Digital Communications Tutorial Cognitive Radio Communications @ Virginia Tech NSF Research Experiences for Undergraduates (REU) Site Ratchaneekorn (Kay) Thamvichai tkay@vt.edu Outline Introduction Fourier Transform Sampling Pulse Amplitude Modulation (PAM) InterSymbol Interference (ISI) Digital Bandpass Modulation Analog vs. Digital Transmitted bits can be detected and regenerated, so noise does not propagate additively. More signal processing techniques are available to improve system performance: source coding, channel (error-correction) coding, equalization, encryption, filtering,… Digital ICs are inexpensive to manufacture Digital communications permits integration of voice, video, and data on a single system (ISDN) Implementation of various algorithms can be done by software instead of hardware Security is easier to implement. Simple Digital Communication System Diagram Digital communications system block diagram Fourier Transform F() is the continuous-time Fourier transform of f(t). The Fourier transformation F(ω) is the frequency domain representation of the original function f(t). It describes which frequencies are present in the original function. Example 1: Ex A: Find the Fourier Transform of x(t) = (t) x(t) t=0 Ex B: Find the Fourier Transform of x(t) = 0.5cos(500pt) Ex C: Find the Fourier Transform of f(t) = rect(t/) t F rect / 2 jt jt dt e dt e / 2 2 sin sin 1 2 2 sinc e j / 2 e j / 2 j 2 2 sinc function sinc(x) 1 x 3p 2p p 0 p 2p sinc(x) = sin(x) x even function zero crossings at x p , 2p , 3p , ... Amplitude decreases proportionally to 1/x 3p Ex D: Pulsed Cosine: cos(0t)rec(t/T) <=> (T/2) sinc(0 T + sinc(+0T 2 2 Linear Time-Invariant (LTI) system h(t) Convolution: y(t) = x(t)*h(t) Its Fourier Transform: Y(ω) = X(ω)H(ω) where H(ω) is a frequency response or a transfer function of a system h(t). Ideal filters A filter is used to eliminate unwanted parts of the frequency spectrum of a signal. A filter is LTI system with an impulse response h(t). The output y(t) of a filter can be founded in time domain using a convolution. However, it is easier to do it in a frequency domain: Y(ω) = X(ω)H(ω) Low Pass Filter with a cutoff frequency c High Pass Filter Example 2: Given x(t) = cos(500pt)cos(1000pt), find an impulse response h(t) of a low-pass filter that passes the low frequency component of the signal. x(t) Low-pass filter h(t) y(t) = low freq. component of x(t) H() Y() = H()X3() = p/2[(-500p) + (+500p)] => y(t) = 0.5cos(500pt) H( ) = rect(/2000p => h(t) = 1000sinc(1000t) Outline Introduction Fourier Transform Sampling Pulse Amplitude Modulation (PAM) InterSymbol Interference (ISI) Digital Bandpass Modulation Sampling Continuous-Time signals Sampling – generating of an ordered number of sequence by taking values of f(t) as specified instants of time i.e. f(t1), f(t2), f(t3), … where tm are instants at which sampling occurs. Sampling operation is implemented in hardware by an analog-to-digital converter (ADC) – electronic device used to sample physical voltage signals. In most cases, continuous-time signals are sampled at equal increments of time. The sample increment, called sample period, is usually denoted as Ts. Impulse sampling Define the continuous time impulse train as: p(t) is an infinite train of continuous time impulse functions, spaced Ts seconds apart. Let x(t) be a continuous time signal we wish to sample. We will model sampling as multiplying a signal x(t) by p(t). Sampling Theorem let P(ω) be a Fourier Transform of p(t), X(ω) be a Fourier Transform of x(t), Xs(ω) be a Fourier Transform of xs (t), Since xs(t) = x(t)p(t) by a multiplication property (Fourier Transform), where Ck are the Fourier Series coefficients of the periodic signal. 21 We see that an impulse train in time, p(t), has a Fourier Transform that is an impulse train in frequency, P(). The spacing between impulses in time is Ts, and the spacing between impulses in frequency is ω0 = 2p/Ts. Note: If we increase the spacing in time between impulses, this will decrease the spacing between impulses in frequency, and vice versa. Spectrum of a sampled signal replicated scaled versions of X(), spaced every 0 apart in frequency Time-domain Frequency-domain ω0 = 2p/Ts If 0c < c , ALIASING (overlap area) occurs If 0c ≥ c , Note: if ω0 - ωc ≥ ωc or ω0 ≥ 2ωc, there is no aliasing Sampling Theorem Let x(t) be a band-limited signal with X(ω) = 0 for |ω| > ωc. Then x(t) is uniquely determined by its samples x(nTs), n = 0, ±1, ± 2, … if ω0 ≥ 2ωc where ω0 = 2p/Ts. This is how to choose a sampling frequency (fs = 1/Ts) or period (Ts) such that an original continuous-time signal x(t) can be recovered from a sampled version xs(t). => a sampling rate (ω0) MUST be at least twice the highest frequency (ωc) of a signal to avoid aliasing problem. To recover x(t) from its sampled version xs(t), we use a low pass filter (reconstruction filter) to recover the center island of Xs(): Ex: Given a signal x(t) with Fourier Transform with cutoff frequency ωc as shown: Given three different pulse trains with periods Draw the sampled spectrum in each case. Which case(s) experiences aliasing? Aliasing Phenomenon Sampling theorem: the signal is strictly band-limited (c). However, in practice, no information-bearing signal is strictly band-limited. Aliasing is the phenomenon of a high-frequency component in the spectrum of the signal seemingly taking on the identify of a lower frequency in the spectrum of its sampled version. To prevent the effects of aliasing in practice Prior to sampling : a low-pass anti-alias filter is used to attenuate those high-frequency components of a message signal that are not essential to the information being conveyed by the signal. The filtered signal is sampled at a rate slightly higher than the Nyquist rate. Example: Why 44.1 kHz for Audio CDs? Sound is audible in 20 Hz to 20 kHz range: fmax = 20 kHz and the Nyquist rate 2fmax = 40 kHz What is the extra 10% of the bandwidth used? Rolloff from passband to stopband in the magnitude response of the anti-aliasing filter. Okay, 44 kHz makes sense. Why 44.1 kHz? At the time the choice was made, only recorders capable of storing such high rates were VCRs. NTSC: 60-Hz video (30 frames/s) - 490 lines per frame or 245 lines per field, 3 audio samples per line the sampling rate is 60 X 245 X 3 = 44.1 KHz Outline Introduction Fourier Transform Sampling Pulse Amplitude Modulation (PAM) InterSymbol Interference (ISI) Digital Bandpass Modulation Pulse-Amplitude Modulation (PAM) The amplitude of regularly spaced pulses are varied in proportion to the corresponding sample values of a continuous message signal. Two operations involved in the generation of the PAM signal Instantaneous sampling of the message signal m(t) every Ts seconds, Lengthening the duration of each sample, so that it occupies some finite value T. Sample-and-Hold Filter : Analysis The PAM signal is s(t ) m(nT )h(t nT ) s n s The h(t) is a standard rectangular pulse of unit amplitude and duration T 1, 0 t T t 2 1 , t 0, t T h(t ) rect T 2 0, otherwise The instantaneously sampled version of m(t) is m (t ) m(nT ) (t nT ) n s s To modify mδ(t) so as to assume the same form as the PAM signal: s(t ) m (t ) h(t ) m ( )h(t )d m(nT )h(t nT ) n s s The PAM signal s(t) is mathematically equivalent to the convolution of mδ(t) , the instantaneously sampled version of m(t), and the pulse h(t). Its Fourier Transform: S( f ) fs M ( f kf )H ( f ) k s One benefit of PAM It enables the simultaneous transmission of multiple signals using time-division multiplexing (TDM). User 1 User 2 Quantization Process Amplitude quantization: The process of transforming the sample amplitude m(nTs) of a baseband signal m(t) at time t=nTs into a discrete amplitude v(nTs) taken from a finite set of possible levels. It will be represented by binary number(s) I k : {mk m mk +1}, k 1,2,...,L 39 Outline Introduction Fourier Transform Sampling Pulse Amplitude Modulation (PAM) MATLAB! InterSymbol Interference (ISI) Digital Bandpass Modulation Baseband Transmission of Digital Data The transmission of digital data over a physical communication channel is limited by two unavoidable factors 1.Intersymbol interference 2.Channel noise 42 The level-encoded signal and the discrete PAM signal are + 1 if t heinput bk is symbol1 ak 1 if t heinput bk is symbol0 The transmitted signal is s(t ) The channel output is a g (t kT ) k k b x(t ) s(t ) h(t ) The output from the receive-filter is y(t ) x(t ) q(t ) s(t ) * h(t ) * q(t ) 43 The InterSymbol Interference (ISI) Problem We may express the receive-filter output as the modified PAM signal a y(t ) k where After sampling: k p(t kTb ) p(t ) g (t ) h(t ) q(t ) y(iTb ) a k a k p[(i k )Tb ], i 0,1,2,... pi p(iTb ) yi y (iTb ) yi k k pi k , i 0,1,2,... 44 ISI (cont.) p0 p(0) E Define where E is the transmitted signal energy / bit (symbol). What we desire is However, from yi yi Eai , for all i a k yi E ai + k pi k , a k k i k i 0,1,2,... pi k , i 0,1,2,... Residual phenomenon, intersymbol interference (ISI) 45 Pulse-shaping Given the channel transfer function, determine the transmit-pulse spectrum and receive-filter transfer function so as to satisfy two basic requirements: 1.Intersymbol interference (ISI) is reduced to zero. 2.Transmission bandwidth is conserved. The Nyquist Channel The optimum solution for zero ISI at the minimum transmission bandwidth possible in a noise-free environment For zero ISI, it is necessary for the overall pulse shape p(t) and the inverse Fourier transform of the pulse spectrum P(f) to satisfy the condition E , for i 0 pi p(iTb ) for all i 0 0, i sin c(2 B0t i) p(t ) p i 2 B0 B0 1 2Tb E sin(2pB0t ) Popt (t ) E sin c(2B0t ) 2pB0t 47 The overall pulse spectrum is defined by the optimum brickwall function: E , for B0 f B0 Popt ( f ) 2 B0 0, ot herwise The brick-wall spectrum defines B0 as the minimum transmission bandwidth for zero intersymbol interference. The optimum pulse shape is the impulse response of an ideal low-pass channel with an amplitude response Popt(f) in the passband and a bandwidth B0 48 49 Symbol 1 Symbol 2 Symbol 3 Two difficulties that make its use for a PAM system impractical: 1. The system requires that the spectrum P(f) be flat from –B0 to B0, and zero elsewhere 2. The time function p(t) decreases as 1/|t| for large |t|, resulting in a slow rate of decay 51 Raised-Cosine Pulse Spectrum To ensure physical realizability of the overall pulse spectrum P(f), the modified P(f) decreases toward zero gradually rather than abruptly 1. Flat portion, which occupies the frequency band 0≤|f| ≤f1 for some parameter f1 to be defined 2. Roll-off portion, which occupies the frequency band f1 ≤|f| ≤2B0-f1 E , 0 f f1 2 B0 p ( f f1 ) E p( f ) 1 + cos , f1 f 2 B0 f1 2( B0 f1 ) 4 B0 0, 2 B0 f1 f 52 The roll-off factor: f1 1 B0 Time-domain of the overall channel cos(2pB0t ) p(t ) E sin c(2 B0t ) 2 2 2 1 16 B0 t The amount of intersymbol interference resulting from a timing error ∆t decreases as the roll-off factor is increased form zero to unity. 53 Frequency domain P(f) Time domain p(t) 54 Transmission-Bandwidth Requirement The transmission bandwidth required by using the raised-cosine pulse spectrum is BT 2B0 f1 BT B0 (1 + ) Excess channel The transmission bandwidth requirement of the raised-cosine spectrum exceeds that of the optimum Nyquist channel by the amount f v B0 1. When the roll-off factor is zero, the excess BW is reduced to zero 2. When the roll-off factor is unity, the excess BW is increased to B0. 55 Summary (ISI) The intersymbol interference problem, which arises due to imperfections in the frequency response of the channel ISI refers to the effect on that pulse due to cross-talk or spillover from all other signal pulses in the data stream applied to the channel input A corrective measure widely used in practice is to shape the overall pulse spectrum of the baseband system, starting from the source of the message signal all the way to the receiver. ISI is a signal-dependent phenomenon, it therefore disappears when the information-bearing signal is switched off. Noise is always there, regardless of whether there is data transmission or not. Another corrective measure for dealing with the ISI: channel equalization. 56 Outline Introduction Fourier Transform Sampling Pulse Amplitude Modulation (PAM) InterSymbol Interference (ISI) Digital Bandpass Modulation Techniques Digital band-pass modulation techniques Baseband Communication: Signals are transmitted without any shift in the range of frequency of the signal. Band-pass Communication: Uses modulation to shift the frequency spectrum of a (carrier) sinusoidal signal. Usually, one of the basic parameters (amplitude, frequency, or phase) of the carrier signal is varied in proportion to the baseband signal (information-bearing data stream). Why modulate signals? Convert signals to a form that is suitable for transmission Sharing the frequency band with other stations Three basic modulation schemes: Amplitude-shift keying (ASK) Phase-shift keying (PSK) Frequency-shift keying (FSK) 58 Given a binary source The modulation process involves switching or keying the amplitude, phase, or frequency of a sinusoidal carrier wave between a pair of possible values in accordance with symbol (bit) 0 and 1. c(t ) Ac cos(2pf ct + c ) Examples of a band-pass process 1. Binary amplitude shift-keying (BASK) The carrier amplitude is keyed between the two possible values used to represent symbols 0 and 1 2. Binary phase-shift keying (BPSK) The carrier phase is keyed between the two possible values used to represent symbols 0 and 1. 3. Binary frequency-shift keying (BFSK) The carrier frequency is keyed between the two possible values used to represent symbols 0 and 1. 60 In digital comm., the usual practice is to assume that the carrier c(t) has unit energy measured over one symbol (bit) duration (Tb). 2 c(t ) cos(2pf ct + c ) Tb where Ac 2 Tb Decreasing the bit duration Tb has the effect of increasing the transmission bandwidth requirement of a binary modulated wave. (Fourier Transform property). 61 Band-Pass Assumption The spectrum of a digital modulated wave s(t) is centered on the carrier frequency fc s(t ) b(t )c(t ) where b(t) is an incoming binary stream with bandwidth W. 2 s(t ) b(t ) cos(2pf ct ) Tb Assumption: fc>> BW, There will be no spectral overlap in the generation of s(t) The transmitted signal energy per bit can be approximated as: Tb Eb s(t ) dt 2 0 2 Tb Tb 0 b(t ) cos2 (2pf ct )dt 2 1 Tb Tb 0 2 b(t ) dt 62 Binary Amplitude-Shift Keying (BASK) The ON-OFF signaling variety Eb , for binary symbol1 b(t ) for binary symbol0 0, 2 Eb cos(2pf ct ), for symbol1 s(t ) Tb 0, for symbol0 The average transmitted signal energy is (the two binary symbols must be equi-probable) Eb Eav 2 64 fc = 8 Hz, Tb = 1s 65 fc = 8 Hz, Tb = 0.5 s 66 From figures: The spectrum of the BASK signal contains a line component at f=fc When the carrier is fixed and the bit duration is halved, the width of the main lobe of the sinc function defining the envelope of the BASK spectrum is doubled, which, in turn, means that the transmission bandwidth of the BASK signal is doubled. Tb halved <= => W is doubled The transmission bandwidth of BASK, measured in terms of the width of the main lobe of its spectrum, is equal to 2/Tb, where Tb is the bit duration. 67 Phase-Shift Keying Binary Phase-Shift Keying (BPSK) The pair of signals used to represent symbols 1 and 0, 2 Eb cos(2pf c t ), for symbol1 Tb si (t ) 2 Eb cos(2pf t + p ) 2 Eb cos(2pf t ), for symbol0 c c Tb Tb An antipodal signals A pair of sinusoidal wave, which differ only in a relative phase-shift of p radians. Note: The transmitted energy per bit, Eb, is constant. Equivalently, the average transmitted power is constant. 68 Signal Space diagram of BPSK 70 fc = 8 Hz, Tb = 1s 71 fc = 8 Hz, Tb = 0.5 s 72 From figures: BASK and BPSK signals occupy the same transmission bandwidth (2/Tb), which defines the width of the main lobe of the sinc-shaped power spectra. The BASK spectrum includes a carrier component, whereas this component is absent from the BPSK spectrum. 73 Quadriphase-Shift Keying (QPSK) An important goal of digital communication is the efficient utilization of channel bandwidth. In QPSK, the phase of the sinusoidal carrier takes on one of the four equally spaced values, such as p/4, 3p/4, 5p/4, and 7p/4 2E p cos2pf ct + (2i 1) , 0 t T si (t ) T 4 elsewhere 0, Each one of the four equally spaced phase values corresponds to a unique symbol which is a pair of bits (00, 01, 10, 11). Symbol duration T 2Tb 74 p 2E p 2E sin (2i 1) sin(2pf ct ) cos(2i 1) cos(2pf ct ) si (t ) 4 T 4 T 1. In reality, the QPSK signal consists of the sum of two BPSK signals. 2. One BPSK signal, represented by the first term defined the product of modulating a binary wave by the sinusoidal carrier p 2 E / T cos(2i 1) cos(2pf ct ), 4 p E / 2 for i 1, 4 E cos(2i 1) 4 E / 2 for i 2 , 3 3. The second binary wave p 2 E / T sin (2i 1) sin( 2pf c t ), 4 p E / 2 for i 1, 2 E sin (2i 1) 4 E/2 for i 3, 4 75 76 Signal Space diagram of QPSK 78 QPSK Transmitter QPSK Receiver 80 fc = 8 Hz, Tb = 1s BW = 1/Tb 81 fc = 8 Hz, Tb = 0.5 s BW = 1/Tb 82 Frequency-Shift Keying Binary Frequency-Shift Keying (BFSK) Each symbols are distinguished from each other by transmitting one of two sinusoidal waves that differ in frequency by a fixed amount: 2 Eb cos(2pf1t ), for symbol1 Tb si (t ) 2 Eb cos(2pf t ), for symbol0 2 Tb Sunde’s BFSK When the frequencies f1 and f2 are chosen in such a way that they differ from each other by an amount equal to the reciprocal of the bit duration Tb 83 84 fc = 8 Hz, Tb = 1s f = fc ± 1/(2Tb) BW = 3/Tb 85 Bandwidth 2/Tb 2/Tb 3/Tb for f1,2 = fc ± 1/(2Tb) QPSK 1/Tb 86 M-ary Digital Modulation Schemes We send any one of M possible signals during each signaling interval of duration T. The requirement is to conserve bandwidth at the expense of both increased power and increased system complexity. When the bandwidth of the channel is less than the required value, we resort to an M-ary modulation scheme for maximum bandwidth conservation 87 M-ary Phase-Shift Keying If we take blocks of m bits to produce a symbol and use an M-ary PSK scheme with M=2m and symbol duration T=mTb The bandwidth required is proportional to 1/(mTb). The use of M-ary PSK provides a reduction in transmission bandwidth by a factor m=log2M over BPSK. 88 89 M-ary Quadrature Amplitude Modulation (QAM) The mathematical description of the new modulated signal i 0,1,...,M 1 2E0 2E0 si (t ) ai cos(2pf ct ) bi sin(2pf ct ), 0t T T T The level parameter for in-phase component and quadrature component are independent of each other for all i. M-ary QAM is a hybrid form of M-ary modulation. M-ary amplitude-shift keying (M-ary ASK) si (t ) 2 E0 ai cos(2pf c t ) i 0,1,...,M 1 T 90 Signal-Space Diagram Figure 7.21 is the signal-space representation of M-ary QAM for M=16 Unlike M-ary PSK, the different signal points of M-ary QAM are characterized by different energy levels Each signal point in the constellation corresponds to a specific quadbit 92 Bit Error Rate Average bit error rate (BER) Let n denote the number of bit errors observed in a sequence of bits of length N; then the relative frequency definition of BER is n BER lim N N BER goal: For data transmission over wireless channels, a bit error rate of 10-5 to 10-6 For video transmission, a BER of 10-7 to 10-12 depending upon the quality desired and the encoding method. 93 Signal to Noise Ratio (SNR) The ratio of the modulated energy per information bit to the one-sided noise spectral density; namely, Modulated energy per bit Noise spectral density Eb N0 SNR digital ref The reference SNR is independent of transmission rate. Since it is a ratio of energies, it has essentially been normalized by the bit rate. where Q x 1 2p x exp( s 2 / 2)ds 95 BASK Pe = 0.5P(0 decided| 1 is trans.) + 0.5P(1 decided|0 is trans.) Pe 0.5Q + 0.5Q Q for /2 2 Pe Q Q 2 Eb N0 where Q N 0T / 2 and AT 1 2p exp(( y ) 2 / 2 2 )dy 2 EbT 97 Real-world use (Tidbits) The wireless LAN standard, IEEE 802.11b-1999, uses a variety of different PSKs depending on the data-rate required. - Basic-rate of 1 Mbit/s, DBPSK - Extended-rate of 2 Mbit/s, DQPSK - 5.5 Mbit/s and the full-rate of 11 Mbit/s, QPSK is used with complementary code keying. The higher-speed wireless LAN standard, IEEE 802.11g2003[1][3] has eight data rates: 6, 9, 12, 18, 24, 36, 48 and 54 Mbit/s. - The 6 and 9 Mbit/s modes, OFDM modulation where each sub-carrier is BPSK modulated. - The 12 and 18 Mbit/s modes use OFDM with QPSK. - The fastest four modes use OFDM with QAM BPSK is appropriate for low-cost passive transmitters, and is used in RFID standards. Bluetooth uses p/4-DQPSK for the rate 2 Mbit/s and 8-DPSK at its higher rate (3 Mbit/s) IEEE 802.15.4 (the wireless standard used by ZigBee) also relies on PSK. It has two frequency bands: - 868–915 MHz using BPSK and - 2.4 GHz using OQPSK References: Simon Haykin and Michael Moher, “Introduction to Analog and Digital Communications,” 2nd ed., John Wiley & Sons, Inc., 2007. Charles L. Phillips, John M. Parr, Eve A. Riskin, “Signals, Systems, and Transforms,” 4th ed., Pearson/Prentice Hall, 2008.