TV Black-space Spectrum Access for Wireless Local Area and Cellular Networks

Transcription

1 A AALTO UNIVERSITY SCHOOL OF ELECTRICAL ENGINEERING Department of communications and Networking (Comnet) Yihenew Dagne Beyene TV Black-space Spectrum Access for Wireless Local Area and Cellular Networks A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Technology. Helsinki, August 10, 2012 Supervisor: Instructor: Riku Jäntti, Professor, D.Sc Kalle Ruttik, D.Sc

2 AALTO UNIVERSITY SCHOOL OF ELECTRICAL ENGINEERING Author: Yihenew Dagne Beyene ABSTRACT OF THE MASTER S THESIS Name of the Thesis: TV Black-space Spectrum Access for Wireless Local Area and Cellular Networks Date: August 10, 2012 Number of pages: 72 Department: Professorship: S-72 Supervisor: Instructor: Department of Communications and Networking Riku Jäntti, Professor, D.Sc Kalle Ruttik, D.Sc This thesis presents a black-space spectrum access scheme for overlay cognitive radio in TV band. Physical layer implementations of secondary systems using software defined radio technology have been proposed. We consider two types of secondary transmitters with WLAN-type and LTE-type frame structures in order to study the impact of secondary transmission over primary pilot carriers on performances of channel estimation and interference cancellation algorithms. Bit error rate has been used as performance metric, and performances of the two secondary systems have been presented in different scenarios. The effect of secondary transmitters interference power level on performances of primary and secondary receivers has been investigated. Keywords: Cognitive radio, overlay, Interference channel, GNU Radio, DVB, WLAN, LTE ii

3 Acknowledgements This thesis is part of End-to-End Cognitive Radio Testbed (EECRT) project at Department of Communications and Networking, Aalto University. I would like to thank my supervisor, professor Riku Jäntti, for giving me the chance to work in this project. Moreover, I owe my gratitude to my instructor, Dr. Kalle Ruttik, for his invaluable advices and comments throughout the course of my work. I also acknowledge Viktor Nässi who gave me technical supports during my stay at communications laboratory. I greatly appreciate Juha Vierinen and Lassi Roininen, researchers at Sodankylä Geophysical Observatory, for recommending me to study in Finland which is known for its leading-edge research in telecommunications engineering. Finally, thanks to all who deserve gratitude and blessings in more than words. Otaniemi, August 10, 2012 Yihenew Dagne Beyene iii

4 Contents List of Abbreviations List of Notations List of Figures List of Tables vii x xv xvi 1 Introduction Cognitive radio Scope of the thesis Thesis organization OFDMA Systems OFDM principles OFDM in wireless channels DVB system IEEE OFDM physical layer LTE physical layer (Release 8) Cognitive Radios - Interference Channels Secondary spectrum access Interweave system Underlay system Overlay system iv

5 3.2 Interference channel Orthogonal channel access General Gaussian interference channel Overlay Cognitive Radio Overlay cognitive radio Gaussian channel Achievable capacity Practical considerations Overlay transmitter Overlay receiver Implementation Implementation platform GNU Radio and USRP WLAN-type overlay transmitter Packet encoder and scrambler OFDM mapper Insert preamble IFFT Cyclic prefix adder Gain USRP WLAN-type overlay receiver USRP Channel filter Synchronizer Sampler FFT DVB equalizer DVB demodulator Frame acquisition WLAN demodulator v

6 Message decoder and descrambler LTE-type overlay transmitter Scrambler Modulation mapper Layer mapper Precoder Resource element mapper Insert reference signals IFFT USRP LTE overlay receiver FD-sync Equalizer LTE demodulator Message decoder Results and Discussion Hardware characteristics Measurement setup Results Conclusion 67 vi

7 List of Abbreviations AGC AWGN BER BW CRC DL DSSS DVB DVB-T DVB-H FCC FDD FDMA ICI IDFT IFFT IFO Automatic Gain Control Additive White Gaussian Noise Bit-Error-Rate BandWidth Cyclic Redundancy Check DownLink Direct Sequence Spread Spectrum Digital Video Broadcast DVB-Terrestrial DVB-Handheld Federal Communications Commission Frequency Division Duplexing Frequency Division Multiple Access Inter-Carrier Interference Inverse Discrete Fourier Transform Inverse Fast Fourier Transform Integer Frequency Offset vii

8 IR ISI ISM LNA LS LTE MAC MIMO MPDU MMSE MSE OFDM OFDMA PLCP PMD PLME PPDU PRBS PSDU RMS SDR SINR InfraRed Inter-Symbol Interference Industrial, Scientific and Medical Low Noise Amplifier Least Squares Long Term Evolution Medium Access Control Multiple-Input-Multiple-Output MAC Protocol Data Unit Minimum Mean-Square Error Mean-Square Error Orthogonal Frequency Division Multiplexing Orthogonal Frequency Division Multiple Access Physical Layer Convergence Procedure Physical Medium Dependent Physical Layer Management Entity PLCP Protocol Data Unit Pseudo-Random Binary Sequence PLCP Service Data Unit Root-Mean-Square Software Defined Radio Signal-to-Interference-Noise Ratio viii

9 SIR SNR TDD TDMA TPS TU UL UHD USB USRP WLAN ZF Signal-to-Interference Ratio Signal-to-Noise Ratio Time Division Duplexing Time Division Multiple Access Transmission Parameter Signalling Typical Urban UpLink Universal Hardware Driver Universal Serial Bus Universal Software Radio Peripheral Wireless Local Area Network Zero-Forcing ix

10 List of Notations a k a n modulating complex symbol Amplitude response of n th multipath component a ij Magnitude-square of c ij c ij C C causal Standard form transformation of h ij Channel capacity Channel capacity for causally known interference g, G Amplifier gain h ij h(t) h(t, τ) Im{.} k f M P P τn p m,i,k Channel gain between i th transmitter and j th receiver Instantaneous time-domain channel response Multipath channel response at time t for impulse transmitted at time t τ Imaginary part Frequency-dependent constant Transmitted message Signal power Normalized power of n th multipath component Pilot signal on k th subcarrier of I th symbol in m th DVB frame x

11 Q R Re{.} S T U T S w k x(t) X(f) x T (t) X T (f) X X X c X p X s Y Y Z Average power of known Gaussian interference signal Transmission rate over a channel Real part Known Gaussian interference signal Useful time duration of OFDM symbol Total OFDM symbol duration k th value in PRBS sequence Continuous-time signal Fourier-transform of x(t) Time-domain rectangular signal of duration T Fourier transform of x T (t) Channel input signal Channel input message alphabet set Cognitive transmitter s own signal Primary transmitted signal Secondary transmitted signal Channel output signal Channel output message alphabet sets Gaussian noise xi

12 γ τ(t) f τ τ 2 τ n τ rms φ n Signal-to-Interference Ratio OFDM guard interval Excess delay Inter-carrier spacing Weighted first moment of channel delay Weighted second moment of channel delay Time delay caused by n th multipath component RMS delay spread Phase response of n th multipath component xii

13 List of Figures 2.1 Time-domain representation of a single subcarrier. T is the symbol duration Frequency spectrum of a single subcarrier (sinc function). T is the symbol duration in time domain Orthogonal multiplexing of subcarriers. At the peak of each subcarrier, interference from other sub-carriers is zero General block diagram of OFDM signal generation using IDFT ISI-free sampling of OFDM signal in multipath channel Illustration of multipath propagation Normalized channel power delay profile and magnitude response for TU Functional block diagram of DVB transmitter [1] PPDU frame structure of OFDM physical layer [2] Block diagram of uplink physical channel processing Block diagram of downlink physical channel processing LTE frame structure (type-1) Time-frequency resource grid structure of LTE for normal cyclic prefix configuration. In multi-antenna transmission, each antenna port has one resource grid Interweave cognitive radio operation - secondary spectrum access on white spaces xiii

14 3.2 Underlay cognitive radio - secondary signal is spread over wide band with low power constrained by interference level tolerated by other non-cognitive users Overlay cognitive radio - secondary signal is added on top of the primary signal at relatively low power such that SNR requirement of the primary users is satisfied Interference channel with two inputs and two outputs Sequential decoding schemes of Gaussian interference channel Bounds for achievable rate regions for general Gaussian interference channel: (a) P 1 = P 2 = 6, a 12 = a 21 = 0.1, (b) P 1 = P 2 = 6, a 12 = a 21 = 0.55, (c) P 1 = P 2 = 6, a 12 = a 21 = 5 (d) P 1 = P 2 = 6, a 12 = a 21 = Concurrent operation of secondary system (cognitive) within the coverage area of primary system A two-input-two-output Gaussian interference channel with the second transmitter having non-causal messages of the primary Dirty-paper coding with additive Gaussian noise and Gaussian distributed interference, S, known by the transmitter. Power of the transmitted signal, X = U αs, is constrained by (1/n) n i=1 X2 i P Illustration of DVB signal pilot structure. Unlike scattered pilots, continual pilots have fixed locations Proposed WLAN-type frame structure for orthogonal pilot overlay cognitive radio LTE-type frame structure with synchronization symbols and reference pilots xiv

15 4.7 Overlay transmitter design where W p is the primary message encoded to X p, and W s is the secondary message encoded to X c. The scaling factors, g and G, are calculated based on the power constraints, P p and P s, and relaying fraction, α, obtained in Equation (4.6) Sequential decoding of secondary message. The equalizer estimates the channel using pilots of the primary signal Time-domain interpolation using (a) zero-order and (b) firstorder polynomial filters [3] Software define radio Interaction among USRP hardware, GNU Radio and upper layers WLAN-type OFDM transmitter implementation using GNU Radio platform GNU Radio implementation of WLAN-type overlay receiver GNU Radio implementation of LTE DL overlay transmitter GNU Radio implementation of LTE-type overlay receiver Measured BER for primary and secondary signals, γ p = 7dB Measured BER for primary and secondary signals, γ p = 12dB Measured BER for primary and secondary signals, γ p = 17dB. 66 xv

16 List of Tables 2.1 Constellation points for uniform and non-uniform modulations in DVB [1] Comparison of OFDM parameters of DVB-T, WLAN and LTE physical layers. The parameter l is symbol number in an LTE time slot RF characteristics of hardware used during measurement xvi

17 Chapter 1 Introduction The radio spectrum is, nowadays, a scarce resource. On the other hand, not all licensed frequency bands are fully utilized. The spectrum has been divided into segments by the regulatory bodies such as FCC (Federal Communications Commission) of United States. These spectral segments are assigned for different kinds of wireless services such as radio and TV broadcasts, cellular networks, satellite communication, and others. Each frequency band segment is then divided into many channels which are granted to operators and other commercial and governmental institutions. This fixed allocation of resources ensures that there will be no interference between licensed users. However, it causes inefficient usage of the resources as some frequency bands may be highly occupied while there is little or no transmission on other bands. Underutilization of the radio resource can be in spatial, spectral and/or temporal domain. The fact that demand for wireless traffic is increasing exponentially while there is limited radio resource suitable for wireless communication has been a driving force for researchers to come up with efficient modulation and coding schemes, protocols and multiple antenna diversity and multiplexing techniques. Optimization of radio access technologies was the main focus in this research. On the other hand, some radio spectra such as TV channels have been found to be inefficiently utilized [4]. Therefore, secondary networks can be deployed in such bands to access unused radio resources whenever available. Such oppor- 1

18 CHAPTER 1. INTRODUCTION 2 tunistic spectrum access requires reconfigurable radio system architecture. 1.1 Cognitive radio Cognitive radio is a candidate technology that enables coexistence of multiple networks operating in the same frequency band. The concept was originally introduced by J. Mitola [5], and it has been defined in various ways since then [6], [7], [8]. In general, it is a radio system having some form of intelligence. The intelligence can be ability to sense signal spectrum and use this information to decide how and when to transmit over the given band. Unlike ordinary radios, cognitive radio has the capability of adapting to changes in its environment by changing its parameters such as modulation and coding schemes, frequency and power. Cognitive radio can be employed to increase efficiency of utilization of the radio spectrum by allowing flexible sharing of the resource among multiple users without interfering each other. Realizing such system demands sophisticated signal processing by the cognitive transmitters and receivers in a way that maximal usage of the resource is achieved while making sure that other users will not suffer from interference caused by the cognitive transmitter. Reconfigurable signal processing needs of cognitive radios can be enabled via software defined radio (SDR) [9], [10] technology using high-speed programmable digital hardware and general purpose computers available today. In chapter 5 we will discuss more on practical implementation of overlay cognitive radio on an open-source software defined radio platform [11]. 1.2 Scope of the thesis Research in cognitive radio systems has received a lot of attention due to spectrum scarcity, and many articles on the topic have been published in recent days [8], [7], [12], [13]. Many of them are, however, based on a set of assumptions such as perfect knowledge of channel coefficients which might not exist in practical wireless communication systems. The thesis focuses

19 CHAPTER 1. INTRODUCTION 3 on a particular class of cognitive radio, specifically overlay cognitive radio. We will look into black-space spectrum 1 access scheme in DVB (Digital Video Broadcast) channels. We study how the quality of pilot symbols impacts secondary system s performance. Practical implementation of overlay system using LTE (Long Term Evolution) type and WLAN (Wireless Local Area Network) type frame structures using a software defined radio platform, and performance measurements results are presented. 1.3 Thesis organization The thesis starts with defining fundamental concepts and mathematical models. Later practical implementation of overlay secondary sharing algorithm and discussion of measurement results are presented. Chapter 2 gives overview of OFDM (Orthogonal Frequency Division Multiplexing) modulation scheme which is used in latest radio access technologies. The chapter also briefly points out physical layer structures of targeted OFDMA systems, DVB, LTE and WLAN, in practical implementation. Chapter 3 introduces secondary spectrum access methods for cognitive radios. Chapter 4 discusses overlay cognitive radio channel and coexistence requirements for the cognitive radio. Later cognitive transmitter and receiver designs are presented. Chapter 5 focuses on practical realization of overlay cognitive radio on an open-source software defined radio platform. Measurement results of the system performance have been discussed in Chapter 6. Finally, Chapter 7 concludes the main observations on practical performance results and limitations of the system. 1 It refers to an already occupied channel as seen by another secondary transmitter [7].

20 Chapter 2 OFDMA Systems Orthogonal Frequency Division Multiplexing (OFDM) is a multi-carrier modulation scheme where a set of narrow-band orthogonal subcarriers are used to carry information. Due to its high spectral efficiency it has been adopted in latest cellular and local area networks. The transmitted signal is cyclically extended in time such that multipath components in fast fading environment can be combined without causing ISI (Inter-Symbol Interference). The overlay cognitive radio implementation targets two OFDMA based cellular and WLAN systems as secondary transmitters in DVB (Digital Video Broadcast) channels. In this chapter, basics of OFDM and its advantage in fading environments will be discussed. Later, the physical layer structures of DVB, LTE and WLAN systems will be briefly presented. 2.1 OFDM principles In OFDM, the data stream is converted into low-rate parallel streams each modulating a subcarrier. In frequency domain, the subcarriers are sinc waveforms which are rectangular pulses in time-domain. Figure 2.1 shows time domain representation of one subcarrier. 4

21 CHAPTER 2. OFDMA SYSTEMS 5 1/T x 0 (t) T/2 0 T/2 Figure 2.1: Time-domain representation of a single subcarrier. T is the symbol duration. x T (t) = 1 ( t T rect T = 1, T ) T 2 t < T 2 0, elsewhere (2.1) (2.2) The frequency domain representation of the subcarrier is shown in Figure 2.2. X 0 (f) 1/T 0 1/T f Figure 2.2: Frequency spectrum of a single subcarrier (sinc function). T is the symbol duration in time domain. X T (f) = sinc(ft ) (2.3) Such sub-carriers are put aside to each other orthogonally while spectral overlap can be allowed. The orthogonality is maintained by centering each subcarrier at frequencies where other subcarriers have zero spectral component. In other words, we put them at intervals of 1/T Hz. In OFDM, each subcarrier carries one modulating symbol (complex number)

22 CHAPTER 2. OFDMA SYSTEMS 6 so that one sample from each subcarrier is sufficient to recover the transmitted information. Orthogonality of carriers makes sure that there will be no interference from neighboring subcarriers at the instant of sampling as shown in Figure 2.3. The subcarrier spacing, f, is inversely proportional to the symbol period, T. X(f) f = 1/T 0 f Figure 2.3: Orthogonal multiplexing of subcarriers. At the peak of each subcarrier, interference from other sub-carriers is zero. X(f) = M k= M ( X T f k ) T (2.4) In general, each subcarrier, X k, is modulated by a complex symbol, a k, so that the baseband signal spectrum can be written as X(f) = M k= M ( a k X T f k ) T (2.5) The corresponding time-domain signal will be x(t) = 1 T M ( ) a k e j2πk t t T rect T k= M (2.6) The signal is bounded by finite time duration of T seconds. For time interval

23 CHAPTER 2. OFDMA SYSTEMS 7 t [ T/2, T/2], x(t) can be expressed as x(t) = 1 T M k= M a k e j2πk t T (2.7) The orthogonal subcarriers are uniquely defined by the complex symbols, a k, k { M, M + 1,, M}. This implies that finite time-domain samples are sufficient for describing the signal. The minimum number of samples is equal to the number of subcarriers (or number of complex symbols) which is equal to 2M + 1. If we divide the symbol duration, T into 2M + 1 equally spaced bins ( T/(2M + 1)) we get discrete samples as sampling rate of (2M + 1)/T samples/second. This conforms to the Nyquist criterion. The discrete-time version of the signal is x[n] = x(nt/(2m + 1)) (2.8) = 1 M a k e j2πnk/(2m+1) (2.9) T k= M where x[n] is the n th n {0, 1,, 2M}. discrete sample taken at t = nt/(2m + 1) seconds, The signal x[n] is equivalent to Inverse Discrete Fourier Transform (IDFT) of the complex symbols, a k. Therefore, OFDM can easily implemented as IDFT operation on the modulating complex symbols. Figure 2.4 shows IDFT implementation of OFDM signal. OFDM signal bits Modulator Serial to symbols parallel symbols IDFT samples Converter Parallel to serial converter samples Data Frequency domain Time Domain Figure 2.4: General block diagram of OFDM signal generation using IDFT.

24 CHAPTER 2. OFDMA SYSTEMS 8 Cyclic prefix It can be seen that the time-domain OFDM signal, x[n], is periodic. There are 2M + 1 samples in one period. x[n (2M + 1)] = 1 T = 1 T = 1 T M a k e j2πk( n 1) 2M+1 (2.10) k= M M k= M M k= M e j2πk a k e j2πnk/(2m+1) (2.11) a k e j2πnk/(2m+1) (2.12) x[n (2M + 1)] = x[n] (2.13) Therefore, we can extend symbol duration to the left and right of the minimum duration, [ M, M +1,, M]. Let us add one sample at n = M 1. From the property of periodicity, we have x[ M 1] = x[ M 1 + (2M + 1)] = x[m] (2.14) In general, x[ M l] = x[m + 1 l], l = 1, 2,, L (2.15) The symbol duration is extended to the left by L samples taken from the last L samples of the interval { M, M + 1,, M}. The cyclically added group of L extra samples is called cyclic prefix. From the extended symbol duration, { M L, M L + 1,, M}, any 2M + 1 consecutive samples can be used to describe the signal. Different time shifts, however, will cause phase offset in frequency domain. The advantage of added cyclic prefix is to combine different multipath components of the received signal without causing ISI as shown in Figure 2.5. However, this is achieved at the cost of extra guard period which is not used to transmit information.

25 CHAPTER 2. OFDMA SYSTEMS 9 Figure 2.5: ISI-free sampling of OFDM signal in multipath channel. 2.2 OFDM in wireless channels In wireless communication, electromagnetic waves propagate from the transmitter antenna to receiver antenna through air. Due to spreading of signal energy to wider area as it travels far away from the transmitter, the signal strength is attenuated exponentially. A simple free-space channel model is expressed as ( 2 1 h(t) = k f (2.16) d) where h(t) is the instantaneous (at time t) channel gain at a distance d, and k f is a frequency-dependent constant. In practical wireless communications, the communication channel is far from free space. Signal propagation is also affected by antenna type, antenna height and obstacles [14]. Moreover, the received signal is superposition of various signals arriving at different time instants from different directions. The signal from each path is a duplicate of the transmitted signal being distorted in amplitude, frequency and phase. As shown in Figure 2.6 the transmitted signal is reflected from the surrounding objects before reaching at the receiver. The reflection and diffraction effects cause phase shift on the signal. The phase shift is time-variant if there is mobility in the transmitter, reflectors, and/or receiver. A finite multipath component approximation of the channel is given by [3] h(τ, t) = N a n (t)e jφn(t) δ(τ τ n (t)) (2.17) n=1

26 CHAPTER 2. OFDMA SYSTEMS 10 Transmitter Receiver Figure 2.6: Illustration of multipath propagation. where h(t, τ) is the channel response at time t for the impulse transmitted at time t τ, and a n, φ n, τ n are time-varying amplitude response, phase shift and time delay caused by the n th path. The excess delay τ(t) is defined as time difference between the first and the last arriving multipath components. Root-mean-square (RMS) delay spread RMS delay spread is defined as weighted RMS delay given by Equation It is expressed as τ rms = τ 2 ( τ) 2 (2.18) where τ and τ 2 are weighted first and second moments of delays. The weights are normalized powers of multipath components [3]. Normalized power of n th multipath component given by p τn = E { a n (t) 2 } N n=1 E { a n(t) 2 } (2.19) The RMS delay spread, hence, takes the form ( ) τ rms = N 2 N τ n τ n p τn p τn (2.20) n=1 n=1

27 CHAPTER 2. OFDMA SYSTEMS 11 where N p τn = 1 (2.21) n=1 RMS delay spread is a measure of frequency selectivity of the channel. A value larger than the symbol duration, causes Inter Symbol Interference (ISI). Equivalently, the channel s frequency response is not flat over the signal bandwidth. For communication systems with symbol duration smaller than the RMS delay spread, the channel is said to be frequency-selective; otherwise, it is called frequency-flat channel. In practical systems, the symbol duration is made sufficiently large. Figure 2.7 shows channel impulse response and frequency response for a typical urban (TU) channel model [15]. The RMS delay spread in this model is 0.5µs. In LTE downlink, the typical symbol duration 0.35 Power delay profile 1 Magnitude response Normalized average power p τ Time delay, τ (µ s) H(f) Frequency, f (MHz) Figure 2.7: Normalized channel power delay profile and magnitude response for TU. is 66.7µs for type-1 frame structure [16]. The shortest cyclic prefix duration is 4.7µs which is much larger than the channel delay spread. Therefore, multipath components can be effectively combined without causing ISI. 2.3 DVB system The physical layer standard for digital terrestrial broadcast system defines OFDM based frame structure of the baseline transmitter [1]. The OFDM system is specified for 6 MHz, 7 MHz and 8 MHz channels. While the same specification applies to these options in frame structure, the sample rate varies

28 CHAPTER 2. OFDMA SYSTEMS 12 accordingly. Hierarcheal modulation and coding are used to multiplex two MPEG transport streams using uniform and multi-resolution constellation. Figure 2.8 shows functional block diagram of the transmitter. The receiver uses the same set of decoders and de-interleavers for each stream. Figure 2.8: Functional block diagram of DVB transmitter [1]. MPEG-2 transport multiplex (MUX) packet is 188 bytes. Inner interleaver consists of bit-wise interleaving followed by symbol interleaving; both of which are block based. All data carriers in one OFDM frame are modulated using either QPSK, 16-QAM, 64-QAM, non-uniform 16-QAM or non-uniform 64-QAM. Non-uniform constellation apply only for hierarchical transmission. DVB frame structure One OFDM frame comprises 68 OFDM symbols each with duration T S = T U + where T U is the useful part, and is the guard interval. Four guard interval values are defined: /T U = 1/4, 1/8, 1/16, 1/32. Two transmission modes, 2K mode and 8K mode, are available for DVB-T (DVB- Terrestrial) and DVB-H (DVB-Handheld) systems where an OFDM symbols contains 6817 carriers in the 8K mode and 1705 carriers in 2K mode. Each 2 carrier is assigned to either data symbols, scattered pilots, continual pilots or TPS (Transmission Parameter Signalling) pilots. One MPEG-2 transport multiplex packet containing 188 bytes is mapped to each OFDM symbol. All data carriers in one OFDM frame are modulated using either QPSK, 16-QAM, 64-QAM, non-uniform 16-QAM or non-uniform

29 CHAPTER 2. OFDMA SYSTEMS 13 Modulation Normalized constellations: n + jm QPSK n, m { 1 2, 1 2 } 16-QAM (non-hierarchical and hierarchical with α = 1) n, m { 3 10, 1 10, 1 10, 3 10 } Non-uniform 16-QAM with α = 2 n, m { 4 20, 2 20, 2 20, 4 20 } Non-uniform 16-QAM with α = 4 n, m { 6 52, 4 52, 4 52, 6 52 } 64-QAM n, m { 7 42, 5 42, 3 42, (non-hierarchical and hierachical with α = 1) 1 42, 1 42, 3 42, 5 42, 7 42 } Non-uniform 64-QAM with α = 2 n, m { 8 60, 6 60, 4 60, 2 60, , 6 60, 8 60, 60 } Non-uniform 64-QAM with α = 4 n, m { , 8 108, 6 108, 4 108, 4 108, 6 108, 8 108, } Table 2.1: Constellation points for uniform and non-uniform modulations in DVB [1]. 64-QAM. Non-uniform constellation apply only for hierarchical transmission. Table 2.1 shows constellation points for different modulation schemes in DVB. The parameter α is the ratio of the minimum distance separating two constellation points having different high priority bit stream values and the minimum distance between any two constellation points [1]. Subcarriers (cells) containing reference signals (scattered pilots and continual pilots) are transmitted at boosted power levels. The pilot information is derived from a Pseudo-Random Binary Sequence (PRBS). The locations of scattered pilots changes from one symbol to another. On the other hand, continual pilots have fixed locations, and there are 45 continual pilots in 2K mode and 177 in 8K mode. The modulation value of continual and scattered pilots is a complex number, p m,i,k, given by Re{p m,i,k } = 4/3 2(1/2 w k ) (2.22) Im{p m,i,k } = 0 (2.23)

30 CHAPTER 2. OFDMA SYSTEMS 14 where m is the frame index, k is the subcarrier index, I is the symbol index and w k is a PRBS sequence generated by a polynomial x 11 + x TPS carriers are used for signaling transmission information: modulation, hierarchy, guard interval, code rate, transmission mode, frame number and cell identification. The same TPS information is transmitted on 17 carriers (2K mode) and 68 carriers (8K mode) in every OFDM symbol. 2.4 IEEE OFDM physical layer IEEE is a standard for wireless local area networks (WLANs) that defines physical layer and MAC layer functions. It is the most widely deployed wireless access technology worldwide. Different types of physical layers are defined to work with the same MAC layer protocol. The physical layers operate in the infrared (IR) and ISM (Industrial, Scientific and Medical) radio bands. The first standard that was published in 1997 defines three physical layer types [17]. Later, a revised version [2] was released with additional amendments that were published to date. IEEE a, IEEE b and IEEE g are the most common amendments included in this revision. Except for the IEEE b, which uses DSSS (Direct Sequence Spread Spectrum) modulation, the remaining two are OFDM (Orthogonal Frequency Division Multiplexing) systems. In addition to the revised physical and MAC (Medium Access Control) layer specification, enhancements have been published since then. IEEE n is the newest OFDM system with MIMO (Multiple-Input-Multiple- Output) capabilities. The OFDM physical layer is divided into three sublayers [2]. PLCP (physical layer convergence procedure) sublayer is the upper layer providing services to the MAC layer, and PMD (physical medium dependent) is responsible for transmission and reception of PPDU (PLCP protocol data unit) over the physical medium. The third sublayer is PLME (physical layer management entity) which manages the physical layer functions.

31 IEEE PART 11: WIRELESS LAN MAC AND PHY SPECIFICATIONS Std PLCP frame format Figure 17-1 shows the format for the PPDU including the OFDM PLCP preamble, OFDM PLCP header, PSDU, tail bits, and pad bits. The PLCP header contains the following fields: LENGTH, RATE, a reserved bit, an CHAPTER even parity bit, 2. and OFDMA the SERVICE SYSTEMS field. In terms of modulation, the LENGTH, RATE, reserved 15bit, and parity bit (with 6 zero tail bits appended) constitute a separate single OFDM symbol, denoted SIGNAL, which is transmitted with the most robust combination of BPSK modulation and a coding rate of R = 1/2. The SERVICE Frame structure field of the PLCP header and the PSDU (with 6 zero tail bits and pad bits appended), denoted as DATA, are transmitted at the data rate described in the RATE field and may constitute multiple OFDM symbols. The The PLCP tail bits sublayer in the communicates SIGNAL symbol enable with the decoding MACof layer the RATE via and PSDU LENGTH (PLCP fields immediately after the reception of the tail bits. The RATE and LENGTH fields are required for decoding the DATA service part of data the packet. unit), In addition, also called the CCA MPDU mechanism (MAC can be protocol augmented data by predicting unit) from the duration MACof the packet layerfrom side. the contents Duringof transmission, the RATE and LENGTH PPDUfields, is formed even if the from data rate PSDU is not supported and PLCP by the STA. Each of these fields is described in detail in , , and preamble. Figure 2.9 shows frame format for OFDM physical layer. PLCP Header RATE 4 bits Reserved 1 bit LENGTH 12 bits Parity Tail SERVICE PSDU Tail 1 bit 6 bits 16 bits 6 bits Pad Bits Coded/OFDM Coded/OFDM (BPSK, r = 1/2) (RATE is indicated in SIGNAL) PLCP Preamble 12 Symbols SIGNAL One OFDM Symbol DATA Variable Number of OFDM Symbols Figure 17-1 PPDU frame format Figure 2.9: PPDU frame structure of OFDM physical layer [2] Overview of the PPDU encoding process PLCP preamble The encoding process is composed of many detailed steps, which are described fully in later subclauses, as noted below. The following overview intends to facilitate understanding the details of the convergence procedure: It comprises 12 OFDM symbols containing training sequences used for synchronization and channel estimation. The first 10 symbols are repetitions of a) Produce the PLCP Preamble field, composed of 10 repetitions of a short training sequence (used a short for AGC sequence convergence, for AGC diversity (automatic selection, timing gainacquisition, control) and convergence, coarse frequency diversity acquisition se-ilection, frequency time acquisition synchronization in the receiver), andpreceded coarseby frequency a guard interval acquisition. (GI). Refer to After afor guard the receiver) and two repetitions of a long training sequence (used for channel estimation and fine details. interval, two long sequences are transmitted to aid the receiver estimate the b) Produce the PLCP header field from the RATE, LENGTH, and SERVICE fields of the channel TXVECTOR and fine by filling frequency the appropriate offset [2]. bit fields. The RATE and LENGTH fields of the PLCP header are encoded by a convolutional code at a rate of R = 1/2, and are subsequently mapped onto a single BPSK encoded OFDM symbol, denoted as the SIGNAL symbol. In order to facilitate a PLCP reliable header and timely detection of the RATE and LENGTH fields, 6 zero tail bits are inserted into the PLCP header. The encoding of the SIGNAL field into an OFDM symbol follows the same steps for convolutional encoding, interleaving, BPSK modulation, pilot insertion, Fourier transform, and The header contains control information appended to PSDU in order to help prepending a GI as described subsequently for data transmission with BPSK-OFDM modulated at coding rate 1/2. The contents of the SIGNAL field are not scrambled. Refer to for details. receiver extract subcarrier modulation type, code rate, payload length and scrambling information. c) Calculate from RATE field of the TXVECTOR the number of data bits per OFDM symbol (N DBPS ), the coding rate (R), the number of bits in each OFDM subcarrier (N BPSC ), and the number of coded bits per OFDM symbol (N CBPS ). Refer to for details. Data PSDU from MAC layer is appended to SERVICE field of PLCP header, and Copyright 2007 IEEE. All rights reserved. 595 then extended with tail bits. The data is then scrambled and encoded with

32 CHAPTER 2. OFDMA SYSTEMS 16 Rate 1/2 convolutional encoder. Encoded bits are divided into groups each being mapped to complex symbols. Finally, OFDM symbols are formed taking 48 symbols at a time, applying IFFT (Inverse Fast Fourier Transform) and extending them with cyclic prefix. The number of OFDM symbols varies depending on the PSDU length. 2.5 LTE physical layer (Release 8) LTE uses OFDM modulation at the physical layer. This brings many advantages over modulation techniques used by earlier versions of 3GPP standards such as HSPA/WCDMA and GSM. Since the system bandwidth is divided into independently modulated narrow-band subcarriers, channel fading can be easily mitigated. Over the subcarrier band, the channel response is almost flat so that simple 1-tap channel equalizer can effectively compensate for magnitude and phase distortion. With extended symbol duration (adding cyclic prefix in time domain), multipath fading is avoided. However, OFDM is very sensitive to carrier frequency offset which causes severe inter-carrier interference (ICI). Hence, robust frequency synchronization is required at the receiver. The physical layer is responsible for mapping of transport layer control information and user data to real physical resources (downlink), and construction of higher level data from electromagnetic signal received via antennas. According to the physical channels and modulation specification of LTE [16], the basic components of LTE physical layer on transmission side are channel coding, modulation, precoding, and mapping of symbols onto OFDM subcarriers (resources) of each antenna port. On the other hand, the receiver structure can have different levels of complexity depending on the implementation algorithms. Efficient channel estimation and equalization algorithms are needed in order to achieve good performance in fading environments. The physical layer specification of LTE [16] defines uplink and downlink physical layer processing functions. Figures 2.10 and 2.11 show the generic block diagrams for uplink (UL) and downlink (DL) transmissions respectively.

33 CHAPTER 2. OFDMA SYSTEMS 17 Scrambling Scrambling Modulation Modulation mapper mapper Transform Transform precoder precoder Resource Resource element element mapper mapper SC-FDMA SC-FDMA signal signal gen. gen. Figure 2.10: Block diagram of uplink physical channel processing. codewords layers antenna ports Scrambling Scrambling Modulation mapper Modulation mapper Layer mapper Precoder Resource element mapper Resource element mapper OFDM signal generation OFDM signal generation Figure 2.11: Block diagram of downlink physical channel processing. LTE frame structure Resources are divided both in time and frequency domain. In time domain, a 10 ms duration is called one frame. In LTE there are two types of frame structures depending whether FDD (Frequency Division Duplexing) or TDD (Time Division Duplexing) is used in UL/DL communication. From now on, we will consider the type-1 (FDD) frame structure shown in Figure A frame is divided into 10 subframes each of which is 1 ms long. There are two 0.5 ms long time slots in every subframe. The 20 time slots in a frame are numbered from 0 to Frame = 10ms slot = 0.5 ms 1 subframe = 1ms Figure 2.12: LTE frame structure (type-1). Depending on the cyclic prefix length, there are 6 or 7 OFDM symbols per time slot. Under normal cyclic prefix configuration, there are 7 OFDM symbols per slot. Except for the first symbol, which has 5.2 µs long cyclic prefix, the cyclic prefix is 4.7 µs long for the rest 6 symbols. The useful symbol duration is then 1/15 ms, and hence the subcarrier spacing is 15 khz. Unless mentioned, a

34 CHAPTER 2. OFDMA SYSTEMS 18 normal cyclic prefix configuration is assumed. 1 Resource block = 12 x 7 resource elements ( 180 khz over 0.5 ms duration) One OFDM symbol Symbol perioad = 1/15 ms Cyclic prefix = 5.2 µs for the first symbol, and 4.7µs for the rest 6 symbols Subcarrier Spacing = 15 khz slot = 0.5 ms 1 subframe = 1ms 1 Frame = 10ms Figure 2.13: Time-frequency resource grid structure of LTE for normal cyclic prefix configuration. In multi-antenna transmission, each antenna port has one resource grid. Each OFDM symbol carries a set of subcarriers. One subcarrier on a time slot is the smallest time-frequency unit called resource element. A group of 12 x 7 resource elements is called a resource block. This corresponds to 12 subcarriers ( or 180 khz bandwidth) over 7 OFDM symbols as shown in Figure The resource elements are mapped to physical channels and physical signals. Reference signals and synchronization signals are the two physical signals used by the physical layer. On the other hand physical channels carry information from the higher layers. The channels carry user data, control information and other signals to the user. So far we have seen physical layer frame structures of typical OFDM systems which are used in broadcast, wireless local area and cellular networks. Table 2.2 compares OFDM parameters of these systems. Later, Chapter 4 will investigate coexistence issues of these different systems in a shared radio resource.

35 CHAPTER 2. OFDMA SYSTEMS 19 Parameter DVB-T IEEE LTE (Release 8) Channel spacing (MHz) 6, 7, , 3, 5, 10, 15, 20 FFT size 2048 (2k mode) ******** , 256, 512, 1024, Useful subcarriers Useful symbol duration, T U (µs) Guard interval (µs) 8192 (8k mode) 1705 (2k mode)******** 6817 (8k mode) 224 (2k mode)********** 896 (8k mode) 1 4 T U, 1 8 T U, 1 16 T U, 1 32 T U Subcarrier spacing (KHz) (2k mode)****** Subcarrier modulation (8k mode) QPSK, 16QAM, 64QAM, non-uniform 16QAM,**** non-uniform 64QAM 1536, , 180, 300, 600, 900, T U 5.2 for l = 0********* 4.7 for l = 1, 2,, BPSK, QPSK, ***** 16QAM, 64QAM QPSK, 16QAM, **** 64QAM ***** Table 2.2: Comparison of OFDM parameters of DVB-T, WLAN and LTE physical layers. The parameter l is symbol number in an LTE time slot.

36 Chapter 3 Cognitive Radios - Interference Channels In this chapter we will discuss coexistence issues of multiple networks in the shared radio resource. While the frequency band is primarily granted for a licensed user, other secondary users (cognitive radios) access the resource without affecting the primary users. Secondary systems may transmit when other transmitters are not active. Alternatively, the secondary system may transmit simultaneously with the primary transmitter while making sure that the primary system s performance is not degraded. In this type of cognitive radio system, the channel is modeled as interference channel. In the following sections we will briefly discuss forms of secondary spectrum access by cognitive radios. Later, detailed analytical modeling of interference channels and their theoretically achievable capacities are reviewed. 3.1 Secondary spectrum access The term secondary spectrum access refers to the case where other secondary users try to take advantage of underutilization of the spectrum by the incumbent (primary) users. While it is used by primary users (mostly licensed users), other secondary users (unlicensed users) can access the spectrum without causing intolerable interference to the primary users. Secondary spectrum access mechanisms are devised based on nature of availability of spectrum 20

37 CHAPTER 3. COGNITIVE RADIOS - INTERFERENCE CHANNELS 21 and SNR (Signal-to-Noise-Ratio) headroom. In general, the radio frequency spectra can be divided into three groups [7]: 1. Black spaces - are occupied bands by high-power users. 2. Grey spaces - are partially occupied bands (adjacent bands). 3. White spaces - are unoccupied bands by any potential transmitter. Based on the spectrum occupancy and available information to the cognitive transmitter, a cognitive radio system may interweave, underlay or overlay to the primary transmission Interweave system In this type of cognitive radio, cognitive transmitter first detects whether the spectrum is occupied by other potential transmissions. Detection of these white spaces can be, for instance, done by measuring the signal power over a chosen sensing period and comparing it to a threshold value defined. When other sided information about ongoing transmission is known, advanced and more reliable spectrum sensing algorithms can be employed. For example, cyclostationary signals can be detected by autocorrelation of received signals [18], [19]. Moreover, channels occupied by static transmitters such as TV broadcasting stations are known, and geo-location database of these channels is used together with signal detection mechanisms for identifying white spaces [20], [21], [22]. Once white spaces are identified, cognitive transmitters will be able to use the spectrum for some period of time. This kind of opportunistic spectrum access is accompanied with either continuous spectrum sensing or local database (if applicable) about activity of primary transmitters on that geographical location. The later approach is applied when transmission times of primary users are known beforehand. TV broadcast stations are one of such systems where some channels could be unoccupied for known period of time or in some geographical areas. Figure 3.1 shows possible interweave operation of cognitive radios in white spaces. Classification of a given band as white space or occu-

38 CHAPTER 3. COGNITIVE RADIOS - INTERFERENCE CHANNELS 22 Primary signals Power density White space Frequency Figure 3.1: Interweave cognitive radio operation - secondary spectrum access on white spaces. pied band is made based on chosen threshold. The decision boundary in this figure is illustrated based on signal power. Other techniques can be used in conjunction to this for more accurate signal detection Underlay system In this type of operation, transmissions from cognitive users can occur while the channels is used by primary transmitters. Transmission power of secondary users is set below the interference threshold for primary receivers [23], [24], [8]. The signal is typically spread over a wide band that could span multiple channels as shown in Figure 3.2. Despreading of secondary signal increases the SNR (Signal-to-Noise-Ratio) at the receiver. Power density Interference threshold Primary signals Underlay signal Frequency Figure 3.2: Underlay cognitive radio - secondary signal is spread over wide band with low power constrained by interference level tolerated by other noncognitive users.

39 CHAPTER 3. COGNITIVE RADIOS - INTERFERENCE CHANNELS Overlay system This is a more complex type of cognitive radio which operates at the same frequency band simultaneously with the primary system but at low power as shown in Figure 3.3. The overlay system is a special class of interference channel discussed in the following section. Operation of overlay cognitive radio depends on the level of cooperation between the primary and secondary transmitters. It is usually assumed that the secondary transmitter has non-causal message and codebook of the primary transmitter so that it can adapt its transmission accordingly. The secondary transmitter may allocate fraction its power for primary signal to compensate for interference caused by its transmission to the primary receiver. Then, the secondary receiver applies interference cancellation algorithms to remove the primary signal. The capacity gain can be further improved when it is possible to have cooperative coding between the primary and the secondary systems. Detailed analysis of capacity gains of overlay system has been addressed in Chapter 4. Power density Tolerable interference constraint Primary signals Overlaid signal Frequency Figure 3.3: Overlay cognitive radio - secondary signal is added on top of the primary signal at relatively low power such that SNR requirement of the primary users is satisfied. 3.2 Interference channel In overlay cognitive radio systems both the primary and the secondary users transmit simultaneously over a shared radio spectrum thereby interfering each other. The channel of such systems belongs to the class of interference channel with two transmitters and two receivers. This section presents mathematical model of a Gaussian interference channel.

40 CHAPTER 3. COGNITIVE RADIOS - INTERFERENCE CHANNELS 24 Consider a discrete memoryless two-input-two-output Gaussian interference channel shown in Figure 3.4. The channel has input signals X 1, X 2 and outputs Y 1, Y 2 where the channel has coefficients h ij R, i, j {1, 2} with additive Gaussian noises Z 1 N (0, N 1 ), Z 2 N (0, N 2 ). The transmit symbols X i are random variables with the following power constraints. E { X 2 1} P1 (3.1) E { X 2 2} P2 (3.2) Z1 X1 h11 + Y1 h21 h12 X2 h22 + Y2 Z2 Figure 3.4: Interference channel with two inputs and two outputs. Outputs of the channel are expressed as Y 1 = h 11 X 1 + h 21 X 2 + Z 1 (3.3) Y 2 = h 12 X 1 + h 22 X 2 + Z 2 (3.4) where the channel coefficients are assumed to be globally known. Let us define the following four variables: SNR 1 = h 11 2 P 1 N 1 SNR 2 = h 22 2 P 2 N 2 (3.5) INR 1 = h 21 2 P 2 N 1 INR 2 = h 22 2 P 1 N 2 (3.6) where SNR i is the signal-to-noise ratio at receiver i, and INR i is interferenceto-noise ratio at receiver i, i {1, 2}. The channel can be fully expressed