӭᒡΕӭᒡр҅ҬϩᓎӭπՉ୏ᆛၡֻᢀϺጕϩ໣

ӝٳᐒڋϐࣴز!

ᏢғǺڬণᐇ!

!

!!!ࡰᏤ௲௤ǺЦᆿ։!

୯ ҥ Ҭ ೯ ε Ꮲ ႝ ᐒ Ꮲ ଣ ႝ ߞ Ꮲ ำ ᅺ γ ੤

ᄔ

ा

! ! ! ! ! !

ӧ҂ٰ҅Ҭϩᓎӭπ(Orthogonal Frequency Division Multiplexing)Չ୏ᆛၡύǴ୷

ӦѠஒ཮௦ҔӭᒡΕӭᒡр(

Multi-input Multi-output

)

Ϻጕמೌٰׯ๓ჹ٬Ҕޣޑ໺

ᒡ৒ໆ(Throughput)ک᜘่ё᎞ࡋ(Link Reliability)Ƕ྽୷ӦѠ٬ҔϺጕϐޜ໔ӭ

πמೌǴӧΠՉೱ่ޑ௃ݩΠǴᎃ߈Ҟ߻୍ܺ୷ӦѠޑ٬Ҕޣޑ໺ᒡ৒ໆ཮೏ε

ໆޑቚуǴՠࢂ୷ӦѠӄ೽ޑ໺ᒡф౗೏ϩණډӚঁ໺ଌϺጕǴૻဦυᘋуᚇૻ

К౗(Signal to Interference-and-Noise Ratio)཮ᒿ๱໺ଌϺጕঁኧቚуԶ΋ޔफ़

եǴ੝ձࢂӧಒझᜐࣚޑӦБǴ཮೷ԋ໺ᒡ৒ໆ཮ᝄख़Πफ़ǴᏤठಒझ୍ܺጄൎ

ޑᕭ෧ǴࣁΑ఼ᇂচӃޑ୍ܺጄൎǴႝߞ཰ޣሡाթࡌ׳ӭޑ୷ӦѠǴ೭ኬ཮ቚ

у཰ޣ୷ӦѠթࡌޑԋҁکቚу٬ҔޣϪඤډόӕ୷ӦѠޑԛኧǶӧჴሞޑᕉნ

ύǴҗܭϺጕϐ໔ؒԖى୼ޑຯᚆکલЮ೯ၰޑණ৔ਏᔈ(Scatter Effect)ǴϺጕ࣬

ᜢ܄཮Ӹӧܭ୷ӦѠ܈٬ҔޣǴऩ໺ଌϺጕᆄӸӧϺጕ࣬ᜢ܄ǴӭᒡΕӭᒡрϺ

ጕϐޜ໔ӭπਏૈ཮εεޑڙډቹៜǴӵԜǴाှ،ӭᒡΕӭᒡрϺጕϐޜ໔ӭ

πޑ఼ᇂୢᚒ཮ᡂԋ΋ঁ׳֚ᜤޑπբǶ

ךॺӧፕЎύගрٿᅿֻᢀϺጕϩ໣่ӝᐒڋǴቚу٬ҔӭᒡΕӭᒡрϺጕ

ϐޜ໔ӭπ҅ҬϩᓎӭπՉ୏ᆛၡ୷ӦѠᜐࣚޑ໺ᒡ৒ໆک຾΋؁ׯ๓຾Չඤ

Ћ (Handover) ޑ ໺ ᒡ ৒ ໆ Ǵ ΋ ঁ ࢂ ޜ ᓎ ጓ ዸ ֻ ᢀ ϩ ໣ ่ ӝ ޜ ໔ ӭ π ᐒ ڋ

(Space-Frequency Block Code Macro-Diversity Combining Spatial Multiplexing)

Ǵќ

΋ঁࢂൻᕉۯᒨֻᢀϩ໣่ӝޜ໔ӭπᐒڋ (Cyclic Delay Macro-Diversity

Combining Spatial Multiplexing)

Ǵӧ࣬ᎃޑ୷ӦѠϐ໔ቚу໺ଌϩ໣٠่ӝ୷Ӧ

Ѡҁي٬ҔޑϺጕϐޜ໔ӭπמೌǴӧόቚу໺ଌϺጕ܈୷ӦѠޑঁኧΠǴךॺ

ёаӕਔளډӭᒡΕӭᒡрϐޜ໔ӭπکޜ໔ϩ໣ޑᓬ༈Ǵ྽٬Ҕޣ۳୷ӦѠᜐ

ࣚ౽୏Ǵ࣬ӕၗ਑࿶җޜᓎጓዸ܈ൻᕉۯᒨፓᇙӧ࣬ᎃޑ୷ӦѠа࣬ӕޑᓎ౗໺

ଌǴ٬Ҕޣޑௗԏᆄёа୺ՉϺጕϩ໣่ӝޑᐒڋᕇளϩ໣ቚ੻Ǵගٮ׳ଯޑ໺

ᒡ৒ໆ٠ׯ๓ֻᢀϩ໣Ҭሀ(Macro-Diversity Handover)ޑਏૈǶ

ኳᔕ่݀ᡉҢǴ୷ܭफ़եس಍ፄᚇ܄аϷჹ౜Ӹคጕ೯ૻس಍ޑঋ৒܄Ǵך

ॺёаᒧ᏷௦ҔൻᕉۯᒨֻᢀϺጕϩ໣่ӝޜ໔ӭπᐒڋٰׯ๓٬Ҕޣӧ୷Ӧ

Ѡᜐࣚय़ᖏ໺ᒡ৒ໆό٫ޑୢᚒǶٿᅿֻᢀϺጕϩ໣่ӝᐒڋӧ໺ଌᆄϺጕ࣬ᜢ

܄ၨեޑచҹΠǴૈεໆׯ๓٬Ҕޣӧ୷ӦѠᜐࣚޑ໺ᒡ৒ໆǶ྽໺ଌϺጕ࣬ᜢ

܄ࣁ0.1ਔǴޜᓎጓዸֻᢀϺጕϩ໣่ӝޜ໔ӭπᐒڋёаගٮКচӃϺጕϐޜ

(4)

VII

໔ӭπᐒڋӭ2 bit/s/Hsޑ໺ᒡ৒ໆǴ྽໺ଌϺጕ࣬ᜢ܄ࣁ0.9ਔǴΨᗋቚу0.5

bit/s/Hs

ޑ໺ᒡ৒ໆǶԶൻᕉۯᒨֻᢀϺጕϩ໣่ӝޜ໔ӭπᐒڋ྽໺ଌϺጕ࣬ᜢ

܄ࣁ0.1ਔǴёаගٮКচӃϺጕϐޜ໔ӭπᐒڋӭ1 bit/s/Hsޑ໺ᒡ৒ໆǴ྽໺ଌ

Ϻጕ࣬ᜢ܄ࣁ0.9ਔǴΨᗋቚу0.5 bit/s/Hsޑ໺ᒡ৒ໆǶᒿ๱ӧ୷ӦѠ໺ଌϺጕ࣬

ᜢ܄ޑቚуǴϝฅёаගٮКচҁ٬ҔϺጕϐޜ໔ӭπޑᐒڋ׳ଯޑ໺ᒡ৒ໆǴ

ԶЪޜᓎጓዸֻᢀϺጕϩ໣่ӝޜ໔ӭπᐒڋکൻᕉۯᒨֻᢀϺጕϩ໣่ӝޜ

໔ӭπᐒڋϐ໔ჹ໺ᒡ৒ໆׯ๓ޑৡ౦ຫٰຫλǶ

(5)

Macro-Diversity Antenna Combining for MIMO

OFDM Cellular Mobile Networks

Student

ǺZhe-Hua Chou

Advisor

ǺLi-Chun Wang

Degree Program of Electrical and Computer Engineering

National Chiao Tung University

ABSTRACT

In the future orthogonal frequency division multiplexing (OFDM) cellular

network, base station will adopt multi-input multi-output (MIMO) techniques to

improve throughput and link reliability for mobile stations. When the base station

using spatial multiplexing (SM) in the downlink case, the throughput of mobile station

can be hugely increased nearby the serving base station. However, the total transmit

power is split uniformly across the transmit antennas, the signal to

interference-and-noise ratio (SINR) degreases with the increasing number of transmit

antennas. Additionally, the capacity will seriously decrease at the cell boundary. Thus,

SM will reduce the cell coverage in this kind of MIMO OFDM system, the

telecommunications operators need to set up more base stations for covering the

service areas. This will raise the base station equipment cost of the operators and

increase the handover frequency. In a real environment, spatial correlation exists at

base station or mobile station due to insufficient antenna separation and the lack of

scatter effect. The performance of SM degreases seriously with non-trivial spatial

correlation among the transmit antennas, therefore, it becomes a more difficult task to

overcome the problem of cell coverage reduced by SM.

In this thesis, we introduce two kinds of macro-diversity combining techniques

for spatial multiplexing based MIMO OFDM cellular networks to increase cell

boundary throughput and improve soft handover performance. The one is

space-frequency block code macro-diversity combining spatial multiplexing scheme.

The other is cyclic delay macro-diversity combining spatial multiplexing scheme.

Applying transmit diversity at adjacent base station sides and using SM of each base

station, we can take both advantages of spatial multiplexing and spatial diversity

without increasing the number of transmit antennas or base stations. When a mobile

station moves near the cell boundary, the adjacent base stations transmit the same data

encoded by SFBC or CDD at the same frequency. At the receiver of the mobile station

can perform diversity combining to get the macro-diversity gain, thereby improving

throughput and the handover performance.

Because of lower complexity and compatibility to existing wireless

communication systems, we conclude that adopting the CDD macro-diversity

(6)

IX

combining with SM scheme is a preferable scheme to improve the poor throughput of

mobile stations near the cell boundary. These two macro-diversity combining schemes

can provide a high throughput improvement of mobile station at the cell area in low

spatially-correlated channels. As transmit antenna spatial correlation is equal to 0.1,

SFBC macro-diversity combining with SM scheme can improve 2 bit/s/Hz more than

the conventional SM scheme. As transmit spatial correlation is equal to 0.9, it can still

improve 0.5 bit/s/Hz higher than the conventional SM scheme. As transmit antenna

spatial correlation is equal to 0.1, CDD macro-diversity combining with SM scheme

can improve 1 bit/s/Hz higher than the conventional SM scheme. As transmit spatial

correlation is equal to 0.9, it can also improve 0.5 bit/s/Hz more than the conventional

SM scheme. With spatial correlation increased at each transmitters of each base

station, they also can provide higher throughput than the conventional SM scheme.

The variations of throughputs between CDD macro-diversity combining with SM

scheme and SFBC macro-diversity combining with SM scheme are getting small at

the cell boundary.

(7)

Acknowledgments

I would like to thank my parents and my wife. They always give me endless supports and have confidence in me. I especially would like to thank professor Li-Chun Wang who gave me many valuable suggestions in the research during these years and take care of me. I would not finish this work without his guidance and comments.

In addition, I am deeply grateful to my laboratory mates, Chiao Lee, Hsien-Wen Chang, Ang-Hsun Tasi and junior laboratory mates at Wireless System Laboratory at the Department of Communications in National Chiao-Tung University. They provide me much assistance and share much happiness with me.

(8)

XI

List of Tables

2.1 Alamouti Scheme Table . . . 8 2.2 Comparison of Recent Research Works . . . 11 3.1 Multi-Path Delay and Power Profile of The ITU PEDESTRIAN B Channel

Model . . . 14 3.2 Reference EESMβ Values for ITU Pedestrian B Channel . . . 19 4.1 Simulation Parameters . . . 26 4.2 Throughputs Improvement by DSFBC-SM Compared with The

Conven-tional SM. . . 33 5.1 Throughputs Improvement by DCDD-SM Compared with The

(11)

XIV

List of Figures

2.1 Principle of HO initialization in MDHO and FBSS. Black arrows present the time instance of initialization of HO. . . 7 3.1 The System Model of Macro-Diversity Combining for MIMO OFDM

Cel-lular Mobile Network in The Downlink Case. . . 13 3.2 Block Diagram of a Conventional OFDM System Model. . . 16 4.1 The Block Diagram of Space-Frequency Block Code Macro-Diversity

Com-bining with Spatial Multiplexing Scheme. . . 23 4.2 DSFBC-SM Simplified System Model . . . 23 4.3 Throughputs of SM 2x2 , SFBC 2x2 and DSFBC-SM 4x2 withρtx= 0.1 at

each BS. . . 27 4.4 Throughputs of SM 2x2 , SFBC 2x2 and DSFBC-SM 4x2 with_ρtx= 0.3 at

each BS. . . 28 4.5 Throughputs of SM 2x2 , SFBC 2x2 and DSFBC-SM 4x2 withρtx= 0.5 at

each BS. . . 31 4.8 Throughputs of SM 2x2, SFBC 2x2 and DSFBC-SM 4x2 schemes for

(12)

4.9 Throughputs of SM 2x2, SFBC 2x2 and DSFBC-SM 4x2 schemes for dif-ferent_ρtxwith the distance from the serving BS equal to 750m (Rc). . . 33

5.1 The Block Diagram of Cyclic Delay Macro-Diversity Combining with Spa-tial Multiplexing Scheme. . . 35 5.2 DCDD-SM Simplified System Model. . . 36 5.3 Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 withρtx= 0.1 at each BS. . . 39 5.4 Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 withρtx= 0.3 at each BS. . . 40 5.5 Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 withρtx= 0.5 at each BS. . . 41 5.6 Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 with_ρtx= 0.7 at each BS. . . 42 5.7 Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 with_ρtx= 0.9 at each BS. . . 43 5.8 Throughputs of SM 2x2, SFBC 2x2, DSFBC-SM 4x2 and DCDD-SM 4x2 schemes for different ρtx with the distance from the serving BS equal to

500m (2Rc/3). . . 44 5.9 Throughputs of SM 2x2, SFBC 2x2, DSFBC-SM 4x2 and DCDD-SM 4x2

schemes for different ρtx with the distance from the serving BS equal to

750m (Rc). . . 45

(13)

(14)

1

CHAPTER 1 Introduction

It is well-known that multi-input multi-output (MIMO) antenna transmission systems can provide spatial multiplexing gain and diversity gain to increase spectral efficiency and link reliability [1] [2] [3] [4]. For future broadband wireless systems, in order to meet the data rate and quality-of-service (QoS) requirements of mobile stations, spectral efficiency and link reliability should be improved. Spatial multiplexing techniques in MIMO systems can provide huge capacity gains by creating a number of spatial pipes, through which several independent information streams can be transmitted at the same frequency. MIMO systems also provide independent fading channels between the transmit antennas and receiver an-tennas. The replicas of the same signal yield the transmit and receive diversity techniques. Orthogonal frequency division multiplexing (OFDM) is a very popular modulation technique for transmitting broadband signals in multi-path fading environments, which converts a frequency selective fading channel into a parallel collection of frequency flat fading sub-channels. OFDM can overcome inter-symbol interference (ISI) by applying a cyclic symbol extension. Besides, orthogonal sub-carriers with overlapped spectra can achieve high spectral efficiency. However, OFDM based transmission schemes has no built-in spatial diversity, therefore combining MIMO with OFDM (MIMO OFDM) has been chosen as an attractive air-interface solution for the next generation high speed wire-less systems [5], including the IEEE 802.16e/WiMAX and the 3GPP long-term evolution (LTE).

(15)

1.1 Problem and Solution

MIMO technology exploits the spatial components of the wireless channel to improve throughput and bit error rate (BER) performance of communications systems, through multiplexing or diversity transmission techniques. In future cellular mobile networks, sup-porting spatial multiplexing [6], the mobile station can receive sub-streams from multiple transmit antennas on one or more base stations. When the base station uses spatial mul-tiplexing in MIMO OFDM system, the total transmit power of the base station is split uniformly across transmit antennas. Increasing the number of transmit antennas leads to lower signal to interference-and-noise ratio (SINR) per degree of freedom. Additionally, the capacity decreases seriously at the cell boundary. Thus, SM technique will decrease the cell coverage in this kind of MIMO OFDM systems [7]. In order to provide radio services to mobile station, the telecommunications operators need to set up more base stations for covering the service areas. This leads to raise the base station equipment cost of the op-erators and increases the handover frequency. In the hard handover case, it increases the probability of service disruption and ping-pong effect [8]. Soft handover can improve this drawback of hard handover.

Besides, the increase in throughput using spatial multiplexing relies on the inde-pendence of the channels from a transmitter to a receiver. If we consider the spatially correlated channel, the impacts of spatial correlation on spatial multiplexing can be dra-matic [9]. Spatial correlation at the transmitter increases the linear dependence of the input streams response, which makes stream separation and decoding processes more difficult.

Through spatial multiplexing, the MIMO OFDM cellular mobile network can pro-vide a huge increase in throughput for the mobile station close to the serving base station, but it is quite difficult to improve or maintain the throughput of mobile station at the cell boundary, especially in spatially-correlated channels. The performance of spatial multi-plexing degrades seriously. In order to overcome the drawbacks of spatial multimulti-plexing, we

(16)

provide a simple and effective method to increase the SINR by applying transmit diversity techniques at the adjacent base station sides instead of the transmitters of each base station. When the mobile station moves near the serving base station, the huge throughput is pro-vided by the serving base station using SM. When the mobile station moves near the cell boundary, the throughput is improved by applying transmit diversity techniques with the cooperation of the adjacent base stations.

1.2 Challenges

In order to provide required data throughput of mobile station around the cell edge, the serving base station performs spatial multiplexing to offer more capacity for the nearby mobile station. Due to path loss and spatial correlation, the SINR decreases seriously as the distance from the serving base station. Around the cell edge, the throughput per-formance of spatial multiplexing is even worse than spatial diversity, because SM needs high SINR for reliable detection. Both [10] [11] have shown that space-time block code achieves higher capacity and throughput than spatial multiplexing at low SINR, especially when there is non-trivial spatial correlation among the transmit antennas. The capacity of spatial multiplexing decreases significantly at low SINR. Because the total transmit power is limited for each base station in a cellular network to avoid interference to other cells and system constrains, increasing the transmit power is not a reasonable and efficient method to overcome this difficulty. We need to find another method suitable for a cellular system to increase the SINR of mobile station around cell edge. To increase spatial diversity and maintain the number of transmit antennas of a base station, we can apply transmit diversity scheme at the adjacent base station sides. When a mobile station moves to cell boundary, the adjacent base station cooperates with the serving base station to transmit the same sig-nals, and the mobile station combines the received signals to get the spatial diversity for increasing SINR. The required data throughput of the mobile station can be achieved and

(17)

the problem of cell coverage reduced due to SM is solved.

In this thesis, we introduce two kinds of macro-diversity combining schemes for spatial multiplexing based MIMO OFDM cellular mobile networks to increase throughput at the cell boundary. One is the space-frequency block code macro-diversity combining with spatial multiplexing scheme and the other is cyclic delay macro-diversity combining with spatial multiplexing scheme, these two schemes are investigated under a more realistic channel model in which spatial correlation due to insufficient antenna separation at the transmitters of each base station and the lack of scatter effects of the channel. Through a system performance analysis approach, we compare the performances of the two different macro-diversity combining schemes and their improvements, which are evaluated in terms of the distance from the desired base station versus throughput.

1.3 Thesis Outline

The rest of this thesis is organized as follows. Chapter 2 introduces the backgrounds of macro-diversity applications in CDMA and OFDM systems and transmit macro-diversity techniques for MIMO OFDM cellular mobile networks. Chapter 3 shows the system model of macro-diversity scheme for MIMO OFDM cellular network. The effects of spatial corre-lation are considered by applying spatially-correlated Rayleigh MIMO channel. The Expo-nential Effective SNR Mapping (EESM) method is introduced for system level evaluation to simplify simulation complexity. In Chapter 4, we analyze the SFBC macro-diversity combining with SM scheme and discuss the simulation results. In Chapter 5, we analyze CDD macro-diversity combining with SM scheme and discuss the simulation results com-pared with SFBC macro-diversity combining with SM scheme. Conclusions are given in Chapter 6.

(18)

5

CHAPTER 2 Background

In this chapter, we discuss the macro-diversity applications in code division multiple access (CDMA) [12] and OFDM [13] systems and the differences between them. Then, we in-troduce the transmit diversity techniques in OFDM-based systems. The transmit diversity techniques are applied at the adjacent base station sides and the receiver of mobile station performs macro-diversity combining in MIMO OFDM cellular mobile network.

2.1 Macro-Diversity Applications in CDMA and OFDM

Systems

In CDMA systems, macro-diversity is provided for the downlink in a soft handover case. Diversity gain is achieved by using Rake receiver with maximal-ratio combining (MRC) at mobile station. Rake receiver can combine the multi-path signals, which is the time-delayed versions of the original signal. When the one of Rake receiver fingers in the user is receiving signal from one base station, and another might be receiving signal from another base station, the Rake receiver performs maximal-ratio combining to get the diversity gain [14] .

For OFDM systems, there are two types of soft handover defined in IEEE 802.16e [15]: Fast Base Station Switching (FBSS) and Macro Diversity Handover (MDHO). Their initiation criteria are assumed by the IEEE 802.16 management messages. one is

(19)

chan-nel quality indicators, such as CINR (Carrier to Interference Noise Ratio) or the signal strength. Another is QoS characterized by service level prediction, or other criteria such as the bit error rate(BER) [13]. Mobile station and base stations maintain the list of base stations involved in the handover (HO) procedure, which is called diversity set. There are two threshold defined: Add Threshold (ADD T) and Delete Threshold (Delete T) as Fig. 2.1. While the Add Threshold is used to decide the absolute CINR level for adding base station into the diversity set, and the Delete Threshold is used to decide the absolute CINR level for removing base station into the diversity set. When one of these threshold is met by the adjacent base station, the HO procedure is started. For downlink, in case of FBSS, the mobile station communicates only with so called anchor base station from the diversity set, which the mobile station is currently synchronized and registered. mobile station performs ranging with anchor base station and monitors the downlink channel for control information. The anchor base station can be switched among all base stations in the Diversity set on frame by frame basis. It reduces the HO procedure for switching of anchor base stations, In case of MDHO, the mobile station communicates with all base stations in the diversity set, and all base stations transmit the same data to the mobile station such that the mobile station can perform the diversity combining. Because FBSS mobiles does not perform diversity combining, we focus on MDHO case which can provide diversity gain to improve the performance at the cell boundary.

2.2 Transmit Macro-Diversity Techniques for MIMO OFDM

Cellular Mobile Networks

Open loop transmit diversity techniques are used to realize the reliability benefits of multi-antenna systems in flat fading environment without channel state information (CSI) avail-able at the transmitter side. They transmit multiple signals conveying the same information

(20)

Figure 2.1: Principle of HO initialization in MDHO and FBSS. Black arrows present the time instance of initialization of HO.

over different spatial channels. Alamouti space-frequency block code (SFBC) [16] and cyclic delay diversity (CDD) [17] are widely used for 2-transmit antenna diversity sys-tems. Here we apply the two transmit diversity techniques at the adjacent base stations side instead of multiple antennas for one base station to perform transmit macro-diversity in OFDM-based system, and the receiver of the mobile station can get the spatial diversity gain to improve SINR for better performance.

2.2.1 Space-Frequency Block Code

The signals of SFBC are encoded into multiple sets of orthogonal signals in the frequency domain for the transmit antennas, and superimposed in the receiving antennas. After

(21)

ceiver decodes signals from channels with different frequencies, they are combined to re-construct the original signals. Because OFDM can convert the frequency selective channel to a set of flat sub-channels, SFBC schemes can be extended to frequency selective channel in OFDM systems.

In Alamouti SFBC scheme, two consecutive signals are encoded for two transmit antennas. Let the two consecutive signals in frequency domain be_X₁(m) and X₂(m). They are coded as Table 2.1.

Table 2.1: Alamouti Scheme Table OFDM Transmit Transmit sub-carrier index antenna 1 antenna 2

ith_sub-carrier _X

1(m) X2(m)

(i + 1)th_sub-carrier _−X

2(m)∗ X1(m)∗

2.2.2 Cyclic Delay Diversity

CDD is a simple approach to introduce spatial diversity in an OFDM based transmission that itself has no built-in diversity and can be applied at the transmitter or receiver [18]. The signal is not truly delayed, but cyclic shifted between the respective antennas. It actually provides virtual echos and thus increases the frequency selectivity of the channel seen by the receiver. When CDD inserts the delays in time domain, it appears as different phase angles in different sub-carriers in frequency domain after FFT. For Mt transmit antennas

the transmitted signal is given by

x(t)CDD = √1 Mt Mt i=1 x (t − τi) , (2.1)

wherex(t) is the original signal, τiis the cyclic delay at theithtransmit antenna, andNF F T

is the FFT size. The frequency domain representation of the signal with cyclically shifted is given by,

(22)

YCDD(f) = √1 Mt Mt i=1 Hi(f)X(f)e −j2πfτi NF F T + N₀_. _(2.2)

where_Hi(f) is channel response and N0is AWGN. The composie channel responseHCDD(f)

can be modified as HCDD(f) = √1 Mt Mt i=1 Hi(f)e −j2πfτi NF F T _, _(2.3) and ( 2.2) becomes YCDD(f) = HCDD(f)X(f) + N0 . (2.4)

2.2.3 Comparsion

SFBC is easy to encode at the transmitter side and decode at the receiver side while still achieving the optimal diversity gain in 2x1 Rayleigh fading channels. Hence, SFBC is very suitable to increase spatial diversity at the adjacent base stations performing cooperation. In most cases, SFBC outperforms CDD. But the main advantage of CDD remains in its structural simplicity and much lower complexity compared to SFBC. When the number of transmit antennas is large than two, no orthogonal schemes of SFBC have been found achieving the full rate [4]. CDD can be designed for arbitrary number of transmit antennas. However, block-error rate (BLER) performance of CDD is worse than SFBC. Besides, CDD has some drawbacks such as frequency-selective nulls [19] in spatially-correlated channels. It means CDD can not achieve the same diversity gain as obtained in antenna-uncorrelated channels [20]. Fortunately, the distance between the adjacent base stations is far enough, we can apply CDD at the adjacent base stations free from spatial correlation effects and get the diversity gain.

(23)

2.3 Literature Survey

In the literature, the bit error rate (BER) performance and throughput of the mobile sta-tion nearby the cell boundary are studied in some depth. The work in [21] only investi-gated the BER improvement of a CDMA cell system. That applied space-time block code (STBC) between the base stations and considers the inter-cell-interference (ICI) in each cell. In [22], the BER performance of OFDM cellular network was investigated, CDD was applied between the adjacent base stations to improve BER of mobile station at the cell boundary. Both [21] and [22] consider the case that each base station is equipped a single antenna without the implementation of MIMO techniques. Hence, the throughput of mobile stations can not be improved greatly. In [23], the throughput improvement for the distributed MIMO OFDM system was investigated. Each cell is divided to two parts: an inner cell area and an outer cell area. Depending on whether the mobile station is in the inner cell or outer cell, the system selects either a localized or distributed antenna con-figuration based on SNR measurement. It assumed that each base station is equipped two antennas and uses STBC in the serving base station or between the adjacent base stations according the location of mobile station. Different from [21]- [23], this paper developed the macro-diversity combining for MIMO OFDM cellular mobile networks with SM. The serving base station applying SM can offer huge throughput for the nearby mobile sta-tion. We adopt transmit macro-diversity techniques at the adjacent base stations side to increasing SINR of mobile station around cell edge and overcome the drawback of SM. In addition, the impacts of spatial correlation were investigated. The related literature survey is summarized in Table 2.2. It is shown that none of existing work has considered the OFDM-based SFBC or CDD macro-diversity in SM MIMO systems.

(24)

Table 2.2: Comparison of Recent Research Works

CDMA / OFDM Macro-Diversity MIMO in The impact of system method single base station spatial correlation [Fujii, 05] [21] CDMA STBC X X

[Dammann ,07] [22] OFDM CDD X X [Zhou, 06] [23] OFDM STBC STBC X Our work OFDM SFBC and CDD SM O

(25)

12

CHAPTER 3 System Model and Problem Formulation

In this chapter, we consider the MIMO OFDM cellular system in the downlink case. In the distributed macro-diversity scheme, cooperation among base stations is very important. Hence, the system must have a base station controller, which implements functions such as power control, system synchronization and radio resource allocation, etc.

When a mobile station moves nearby the cell boundary of the serving base station, the adjacent base stations transmit the same signals at the same time and over the same frequency to the mobile station due to the MDHO procedure started. It can be regard as a Dynamic Single Frequency Networks (DSFN) [24]. Consider a mobile station in the target Cell 0. Each base station and mobile station are equipped with multi antennas as depicted in Fig. 3.1. There are Mt transmit antennas at each base station and Mr receive

antennas at a mobile station. The mobile station in the cell receives the desired signals form its serving base station and the adjacent base station. Here we consider the impacts of path-loss, multi-path fading channel and spatially-correlated Rayleigh Channel as follows.

(26)

Figure 3.1: The System Model of Macro-Diversity Combining for MIMO OFDM Cellular Mobile Network in The Downlink Case.

3.1 Propagation Model

3.1.1 Path Loss

To begin with, let _PT is the total transmit power of each base station. Then the received

signal after path loss becomes

Pb = P T/Mt

L(db) ,

(3.1) where_L(db) is the path loss and db is the distance from the b-th base station to the receiver

of mobile station. In the following simulations, we will adopt the path loss model defined in [25] as

L(db) = 130.62 + 37.6 log(db) . (3.2)

(27)

3.1.2 Multi-Path Fading Channel Model

Table 3.1 shows that multi-path fading channel model described as ITU Pedestrian-B chan-nel [25], which defines six multi-path delays and power profiles with a velocity of 3 km/hr of the mobile station.

Table 3.1: Multi-Path Delay and Power Profile of The ITU PEDESTRIAN B Channel Model

Number of Paths Power of the each path (dB) Path Delay (ns)

1 0 0 2 -0.9 200 3 -4.9 800 4 -8 1200 5 -7.8 2300 6 -23.9 3700

3.1.3 Spatially-Correlated Rayleigh MIMO Channel

In the real-world wireless channel, spatial correlation can occur at a base station or a mo-bile station due to insufficient antenna separation and scattering environment. Correlation results in diversity loss and performance degradation. In this paper, a downlink case in the MIMO OFDM system is considered. Spatially-correlated Rayleigh MIMO channel is constructed by using the Jakes fading model [26] widely adopted [20] [27]. The simple model can illustrate clearly the impacts of spatial correlation in relative performance. In the following_NtandNr are denoted as the number of transmit and receive antennas, and

H is theNrxNttransfer function of the MIMO channel. Thus, we have

Hc = R1/2HS1/2, (3.3)

where H ∈ CNrxNt _{contains independent entries and distribution is} _{CN(0, 1), and S and}

(28)

correlation model at the transmitter and receiver: _Ri.j = ρ|i−j|rx , Si.j = ρ|i−j|tx . ρrx andρtx

are the receive and transmit spatial correlation coefficients between adjacent antennas.

3.2 OFDM Signal Model

OFDM is a multiplexing technique that subdivides the bandwidth into multiple frequency sub-carriers as Fig. 3.2. The input stream is divided into several sub-streams to reduce data rates and each stream is modulated and transmitted on a separate orthogonal sub-carrier. The symbol duration is increased by reducing data rates. OFDM is more robust to the delay spread, and the introduction of cyclic prefix (CP) can preserves the orthogonality of the sub-carriers and completely avoid the Inter-Symbol Interference (ISI) when the CP duration is longer than the channel delay spread, the equalization at the receiver is low-complexity. In this system model, there are the propagation delays for those signals. A simple FFT window is utilized such that the first signal received from the serving base station (BS 0) is used as the reference signal. All signals from the adjacent base stations will arrive later, and the signals from the adjacent base stations will align to the first received signal.

For kth _{sub-carrier, the received signal} _{R(k) at the location between the serving}

base station and the adjacent base stations is

R(k) = NBS b=0 √ Pb ⎡ ⎢ ⎢ ⎢ ⎣ H_1,1b (k) · · · H_1,Mb t(k) .. . · · · ... HMb r,₁(k) · · · HMb r,Mt(k) ⎤ ⎥ ⎥ ⎥ ⎦ ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ Sb 1(k) Sb 2(k) .. . Sb Mt(k) ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ + ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ Nb 1(k) Nb 2(k) .. . Nb Mt(k) ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ = He_(k)Sb_{(k) + N} 0(k) , (3.4)

whereNBS + 1 is the number of effective base stations, R(k) is a received vector, Hj,ib is

the channel frequency response from the_ith_{transmit antenna of the b-th base station to the}

(29)

Figure 3.2: Block Diagram of a Conventional OFDM System Model.

jth _{receiver of mobile station and BS 0 is the serving base station,}_Sb _{is the signal vector}

of the_bth_{base station,}_N

0 is AWGN of variance, andPb is the received power from thebth

base station represented as ( 3.1).

3.2.1 Signal to Interference-and-Noise Ratio

The composite channel frequency response He_{(k) can also be represented as the Fourier}

transform of composite channel impulse response

he(t) = P₋₁ p₌₀ Mr j Mt i hpj,iδ(t − τp) , (3.5)

where P is total number of paths from different base station, hpj,i and τp are the complex

path-gain and non-negative path-delay associated with the pth _{path, respectively. Those}

(30)

defined as w(t) = ⎧ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎩ 1 _{0 ≤ t < T}g Ts−(t−Tg) Ts Tg ≤ t < Ts+ Tg 0 _{t ≥ T}s+ Tg (3.6)

whereTsandTg is the symbol duration and guard interval in OFDM systems, respectively.

Then the expression for SINR atkth_{sub-carrier can be expressed as}

γ(k) = |H e_(k)|2 Pi+ N_E_s0 , (3.7) where He(k) = (w(t)he(t)) (3.8) and Pi = P p=1 Mr j Mt i (1 − w2_(τ p))hpj,i 2 (3.9) is the ISI caused by paths arriving late and_Esis the transmit energy per QAM symbol. The

ISI impact is avoided by CP design of the OFDM system if it can be assured that all multi-path signals of the target mobile station arrive within the guard interval. For simplicity,_Pi

is ignored in the following calculation of the distributed macro-diversity schemes.

3.2.2 Exponential Effective SINR Mapping (EESM)

EESM is an effective method in predicting link-level performance in a system level sim-ulations [28] [29], and can be generalized to the multi-state channel with different SINR values at each sub-carrier or each symbol transmission time interval (TTI). EESM maps power level and modulation coding schemes (MCS) level to SINR values in AWGN chan-nel domain, which allows using this mapping along with AWGN assumptions in order to predict the effects of MCS. When only the influence of channel frequency selectivity is considered, the equation of EESM is given by [30]

(31)

ref f = EESM(r, β) = −β ln 1 N N i=1 e−riβ , (3.10) where γ is vector [r1, r2, ...rN] of the per-tone SINR values, ref f is the effective SINR

mapping, while_{β is a parameter dependent on the adopted MCS.}

When doing system level simulations, we no longer worry the perfect channel con-dition at the link level. It is only necessary to know AWGN performance of different modulation and coding rates and their corresponding factors_{β. For our simulations, a set} ofβ calibrated for link-level interface in the IEEE 802.16e (WiMAX) is adopted and listed in Table 3.2 [25]. Therefore, interface from link level to system level simulation will be greatly simplified. Once the SINR at a given location is calculated as described above, spectral efficiency (SE) can be obtained for a dedicated MCS and throughput is directly related to the corresponding spectral efficiency (SE). Then the system performance can be measured by using the EESM method.

3.3 Problem Formulation

In future cellular networks, base stations will be equipped with multi-antenna and adopt MIMO techniques to provide higher data throughput and link reliability for better ser-vices or new applications. By using spatial multiplexing (SM), base stations can offer huge throughput for mobile station at neighboring area. The total transmit power of the base station is split uniformly across transmit antennas, and it leads to lower SINR. Since SM needs high SINR for reliable detection, the throughput of mobile stations around cell edge decreases seriously, it is even worse than using spatial diversity (SD). Hence, the cell coverage of the serving base station is reduced. To cover the service area, the telecommu-nications operators need to set up more base stations, this causes higher equipment cost and increases the handover frequency of mobile station due to smaller cell size. In reality,

(32)

Table 3.2: Reference EESMβ Values for ITU Pedestrian B Channel

Code Rate Spectrum Minimum EESM Modulation (Repetition: Efficiency SINR factor

default=1) σ (bit/s/Hz) β QPSK 1/2(4) 0.25 -2.5 dB 2.18 QPSK 1/2(2) 0.5 0.5 dB 2.28 QPSK 1/2 1 3.5 dB 2.46 QPSK 3/4 1.5 6.5 dB 2.56 16-QAM 1/2 2 9 dB 7.45 16-QAM 3/4 3 12.5 dB 8.93 64-QAM 1/2 3 14.5 dB 11.31 64-QAM 2/3 4 16.5 dB 13.8 64-QAM 3/4 4.5 18.5 dB 14.71 19

(33)

spatial correlation can occur at a base station or mobile stations due to insufficient separa-tion and scattering environment, and the correlasepara-tion causes diversity loss and performance degradation. For SM especially, the increased throughput depends upon the fact that the channels from a transmitter to a receiver follow independent paths. Increasing the linear dependence of the input streams response make it more difficult to separate streams and decode signals. The impacts of spatial correlation in SM MIMO systems is more dramatic than that in SD MIMO systems.

In this thesis, we want to get the multiplexing gain of SM and improve the per-formance around the cell edge. Due to the limitation of transmit power for avoiding in-terference and system constrains, increasing transmit power or transmit antennas is not a feasible way to overcome this problem. Therefore, we purpose to apply transmit diversity at the base stations sides for increasing SINR of mobile station at the cell area. With MDHO, when the mobile station moves around the cell edge, the adjacent base stations in diversity set are synchronized to transmit the same signals at the same time over the same frequency to mobile station, seen as a DSFN. SFBC and CDD techniques are adopted for transmit macro-diversity at the base station sides. Each base station performs SM and the impacts of spatial correlation are considered. The performances of the transmit macro-diversity combining with SM are investigated and discussed with simulation results.

(34)

21

CHAPTER 4 Space-Frequency Block Code Macro-Diversity

Combining for MIMO OFDM Cellular Mobile

Networks

Spatial multiplexing (SM) can be applied as a promising and powerful method to dramati-cally increase the system capacity. In a rich scattering environment, the independent spatial channels can be exploited to transmit multiple signals at the same time and frequency re-sulting in higher spectral efficiency. Due to path loss and spatial correlation, the SINR of mobile station degrades when it moves away from the serving base station, and thus decreasing throughput seriously. Spatial Diversity (SD) methods can be used to improve signal quality, which is more robust than SM for the spatially correlated channel. Alamouti transmission structure [16] can be applied in the space-time and space-frequency domain in OFDM systems. By using STBC and SFBC in OFDM cellular networks, the SINR can be increased at the receiver and cell coverage is extended. In contrast to this, using SM in the OFDM cellular networks reduces the cell coverage and needs high receive SINR for reliable detection. Thus it is required to combine the SM and SD MIMO techniques in the cellular network. Both spatial diversity and spatial multiplexing gains can be achieved at farther locations from the serving base station. Because the encoding of SFBC is done across the sub-carriers inside the OFDM symbol and STBC applies encoding across the OFDM symbol, there is an inherent processing delay unavoidable in STBC system. Delay

(35)

spreads in frequency selective fading channels destroy the orthogonality of the receive sig-nals [31]. Thus in OFDM systems the sub-carrier spacing is usually small and symbol time is long. Thus it is more reasonable to use SFBC instead of STBC in the OFDM cellular networks with moving mobiles. In this chapter, we introduce the space-frequency block code macro-diversity combining with spatial multiplexing scheme.

4.1 SFBC Macro-Diversity Scheme

In this section, we introduce the space-frequency block code macro-diversity combining to improve the throughput of the mobile station around cell edge with the cooperation of the adjacent suitable base stations. When the mobile station moves from the serving base station (BS 0) to the target base station (BS 1) as depicted in Fig. 3.1. When reaching the cell boundary, the mobile station receives the signals from two groups of antennas located at adjacent base stations by MDHO. The adjacent base stations transmit the same data encoded by SFBC at the same frequency as depicted in Fig. 4.1. The receiver of the mobile station can performs diversity combining to get the diversity gain. Hence we can increase the SINR of the mobile station around cell edge to provide higher throughput. At the receiver, each antenna receives the MIMO OFDM signals. The guard interval is removed and fast Fourier transform on each receive antenna is performed to get the output signals of OFDM demodulation.

4.2 SINR Performance

When applying SFBC among the adjacent base stations, each base station uses two transmit antennas to perform spatial multiplexing as depicted Fig. 4.2. For N used sub-carriers in the OFDM system, we denote those modulated symbols as sm. Then, the sequences ofsm

(36)

Figure 4.1: The Block Diagram of Space-Frequency Block Code Macro-Diversity Com-bining with Spatial Multiplexing Scheme.

S1:

Figure 4.2: DSFBC-SM Simplified System Model

S0 =s1 s2− s∗3 − s∗4 s5 s6− s∗7− s∗8... s2N−3 s2N−2 − s∗2N−1− s∗2N ; S1 =s3 s4 s∗1 s∗2 s7 s8 s∗5s∗6... s2N−1 s2N s∗2N−3s∗2N−2 .

For the next SM branches,S0is split into two vectorsS₁0andS₂0. Also,S1 is split into two vectors_S₁1 and_S₂1 for the transmit antennas of each base station. Thus, the output signals

(37)

of the SM modules become S₁0 =s1 − s∗3 s5 − s∗7... s2N−3 − s∗2N−1 S₂0 = [s2 − s∗4 s6 − s∗8... s2N−2 − s∗2N] S₁1 =s3 s∗1 s7 s∗5... s2N−1 s∗2N−3 S21 = s4 s∗2 s8 s∗6... s2N s∗2N−2 . ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ R₁_{(2m − 1)} R₁(2m)∗ R₂_{(2m − 1)} R₂(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ H0 11(2m − 1) H012(2m − 1) H111(2m − 1) H112(2m − 1) H1 11(2m)∗ H112(2m)∗ −H011(2m)∗ −H012(2m)∗ H0 21(2m − 1) H022(2m − 1) H121(2m − 1) H122(2m − 1) H1 21(2m)∗ H122(2m)∗ −H021(2m)∗ −H022(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ × ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ S(4(m − 1) + 1) S(4(m − 1) + 2) S(4(m − 1) + 3) S(4(m − 1) + 4) ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ + ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ n₁_{(2m − 1)} n₁(2m)∗ n₂_{(2m − 1)} n₂(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ . (4.1) In short, we can represent ( 4.1) as

Rdsf bcsm(m) = Hdsf bcsm(m)Sdsf bcsm(m) + Ndsf bcsm(m) , (4.2) Rdsf bcsm(m) = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ R1(2m − 1) R1(2m)∗ R2(2m − 1) R2(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ , Hdsf bcsm(m) = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ H0 11(2m − 1) H012(2m − 1) H111(2m − 1) H112(2m − 1) H1 11(2m)∗ H112(2m)∗ −H011(2m)∗ −H012(2m)∗ H0 21(2m − 1) H022(2m − 1) H121(2m − 1) H122(2m − 1) H1 21(2m)∗ H122(2m)∗ −H021(2m)∗ −H022(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ,

(38)

and Sdsf bcsm(m) = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ S(4(m − 1) + 1) S(4(m − 1) + 2) S(4(m − 1) + 3) S(4(m − 1) + 4) ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ , Ndsf bcsm(m) = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ n1(2m − 1) n1(2m)∗ n2(2m − 1) n2(2m)∗ ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ .

where_{m = 1, 2, 3, ....,}N₂. In a linear MMSE receiver, the receive signal vectorRdsf bcsm(m)

is multiplied with GM M SE_{(m). The MMSE detector minimizes the mean square error}

between the actually transmitted symbols and the output of the receiver, i.e., ˆ Sdsf bcsm(m) = GM M SE(m)Rdsf bcsm(m) (4.3) and GM M SEe (m) = (Hdsf bcsm∗ (m)Hdsf bcsm(m) + (MtN0 Es)I2Mt )−1_H∗ dsf bcsm(m) (4.4)

whereH∗ is the Hermitian transpose of H. The total power transmitted onMtantennas at

one symbol time is _Es at each base station. We extend SINR calculations in the

narrow-band MIMO systems [32] to the SFBC combining with SM in OFDM system. Given the equivalent MIMO channel_Hdsf bcsm(m) the per-tone SINR calculation at the sthsub-carrier

is given by SINRsM M SE = Es MtN0 (H∗ dsf bcsm(m)Hdsf bcsm(m) + (MtN_E0_s)I2Mt)−1 s,s − 1 . (4.5)

After the SINR evaluation of MMSE detector, we can apply the EESM approximation method to get the effective SINR _ref f from ( 4.5) and then decide the MCS to the same

block error rate (BLER) by a lookup table through an AWGN curves as Table 3.1. We can find the throughputs with the decided MCS of the SFBC macro-diversity combining with SM scheme.

(39)

4.3 Simulation Results

The system level performance of the SFBC macro-diversity combining with SM is inves-tigated in this section. The parameter used in the simulation are listed in Table 4.1 [25]. When the mobile station moves from the serving base station (BS 0) to the adjacent base station (BS 1) as shown in Fig. 3.1, we calculate the throughputs under different spatial correlation at the transmitter of each base station and compare with the single base station using SM or SFBC. The separation of each base station is far enough to ignore the spatial correlation between the transmit antennas of BS 0 and BS 1.

This simulation is investigated to show the throughput gain by adopting SFBC macro-diversity combining with SM around cell edge compared with that of the conven-tional centralized MIMO-OFDM scheme and then reform the performance of MDHO. Per-fect channel state information at receivers is assumed during the simulations.

Table 4.1: Simulation Parameters

Parameter Value Remark Center frequency 2.5 GHz

Channel bandwidth 10 MHz FFT size 1024 Cyclic prefix ratio 1/8

Guard sub-carriers (_Ng) 159 Right : 80 , Left : 79

Used sub-carriers (_Nu) 865

Site-to-site distance 1.5 km Transmit power 37 dBm (5W)

Path loss 130.62+37.6log(d) d:km Thermal noise density -174 dBm

Channel model ITU Pedestrian B

From Fig. 4.3, with spatial correlation at the transmitter _ρtx = 0.1, it is observed

that mobile station can get huge throughput close to the serving base station using SM, but the throughput decreases seriously with the distance from the serving base station. The throughput performance of SM reaches 1 bit/s/Hz around cell edge which is similar

(40)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Distance from the serving BS (m)

Throughputs (bit/s/Hz) SM 2x2 corr=0.1 SFBC 2x2 corr=0.1 DSFBC−SM 4x2 corr=0.1 1 bit/s/Hz 1 bit/s/Hz

Figure 4.3: Throughputs of SM 2x2 , SFBC 2x2 and DSFBC-SM 4x2 with _ρtx = 0.1 at

each BS.

to SFBC. The poor throughput around the cell edge is improved obviously by applying distributed SFBC technique at the base station sides. Because of the equal receive signal strength from each base station, using transmit diversity at the base station sides increases SINR quite effectively around the cell edge, thereby allowing for choosing higher order MCS in Table II for higher spectrum efficiency. Therefore, the SFBC macro-diversity combining with SM scheme can overcome the drawbacks of the single base station using SM and maintain the large throughput for mobile station around cell edge. Even if the offered throughput satisfies the requirement of a mobile station, call-dropped frequency

(41)

can be reduced at the cell boundary and the ping-pong effect [12] can be relieved. Here the ping-pong effect means that the mobile station handovers back and forth several times between the adjacent base stations. Hence, the handover performance in OFDM cellular network is also enhanced greatly.

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

each BS.

The effects of increasing the spatial correlation of each transmitters at each base station are showed in Figs. 4.4, 4.5, 4.6 and 4.7. In a rich spatially-correlated channel, the throughput of the single base station using SM is worse as [11] [33] [34] and is even lower than that using SFBC around the cell edge. The SFBC macro-diversity combining with SM

(42)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

each BS.

scheme can also provide higher throughput than the conventional SFBC and SM schemes with_ρtx= 0.7. Forρtx = 0.9, the throughput of the SFBC macro-diversity combining with

SM scheme is the same as that of the conventional SFBC scheme. It is evident that the performance of SFBC macro-diversity combining with SM scheme is more robust than the conventional SM scheme as spatial correlation increases.

When the distance between a mobile station and the serving base station is two thirds of the cell radius, Fig. 4.8 shows that SFBC macro-diversity combining with SM scheme can provide higher throughput than the conventional SM and SFBC schemes for

(43)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

each BS.

ρtx ≤ 0.7. For ρtx = 0.9, the throughput of the conventional SFBC scheme is higher

than SFBC macro-diversity combining with SM scheme and the conventional SM scheme. At the two thirds of the cell radius, the performance of SFBC macro-diversity combining with SM scheme is not effective due to the longer distance from the adjacent base station. Because of path loss, the received signal strength from the adjacent base station is decreased seriously. When the distance between a mobile station and the serving base station is a cell radius. Fig. 4.9 shows that the throughput is quickly getting worse for the conventional SM and SFBC schemes. However, the performance of SFBC macro-diversity combining

(44)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughputs (bit/s/Hz)

SM 2x2 corr=0.9 SFBC 2x2 corr=0.9 DSFBC−SM 4x2 corr=0.9

0.5 bit/s/Hz

each BS.

with SM scheme at the cell boundary is better than that at the two thirds of the cell radius. Because the distance from the serving base station is almost equal to the distance from the adjacent base station, the mobile station receives the equal signal strength from the two base stations. At the cell boundary, SFBC macro-diversity combining with SM scheme can offer higher throughputs than the conventional SM and SFBC schemes under different spatial correlations. Obviously the improvement of distributed SFBC is great,

In Table 4.2, we show increased throughputs of SFBC macro-diversity combining with SM scheme compared with the conventional SM scheme when a mobile station is at

(45)

0 0.1 0.3 0.5 0.7 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Transmit antenna spatial correlation

SM 2x2 SFBC 2x2 DSFBC−SM 4x2

Figure 4.8: Throughputs of SM 2x2, SFBC 2x2 and DSFBC-SM 4x2 schemes for different ρtx with the distance from the serving BS equal to 500m (2Rc/3).

the two-thirds of the cell radius and a cell radius (Rc). It is shown that the improvements of throughput by SFBC macro-diversity combining with SM scheme is remarkable under spatially correlated channels. SFBC macro-diversity combining with SM scheme can offer higher data throughput for a mobile station at the cell boundary. Hence, the handover performance can be enhanced greatly.

(46)

0 0.1 0.3 0.5 0.7 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

SM 2x2 SFBC 2x2 DSFBC−SM 4x2

Figure 4.9: Throughputs of SM 2x2, SFBC 2x2 and DSFBC-SM 4x2 schemes for different ρtx with the distance from the serving BS equal to 750m (Rc).

Table 4.2: Throughputs Improvement by DSFBC-SM Compared with The Conventional SM.

Increased Throughputs

by DSFBC-SM (bit/s/Hz) ρtx= 0.1 ρtx= 0.3 ρtx= 0.5 ρtx= 0.7 ρtx= 0.9

Distance from the serving BS 500m (2Rc/3) 1 1 1 1 0 Distance from the serving BS 750m (Rc) 2 1 1 1 0.5

(47)

34

CHAPTER 5 Cyclic Delay Macro-Diversity Combining for

MIMO OFDM Cellular Mobile Networks

In this chapter, we replace space-frequency block code (SFBC) with cyclic delay diversity (CDD) to increase transmit diversity at the adjacent suitable base stations. CDD can be viewed as a space-time block code. In contrast to SFBC, it is not necessary to assume constant channel properties over several sub-carriers or symbol and the number of transmit antennas. CDD is an efficient way to achieve diversity in a flat fading channel. Applying CDD only requires changes at the transmitter and the receiver remains changeless. The complexity of implementing CDD is very minimal. CDD inserts virtual echoes on the channel response. That increases the frequency selectivity of the channel seen by the re-ceiver [17] [22]. Thus, it reduces the likelihood of deep fading. CDD can achieve desirable transmit diversity gain over uncorrelated channel. Because the distance between the adja-cent is far enough for CDD free from spatial correlation, applying CDD at the adjunct base station sides is a more efficient way than at the transmit antenna sides of each base station. Hence, we introduce the cyclic delay macro-diversity combining with spatial multiplexing scheme.

(48)

Figure 5.1: The Block Diagram of Cyclic Delay Macro-Diversity Combining with Spatial Multiplexing Scheme.

5.1 CDD Macro-Diversity Scheme

Fig. 5.1 shows the block diagram of two adjacent base stations apply CDD technique and each base station uses two antennas to perform spatial multiplexing. The signals are trans-mitted over the adjacent base stations, whereas it is only different with a specific cycle shift. After cyclic shifting, a cyclic prefix in add to avoid the ISI and maintain sub-carriers orthogonality for multi-path channels. δi denotes the cyclic shift of theith base station in

the time domain and it can be regard as a phase factor as θk = e−j

2π

NF F Tδik_. _(5.1)

This phase factor linearly increase with the sub-carrier index k for each base station.

5.2 SINR Performance

Both the serving base station (BS 0) and the target base station (BS 1) use two transmit antennas to perform spatial multiplexing. Because we apply CDD with a cyclic shift at the

(49)

adjacent base stations side, the cyclic delay could be chosen according to [35]. δ0 = 0 ,

δi = N

Nbs + δ

i−1 . (5.2)

Figure 5.2: DCDD-SM Simplified System Model.

As depicted in Fig. 5.2,N is the number of used sub-carriers in the OFDM system and _Nbs is the number of base stations in the scheme. We can get the cyclic shift δ1 =

N/Nbsfor BS 1, and denote those modulated symbols assm. The sequences ofsmencoded

by CDD can be separated into two groups for each base station as vectors_S0and_S1 : S0 = [s1 s2 s3 s4... s2N−3 s2N−2 s2N−1s2N] ;

S1 =s1e−j2πkδ1N s₂e−j2πkδ1N ... s_2N−1e−j2πkδ1N s_2Ne−j2πkδ1N

.

For the next SM branches,S0is split into two vectorsS₁0andS₂0. Also,S1 is split into two vectors_S₁1 and_S₂1 for the transmit antennas of each base station. Thus, the output signals of the transmitters becomes

S₁0 = [s1 s3 s5 s7... s2N−3 s2N−1] S₂0 = [s2 s4 s6 s8... s2N−2 s2N] S₁1 = s1e−j2πkδ1N _s₃_e−j2πkδ1N _s₅_e−j2πkδ1N _s₇_e−j2πkδ1N _{... s}_2N−3_e−j2πkδ1N _s_2N−1_e−j2πkδ1N S₂1 = s2e−j2πkδ1N s₄e−j2πkδ1N s₆e−j2πkδ1N s₈e−j2πkδ1N ... s_2N−2e−j2πkδ1N s_2Ne−j2πkδ1N .

(50)

where_{k = 1, 2, 3, ..., N} ⎡ ⎣ R1(k) R2(k) ⎤ ⎦ = ⎡ ⎣ H110 (k) + H111 (k)e−j2πkδ1N H120 (k) + H121 (k)e−j2πkδ1N H₂₁0 (k) + H₂₁1 (k)e−j2πkδ1N _H0 22(k) + H221 (k)e−j 2πkδ1 N ⎤ ⎦ × ⎡ ⎣ S(2k − 1) S(2k) ⎤ ⎦ + ⎡ ⎣ n1(k) n2(k) ⎤ ⎦ . (5.3) In short, we can represent ( 5.3) as

Rdcddsm(k) = Hdcddsm(k)Sdcddsm(k) + Ndcddsm(k) , (5.4) Rdcddsm(k) = ⎡ ⎣ R1(k) R2(k) ⎤ ⎦ , Hdcddsm(k) = ⎡ ⎣ H110 (k) + H111 (k)e−j 2πkδ1 N H0 12(k) + H121(k)e−j 2πkδ1 N H₂₁0 (k) + H₂₁1 (k)e−j2πkδ1N _H0 22(k) + H221(k)e−j 2πkδ1 N ⎤ ⎦ , and Sdcddsm(k) = ⎡ ⎣ S(2k − 1) S(2k) ⎤ ⎦ , Ndcddsm(k) = ⎡ ⎣ n1(k) n2(k) ⎤ ⎦ .

The same way to detect the signals as distributed SFBC combining with SM with a linear MMSE receiver, the receive signal vector Rdcddsm(k) is multiplied withGM M SE(k), the

MMSE detector minimizes the mean square error between actually transmitted symbols and the output of the receiver,

ˆ Sdcddsm(k) = GM M SE(k)Rdcddsm(k) (5.5) GM M SEe (k) = (Hdcddsm∗ (k)Hdcddsm(k) + (MtN0 Es)I2Mt )−1_H∗ dcddsm(k), (5.6)

where H∗ is the Hermitian transpose of H, the total power transmitted onMtantennas at

one symbol time is_Esat each base station. We extend SINR calculations in the narrowband

(51)

MIMO systems [32] to the CDD macro-diversity combining with SM in OFDM system. Given the equivalent MIMO channelHdcddsm(k) , the per-tone SINR calculation at the sth

sub-carrier is given by SINRM M SEs = Es MtN0 (H∗ dcddsm(k)Hdcddsm(k) + (MtN_E0_s)I2Mt)−1 s,s − 1. (5.7)

After the SINR evaluation of MMSE detector, we can apply the EESM approximation method to get the effective SINR ref f from ( 5.7) and then decide the MCS to the same

block error rate (BLER) by a lookup table through an AWGN curves as Table 3.1. We can find the throughputs with the decided MCS of the CDD macro-diversity combining with SM scheme.

5.3 Simulation Results

As SFBC macro-diversity combining with SM scheme in Chapter 4, the parameters used in the simulation are listed in Table 4.1. When the mobile station moves from the serving base station (BS 0) to the adjacent base station (BS 1) as shown in Fig. 3.1, we calculate the throughputs for different spatial correlations of the transmit antennas at each base sta-tion and compare with the single base stasta-tion using SM or SFBC. The separasta-tion of each base station is far enough to ignore the spatial correlation between the transmit antennas of BS 0 and BS 1. The main goal of this simulation is to show that the throughput gain can be achieved by adopting CDD macro-diversity combining with SM around the cell edge compared with that of the conventional centralized MIMO-OFDM scheme and SFBC macro-diversity combining with SM scheme. Perfect channel state information at receivers is assumed in our simulations. Particularly, we are interested in the performance compari-son between CDD macro-diversity combining with SM scheme and SFBC macro-diversity combining with SM scheme.

(52)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughtputs (bit/s/Hz) SM 2x2 corr=0.1 SFBC 2x2 corr=0.1 DSFBC−SM 4x2 corr=0.1 DCDD−SM 4x2 corr=0.1 1 bit/s/Hz 1 bit/s/Hz

Figure 5.3: Throughputs of SM 2x2 , SFBC 2x2 , DSFBC-SM 4x2 and DCDD-SM 4x2 withρtx= 0.1 at each BS.

From Fig. 5.3, the poor throughput of the conventional SM scheme around the cell edge is improved obviously by applying the distributed CDD techniques at the adjacent base stations sides. Because CDD can not achieve the optimal diversity gain as SFBC, distributed SFBC outperforms distributed CDD slightly. Using transmit diversity at the base station side increases SINR effectively around the cell edge. It allows us to choose higher order modulation and coding schemes (MCS). Therefore, the CDD macro-diversity combining with SM scheme can also overcome the drawbacks of the single base station using SM and maintain the large throughput for the mobile station around the cell edge. It

(53)

can be concluded that distributed CDD is another effective scheme to get spatial diversity gain in cellular networks as distributed SFBC.

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughtputs (bit/s/Hz) SM 2x2 corr=0.3 SFBC 2x2 corr=0.3 DSFBC−SM 4x2 corr=0.3 DCDD−SM 4x2 corr=0.3 1 bit/s/Hz 1 bit/s/Hz

As Figs. 5.4, 5.5, 5.6 and 5.7 show the effects of increasing spatial correlation of each transmitters for the CDD macro-diversity combining with SM scheme. It is shown that it can provide higher throughput than the conventional SFBC and SM schemes with ρtx = 0.7. The variations of throughputs between CDD macro-diversity combining with

SM scheme and SFBC macro-diversity combining with SM scheme are getting small at the cell boundary as spatial correlation increases. Forρtx= 0.9, the throughputs of the two

(54)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughtputs (bit/s/Hz) SM 2x2 corr=0.5 SFBC 2x2 corr=0.5 DSFBC−SM 4x2 corr=0.5 DCDD−SM 4x2 corr=0.5 1 bit/s/Hz

schemes are the same at cell boundary.

When the distance between a mobile station and the serving base station is two thirds of the cell radius, Fig. 5.8 shows that the performance of CDD macro-diversity combining with SM scheme is the same as SFBC macro-diversity combining with SM scheme and better than the conventional SFBC and SM schemes with ρtx ≤ 0.3. SFBC

macro-diversity combining with SM scheme also outperforms the CDD macro-diversity combining with SM scheme for higher _ρtx values. At the cell boundary, Fig. 5.9 shows

that the CDD macro-diversity combining with SM scheme performs similar to the SFBC

(55)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughtputs (bit/s/Hz) SM 2x2 corr=0.7 SFBC 2x2 corr=0.7 DSFBC−SM 4x2 corr=0.7 DCDD−SM 4x2 corr=0.7 1 bit/s/Hz

macro-diversity combining with SM scheme. It can also provide higher throughput than the conventional SFBC and SM schemes.

Table 5.1 shows the increased throughputs of CDD macro-diversity combining with SM scheme compared with the conventional SM scheme. It shows that CDD macro-diversity combining with SM scheme can also effectively improve the throughput under spatially correlated channels and can offer higher data throughput for a mobile station at the cell boundary. Hence, the handover performance can be enhanced greatly.

(56)

0 250 500 750 1000 1250 1500 0 1 2 3 4 5 6 7 8 9

Throughtputs (bit/s/Hz) SM 2x2 corr=0.9 SFBC 2x2 corr=0.9 DSFBC−SM 4x2 corr=0.9 DCDD−SM 4x2 corr=0.9 0.5 bit/s/Hz

Table 5.1: Throughputs Improvement by DCDD-SM Compared with The Conventional SM.

Increased Throughputs

by DCDD-SM (bit/s/Hz) _ρtx= 0.1 ρtx= 0.3 ρtx= 0.5 ρtx= 0.7 ρtx= 0.9

Distance from the serving BS 500m (2Rc/3) 1 1 0 0 0 Distance from the serving BS 750m (Rc) 1 1 1 1 0.5

(57)

0 0.1 0.3 0.5 0.7 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

SM 2x2 SFBC 2x2 DSFBC−SM 4x2 DCDD−SM 4x2

Figure 5.8: Throughputs of SM 2x2, SFBC 2x2, DSFBC-SM 4x2 and DCDD-SM 4x2 schemes for differentρtxwith the distance from the serving BS equal to 500m (2Rc/3).

(58)

0 0.1 0.3 0.5 0.7 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

SM 2x2 SFBC 2x2 DSFBC−SM 4x2 DCDD−SM 4x2

Figure 5.9: Throughputs of SM 2x2, SFBC 2x2, DSFBC-SM 4x2 and DCDD-SM 4x2 schemes for differentρtxwith the distance from the serving BS equal to 750m (Rc).