Pulse Digital Modulation Introduction * There are three types of modulation (i) Amplitude modulation (i) Angle modulation (iii) Pulse modulation * Pulse modulation can be further ‘classified as, (i) Pulse analog modulation (ii) Pulse digital modulation © The above two techniques can be further classified as, [ Pulse modulation = () Pulse amplitude modulation (0 Pulse cade modulation (i) Pulse position modulation (i) Delta modulation (ii) Pulse duration modulation (ii) Adaptive delta modulation (iv) Differential pulse code médulation «In the above techniques following points are studied : (i) Principle of operation (ii) Transmitter and receiver block diagram (ii) Error analysis (iv) Signal to quantization noise ratio. a-1)Digital Communications. 1-2 Pulse Digital Modulation 1.1 Advantages of Digital! Communication System Presently most of the communication is digital. For example cellular (mobile phone) communication, satellite communication, radar and sonar signals, Facsimile, data transmission over internet etc all use digital communication, Paractically, after 20 years, analog communication will be totally replaced by digital communication. Why digital communication is so popular ? There are few reasons due to which people are prefering digital communication over analog communication. 1. Due to advancements in VLSI technology, it is possible to manufacture very high speed embedded circuits, Such circuils are used in digital communications. 2. High speed computers and powerful software design tools are available. They make the development of digital communication systems feasible. 3. Internet is spread almost in every city and towns. The compatibility of digital communication systems with internet has opened new area of applications. Advantages and Disadvantages of Digital Communication Advantages : 1. Because of the advances in digital IC technologies and high speed computers, digital communication systems are simpler and cheaper compared to analog systems. 2. Using data encryption, only permitted receivers can be allowed to detect the transmitted data. This is very useful in military applications. 3. Wide dynamic range is possible since the data is converted to the digital form. 4. Using multiplexing, the speech, video and other data can be merged and transmitted over common channel. 5. Since the transmission is digital and channel encoding is used, the noise does not accumulate from repeater to repeater in long distance communication. 6. Since the transmitted signal is digital, a large amount of noise interference ean be tolerated. 7, Since channel coding is used, the errors can be detected and corrected in the Teceivers. 8. Digital communication is adaptive to other advanced branches of data Processing such as digital signal processing, image processing, data compression etc.Digital Communications 1-3 Pulse Digital Modulation Disadvantages : Eventhough digital communication offer many advantages as given above, it has some drawbacks also, But the advantages of digital communication outweigh disadvantages. They are as follows - 1. Because of analog to digital conversion, the data rate becomes high. Hence more transmission bandwidth is required for digital communication. 2. Digital communication needs synchronization in case of synchronous modulation. 1.2 Elements of Digital Communication System Fig. 1.2.1 shows the basic operations in digital communication system. The source and the destination are the two physically separate points. When the signal travels in the communication channel, noise interferes with it, Because of this interference, the smeared or disturbed version of the input signal is received at the receiver. Therefore the signal received may not be correct. That is errors are introduced in the received signal. Thus the effects of noise due to the communication channel limit the rate at which signal can be transmitted. The probability of error in the received signal and transmission rate are normally used as performance measures of the digital communication system. Fig. 1.2.1 Basic dialtal communication system 1.2.1 Information Source The information source generates the message signal to be transmitted. In case of analog communication, the information soyrce is analog. In case of digital communication, the information source prodtices a message signal which is not continuously varying with time. Rather the message signal is intermittent with respect to time. The examples of discrete information sources are data from computers,Digital Communications 1-4 Pulse Digital Modulation teletype etc. Even the message containing text is also discrete. The analog signal can be converted to discrete signal by sampling and quantization. In sampling, the analog signal is chopped off at regular time intervals. Those chopped samples form a discrete signal. The discrete information sources have following important parameters : a) Source alphabet : These are the letters, digits or special characters available from the information source. b) Symbol rate ; It is the rate at which the information source generates source alphabets. It is normally represented in symbols/sec unit. Source alphabet probabilities : Each source alphabet from the source has independent occurrence rate in the sequence. For example, letters A, E, I etc. occur frequently in the sequence. Thus probability of the occurrence of each source alphabet can become one of the important property which is useful in digital communication. a d Probabilistic dependence of symbols in a sequence : The information carrying capacity of each source alphabet is different in a particular sequence. This parameter defines average information content of the symbols. The entropy of a source refers to the average information content per symbol in long messages. Entropy is defined in terms of bits per symbol. Bit is the abbreviation for binary digit. The source information rate is thus the product of symbol rate and source entropy i.e.. Information rate = Symbol rate x Source entropy (Bits/sec) (Symbols/sec) (Bits /Symbol) The information rate represents minimum average data rate required to transmit information from source to the destination, 1.2.2 Source Encoder and Decoder The symbols produced by the information source are given to the source encoder. ‘These symbols cannot be transmitted directly. They are first converted into digital form (i.e. Binary sequence of 1's and 0's) by the source encoder. Every binary ‘1’ and ‘0’ is called a bit. The group of bits is called a codeword. The source encoder assigns codewords to the symbols. For every distinct symbol there is a unique codeword. The codeword can be of 4, 8, 16 or 32 bits length. As the number of bits are increased in each codeword, the symbols that can be represented are increased. For example, 8 bits will have 2° = 256 distinct codewords. Therefore 8 bits can be used to represent 256 symbols, 16 bits can represent 2'° = 65536 symbols and so on. In both of the above examples the number of bits in every codeword is same throughout. That is 8 in first case and 16 in next case respectively, This is called fixed length coding. Fixed length coding is efficient only if all the symbols occur with equalDigital Communications 41-5 Pulse Digital Modulation probabilities in a statistically independent sequence. In the practical situations, the symbols in the sequence are statistically dependent and they have unequal probabilities of occurrence. For example, let us assume that the symbol sequence tepresents the percentage marks of the students. The 02%, 08%, 20%, 98%, 99% etc. symbols will have minimum probability of occurrence. But 60%, 55%, 70%, 75% will have more probability. For such symbols normally variable length codewords are assigned. More bits (More length) are assigned to rarely occurring symbols and less bits are assigned to frequently occurring symbols. Typical source encoders are pulse code modulators, delta modulators, vector quantizers etc. We will come across these codewords in detail in the subsequent chapters. Source encoders have following important parameters. a) Block size : This gives the maximum number of distinct codewords that can be represented by the source encoder. It depends upon maximum number of bits in the codeword. For example, the block size of 8 bits source encoder will have 28 =256 codewords. b) Codeword length : This is the number of bits used to represent each codeword. For example, if 8 bits are assigned to every codeword, then codeword length is 8 bits. ©) Average data rate : It is the output bits per second from the source encoder. The source encoder assigns multiple number of bits to every input symbol. Therefore the data rate is normally higher than the symbol rate. For example let us consider that the symbols are given to the source encoder at the rate of 10 symbols/sec and the length of codeword is 8 bits. Then the output data rate from the source encoder will be, Date rate = Symbol rate x Codeword length = 10 x 8 = 80 bits/sec Information rate is the minimum number of bits per second needed to convey information from source to destination as stated earlier. Therefore optimum data rate is equal to information rate. But because of practical limitations, designing such source encoder is difficult. Hence average data rate is higher than information rate and hence symbol rate also. d) Efficiency of the encoder : This is the ratio of minimum source information rate to the actual output data rate of the source encoder. At the receiver, some decoder is used to perform the reverse operation to that of source encoder. It converts the binary output of the channel decoder into a symbol sequence. Both variable length and fixed length decoders are possible. Some decoders use memory to store codewords. The decoders and encoders can be synchronous or asynchronous.Digital Communications. 1-6 Pulse Digital Modulation 1.2.3 Channel Encoder and Decoder At this stage we know that the message or information signal is converted in the form of binary sequence (ie, 1’s and 0's). The communication channel adds noise and interference to the signal being transmitted. Therefore errors are introduced in the binary sequence received at the receiver. Hence errors are also introduced in the symbols generated from these binary codewords. To avoid these errors, channel coding is done. The channel encoder adds some redundant binary bits to the input sequence: These redundant bits are added with some properly defined logic. For example consider that the codeword from. the source encoder is three bits long and one redundant bit is added to make it 4-bit long. This 4" bit is added (either 1 or 0) such that number of 1’s in the encoded word remain even (also called even parity). Following table gives output of source encoder, the 4" bit depending upon the parity, and output of channel encoder. ‘Output of source Bit to be added by channel | Output of channel encoder encoder for even parity encoder bs be by bo bs be by bo 11410 Q 1100 o10 4 o17om oo0 0 ooo0o0 vid 1 aid Table 1.2.1 Even parity coding Observe in the above table that every codeword at the output of channel encoder contains “even” number of 1's. At the receiver, if odd number of 1's are detected, then receiver comes to know that there is an error in the received signal. The channel decoder at the receiver is thus able to detect error in the bit sequence, and reduce the effects of channel noise and distortion. The channel encoder and decoder thus serve to increase the reliability of the received signal. The extra bits which are added by the channel encoders carry no information, rather, they are used by the channel decoder fo detect and correct errors if any. These error correcting bits may be added recurrently after the block of few symbols or added in every symbol as shown in Table 1.2.1. The example of parity coding given above is just illustrative. There are many advanced and efficient coding techniques available. We will discuss them in the book. The coding and decoding operation at encoder and decoder needs the memory (Storage) and processing of binary data. Because of microcontrollers and computers, the complexity of encoders and decoders is nowadays very much reduced. The important parameters for channel encoder are -Digital Communications 1-7 Pulse Digital Modulation The method of coding used. b) Coding rate, which depends upon the redundant bits added by the channel encoder. c) Coding efficiency, which is the ratio of data rate at the input to the data rate at the output of encoder. d) Error control capabilities, ie. detecting and correcting errors ¢) Feasibility or complexity of the encoder and decoder. The time delay involved in the decoding is also an important parameter for channel decoder. 1.2.4 Digital Modulators and Demodulators Whenever the modulating signal is discrete (ie. binary codewords), then digital modulation techniques are used. The carrier signal used by digital modulators is always continuous sinusoidal wave of high frequency. The digital modulators maps the input binary sequence of 1’s and 0's to analog signal waveforms. If one bit at a time is to be transmitted, then digital modulator signal is s\(f) to transmit binary ‘0’ and s2(!) to transmit binary ‘1’. For example consider the output of digital modulator shown in Fig. 12.2. Saft) sal) 50 set) s4(t) s(t) Fig. 1.2.2 Frequency modulated output of a digital modulator The signal s1(t) has low frequency compared to signal s2(t). It is frequency modulation (FM) in two steps corresponding to binary symbols ‘0’ and ‘I’. Thus even though the modulated signal appears to be continuous, the modulation is discrete (or in steps). Single carrier is converted into two waveforms s;(é) and s2(t) because of digital modulation. If the codeword contains two bits and they are to be transmitted at a time, then there will be M=2? =4 distinct symbols (or codewords). These four codewords will require four distinct waveforms for transmission. Such modulators are called M-ary modulators. Frequency Shift Keying (FSK), Phase Shift Keying (PSK), Amplitude Shift Keying (ASK), Differential Phase Shift Keying (DPSK), Minimum Shift Keying (MSK) are the examples of various digital modulators. Since these modulators use continuous carrier wave, they are also called. digital CW modulators.Digital Communications 1-8 Pulse Digital Modulation In the receiver, the digital demodulator converts the input modulated signal to the sequence of binary bits. The most important parameter for the demodulator is the method of demodulation. The other parameters for the selection of digital modulation method are, a) Probability of symbol or bit error. b) Bandwidth needed to transmit the signal. ¢) Synchronous or asynchronous method of detection and d) Complexity of implementation. 1.2.5 Communication Channel As we have seen in the preceding sections, the connection between transmitter and receiver is established through communication channel. We have seen that the communication can take place through wirelines, wireless or fiber optic channels. The other media such as optical disks, magnetic tapes and disks etc. can also be called as communication channel, because they can also carry data through them. Every communication channel has got some problems. Following are the common problems associated with the channels : a) Additive noise interference : This noise is generated due to internal solid state devices and resistors etc. used to implement the communication system. b) Signal attenuation : It occurs due to internal resistance of the channel and fading of the signal. ©) Amplitude and phase distortion : The signal is distorted in amplitude and phase because of non-linear characteristics of the channel. Multipath distortion : This distortion occurs mostly in wireless communication channels. Signals coming from different paths tend to interfere with each other. . There are two main resources available with the communication channels. These two resources are = 4) a) Channel bandwidth : This is the maximum possible range of frequencies that can be used for transmission. For example, the bandwidth offered by wireline channels is less compared to fiber optic channels. b) Power in the transmitted signal : This is the power that can be put in the signal being transmitted. The effect of noise can be minimized by increasing the power. But this cannot be increased to very high value because of the equipment and other constraints. For example, the power in the wireline channel is limited because of the cables. The power and bandwidth limit the data rate of the communication channel. As we know, the fiber optic channel transports light signals from one place to anather just like a metallic wire carriers an electric signal. There is no current or metallic conductor in optical fiber. The optical fiber has following advantages :Digital Communications 4-9 Pulse Digital Modulation a) Very large bandwidths are possible. b) Transmission losses are very small. c) Electromagnetic interference is absent. d) They have small size and weight. e) They offer ruggedness and flexibility. f) Optical fibers are low cost and cheap. Satellites essentially perform wireless communication. Mainly satellites are repeaters. Broad area coverage is the main advantage of satellites. The power requirement is also less, since solar energy is used by satellites. Global communication is very easily possible through satellite channel. The interference on satellite channels is present but it is minimum. Theory Question 1. Explain with neat block diagram the essential and non essential features of a digital communication system. 1.3 Sampling Process 1.3.1 Representation of CT Signals by its Samples Why CT signals are represented by samples 7 * ACCT signal cannot be processed in the digital processor or computer. © To enable digital transmission of CT signals. Fig. 1.3.1 shows the CT signal and its sampled DT signal. In this figure observe that the CT signal is sampled at t = 0, T,, 2T, 3T,, ... and so on. Fig. 1.3.1 CT and its DT signalDigitat Communications 1-10 Pulse Digital Modulation * Here sampling theorem gives the criteria for spacing ‘T,’ between two successive samples. * The samples x(t) must represent all the information contained in x(t). The sampled signal x5(t) is called discrete time (DT) signal. It is analyzed with the help of DTFT and z-transform. 1.3.2 Sampling Theorem for Lowpass (LP) Signals A lowpass or LP signal contains frequencies from 1 Hz to some higher value. Statement of sampling theorem 1) A band limited signal of finite energy, which has no frequency components higher than W Hertz, is completely described by specifying the values of the signal at instants of time separated by 51; seconds and 2) A band limited signal of finite energy, which has no frequency components higher than W Hertz, may be completely recovered from the knowledge of its Samples taken at the rate of 2W samples per second. The first part of above: statement tells about sampling of the signal and second part tells about reconstruction of the signal. Above statement can be combined and stated alternately as follows : A continuous time signal can be completely represented in its samples and recovered back if the sampling frequency is twice of the highest frequency content of the signal. i.c., f = 2W Here f, is the sampling frequency and W is the higher frequency content Proof of sampling theorem There are two parts : (1) Representation of x(t) in terms of its samples (@) Reconstruction of (1) from its samples. Part I : Representation of x(f) in its samples x(nT,) Step 1: Define x3() Step 2: Fourier teansforia of 5() ic: Xs() ‘Step 3: Relation between X(f) and Xa(/)_ “Step 4: Relation between x() and x7.) EiDigital Communications 1-11 Pulse Digital Modilation Step 1: Define x5{t) Refer Fig. 1.3.1. The sampled signal xs(¢) is given as, xs() = Sy se=n7,) +: 13.1) azo Here observe that xg(t) is the product of xs and impulse train 6(f) as shown in Fig. 1.3.1. In the above equation 8(!~nT,) indicates the samples placed at “tT, £2T,, 37, ... and so on. Step 2: FT of xg(t) ie. X5(f) Taking FT of equation (1.3.1). | Soue-xn} Xa(f) = FF {Product of x(f) and impulse train} We know that FT of product in time domain becomes convolution in frequency domain. ie, Xe) = FE G()}¢FTB(-a7,)) 2-132) By definitions, x() #24 x(f) and 3¢-nt,) Log Say-nf) Hence equation (1.3.2) becomes, XW = XD TAG-mf) nene Since convolution is linear, XN =k DXN*8U—M) =f SE) By shifting property of impulse fiction = AXP -26)+6 XU fe +f XK XU 4h XU “2hDigital Communications 1212 Pulse Digital Modulation Comments () The RHS of above equation shows that X(f) is placed att fe, £2f.23f,.-- (ii) This means X(9 is periodic in f,. (iii) If sampling frequency is f, = 2W, then the spectrums X(f) just touch each other. 1 jdt Fig. 1.3.2 Spectrum of original signal and sampled signal Step 3 : Relation between X(f) and X5(f) Important assumption : Let us assume that f, = 2W, then as per above diagram. Xs = LX for -W).Digital Communications 1-16 Pulse Digital Modulation 1.3.3 Effects of Undersampling (Aliasing) While proving sampling theorem we considered that f, = 2W. Consider the case of f; <2W. Then the spectrum of X5(f) shown in Fig. 1.3.4 will be modified as follows : =e I Fig. 1.3.4 Effects of undersampling or aliasing Comments : i) The spectrums located at X(f),X(f-f,),X(f-2f,),... overlap on each other. ii) Consider the spectrums of X(f) and X(f-/,) shown as magnified in above figure. The frequencies from (f, -W) to W are overlapping in these spectrums. iii) The high frequencies near ‘a’ in X(f-f,) overlap with low frequencies (/, - W) Effects of aliasing : i) Since high and low frequencies interfere with each other, distortion is generated. ii) The data is lost and it cannot be recovered. Different ways to avoid aliasing Aliasing can be avoided by two methods : i) Sampling rate f, 2 2W When the sampling rate is made higher than 2W, then the spectrums will not overlap and there will be sufficient gap between the individual spectrums. This is shown in Fig. 135.Digital Communications 1-17 Pulse Digital Modulation Fig. 1.3.5 f, 2 2W avoids aliasing by creating a bandgap ii) Bandlimiting the signal The sampling rate is, f, = 2W. Ideally speaking there should be no aliasing. But there can be few components higher than 2W. These components create aliasing. Hence a lowpass filter is used before sampling the signals as shown in Fig. 1.3.6. Thus the output of lowpass filter is strictly bandlimited and there are no frequency components higher than 'W’. Then there will be no aliasing. x()- Fig. 1.3.6 Bandlimiting the signal. The bandlimiting LPF is called prealins filter 1.3.4 Nyquist Rate and Nyquist Interval Nyquist rate : When the sampling rate becomes exactly equal to ‘ZW’ samples/sec, for a given bandwidth of W Hertz, then it is called Nyquist rate. Nyquist interval ; It is the time interval between any two adjacent samples when sampling rate is Nyquist rate. 2W Hz we (LB) Nyquist interval = Ay seconds 1.3.8) Nyquist rate 4 1.3.5 Reconstruction Filter (Interpolation Filter) Definition In section 1.3.2 we have shown that the reconstructed signal is the succession of sine pulses weighted by a(T,). These pulses are interpolated with the help of a lowpass filter. It is also called reconstruction filter or interpolation filter.Digital Communications 1-18 Pulse Digital Modulation Ideal filter Fig. 1.37 shows the spectrum of sampled signal and frequency response of required filter. When the sampling frequency is exactly 2W, then the spectrums just touch each other as shown in Fig, 1.3.7. The spectrum of original signal, X(f) can be filtered by an ideal filter having passband from -W 0, - Op, This filter provides xeverse action to that of zero-order hold, Fig, 1.310 shows the block diagram with anit-imaging filter. Yell) x(n) Fig. 1.3.10 Block diagram of practical reconstruction 1.3.7 Sampling Theorem in Frequency Domain Statement We have seen that if the bandlimited signal is sampled at the rate of (f, >2W) in time domain, then it can be fully recovered from its samples. This is sampling theorem in time domain. A dual of this also exists and it is called sampling theorem in frequency domain. It states that,Digital Communications 1-241 Pulse Digital Modulation “A timelimited signal which is zero for lier i is uniquely determined by the samples of ite frequency spectrum at interoals less than >. Hertz apart. © Explanation : Thus the spectrum is sampled at f W. * Inphase and quadrature components : This bandpass signal is first represented in terms of its inphase and quadrature components. Let x, () = Inphase component of x!) and XQ (!) = Quadrature component of 2/) Then we can write a(t) in terms of inphase and quadrature components as, (1) = x; (cos (2nf.t)— xg (1) sin (2nf.t) +++ (1.3.10) The imphase and quadrature X (0 &XQN components are obtained ~—by multiplying x) by cos(2nff) and sin (2nf.t) and then suppressing the sum frequencies by means of low-pass filters. ‘Thus jinphase x;(f) and woofow ' quadrature x9 (f) components contain Fig. 1.3.13 Spectrum of inphase and only low frequency components. The quadrature components of bandpass signal spectrum of these components is aed limited between - W to +W. This is shown in Fig. 1.3.13.Digital Communications 1-24 Pulse Digital Modulation + Representation in terms of samples : After some mathematical manipulations, on equation 1.3.10. We obtain the reconstruction formula as, we F *(e) sine(21-3) ow [>a - (13.11) nice Compare this reconstruction formula with that of lowpass signals given by equation (1.3.6). It is clear that x(!) is represented by x (aw) completely. Here, “(ar) and T= x (nT) = Sampled version of bandpass signal a Ww + Thus if 4W samples per second are taken, then the bandpass signal of bandwidth 2W can be completely recovered from its samples, Thus, for bandpass signals of bandwidth 2W, Minimum sampling rate = Twice of bandwidth = 4W samples per second Tm) Example'4.3.1: Show that a bandlimited signal of finite energy which has no frequency components higher than W Fiz is completely described by specifying values of the signals at instants of time separated by 1 / 2W seconds and also show that if the instantaneous values of the signal are separated by intervals larger than Ww seconds, they fail to describe the signal. A bandpass signal has spectral range that extends from 20 to 82 KHlz. Find the acceptable range of sampling frequency f,. Solution : Step 1: Define xg(t). Let x) be the bandlimited signal which has no frequency components higher than W Hz. Let it be sampled by a sampling function xt) = SS 8¢-nt) none The sampling function is the train of impulses with T, as distance between successive impulses. Let x(nT,) be the instantaneous amplitude of signal x(t) at instant t=T,. The sampled version of xf) can be represented as multiplication of x (nT,) and 8) ie.Digital Communications 1-25 Pulse Digital Modulation XO =F xT.) 8-7) + (13.12) Step 2: Fourier transform of xg(t) ie. X5(/) Fourier transform of this sampled signal can be obtained as, Xp) = FT{xs(} =f E XY-m9 a 03.13) neta Here f; is the sampling rate which is given as f= 5 And, X(f=nf,)=X() at nf, =0, tf,22f,,43f,.... ‘Thus the same spectrum X(f) appears at f=0,f=+/,,f=£2f, etc. This means that a periodic spectrum with period equal to f, is generated in frequency domain because of sampling x!) in time domain. Therefore equation 1.3.13 can be written as, XB = fe XN+h XU EAI +S XE) FAX (FESR 4h X (FEA fh) te. w- (13.14) or Xalf = fe X+ PE XG- nf) wes (1.3.15) neces n#0 Step 3: Relation between X(f) and X;(/). By definition of Fourier transform, X(f)= f xthe7/29" at For sampled version of x{), we have t=nT,. Then above equation becomes, = x x(nT,) e PS ++ (13.16) none It is given that the signal is band limited to W Hz and, = a fonda Te = yy seconds, +. fg=q-=2W + (13.17) From equation 1.3.14 we know that Xs (/) is periodic in f. The spectrum X(f) and Xg (fare shown in Fig. 1.3.14.Digital Communications 1-26 Pulse Digital Modulation x X(0) (a) of “woo UW Xy{ipforf, = 2W w, ib) x a * =+3W Fig. 1.3.14 (3) Spectn im of x(t) Spectrum of Xs(f) with f, = 24 Since, =2W; f,-W=W and f, +W=3W ‘Thus the periodic spectrums X(f) just touch £W, £3} , £5W..... etc. Thus there is no aliasing. From equation 1.3.15. we can write, xi) = pxe- 5 X(f- nf) was (1.3.18) nese neo Wh =D tov cain xp = Joxsn- FE xg-m) "0 ie. XY) = ake) For -W x 7,)6(e-nT,)h(—u) du From equation (1.4.14) ¥ x7,) J sun.) (tw au vw (1.4.15) 4 From the sifting property of delta function we know that, day (a) x(t) ~ “ITT 0 Ty me t Any i NN I. h s(U=xg(00 nit) ll . | Fig. 14.7 (a) Baseband signal x(t) | (b) instantaneously sampled signal x; (t) | (c) Constant pulse width function A(t) | (@) Flat top sampled signal s(t) obtained through convolution of A(t)and x; (¢)Digital Communications 1-40 Pulse Digital Modulation J FO8E=tp) = Fg) wn(1.4.16) ‘Using this equation we can write equation 1.4.15 as, s(t) = > x(nT,)A(@t-nT,) w= (14.17) * This equation represents value of s(t) in terms of sampled value x(nT,) and function h (t—T,) for flat top sampled signal. ‘we also know from equation 1.4.12 that, sit) = 350" By taking Fourier transform of both sides of above equation, S) = X;PH s+ (1.4.18) Convolution in time domain is converted to multiplication in frequency domain. Xs (is given as, x= f Exy-nf) = (1.4.19) " «Equation 14.18 becomes, Spectrum of Flat Top Sampled Signal : (=f, z Xf af HA | 1.4.20) This equation represents the spectrum of flat top sampled signal. 1.4.3.4 Aperture Effect Definition ‘The spectrum of flat top sampled signal is given by equation 1.4.20 above. This equation shows that the signal s(t) is obtained by passing through a filter having transfer function H(f). The corresponding impulse response h(t), in time domain is shown in Fig. 1.4:8 (a). This pulse is one pulse of rectangular pulse train shown in Fig. 147 (Q. Every sample of x(t) is convolved with this pulse. Equation 1.4.20 represents that spectrum of this rectangular pulse is multiplied with that of x (#). Fig. 1.48 (b) shows the spectrum of one rectangular pulse of 1(}) The spectrum of a rectangular pulse is given as, H(f) = tsinc(ftye-it A=l vw (4.21)Digital Communications 1941 Pulse Digital Modulation Fig. 1.4.8 (a) One pulse of rectangular pulse train {b) Spectrum of the pulse of Fig. (a) Thus we can see from Fig. 14.8 (b) that by using flat top samples an amplitude distortion is introduced in reconstructed signal x(t) from s(#). The high frequency rolloff of H(f) acts like a lowpass filter and attenuates upper portion of message spectrum. These high frequencies of x(f) are affected. This effect is called aperture effect. Compensation for Aperture Effect As the duration ‘tof the pulse increases, aperture effect is more prominent. Therefore during reconstruction an equalizer is required to compensate for this effect. As shown in Fig, 1.4.9, the receiver consists of lowpass reconstruction filter with cutoff frequency slightly higher than the maximum frequency in message signal. The equalizer compensates for the aperture effect. It also compensates for the attenuation by a low-pass reconstruction filter. Message ‘signal x(t) Fig. 1.4.9 Recovering x(t) From equation 14.21 we now that the sample function f(t) acts like a lowpass filter where Fourier transform is given as, Hf) = tsinc(ft)e7* from equation 1.4.21 w= (14.22)Digital Communications 1-42 Pulse Digital Modulation This-spectrum is plotted in Fig. 1.4.8. Equalizer used in cascade with the Teconstriction filter has the effect of decreasing the inband loss of the reconstruction filter-as the frequency increases in such a manner as to compensate for the aperture effect. The transfer function of the equalizer is given by, Ke- Paha Ha = “Eq o» (1.4.23) Here 't,’ is the delay introduced by lowpass filter which is equal to t/2 Ket tain c (ft) e~ i = —K_ amc o Ha = ww (14.24) This is the transfer function of an equalizer. 14.4 Comparison of Various Sampling Techniques Various sampling techniques can be compared on the basis of their method, noise interference, spectral properties etc. The following table lists some of the important points of comparison. rameter of | ideal or instantaneous | Natural sampling Flat top sampling comparison Principle of It uses multiplication by | It uses chopping It uses sample and sampling an impulse function | principle hhold circuit Circuit of samplor ‘Sampling Discharge: ew nnn an 6, x) xt) © sit) 4 | RealizabilityDigital Communications 1-43 Pulse Digital Modulation ‘Sampling rate tends to | Sampling rate satisfies | Sampling rate satisfies infinity Nyquist criteria Nyquist . criteria Noise interference is Noise interference is | Noise interference is maximum mininumn maximum x= sore S soe = x (aT, )8(-0T,) x (sing (01,0) x(oT,)h(t-nT) glen! ri = = = onan xsnnat, E sna > sine E representation os nee pene Xf-nt, HU) X (fn) sing (nfgt)X (fhe) Table 1.4.1 Comparison of sampling techniques: ump Example 1.4.1: The spectrum of signal x(j) is shown below. This signal is sampled at the Nyquist rate with a periodic train of rectangular pulses of duration SOP milliseconds, Find the spectrum of the sampled signal for frequencies upto 50° Hz giving relevant expression. 7 0010 f Fig. 1.4.10 Solution ; It is clear from Fig, 1.4.10 that the signal is bandlimited to 10 Hz. W = 10Hz 2. Nyquist rate = 2xW=2x10=20Hz Since the signal is sampled at Nyquist rate, the sampling frequency will be, f = 20Hz Rectangular pulses are used for sampling. That is flat top sampling’is used. The spectrum of flat top sampled signal is given by equation 1.4.20 as, si =f Exy-npH w- (1.4.25) nasDigital Gommunications 1044 Pulse Digital Modulation Value of H (f) is given by equation 1.4.21 as, Hf) = tsinc (fr) en inf -~ (1.4.26) Here t is the width of the rectangular pulse used for sampling. The given value of rectangular sampling pulse is 50/3 milliseconds. ie, v= Bios or + = %8 seconds Putting the value of t in equation 1.4.26 we get, Hip = opr sine( SO Je ioosses Put this value of H(f) and f, in equation 1.4.25 si) = 20 5 xy -200% 9 snl SOEs (ince f, = 20 ; s X(f-20rpsine{ SE oss n=-3 0 sn This expression gives the spectrum up to 60 Hz (since n=3) for the sampled signal. tmp Example 1.4.2 : A flat top sampling system samples a signal of maximum 1 Hz with 2.5 Hz sampling frequency. The duration of the pulse is 0.2 seconds. Calculate the amplitude distortion due to aperture effect at highest signal frequency. Also find out the equalization characteristic. Solution : Ik is given that Sampling frequency f, = 25 Hz Maximum signal frequency fia, = 1 Hz Pulse width += 02 sec. By equation 1.4.22, the aperture effect is given by a transfer function H (f) as, H(f) = tsinc(ftye it ftDigital Communications 1-45 Pulse Digital Modulation ‘The magnitude of the above equation is given as, JH (P| = tsinc(f) on (1.4.27) IH (P| = o2sinc(yx02) Aperture effect at highest frequency will be obtained by putting f = fina, =1Hz in above equation ie, 1H()| = 02 sinc (0.2) = 0.18709 ot |H()| = 18.70% ~~» (Ans) From equation 1.4.24 the equalizer characteristic is given as, __k Half = Tsinc (ft) Putting t=(2 second and assuming k=1, the above equation will be, - 1 Ha = 02 sinc(02 f) ~» (1.4.28) This equation is the plot of Hy,(/)Vsf and it represents the equalization characteristic to overcome aperture effect. 1.4.5 Transmission Bandwidth of PAM Signal ‘The pulse duration 't’ is supposed to be very very small compared to time period T, between the two samples. If the maximum frequency in the signal x(t) is 'W' then by sampling theorem, f, should be higher than Nyquist rate or, f, 2 Wor 1 1 . Ts apy since f= 1 and t<< Tssy vo» (1.4.29) If ON and OFF time of the pulse is same, then frequency of the PAM pulse becomes, 1 1 faa 430) : z Thus Fig. 1.4.11 shows that if ON and OFF times of PAM signal are Fig. 4.4.11 Maximum frequency of PAM same, then maximum frequency of sign:Digital Communications 1-46 jigital Modulation | PAM signal is given by equation 1.4.30 ie., 1 fmax = 3g w= (1.4.31) + Bandwidth required for transmission of PAM signal will be equal to maximum frequency fmax given by above equation. This bandwidth gives adequate pulse resolution i.e., Br 2 fax , ” Br 2 w= (1.4.32) . 1 1 . Since t>W ie, Transmission bandwidth of PAM signal : By >>W «+ (14.33) Thus the transmission bandwidth By of PAM signal is very very large compared to highest frequency in the signal x(t). 1.4.6 Disadvantages of PAM 1. As we have seen just now, the bandwidth needed for transmission of PAM signal is very very large compared to its maximum frequency content. 2. The amplitude of PAM pulses varies according to modulating signal. Therefore interference of noise is maximum for the PAM signal and this noise cannot be removed very easily. 3. Since amplitude of PAM signal varies, this also varies the peak power required by the transmitter with modulating signal. — Theory Questions 1. Distinguish between instantaneous sampling, natural sampling and flat top sampling. With functional block diagram explain the working of a circuit that provides flat top sampling. 2. Show that a bandlimited signal of finite energy, whic has no frequency components higher than W Hz may be completely recovered from the knowledge of its samples taken at the rate ‘of 2W samples per second. How the recovered signal will differ in amplitude if samples are taken by (a) Natural sampimg (b) Flat top sampling ? 3. What is aperture effect 7 How it can be reduced 7Digital Communications 1-47 Pulse Digital Modulation 1.5 Other Forms of Pulse Modulation There are two more types of pulse modulation other than PAM : {i) Pulse Duration Modulation (PDM) In this technique the width of the pulse changes according to amplitude of the modulating signal at sampling instant. Fig. 1.5.1 (¢) shows such signal. (il) Pulse Position Modulation (PPM) In this technique the position of the pulse changes according to amplitude of the modulating signal of sampling instant. Fig. 1.5.1(d) shows such signal. x(t) Flat Top PAM (o) (od (a) Fig. 1.5.1 Various pulse modulation methods * Pulse position modulation (PPM) and pulse duration modulation (PDM or PWM) both modulate the time parameter of the pulses. PPM has fixed width pulses where as width of PDM pulses varies. Both the methods ‘are of constant amplitude.Digital Communications 1-48 Pulse Digital Modulation 4.5.1 Generation of PPM and PDM The block diagram of Fig. 1.5.2 (a) shows the scheme to generate PDM and PPM. The corresponding waveforms are shown in Fig. 1.5.2 (b). The scheme of Fig.1.5.2(a) combines both sampling and modulation operation. The sawtooth generator generates the sawtooth signal of frequency f, (je. period T,). The sawtooth signal, also called sampling signal is applied to the inverting input of comparator. ‘Comparator x() © @) ) rae Fig. 4.5.2 Generator of PPM and PDM (a) Block diagram (b) Waveforms The modulating signal x(t) is applied to the noninverting input of the comparator. The output of the comparator is high only when instantaneous value of x(t) is higher than that of sawtooth waveform. Thus the leading edge of PDM signal occurs at the fixed time period ie. kT, the trailing edge of output of comparator depends on the amplitude of signal x(t). When sawtooth waveform voltage is greater than voltage of x() at that instant, the output of comparator remains zero. The trailing edge of the output of comparator (PDM) is modulated by the signal x(f) If the sawtooth waveform is reversed, then trailing edge will be fixed and leading edge will beDigital Communications 1-49 Pulse Digital Modulation modulated. If sawtocth waveform is replaced by triangular waveform, then both leading and trailing edges will be modulated. The pulse duration modulation (PDM) or PWM signal is nothing but output of the comparator. The amplitude of this PDM or PWM signal will be positive saturation of the comparator, which is shown as ‘A' in the waveforms. The amplitude is same for all pulses. To generate pulse position modulation (PPM), PDM signal is used as the trigger input to one monostable multivibrator. The monostable output remains zero untill it is triggered. The monostable is triggered on the falling (trailing) edge of PDM. The output of monostable then switches to positive saturation level 'A'. This voltage remains high for the fixed period then goes low. The width of the pulse can be determined by monostable. The pulse is this delayed from sampling time KT, depending on the amplitude of signal x(t) at kT,. 1.5.2 Transmission Bandwidth of PPM and PDM As can be seen from the waveform, both PPM and PDM possess DC value. The amplitude of all the pulses is same. Therefore nonlinear amplitude distortion as well as noise interference does not affect the detection at the receiver. However both PPM and PDM needs a sharp rise time and fall time for pulses in order to preserve the message information. Rise time should be very very less than T; ie., << T And transmission bandwidth should be, 1 Bre a Thus the transmission bandwidth of PPM and PDM is higher than PAM. The power requirement of PPM is less than that of PDM because of short duration pulses. It can be further reduced by transmitting only edges rather than pulses. Transmission bandwidth of PDM and PPM : By at - (1.5.1) i 1.5.3 Gomparison between Various Pulse Modulation Methods Following table shows the comparison among various pulse modulation techniques.Digital Communications 1-50 Pulse Digital Modulation Pulse Width/Duration Pl Modulation osition Modul: Waveform Waveform . Time oe ‘Time 2 | Amplitude of the pulse is ‘Width of the pulse is ‘The relative position of the Proportional to amplitude of Proportional to amplitude of pulse is proprotional to the modulating signal. modulating signal. amplitude of modulating signal. The bandwidth of the Bandwidth of transmission transmission channel depends] channel depends on rise time of on width of the pulse. the pulse. Bandwidth of transmission channel depends on rising time of the pulse. The instantaneous power of The instantaneous power of the the transmitter varies. The instantaneous power of transmitter varies. the transmitter remains constant. 5 | Noise interference is high. Noise interference is minimum. Noise interference is minimum. 6 | System is complex. Simple to implement. Simple to implement. 7 | Similar to amplitude Similar to frequency modulation. modulation. Similar to phase modulation. I Table 1.5.1 Comparison of PAM, PPM and PDM im Example 1.5.1; For a PAM transmission of voice signal with W = 3 kHz. Calculate Br if f, =8kHz and t=01T,. Solution : T, is given as, T; = z mar 1 8x10° From equation 1.5.1, the transmission bandwidth By is given as, 1 1 By = = 2>—1 __ = 40 kHz 25, ol 8x103 S t= O1T= secDigital Communications 4-51 Pulse Digital Modulation mm Example 1.5.2: For the signal given in example 1.5.1, if the rise time is 1% of the width of the pulse, find out the minimum transmission bandwidth needed for PDM and PPM. Solution : In example 15.1 we oblained the pulse width «== O1 sec. The rise time 10° is given as 1% of width of pulse i.e., 0. M_ Q01 = 125%1077 see 8x10 t= txO01l= We know that transmission bandwidth is given as, 1 1 By 2 =—-2—————— TSB, 351.25%107 24 MHz Theory Questions 1. Compare PAM, PPM and PDM. 2. Explain the scheme to generate PDM and PPM. 3. Explain how to generate PAM signal for various types of sampling techniques. 1.6 Bandwidth Noise Trade-off The noise analysis of PPM and FM have similar results as follows : i) For both systems, the figure of merit is proportional to square of the ratio Br (v7) ii) As the signal to noise ratio is reduced, both the systems exhibit threshold effect. * With digital pulse modulation, the better noise performance than square law can be obtained. + The digital pulse modulation such as pulse code modulation gives negligible noise effect by increasing the average power in binary PCM signal. + With PCM, the bandwidth noise trade-off can be related by exponential law.Digital Communications 4-52 Pulse Digital Modulation 1.7 Time Division Multiplexing (PAM/TDM System) In PAM, PPM and PDM the pulse is present for short duration and form most of the time between the two pulses, no signal is present. This free space between the pulses can be occupied by pulses from other channels. This is called Time Division Multiplexing (TDM). It makes maximum utilization of the transmission channel. 1.7.1 Block Diagram of PAM / TDM Fig.1.7.1 (a) shows the block diagram of a simple TDM system and Fig, 1.7.1 () shows the waveforms of the system. The system shows the time division multiplexing of ‘N’ PAM channels. Each channel to be transmitted is passed through the lowpass filter. The outputs of the lowpass filters are connected to the rotating sampling switch or commutator. It takes the sample from each channel per revolution and rotates at the rate of f,. Thus the sampling frequency becomes j,. The single signal composed due to multiplexing of input channels is given to the transmission channel. At the receiver the decommutator separates (decodes) the time multiplexed input channels. These channel signals are then passed through lowpass reconstruction filters. amc] lie (be) Fig. 1.7.1 TDM system (PAM/TDM system) (a) Block diagram —(b) WaveformsDigital Communications 1+53 Pulse Digital Modulation If the highest signal frequency present in all the channels is ‘W’, then by sampling theorem the sampling frequency f, should be, fz Ww ww (LT) ‘Therefore the time space between successive samples from any one input will be T= t ww (1.7.2) 1 WwW s» (1.73) Thus the time interval T, contains one sample from each input. This time interval is called frame. Let there be 'N’ input channels. Then in each frame there will be one sample from each of the ‘N' channels. That is one frame of T, seconds contain total ‘N’ samples, Therefore pulse to pulse spacing between two samples in the frame will be equal to Z, s Ts 2. Spacing between two samples = % (174) n™ channel pulse (n+1)" channel pulse TIN TN Fig. 1.7.2 Calculation of number of pulses per second in TDM From the above figure we can very easily calculate the number of pulses per second or pulse frequency as, 1 Number of pulses per second= <= Siween two pulses 1 iN azDigital Communications 1-54 Pulse Digital Modulation We know that T, = :. Number of pulses per second = ee Nf wu (17.5) These number of pulses per second is also called signalling rate of TDM signal and is denu.d by 'r ie, Signalling rate = r=N f, + (17.6) Since fi 2 2, then signaling rate becomes, Signalling rate in PAM/TDM system : r 2 2NW we (17.7) The RF transmission of TDM needs modulation. That is TDM signal should modulate some carrier. Before modulation, the pulsed signal in TDM is converted to baseband signal. That is pulsed TDM signal is converted to smooth modulating waveform x; (f); the baseband signal that modulates the carrier. The baseband signal x () passes through all the individual sample values baseband signal is obtained by passing pulsed TDM signal through lowpass filter. The bandwidth of this lowpass filter is given by half of the signalling rate. ic., B, = freiny, (178) . Transmission bandwidth of TOM channel will be equal to bandwidth of the lowpass filter, By = 3N f from above equation If sampling rate becomes equal to Nyquist zate i.e., f;(min) = Nyquist rate = 2W, then By = Nw Minimum transmission bandwidth of TDM channel ; By = NW wn (1.7.9) This equation shows that if there are total 'N' channels in TDM which are bandlimited to 'W' Hz, then minimum bandwidth of the transmission channel will be equal to NW.Digital Communications 1-55 Pulse Digital Modulation ump Example 1.7.1: ‘N’ number of independent baseband signal samples are transmitted over a channel of bandwidth = f. Hz. If each sample is bandlimited to fj, Hz, show that the channel need not have a bandwidth larger than Nf,, in order to avoid crosstalk. Solution : Here we have to show that, the bandwidth of the transmission channel in PAM/TDM system should be minimum of Nj,, in order to avoid crosstalk between successive channel samples, From Fig. 1.7.1 we know that samples from various channels are interlaced one after another. The figure is reproduced here for convenience. Impulses from various channel Fig. 1.7.6 Marker pulses for synchronization in TDM The above figure shows that a marker pulse is inserted at the end of the frame. Because of the marker pulse, synchronization is obtained but number of channels to be multiplexed is reduced by one (i.e. N-1 channels can be multiplexed).jigital Communications 1-59 Pulse Digital Modulation 41.7.3 Crosstalk and Guard Times We have seen that RF transmission of TDM needs modulation. Hence the TDM signal is converted to a smooth modulating waveform (i.e. baseband signal) by passing through a baseband filter. Fig. 1.7.7 shows the TDM transmission with baseband filtering and the baseband waveform. (a) (by iad in Fig. 1.7.7 (a) TDM transmission with baseband filtering (b) Baseband waveform Thus the baseband waveform passes through the values of all the individual samples, The baseband filtering gives rise to interchannel crosstalk from one sample value to the next. In other words crosstalk means the individual signal sample amplitudes interfere with each other. This interference can be reduced by increasing the distance between individual signal samples. The minimum distance between the individual signal samples to avoid crosstalk is called guard time. Now let us derive an expression for guard time in TDM. Let us assume that the transmission channel acts like a first order lowpass filter with 3-dB bandwidth 'B'. And assume that every pulse transmitted in TDM is a rectangular pulse. When this pulse is applied to the channel, its response is shown in Fig. 1.7.8 (b). In the Fig. 1.7.8 observe that even after the pulse is removed, the response of the channel decays from its value of ‘A’. The response then decays for long period. The guard time T, represents the minimum pulse spacing. At the end of guard time, the value of pulse tail is less than A,;,,where it is given as, Ag = Aes sx» (1.7.16)Digital Communications 1-60 Pulse Digital Modulation This decay gives rise to crosstalk Aa Ty Xamon Fig. 1.7.8 (a) A rectangular pulse applied to the lowpass channel $ & Response of the lowpass channel to the rectangular pulse And the cross talk reduction factor is defined as, 2 Ka = rotes( Se) = -54.5BT, dB wu. (L7.A7) This equation shows that to keep cross talk below —30dB,Ty should be greater than J. The guard times are very much important particularly in pulse duration or pulse position modulation techniques. imp Example 1.7.2 : Twelve different message signals, ench of bandwidth 10 kHz are to be multiplexed and transmitted. Determine the minimunt bandwidth required for PAM/TDM system. Solution : Here the number of channels N = 12. Bandwidth of each channel f,, = 10 kHz Minimum channels bandwidth to avoid crosstalk in PAM/TDM system is, fo = Nhu (By equation 1.7.15) 12x10kHz = 120 kHz imp Example 1.7.3: Twenty four voice signals are sampled uniformly and then time division multiplexed. The highest frequency component for each voice signal is 3.4 kHz.Digital Communications 4-61 Pulse Digital Modulation i) If the signals are pulse amplitude modulated using Nyquist rate sampling, what is the minimum channel bandwidth required? ii) If the signals are pulse code modulated with an 8 bit encoder, what is the sampling rate ? The bit rate of system is 1.5x10° bits/sec. Solution: i) We know that if N channels are time division multiplexed, then minimum transmission bandwidth is given as, By = NW Here W is the maximum frequency in the signals. ‘ By = 24x34 kHz=81.6kHz v= (Ans) ii) The signalling rate of the system is given as, r = 1.5x10® bits/sec Since there are 24 channels, the bit rate of an individual channel is, 1.510% 24 62500 bits/sec t (one channel) = Since each sample is encoded using 8 bits, the samples per second will be, (one channel) bits/ see Sample/see = Dits/ sample Samples per seconds is nothing but sampling frequency. . _ 62500 bits/ sec ne fi = Biiscample 7812.5 Hz or samples per second .. (Ans) mm Example 1.7.4: Twenty four voice signals are sampled uniformly and then time division multiplexed. The sampling operation uses flat samples with 1 psec duration. The multiplexing operation provides for synchronization by adding an extra pulse of Iusec duration. Assuming sampling rate of 8 kHz, calculate spacing between successive pulses of multiplexed signal and setup a scheme for accomplishing a multiplexing requirement. Solution ; There are 24 voice signal pulses plus one synchronization pulse. Hence there are total 25 pulses. Sampling rate is 8 kHz. Hence duration of one frame will be, 1_ 1 T= 57 a0 125 psecDigital Communications 1-62 Pulse Digital Modulation Thus in 125 psec time there are 25 pulses at uniform distances. This is illustrated in Fig. 1.7.9 Multiplexing of 24 voice signals As shown in above figure, the pulses are separated by 7H ~ 5 ys. Width of the pulse is 1s. Hence, Spacing between pulses = 5 - 1 = 4 psec. Fig, 1.7.10 shows the multiplexing scheme. x x Voice Multiplexer Mutiplexed signals Hold signa! . “ cirouit 200,000 samples per second %e Synchronization pulse fg= 8 kbtz 24 see Fig. 1.7.10 PAM-TDM system Theory Questions 1. Explain PAM/TDM system for “N’ number of channels. 2. Derive the relation for minimum bandwidth to transmit ‘N’ channels in PAM/TDM system such that crosstalk is avoided. 3. Explain the importance of synchronization in TDM systems.Digital Communications 1-63 Unsolved Examples 1. Twenty four voice signals are sampled uniformly and then time division multiplexed, the sampling operation uses flat top samples with 1 psec duration. The synchronization is provided by adding an extra pulse of 1 usec duration. The highest frequency component of each voice signal is 3.4 kHz. (a) For sampling rate of 8 kHz, calculate spacing between successive pulses of multiplexed signal. (b) For Nyquist rate repeat part (a). 1.8 Pulse Code Modulation 1.8.1 PCM Generator The pulse code modulator technique samples the input signal x(!) at frequency f, 22. This sampled "Variable amplitude’ pulse is then digitized by the analog to digital converter. The parallel bits obtained are converted to a serial bit stream. Fig1.8.1 shows the PCM generator. vdigits {| Parallel | PCM [les r=vf Binary ‘encoder er) converter f,22W Fig. 1.8.1 PCM generator In the PCM generator of above figure, the signal x({) is first passed through the lowpass filter of cutoff frequency 'W' Hz. This lowpass filter blocks all the frequency components above 'W' Hz. Thus x(t) is bandlimited to ‘W' Hz. The sample and hold circuit then samples this signal at the rate of f,. Sampling frequency f, is selected sufficiently above Nyquist rate to avoid aliasing ie., fz Ww In Fig. 1.8.1 output of sample and hold is called x(n7,). This x(n T;) is discrete in time and continuous in amplitude. A q-level quantizer compares input x(1T,) with its fixed digital levels. It assigns any one of the digital level to x(nT,) with its fixed digital levels. It then assigns any one of the digital level to x(n T,) which results in minimum distortion or error. This error is called quantization error. Thus output of quantizer is a digital level called x, (nT,).Digital Communications 1-64 Pulse Digital Modulation Now coming back to our discussion of PCM generation, the quantized signal level x, (nT,) is given to binary encoder. This encoder converts input signal to ‘v' digits binary word. Thus x, (17,) is converted to 'V' binary bits. The encoder is also called digitizer. It is not possible to transmit each bit of the binary word separately on transmission line. Therefore ‘v' binary digits are converted to serial bit stream to generate single baseband signal. In a parallel to serial converter, normally a shift register does this job. The output of PCM generator is thus a single baseband signal of binary bits. An oscillator generates the clocks for sample and hold an parallel to serial converter. In the pulse code modulation generator discussed above ; sample and hold, quantizer and encoder combinely form an analog to digital converter. 1.8.2 Transmission Bandwidth in PCM Let the quantizer use “o’ number of binary digits to represent each level. Then the number of levels that can be represented by ‘0’ digits will be, q= 2 ow (18.1) Here ‘9’ represents total number of digital levels of q-level quantizer. For example if v =3 bits, then total number of levels wili be, q = 23 =8 levels Each sample is converted to 'v' binary bits, ise. Number of bits per sample = v We know that, Number of samples per second = f, . Number of bits per second is given by, (Number of bits per second) = (Number of bits per samples) x (Number of samples per second) = 0 bits per sample xf, samples per second... (1.8.2) The number of bits per second is also called signaling rate of PCM and is denoted by rie, Signaling rate in PCM. fe w= (183) Here f, 22W.Digital Communications 1-65 Pulse Digital Modulation Bandwidth needed for PCM transmission will be given by half of the signaling rate Le, 1 Br2gr wo (18.4) Transmission Bandwidth of PCM : Brzhuf Since f, 22W + (1.8.5) Br2uW + (1.86) 1.8.3 PCM Receiver Fig. 1.8.2 (a) shows the block diagram of PCM receiver and Fig, 1.8.2 (b) shows the reconstructed signal. The regenerator at the start of PCM receiver reshapes the pulses and removes the noise. This signal is then converted to parallel digital words for each sample. Fig. 1.8.2 (a) POM receiver (b) Reconstructed waveformDigital Communications 41-66 Pulse Digital Modulation The digital word is converted to its analog value x, (f) along with sample and hold. This signal, at the output of S/H is passed through lowpass reconstruction filter to get yp (0. As shown in reconstructed signal of Fig. 1.8.2 (b), it is impossible to reconstruct exact original signal x() because of permanent quantization error introduced during quantization at the transmitter. This quantization error can be reduced by increasing the binary levels. This is equivalent to increasing binary digits (bits) per sample. But increasing bits ‘v' increases the signaling rate as well as transmission bandwidth as we have seen in equation 183 and equation 186. Therefore the choice of these parameters is made, such that noise due to quantization error (called as quantization noise) is in tolerable limits. 1.8.4 Uniform Quantization (Linear Quantization) We know that input sample value is quantized to nearest digital level. This quantization can be uniform or nonuniform. In uniform quantization, the quantization step or difference between two quantization levels remains constant over the complete amplitude range. Depending upon the transfer characteristic there are three types of uniform or linear quantizers as discussed next. 1.8.4.1 Midtread Quantizer The transfer characteristic of the midtread quantizer is shown in Fig. 1.8.3. As shown in this figure, when an input is between - 6/2 and + 6/2 then the quantizer output is zero. i.e., For = 8/2 < x(nT,) <8/2; x, (nT) = 0 Here ‘8’ is the step size of the quantizer. for 8/2 S x(OT,) < 36/2; xq (nT) =8 Similarly other levels are assigned. It is called midtread because quantizer output is zero when x{nT,) is zero. Fig.1.8.3 (b) shows the quantization error of midtread quantizer. Quantization error is given as, = x, (XT) - xT) (1.8.7) In Fig. 1.8.3 (b) observe that when x(nT,) = 0, x,(nT,) = 0. Hence quantization error is zero at origin. When x(nT,) = 8/2, quantizer output is zero just before this level. Hence error is 5/2 near this level. From Fig. 1.8.3 (b) it is clear that, -8/2 < €<6/2 .- (1.88) Thus quantization error lies between - 8/2 and + 8/2. And maximum quantization ertor is, maximum quantization error, € max = \3| -. (1.8.9) 2Digital Communications 1-67 Pulse Digital Modulation Ideal transfer characteristic + passes through zer0 cee { 4 Staircase approximation i + Input x{n74) | 782 | bobo Input x(0T,) nee eee i J Fig. 1.8.3 (a) Quantization characteristic of midtread quantizer (b) Quantization error 4842 Midriser Quantizer The transfer characteristic of the midriser quantizer is shown in Fig. 1.8.4. In Fig. 1.8.4 observe that, when an input is between 0 and 8, the output is 6/2. Similarly when an input is between 0 and - 8, the output is - 8/2. ic., For 0 < x (nT) <5; xq (nT) = 5/2 -8s x(WT) <0; (nt) =-5/2 Similarly when an input is between 38 and 4 8, the output is 7 5/2. This is called midriser quantizer because its output is either + 8/2 or - 6/2 when input is zeroDigital Communications 1-68 Pulse Digital Modulation [ 4 7 eal vanster eharaclerisie iL Fig. 1.8.4 (a) Transfer characteristic of midriser quantizer {b) Quantization error Fig. 1.84 (b) shows the quantization error in midriser quantization. When input x(nT,) = 0, the quantizer will assign the level of §/2. Hence quantization error will be, & = x, (nT,) - x (nT) = 8/2-0=8/2 Thus the quantization error lies between - 8/2 and + §/2. ie, -8/2 s esd/2 va (1.8.10) ‘And the maximum quantization error is, emax = | -» (1.8.11)Digital Communications 1-69 Pulse Digital Modulation In both the midriser and midtread quantizers, the dotted line of unity slope pass through origin. It represents ideal nonquantized input output characteristic. The staircase characteristic is an approximation of this line. The difference between the staircase and unity slope line represents the quantization error. 1.8.43 Biased Quantizer Fig. 1.8.5 shows the transfer characteristic of biased uniform quantizer. TT input T,) ee Fig. 1.8.5 (a) Biased quantizer transfer characteristic {b) Quantization error The midriser and midtread quantizers are rounding quantizers. But biased quantizer is truncation quantizer. This is clear from above diagram. When input is between 0 and 8, the output is zero. i.c., for Os x@T) <8; x, @T)=0Digital Communications 4-70 Pulse Digital Modulation Similarly, for -5 3 x(nT,)<0; xq (nT) =-8 Fig. 1.8.5 shows quantization error. When input is 6, output is zero. Hence quantization error is, e Xq (AT) - x(nT,) = 0-6=-5 Thus the quantization error lies between 0 and - & ie., -8s cso vw (1.8.12) And the maximum quantization error is, tmax = 18] ou (1.8.13) Thus the quantization error is more in biased quantizer compared to midriser and midtread quantizers. The unity slope dotted line passes through origin. It represents ideal nonquantized transfer characteristic. The difference between staircase and dotted line gives quantization error. 1.8.5 Quantization Noise and Signal to Noise Ratio in PCM. 4.8.5.1 Derivation of Quantization ErroriNoise or Noise Power for Uniform (Linear) Quantization Step 1: Quantization Error Because of quantization, inherent errors are introduced in the signal. This error is called quantization error. We have defined quantization error as, & Xq(mT,)~ x(n Te) o» (1.8.14) Step 2: Step size Let an input x(n f,) be of continuous amplitude in the range —X qx 10 + Xmax- From Fig. 1.8.4 (a) we know that the total excursion of input x(11T,) is mapped into ‘q' levels on vertical axis. That is when input is 45, output is 25 and when input is -48, output is fa That is +Xmax represents 2 Sand ~xmax represents ~56, Therefore the total amplitude range becomes, Total amplitude range = X;max —(~Ximax) = 2xmax . (1.8.15) If this amplitude range is divided into ‘q' levels of quantizer, then the step size ‘8 is given as, o» (1.8.16)Digital Communications 1-71 Pulse Digital Modulation If signal x(f) is normalized to minimum and maximum values equal to 1, then Xm = 1 xa = -1 ve (1.8.17) Therefore step size will be, b= : (for normalized signal) (1.8.18) Step 3: Pdf of Quantization error If step size ‘8 is sufficiently small, then it is reasonable to assume that the quantization error '‘e’ will be uniformly distributed random variable. The maximum quantization error is given by equation 1.8.11 as, va (1.8.19) fmax = ie. 5 w= (1.8.20) ‘Thus over the interval (- 5.3) quantization error is uniformly distributed random variable. F004 4 E-al~ 2 D fa) tb) Fig. 1.8.6 (a) Uniform distribution (b) Uniform distribution for quantization error In above figure, a random variable is said to be uniformly distributed over an interval (a, b). Then PDF of 'X’ is given by, (from equation of Uniform PDF).Digital Communications 1-72 Pulse Digital Modulation 0 for xSa 1 feos = [Poa for acxsb 0 for x>b vw (1.8.21) Thus with the help of above equation we can define the probability density function for quantization error ‘e’ as, 0 for es® 1 3 & fle) = {8 a a ° for ef ow (1.8.22) Step 4 : Noise Power From Fig, 1.84 (b) we can see that quantization error ‘e! has zero average value. That is mean “m," of the quantization error is zero. ‘The signal to quantization noise ratio of the quantizer is defined as, 'S_ _ Signal power (normalized) (1.8.23) N ~ Noise power (normalized) o If type of signal at input ie., x(t) is known, then it is possible to calculate signal power. The noise power is given as, v2. Noise power = male on 1.8.24) Here V2... is the mean square value of noise voltage. Since noise is defined by random variable ‘e' and PDF f, (€), its mean square value is given as, mean square value = Ele?] = €2 (1.8.25) ‘The mean square value of a random variable "X’ is given as, X? = E[X2]= J x? fc(x)dx By definition a» (1.8.26) Here Ele?] = fe? f @de w= (1.8.27)Digital Communications 1*73 Pulse Digital Modulation From equation 1.8.22 we can write above equation as, Ele] = - @/23 = one = on (1.8.28) «+ From equation 1.8.25, the mean square value of noise voltage is, 2 V2ye = mean square value = 5 When load resistance, R =1 ohm, then the noise power is normalized i.e, ve, Noise power (normalized) = te [with R =1 in equation 1.8.24] ~ &/2_8 “ST Thus we have, Normalized noise power 2 or Quantization noise power = £ ; For linear quantization. or Quantization error (in terms of power) ss (1.8.29) 1.8.5.2 Derivation of Maximum Signa! to Quantization Noise Ratio for Linear Quantization From equation 1.8.23 signal to quantization noise ratio is given as, S$ _ Normalized signal power N © Normalized noise power _ Normalized signal power . (1830) (& /12) The number of bits 'y' and quantization levels ‘q' are related as, qeP +» (1.8.31)Digital Communications 474 Pulse Digital Modulation Putting this value in equation 1.8.16 we have, p= 2m o w» 1.8.32) Putting this value in equation 1.8.30 we get, __ Normalized signal power = Normalized signal power Ce) +12 ra Let normalized signal power be denoted as ‘P’. & N S$. PL Pym N~ Ody 1 a Qa 12 This is the required relation for maximum signal to quantization noise ratio. Thus, Maximum signal to quantization noise ratio : S= 3P at (1.8.33) Xmax wom This equation shows that signal to noise power ratio of quantizer increases exponentially with increasing bits per sample. If we assume that input x(!) is normalized, ie., Sua = 1 -» (1.8.34) Then signal to quantization noise ratio will be, # = 3x 2x P -- (1.8.35) If the destination signal power ‘P" is normalized, i.e, Psi -- 1.8.36) Then the signal to noise ratio is given as, é < 3x2” += (1.8.37) Since Xp, =landPS1, the signal to noise ratio given by above equation is normalized. Expressing the signal to noise ratio in decibels, s s i ' (5 }i6 = 0109 (SJ sine power ato w 10 log yg [3x2"] (48+60)dB inDigital Communications 1-75 Pulse Digital Modulation Thus, Signal to Quantization noise ratio for normalized values of power : (3 jwsasts waB 'P' and amplitude of input x(t) v= (1.8.38) ym Example 1.8.1 : Derive the expression for signal to quantization noise ratio for PCM system that employs linear quantization technique. Assume that input to the PCM system is a sinusoidal signal. OR ‘A PCM system uses a uniform quantizer followed by a v bit encoder. Show that rns signal to quantization noise ratio is approximately given by (1.8 + 60) dB. Solution : Assume that the modulating signal be a sinusoidal voltage, having peak amplitude A,,. Let this signal cover the complete excursion of representation levels. The power of this signal will be, y2 Ps =z Here V = rms value = [A,, / 2] 11839) When R = 1, the power P is normalized, ie., 2 Normalized power : Ps Bs with R =1 in above equation. :. Signal to quantization noise ratio is given by equation 1.8.33 as, Ss _ 3p = SKID N max Ak Here P = AB and max Putting these values in the above equation,Digital Communications 1-76 Pulse Digital Modulation Expressing signal to noise power ratio in dB, s s (= } = Wlogiy (=): 10 10g 4p (1.52) = 1Wlog yg (1.5) +10 log 1p 2” = 1.76+20x10x03 = (1.8.40) ms Example 1.8.2: A Television signal with a bandwidth of 4.2 MHz is transmitted using binary PCM. The number of quantization levels is 512. ‘Calculate, 4) Code word length —_ it) Transmission bandwidth iit) Final bit rate i) Output signall to quantization noise ratio, (March-2003, 10 Marks] Solution : The bandwidth is 4.2 MHz, means highest frequency component will have frequency of 4.2 MHz ie., W = 42 MHz Quantization levels q = 512 i) Number of bits and quantization levels are related in binary PCM as, 2» ie. 2 vlog 2 __ 10g 512 * og? = 9 bits ~~ (Ans) Thus the code word length is 9 bits. if) From equation 1.8.6 the transmission channel bandwidth is given as, By 2 wW = 9x42x106 Hz By 2 378 MHz +» (Ans) iii) The final bit rate will equal to signaling rate. From equation 1.8.3 signaling rate is given as, raofDigital Communications 41-77 Pulse Digital Modulation Sampling frequency f, 2 2W by sampling theorem. f, = 2x42MHz since W = 4.2 MHz “ f 2 84 MHz Putting this value of '/," in equation for signaling rate, 7 = 9x84x106 = 756x106 bits/sec vs (Ans) From equation 1.8.4 transmission bandwidth is also obtained as, 1 By > 5r 2 756x108 bits/sec or By 2 37.8 MHz which is same as the value obtained earlier. iv) The signal to noise ratio s (x}# S 48460 dB S 4846x9 S $88 dB ++ (Ans) ‘a> Example 1.8.3 : The bandwidth of signal input to the PCM is restricted to 4 KHz. ‘The input varies from -38 V to + 3.8 V and has the average power of 30 mW. The required signal to noise ratio is 20 dB. The modulator produces binary output. Assume uniform quantization. 3) Calculate the member of bits required per sarnple. ti) Outputs of 30 such PCM coders are time multiplexed. What is the minimum required transmission bandwidth for the multiplexed signal ? Solution : The given value of signal to noise ratio is 20 dB. 10 10g 19 (jy }-2008 ® — Zio YS 5 " 100Digital Communications 1-78 Pulse Digital Modulation i) The signal to quantization noise ratio is given as, Ss _ 3P.2 . 2s By equation 1.8.33 NO Here Xn = 38V,P= 30m and $= 100 -3 22 . too = 3%30%102 2 (38)? ‘ v= 698 bits = 7 bits ++ (Ans) ii) The maximum frequency is, Ws 4kHz The transmission bandwidth is given by equation 1.8.6 as, By 2 oW Since there are 30 PCM coders which are time multiplexed, the transmission bandwidth will be, ie, By 2 30x0-W 2 30x7*4 kHz % 840 kHz s+ (Ans) Signaling rate is two times the transmission bandwidth as given by equation 1.8.4 Signaling rate r = 840x2 bits/sec = 1680 bits/sec. wm Example 1.8.4: The information in an analog signal voltage waveform is to be transmitted over 2 PCM system with an accuracy of £01% (full scale). The analog voltage waveform has a bandwidth of 100 Hz and an amplitude range of -10 to +10 volts. a) Determine the maximum sampling rate required. ‘b) Determine the number of bits in each PCM word. ¢) Determine minimum bit rate required in the PCM signal. d) Determine the minimum absolute channel bandwidth required for the transmission of the PCM signal.Digital Communications 1-79 Pulse Digital Modulation Solution ; Here an accuracy is given as + 01%. That is quantization error should be £01%. or the maximum quantization error should be +01% or Emax = £01% = +0001 The maximum quantization error for an uniform quantizer is given as, _ (3 r " 2 2 = That is Step sizeé = 2x0.001 = 0.002 The step size, number of levels and maximum value of the signal are related as (By equation 1.8.16) b= 2mm Here |r yax| = 10 volts 2+ Putting values of 6 and Xp.., 0.002 = 2x10 4 20 or 0 = Sa = 10,000 That is the number of levels are 10,000. a) The maximum frequency in the signal is 100 Hz i.e., W = 100 Hz By sampling theorem,minimum sampling frequency should be, £2 > 21002200 Hz ves (AS) b) We know that minimum 10,000 levels should be used to quantize the signal. If binary PCM is used, then number of bits for each samples can be calculated as, qe v Here, 4 = number of levelsDigital Communications 1-81 Pulse Digital Modulation Thus the maximum message bandwidth is 3.57 MHz. b) The modulating wave is sinusoidal. For such signal, the signal to quantization noise ratio is given by (3) dB = 18460 By equation 1.8.40 = 1846x7 (putting for v=7) = 438 4B w+ (Ans) i> Example 1.8.6 : The information in an analog waveform with maximum frequency fy =3KHz is to be transmitted over an M-level PCM system where the number of pulse levels is M=16 The quantization distortion is specified not to exceed 1% of peak to peak analog signal. i) What is the maximum number of bits per sample that should be used in this PCM system 2 ii) What is the minimum sampling rate and what is the resulting bit transmission rate? Solution : i) Since the number of levels given here are M =16, q = M=16 ‘Then bits and levels in binary PCM are related as, qa? v= bits 16 = 2” or ved ii) Since fr = W=3kHz minimum f, 2 2W by sampling theorem * fe & 2x3kHz or > 6kHz ws (Ans) Bit transmission rate or signaling rate is given by equation 1.83 as, revf > 4x6x10% 2 24x10 bits per second a+ (Ans)Digital Communications 1-82 Pulse Digital Modulation imap Example 1.8.7 : A Signal of bandwidth 3.5 kHz is sampled quantized and coded by a PCM system. The coded signal is then transmitted over a transmission channel of supporting a transmission rate of 50 k bits/sec. Calculate the maximum signal to noise ratio that can be obtained by this system. The input signal has peak to peak value of 4 volts and rms value of 0.2 V. Solution : The maximum frequency of the signal is 3.5 kHz, ive. W = 3.5kHz Therefore sampling frequency will be f= Ww 2 2x3.5kHz 2 7kHz The signaling rate is given by equation 18.3 as, rs vf Putting values of r =50x10° bits/sec and f, 27x10? Hz in above equation. 50x103 < v-7x105 v2 7.142 bits a v= &bits The rms value of the signal is 0.2 V. Therefore the normalized signal power will 2 Normalized signal power = or [R = 1 for normalized power] ie, P = 004W The maximum signal to noise ratio is given by S$ _ 3p.28 N xhax Putting the values of P =(04, 0 = 8andx,,.¢ =2 in above equation, S _ 3x004%27*8 N a = 196608 = 33 dB wm Example 1.8.8 : A signal x(0) is uniforntly distributed in the range + 2mqx. Calculate maximum signal to noise ratio for this signal.Digital Communications 1-83 Pulse Digital Modulation Solution : The signal is uniformly distributed in the range +x,,,- Therefore we can write its PDF (using the Standard Uniform Distribution) as, fx) = 0 for x<—Xmay = t for yay << Xmay Pinan max = 0 for x > x, Fig. 1.8.7 shows this PDF, 1 1 MA max Xmax Fax) Fig. 1.8.7 PDF of a uniformly distributed random variable The mean square value of a random variable X is given as, 2 = J? feear Therefore mean square value of x(t) will be, __ Pe ye Os ax nee ax -! eet max 5 —Xmax = Tix 3 2 The signal power P = xe Isince R =1]Digital Communications 1-84 Pulse Digital Modulation = Thx 3 Step size bs Ein By equation 1.8.16 3 : ‘mn = 82 q? “Normalized signal power, P= L==1. = 2 Normalized noise power = e By equation 1.8.29 a ; _. 5 _ Normalized signal power +-Signal to noise power ratio +7 = Sie oralized noise ‘power Ld eC B12 Since q= 2°, above equation will be, SL ow N Ss «(30 10 log yp (2) dB ~ 6v This is the required expression for maximum value of signal to noise ratio. ump Example 1.8.9 ; Consider an audio signal comprised of the sinusoidal term s(t) =3.cos (00m!) i) Find the signal to quantization noise ratio when this is quantized using 10 bit PCM. ii) How many bits of quantization are needed to achieve a signal to quantization noise ratio of atleast 40 dB? Solution ; Here s(t) = 3.cos (6007!) That is sinusoidal signal applied to the quantizer. i) Let us assume that peak value of cosine wave defined by s(t) covers the complete range of quantizer. ie. Am = 3V covers complete range.Digital Communications 1-85 Pulse Digital Modulation We know that signal to noise ratio for sinusoidal signal is given by (x je = 18+60 Here 10 bit PCM is used ie., v= W (& a = 18+6x10 = 61.8 dB ii) For sinusoidal signal again we will use the same relation. i.e. : Sup = ie. (x x = 18+60 dB To get signal to noise ratio of at least 40 dB we can write above equation as, 18+6v > 40dB v 2 6.36 bits =7 bits Thus at least 7 bits are required to get signal to noise ratio of 40 dB. map Example 1.8.10: A 7 bit PCM system employing uniform quantization has an overall signaling rate of 56 k bits per second. Calculate the signal to quantization noise ratio that would result when its input i a sine wave with peak to peak amplitude equal to 5. Calculate the dynamic range for the sine wave inputs in order that the signal to quantization noise ratio may be less than 30 dBs. What is the theoretical maximum frequency that this system can handle ? Solution: The number of bits in the PCM system are v = 7 bits Assume that 5 V peak to peak voltage utilizes complete range of quantizer. Then we can find the signal to quantization noise ratio as, (5 \eo = 18460 dB =18+6x7 43.8 dB i By equation 1.8.3 signaling rate is given as, re of,Digital Communications 1-86 Pulse Digital Modulation Putting r =56% 103 bits/second and v=7 bits in above equation we get, 56x103 = 7+f, .-Sampling frequency, f, = 8103 Hz By sampling theorem, f, = 2W :-Maximum frequency that can be handled is given as, Ws bs Soo W s 4000 Hz (Ans) map Example 1.8.11 : The bandwidth of TV video plus audio signal is 4.5 MHz. If the signal is converted to PCM bit stream with 1024 quantization levels, determine the number of bits/sec generated by the PCM system, Assume that the signal is sarapled at the rate of 20% above nyquist rate. If above linear PCM system is converted to companded PCM, will the output bit rate change? Justify. Solution : The given data is, W = 45 MHz q = 1024 levels The Nyquist rate is, Nyquist rate = 2W = 2x 45 = 9 MHz The sampling rate is 20% above the nyquist rate. i.e. Sampling rate, f, = 12 x9 = 108 MHz We know that quantization levels q and number of bits v are related as, q=2 1024 = 2 o v = 10 bits The number of bits/sec generated by PCM system is called bit rate or signaling rate. ie., Signaling rate, r = vf, 10 x 10.8 x 10° bits/sec. 0 108 x10 bits / sec.Digital Communications 1-87 Pulse Digital Modulation The output bit rate does not change if linear PCM is converted into companded PCM. Companded PCM is used to improve the signal to noise ratio. 1.8.6 Nonuniform Quantization In nonuniform quantization, the step size is not fixed. It varies according to certain law or as per input signal amplitude. Fig. 1.8.8 shows the transfer characteristic and error in nonuniform quantization. ‘Out aT) ‘Small quantization error = atlew input levels rol ine dot +++ it cer im Fig. 1.8.8 (a) Nonuniform quantization transfer characteristic {b) Quantization error In this figure observe that step size is small at low input signal levels. Hence quantization error is also small at these inputs. Therefore signal to quantization noise power ratio is improved at low signal levels. Stepsize is higher at high input levels. Hence signal to noise power ratio remains almost same throughout the dynamic range of quantizer.Digital Communications 1-88 Pulse Digital Modulation 4.8.61 Necessity of Nonuniform Quantization In uniform quantization, the quantizer has a linear characteristics as we have seen in Fig. 1.84 (a). The step size also remains same throughout the range of quantizer. Therefore over the complete range of inputs, the maximum quantization error also remains same. From equation 1.8.11 the quantization error is given as, Maximum quantization error = € na, 5 .. (1.8.41) From equation 1.8.16 step size’S' is given as, j= 2mm 9 If x(0) is normalized, its maximum value ie. Xmay =1- ae vss (1.8.42) q Let us consider an example of PCM system in which v= 4 bits. Then number of levels 4 will be, q = 24 =16 levels. :» From equation 1.8.42 the step size 8 will be, Thus the quantization error x volts of the full range voltage. For simplicity, assume that full range voltage is 16 volts. Then maximum quantization error will be 1 velt. For the low signal amplitudes like 2 volts, 3 volts etc, the maximum quantization error of 1 volt is quite high ie. about 30 to 50%. But for signal amplitudes near 15 volts, 16 volts etc, the maximum quantization error (which is same throughout the range) of 1 volt can be considered to be small. This problem arises because of uniform quantization. Therefore nonuniform quantization should be used in such cases, Another example is discussed next.Digital Communications 4-89 Pulse Digital Modulation 1.8.6.2 Necessity of Nonuniform Quantization for Speech Signal We know that speech and music signals are characterized by large crest factor. That is for such signals the ratio of peak to rms value is very high. Crest factor = Peek value vas (1.8.43) ams value Very high for speech and music. We know that the signal to noise ratio is given as, - = (32% xP) By equation 1.8.35 -. (1.8.44) Expressing in decibels, (5) = Wlos,9 (2x22?) If we normalize the signal power ie. if P=1, then above equation becomes, (a ea = (48+60) dB sas (1.8.45) Here power P is defined as, P = /2.,,1 = mean square value of signal voltage = ey i 7 xt) . Normalized power will be, Pp = <3 (with R=1] P =x) va (1.8.46) Crest factor is given as, i Peak value Xanax Crest factor = Stas ° Gan va (1.8.47) = max since P = x2) (1.8.48) | vP | When we normalize the signal x(0), then i Xmax = 1 .. (1.8.49) Putting above value of Xa, in equation 1.8.48, Crest factor = —L (1.8.50) vPDigital Communications 1-90 Pulse Digital Modulation For a large crest factor of voice (speech) and music signals P should be very very less than one in above equation. ie, Po Fig. 1.8.10 PCM performance with y - law compandingDigital Communications 92 Pulse Digital Modulation It can be observed from above figure that signal to noise ratio of PCM remains almost constant with companding. 1.86.5 A-Law for Companding The A law provides piecewise compressor characteristic. It has linear segment for low level inputs and logarithmic segment for high level inputs. It is defined as, Al fer osixisy Za) = |, Lena .. (1.8.53) 1+In(A la) for Leet T+lnA When A = 1, we get uniform quantization. The practical value for A is 87.56. Both A-law and jt-law companding is used for PCM telephone systems. 1.8.6.6 Signal to Noise Ratio of Companded PCM The signal to noise ratio of companded PCM is given as, S _ 34% Nice ss) Here q = 2° is number of quantization levels. ump Example 1.8.12 ; For a random variable are the mean square value and variance always equal ? Calculate these quantities for the quantization noise or error in PCM system. Solution : For a random variable X, the variance o? is given as, o2 = X?-m2 Here X? is the mean square value and m, is the mean value. Above equation shows that variance (a?) and mean square value { X? ) will be same if mean (1,) is zero, From quantization characteristics of Fig. 1.84 (b) it is clear that quantization error (e) has zero mean or average value. And it follows uniform distribution from 4 to 5 Hence probability density function of quantization error will be,Digital Communications 1-93 Pulse Digital Modulation 1 fee) = 43 By equation 1.8.22 o elsewhere Mean square value can be calculated as, x? = Pehla Putting values in above equation, s = fe ale This is the mean square value of quantization error. ad Example 1.8.13: A compact disc (CD) records audio signals digitally by using PCM, Assume the audio signal bandzwidth to be 15 kHz. (i) What is Nyquist rate ? . (ii) If the Nyquist samples are quantized into L = 65,536 levels and then binary coded, determine the number of binary digits required to encode a sample. (iii) Determine the number of binary digits per second (bits/sec) required to encode the audio signal. (iv) For practical reasons, the signals are sampled at a rate well above Nyquist rate at 44100 samples per second. If L = 65,536, determine number of bits per second required to encode the signal and transmission bandwidth of encoded signal. Solution : (i) To obtain Nyquist rate The bandwidth of the signal is, W = 15 kHz. Nyquist rate = 2W = 2«15kHz = 30 kHzDigital Communications 1-94 Pulse Digital Modulation (ii) To determine number of bits Number of levels,q = L = 65,536 Hence binary digits required to encode each sample will be, q=? or v = logyg = log, 65536 = 16 bits (iii) To determine signaling rate » = 16 bits/sample are used. The samples are taken at the rate of ff, = 30,000 samples/sec. Hence signaling rate will be, r= of = 16x 30,000 = 480 k bits/sec iv) To obtain By if f, = 44.1 kHz Levels used are q = 65,536 ‘ v = logy q = 16 bits f, = 44100 samples/sec From equation 1.85, transmission bandwidth required to encode the signal will be, 1 By = fs > 16x 44100 352.8 kHz mma Example 1.8.14 : The output signal to noise ratio of a 10 bif PCM was found to be 30 dB. The desired SNR is 42 dB. It was decided to increase the SNR to the desired value by increasing the number of quantization levels. Find the fractional increase in transmission bandwidth required for this increase in SNR. Sol. : (i) To obtain no of bits for 42 dB Signal to noise ratio of PCM is given as, § = (5 \e = (48 + 60) dB Above equation shows that signal to noise ratio increases by 6 4B with every bit, It is given that 5 = 30 dB for 10 bits NDigital Communications 1-96 Pulse Digital Modulation Theory Questions 1. With the help of neat diagrams, explain the transmitter and receiver of pulse code modulation. 2. What is uniform (linear) quantization ? 3. Expalin quantization error and derive an expression for maximum signal to noise ratio in PCM system that uses linear quantization. 4. Derive the relations for signaling rate and transmission bandwidth in PCM systern. 5. What is the necessity of nonuniform quantization and explain companding ? Unsolved Examples 1. A 40 MB hard disk is used to store PCM data. The signal is sampled at 8 kHz and the encoded PCM is to have an average signal to noise ratio of at least 30 dB, For how many minutes the PCM data can be stored on the hard disk ? TAns.: 133 min] 2. In the Finary PCM system, find out the minimum number of bits required so that quantizing noise is less than +k percent of the analog level. (Ans. :v 2 log(50/K)] 3. The Gaussian distributed random variable with zero mean and unit variance is applied to the input of uniform quantizer. (a) What is the probability that the amplitude of this input lies outside the range +4 ? (b) Using the result of past (a), find out the signal to quantization naise ratio. Uns. (a) 1 in 10" (b) (SN) dB = 60 -7.208) 1.9 Digital Multiplexers Digital signals are the sequences of binary 1 and 0 symbols. Digital multiplexing technique simultaneously transmits the symbols from many channels by interleaving them This is very much similar to time division multiplexing. In digital multiplexing there are no constraints like periodic sampling and waveform preservation. The digital multiplexing uses a binary multiplexers and their hierarchies. A binary multiplexer merges input bit from different sources into one signal for transmission via a digital communication system. The multiplexing of various digital signals can be bit by bit or by words or by characters.Additional pulses are inserted in the multiplexed data stream to identify the different channels or frames. These are called control bits. The multiplexer performs following operations. 1. Establish the frame as the smallest time interval containing at least one bit from every input, 2. Assign to each input a number of unique bit slots within the frame. 3. Insert control bits for frame identification and synchronization. 4, Make allowance for any variations of the input bit rates.Digi 1.9.1 Types of Digital Multiplexers ‘There are following types of digital multiplexers Communications 1-97 Pulse Digital Modulation Synchronous multiplexers : When single master clock governs all sources synchronous multiplexers are used. Since a single master clock is used there are no bit rate variations. Synchronous multiplexing has highest throughput efficiency. Synchronous multiplexer have the increased complexity because of master clock signal. Asynchronous multiplexers : Asynchronous multiplexers are used for digital data sources that operate in a start/stop mode producing bursts of characters with variable spacing between the bursts. Buffering and character interleaving makes it possible to merge these sources into a synchronous multiplexed bit stream. Quasi synchronous multiplexers : Quasi synchronous multiplexers are used when input bit rates have the same nominal value but vary within specified bounds. These multiplexers arranged in a hierarchy of increasing bit rates, constitute the building blocks of interconnected digital communication system. 1.9.2 Multiplexing Hierarchies Fig. 1.9.1 shows the multiplexing hierarchy for digital communication. anne Voice net 1.5 Mbis eee TEE] Bank ccona a 64 hols First lovo! Voice pom | 64 kos see Fig. 1.9.1 Multiplexing hierarchy for digital communication In this hierarchy the third level is used, for multiplexing purposes and other three levels are designed for point to point transmission and multiplexing, The bit rate at the next level is more than the sum of all the channels multiplexed at the input of that level. Table 1.9.1 shows inputs and rates for a typical digital multiplexer.Digital Communications 1-98 Pulse Digital Modulation Lavels Number of inputs Output bit rate or bits per sec. First level 24 18x 108 Second level Third level Fourth level 6 ar4x 108 Table 1.9.1 Inputs and rates of digital multiplexers for AT and T system From the above table we can see that fourth level of the multiplexer has total number of inputs as, Multiplexed inputs to 4!" level = 24x4x7x6 = 4032 Voice PCM signals Since 4* level is the last level, the digital multiplexer multiplexes total 4032 voice PCM signals. The bit rate of the fourth level is 274x106 bits per second. That is, it is the final output signaling rate, r= 274x106 We know that transmission channel bandwidth B,- should be, By 2 r/2 2 137 MHz 1.9.3 PCM TDM System 1.9.3.1 Multiplexing Hierarchy The PCM-TDM system uses many codecs as shown in Fig, 1.9.2, The codec is basically a PCM encoder (transmitter) and decoder (receiver). Codec generates serial stream of PCM data. At the receiver side, codec receives serial PCM data and generates analog signal. The sampling frequency of PCM can be selected by external clock given to the codec. One codec per channel is used. The outputs from various codecs are combined by the multiplexer into single bit stream. This bit stream is converted to baseband waveform by line waveform generator. The low pass filter (LPF) bandlimits the baseband signal. The waveform regenerator is used at the receiver to construct the input noisy waveform to clear digital signal. The Demux then detects individual channel signals and separates them. The codecs then recover the required analog signal.Digital Communications 1-99 Pulse Digital Modulation Analog input \ Codes Multiplexed ‘data PCM data Line waveform generator N Codee N Channel | Analog | output 2 Puls — regenerator N Fig. 4.9.2 TDM/PCM system 1.9.3.2 Multiple Channel Frame Alignment For TDM / PCM (I, System) The multiple channel alignment is very important in TDM/PCM system. Fig. 19.3 shows the TDM frame format of most widely used T1 system. sutome CEPEEEEE EEE rg var ~ St o — { Lome ee rane TREES PEL EREEEPELEEEEETEN ETE EEN PEEPS EPEEE EL. 2518 198 diots Fig, 1.9.3 Multiple channel frame alignment in T1 systemDigital Communications 1-100 Pulse Digital Modulation As shown in the Fig.1.9.3, this system contains a multiframe of 12 frames. The duration of the multiframe is 1.5 msec. Each frame consists of samples from 24 channels. Thus the samples of 24 channels are Time Division Multiplexed. Each channel sample is encoded into 8 bits. Thus the total bits of 24 channels will be 24x8=192 bits. This indicate the start of the next frame, the frame sync bit or ’S bit is transmitted at the begining of each frame. Thus the total bits in one frame are (24x 8) +1 =193 bits. Calculation of bit rate : Each channel is normally sampled at 8 kHz rate. Thus the time between any two 1 . 3WOHE =125 microseconds. In the TDM system, the samples from each channel is transmitted in each successive frame. Hence the duration of the frame will also be 125 microseconds. This is shown in Fig, 1.9.3. successive samples of single channel will be Bits per frame = 24 channels/frame x 8 bits/channel + 1 frame sync bit = 193 bits . . _ Number of bits per frame _ 193 bits “ Bitrate Ry = ——FrcoFonelrame ~~ 125x10- seconds 1.544106 bits/sec The signaling information is transmitted by replacing the 8" bit (i.e. LSB) in each channel by signaling bit in every sixth frame. Thus, Signaling period = Period of the signaling bit = 125x105 x6 = 750x 10 sec ee signaling period a 750x 10-6 - = 1.3333 Kbps. imp Example 1.9.1: The T, carrier system used in digital telephony multiplexes 24 voice channels based on 8 bit PCM. Each voice signal is usually put through a towpass filter with cutoff frequency 3.4 kHz. The filtered signal is sampled at 8 Kitz, In addition a single bit is added at the end of frame for the purpose of synchronization.Digital Communications 1-104 Pulse Digital Modulation Calculate i) The duration of each bit ii) The resultant transmission rate iii) Minimum required transmission bandwidth. Solution : i) Duration of each bit The T; system is explained just now. The signals are sampled at 8 kEz. Hence the 1 time between any two successive samples of the same channel will be = 125 ps. Fig. 1.9.3 shows the structure of the frame. In one frame of 125 jis, the total bits are, Bits per frame = 24 channels/frame x 8 bits/channel + 1 frame sync bit 193 bits Time duration of one frame Bits per frame Time duration of one bit = 125x10-6 193, 0.6476 sec/bit ii) Transmission rate The transmission rate is the bit rate R,, which is the reciprocal of duration of one bitie, —_! Duration of one bit 1 0.6476 10-6 = 1544x108 bits/sec iii) Transmission bandwidth The transmission bandwidth in PCM must be greater than half of the bit rate. i.e., 1 Br 2 FR, Ry = v $x 1.544% 108 Br 2 772 kHzDigital Communications 1-102 Pulse Digital Modulation um> Example 4.9.2 : Twenty four voice channels of ¢ kHz bandwidth eack sampled at Nyquist rate and encoded into 8 bit PCM are time division multiplexed with 1 bit/frame as synchronization bit. What is bit rate at the output of multiplexers? Solution ; Given data is N = 24 channels Bandwidth W = 4 kHz v =8 bits = 2W =2x«4 kHz =B8kHz Twenty four voice channels are transmitted in one frame.Each of the channel has & bits. One bit is added in every frame for synchronization. Hence, Nyquist rate Bits in one frame = 24 x8+1 ( Synchronization bit ) = 193 bits / frame. Frames are transmitted at the nyquist rate. Hence bit rate at the output of multiplexer will be, Bit rate = number of bits/frame x number of frames/sec. = 193 x 8000 = 1.544 x 10° bits/sec. Thus the bit rate is 1.544 mbps. ma Example 1.9.3: Describe the digital multiplexing of number of telephone channels, data channels, TV channels, Draw an appropriate diagram showing different multiplexing levels of either AT and T or CCIT standard. Solution : Fig. 1.9.4 shows the configuration of digital multiplexer of AT & T standard. In the first level PCM voice channels and digital data channels are multiplexed. The second level multiplexes T, signal and visual telephone data. The maximum bit rate of the forth multiplexing level can be 274 x10® bits per second.Digital Communications 1-103 Pulse Digital Modulation 24 Ty, voice | Channel telephone || Bank channels ——>} Digital —_*] data T channels —!o| Visual telephone] 274 Mbis Ty + channel Fig. 1.9.4 Digital multiplexing of voice telephone channels, digital data, TV etc. for AT & T standard Theory Questions i. Which are the types of digital multiplexers? 2. Explain the frame structure of T1 system in detail. 3. With the help of block diagram explain PCM/TDM system. 1.10 Virtues, Limitation and Modifications of PCM Advantages of PCM (i) Effect of channel noise and interference is reduced. (i) PCM permits regeneration of pulses along the transmission path. This reduces noise interference. (iii) The bandwidth and signal to noise ratio are related by exponential law. (iv) Multiplexing of various PCM signals is easily possible. (v) Encryption or decryption can be easily incorporated for security purpose. Limitations of PCM () PCM systems are complex compared to analog pulse modulation methods. (ii) The channel bandwidth is also increased because of digital coding of analog pulses.Digital Communications 1-104 Pulse Digital Modulation Modifications of PCM @ PCM can be modified to delta modulation. It is more simplified method of implementation. (i) The PCM can be used in wideband communications channels to overcome the bandwidth problem. (iii) With the help of data comparison along with PCM, the redundancy can be removed and data rate can be reduced. 1.11 Differential Pulse Code Modulation 4.11.4 Redundant Information in PCM The samples of a signal are highly corrected with each other. This is because any signal does not change fast. That is its value from present sample to next sample does not differ by large amount. The adjacent samples of the signal carry the same information with little difference. When these samples. are encoded by standard PCM system, the resulting encoded signal contains redundant information. Fig. 1.11.1 illustrates this. Fig. 1.11.1 shows a continuous time signal x(t) by dotted line. This signal is sampled by flat top sampling at intervals T,, 27, , 37, ....17,. The sampling frequency is selected to be higher than nyquist rate. The samples are encoded by using 3 bit (7 levels) PCM. The sample is quantized to the nearest digital level as shown by small x) bits (levels) xT) Fig. 1.11.4 Redundant information in PCMDigital Communications 1-105 Pulse Digital Modulation circles in the diagram. The encoded binary value of each sample is written on the top of the samples. We can see from Fig. 1.11.1 that the samples taken at 4T,,5T, and 67, are encoded to same value of (110). This information can be carried only by one sample. But three samples are carrying the same information means it is redundant. Consider another example of samples taken at 9, and10T,. The difference between these samples is only due to last bit and first two bits are redundant, since they do not change. 1.11.2 Principle of DPCM If this redundancy is reduced, then overall bit rate will decrease and number of bits required to transmit one sample will also be reduced. This type of digital pulse modulation scheme is called Differential Pulse Code Modulation. 1.11.3 DPCM Transmitter The differential pulse code modulation works on the principle of prediction. The value of the present sample is predicted from the past samples. The prediction may not be exact but it is very close to the actual sample value. Fig. 1.11.2 shows the transmitter of Differential Pulse Code Modulation (DPCM) system. The sampled signal is denoted by x(nT,) and the predicted signal is denoted by £(nT,). The comparator finds out the difference between the actual sample value x(n T,) and predicted sample value #(nT,). This is called error and it is denoted by e(nf,), It can be defined as, e(nT,) = x(nT.)- Hn 1) w= (11) Comparator Fig. 1.11.2 Differential pulse code modulation transmitter Thus error is the difference between unquantized input sample x(nT,) and prediction of it (17). The predicted value is produced by using a prediction filter. ‘The quantizer output signal ¢,(1T,) and previous prediction is added and given asDigital Communications 1-106 Pulse Digital Modulation input to the prediction filler. This signal is called x, (0rT,). This makes the prediction more and more close to the actual sampled signal. We can see that the quantized error signal e,(nT,) is very small and can be encoded by using small number of bits. Thus number of bits per sample are reduced in DPCM. The quantizer output can be written as, eg (nT) = (nT) +q(nTs) w= (1.11.2) Here q(nT;) is the quantization error. As shown in Fig. 1.11.2, the prediction filter input x, (1T,) is obtained by sum %(rT,) and quantizer output i.e., x(n T,) = &(nT,)+eq (nT) wo (1.11.3) Putting the value of e,(7,) from equation 1.11.2 in the above equation we get, xy(nT.) = HnT,)te(n T+ q(nT.) . (LALA) Equation 1.11.1 is written as, e(nT,) = x(nT,)-X(nT,) 2 e(nT,)+i(nT,) = x(0T;) a= (LAL5) +. Putting the value of ¢ (1 T,) + £(T,) from above equation into equation 1.11.4 we get, xg(nTe) = x(nT.)+q (nT) oe (1.11.6) Thus the quantized version of the signal x,(1T,) is the sum of original sample value and quantization error 9(vT;). The quantization error can be positive or negative. Thus equation 1.11.6 does not depend on the prediction filter characteristics. 1.11.4 Reconstruction of DPCM Signal Fig. 1.11.3 shows the block diagram of DPCM receiver. DPCM input Fig. 1.11.3 DPCM receiver The decoder first reconstructs the quantized error signal from incoming binary signal. The prediction filter output and quantized error signals are summed up to give the quantized version of the original signal. Thus the signal at the receiver differs from actual signal by quantization error q(7,), which is introduced permanently in the reconstructed signal. QoQDelta Modulation We have seen in PCM that, it transmits all the bits which are used to code the sample. Hence signaling rate and transmission channel bandwidth are large in PCM. To overcome this problem Delta Modulation is used. 2.1 Delta Modulation 2.1.1 Operating Principle of DM Delta modulation transmits only one bit per sample. That is the present sample value is compared with the previous sample value and the indication,whether the amplitude is increased or decreased is sent. Input signal x(#) is approximated to step signal by the delta modulator. This step size is fixed. The difference between the input signal x(!) and staircase approximated signal confined to two levels, ie. 4+8and—3 If the difference is positive, then approximated signal is increased by one step ic. 'S. If the difference is negative, then approximated signal is reduced by ‘5. ‘When the step is reduced, ‘0’ is transmitted and if the step is increased, ‘I’ is transmitted. Thus for each sample, only one binary bit is transmitted. Fig. 2.1.1 shows the analog signal x(() and its staircase approximated signal by the delta modulator. ot TTT) T ] I | x(t) Ppp Binary one bitsequence! Fig. 2.1.1 Delta modulation waveform (2-1)Digital Communications 2+2 Delta Modulation The principle of delta modulation can be explained by the following set of equations. The error between the sampled value of x(t) and last approximated sample is given as, e@t,) = x(n T)-20T) vw» QA) Here, e(nT,) = Error at present sample x(uT,) = Sampled signal of x() 5(nT,) = Last sample approximation of the staircase waveform. We can call u(7,) as the present sample approximation of staircase output. Then, u[(1=1)T,] = ¥(nT,) ws (2.1.2) = Last sample approximation of staircase waveform. Let the quantity b (11T,) be defined as, b(nT,) = 8 sgn[e(7,)] vx (2.13) That is depending on the sign of error ¢(T,) the sign of step size 5 will be decided. In other words, bT,) = +6 if xT) 2 FT) =-8 if x(nT) < 2(nT) ws. (214) ig b(nT,) = +8; — binary ‘I’ is transmitted and if b(n T,) = -8; — binary ‘0’ is transmitted. T, = Sampling interval. 2.1.2 DM Transmitter Fig. 2.1.2 (a) shows the transmitter based on equations 2.13 to 2.1.5. The summer in the accumulator adds quantizer output (+8) with the previous sample approximation. This gives present sample approximation. i.e., u(aT,) =u(nT, -T,)+[£8] or = ul(n-1T.]+b(nT,) s+ (2.1.5) The previous sample approximation u[(n~-1)T,] is restored by delaying one sample period T,. The sampled input signal x(nT,) and staircase approximated signal X(nT,) are subtracted to get error signal e(nT,).Digital Communications 2-3 Delta Modulation Error Sampled ——_g(kT,) input + "Accumulator (a) Accumulator Input Output (b) Fig. 2.1.2 (a) Delta modulation transmitter and (b) Delta modulation receiver Depending on the sign of e(nT,) one bit quantizer produces an output step of +5 or ~6. If the step size is +5, then binary ‘1’ is transmitted and if it is —6, then binary ‘0’ is transmitted. 2.1.3 DM Receiver At the receiver shown in Fig. 2.1.2 (b), the accumulator and low-pass filter are used. The accumulator generates. the staircase approximated signal output and is delayed by one sampling period 7,. It is then added to the input signal. If input is binary ’1’ then it adds +8 step to the previous output (which is delayed). If input is binary ‘0’ then one step ‘8 is subtracted from the delayed signal. The low-pass filter has the cutoff frequency equal to highest frequency in x(). This filter smoothen the staircase signal to reconstruct x().Digital Communications 2-4 Delta Modulation 2.2 Advantages and Disadvantages of Delta Modulation 2.2.1 Advantages of Delta Modulation The delta modulation has following advantages over PCM, 1. Delta modulation transmits only one bit for one sample. Thus the signaling tate and transmission channel bandwidth is quite small for delta modulation. 2. The transmitter and receiver implementation is very much simple for delta modulation. There is no analog to digital converter involved in delta modulation. 2.2.2 Disadvantages of Delta Modulation Granular noise Slope - overioad distortion X(t g Staircase appioximation--—~ a Z oe . uit) t Fig. 2.2.1 Quantization errors in delta modulation The delta modulation has two drawbacks = 2.2.2.4 Slope Overload Distortion (Startup Error) This distortion arises because of the large dynamic range of the input signal. As can be seen from Fig. 2.2.1 the rate of rise of input signal x(t) is so high that the staircase signal cannot approximate it, the step size ‘8’ becomes too small for staircase signal u(t) to follow the steep segment of x(#). Thus there is a large error between the staircase approximated signal and the original input signal x((). This error is called slope overload distortion. To reduce this error, the step size should be increased when slope of signal of x(t) is high. Since the step size of delta modulator remains fixed, its maximum or minimum slopes occur along straight lines. Therefore this modulator is also called Linear Delta Modulator (LDM). 2.2.2.2 Granular Noise (Hunting) Granular noise occurs when the step size is too large compared to small variations in the input signal. That is for very small variations in the input signal, the staircaseDigital Communications 2-5 Delta Modulation signal is changed by large amount (5) because of large step size. Fig. 2.2.1 shows that when the input signal is almost flat, the staircase signal (f) keeps on oscillating by +8 around the signal, The error between the input and approximated signal is called granular noise, The solution to this problem is to make step size small. Thus large step size is required to accommodate wide dynamic range of the input signal (to reduce slope overload distortion) and small steps are required to reduce granular noise. Adaptive delta modulation is the modification to overcome these ‘errors, ims} Example 2.2.1: Using predictability theory, prove that transmission of encoded error signal (rather than encoded signal itself is sufficient for reasonable reconstruction of signal. With the help of block schematic suggest any one technique to transmit and receive encoded errors. What are the limitations and advantages of such techniques with reference to linear or uniform PCM ? Solution : Here the technique that uses predictibility theory is basically delta modulation. The output of the accumulator in DM transmitter is given by equation 2.15 as, unt) = uf(n-T, | +(nT,) 221) Here WT.) = £8 or dSsgnfe(n7,)] Thus b(nT,) basically represents error signal. Sign of step size 'S’ depends upon whether e(nT,) is positive or negative. Now we will show that the signal can be reconstructed only with the help of encoded error signal, ie. L(nT,). The accumulator of Fig. 2.1.2(b) acts as a delta modulation receiver. u(a7.) is the output of accumulator. For simplicity let us drop T, in equation 2.2.1 Then we get, ula) = u(ir—1) +6(n) (222) Observe that this is recursive equation. Hence u(n—1) can be calculated as, u(n—=1) = n(n =2)+(n =1) we (2.2.3) Hence equation 2.22 becomes, u(x) = u(n—2)4(n -1) +b(n) ~~ 24) From equation 22.3 we can calculate u(1t ~2) as, u(t —2) = u(t —3)+6(" —2) Hence equation 2.2.4 becomes, u(n) = u(t -3)+8(n - 2) +0(n -1) (0)Digital Communications 2-6 Delta Modulation Above equation can be generalized as, ui) = Bla) +n 1) 4+ Thus un) can be reconstructed totally from encoded errors b(n), b(n —2), «1. ete. +. 2.3 Adaptive Delta Modulation 2.3.1 Operating Principle To overcome the quantization errors due to slope overload and granular noise, the step size (8) is made adaptive to variations in the input signal x(). Particularly in the steep segment of the signal x(t), the step size is increased. When the input is varying slowly, the step size is reduced. Then the method is called Adaptive Delta Modulation (ADM). The adaptive delta modulators can take continuous changes in step size or discrete changes in step size. 2.3.2 Transmitter and Receiver Fig. 2.3.1 (a) shows the transmitter and 2.3.1 (b) shows receiver of adaptive delta modulator. The logic for step size control is added in the diagram. The step size increases or decreases according to certain rule depending on one bit quantizer output. ‘Accumulator Fig. 2.3.1 Adaptive delta modulator (a) Transmitter (b) ReceiverDigital Communications 2-7 Delta Modulation For example if one bit quantizer output is high (1), then step size may be doubled for next sample. If one bit quantizer output is low, then step size may be reduced by one step. Fig. 2.3.2 shows the waveforms of adaptive delta modulator and sequence of bits transmitted. In the receiver of adaptive delta modulator shown in Fig. 2.3.1 (b) the first part generates the step size from each incoming bit. Exactly the same process is followed as that in transmitter. The previous input and present input decides the step size. It is then given to an accumulator which builds up staircase waveform. The low-pass filter then smoothens out the staircase waveform to reconstruct the smooth signal. Tt "Amplitude: 4—se= |. Binary one bit quantized signi L |! Fig. 2.3.2 Waveforms of adaptive delta modulation 2.3.3 Advantages of Adaptive Delta Modulation Adaptive delta modulation has certain advantages over delta modulation. i.e., 1. The signal to noise ratio is better than ordinary delta modulation because of the reduction in slope overload distortion and granular noise. 2. Because of the variable step size, the dynamic range of ADM is wide. 3. Utilization of bandwidth is better than delta modulation. Plus other advantages of delta modulation are, only one bit per sample is required and simplicity of implementation of transmitter and receiver. nm=> Example 2.3.1: Consider a sine wave of frequency f,, and amplitude A,, applied to a delta modulator of step size 8. Show that the slope overload distortion will occur ifDigital Communications 2-8 Delta Modulation 8 Au > TERT, where T, is the sampling period. (Nov/Dec.-2004, 4 Marks) Solution : Let the sine wave be represented as, 2) = Ay sina fy, 9 Slope of x(t) will be maximum when derivative of x(t) with respect to ‘t’ will be maximum. The maximum slope of delta modulator is given from Fig. 2.1.1 as, Step size Max slope = = pling period é “Ff ws (23.1) Slope overload distortion will take place if slope of sine wave is greater than slope of delta modulator i.e. a 8 max| x0] >F da, 3 marl Ay sin fy 4 or max|A,, 2% fy 60S (2% fry 1 > a 3 Ann > «= (23.2) or imp Example 2.3.2: A delta modulator system is designed to operate at five times the Nyquist rate for a signal with 3 kHz bandwidth. Determine the maximum amplitude of a 2 kHz input sinusoid for which the delta modulator does not have slope over load. Quanttizing step size is 250 mV. Derive the formula that you use. Solution : In example 2.3.1 we have derived the relation for slope overload distortion which will occur if, 8 An > Pe By equation 23.2 ~ (233)Digital Communications 2-10 Delta Modulation 4 f,=2e10 Hz Fig. 2.3.3 Delta modulator Solution : Optimum step size means slope overload will not occur. Please refer to the relationship proved in example 2.3.1 It is given as, slope overload distortion will occur if, 3 An > TERETE Here Am = Ais the amplitude of sine wave 8 = kis the step size T= , = Lis the sampling duration. i Hence above equation can also be written as, slope overload distortion will occur if, A>—t* wu (235) 2H fm Se Here the sine wave is given as, mf) = O.1sin(2n x1071) = Asin(2z x fn t) Hence A=01V fm = 10? Hz and f = 2x104 Hz Fork=4mV From equation 2.35, slope overload will occur if k A> SaRTEDigital Communications 2-11 Delta Modulation 4x103 2ax105 / (2104) > 0.01273 We know that A is 0.1 V which is greater than 0.01273. Hence slope overload distortion will occur for step size of 4 mV. For k = 60 mV From equation 2.35 slope overload will occur if K A> STE > 60x10 2nx 103 /(2«104) > 0.19098 Since A is 0.1 V, the slope overload distortion will not ocur in this case. This is because A(0.1 V) is less than 0.19098 V. Hence slope overload distortion will not occur for step size of 60 mV. Sr. No. Step size k Slope overload distortion [+ [asamy | sep oto anerion ssa [2 [isco [se etn enw tr Table 2.3.1 Results: ww Example 2.3.4: Find the signal amplitude for minimum quantization error in a delta modulation system if step size is 1 volts having repetition period 1 ms. The information signal operates at 100 Hz. Solution ; The quantization error is minimum when slope overload distortion is absent. Then the quantization error is due to granular noise only. We know that the slope overload distortion is absent if, Here An is the signal amplitude 3 is step size. Given 1V fn is signal frequency. Given 100 Hz T, is sampling duration. Given 1 ms.Digital Communications 2-12 Delta Modulation Hence signal amplitude becomes, 1 2nx100x1x10 ie. An S 16V That is, the signal amplitude should be less than 1.6 volts to have minimum quantization error. ‘mp Example 2.3.5 : Derive an expression for signal to quantization noise power ratio for delta modulation. Assume that no slope overload distortion exists. [March-2006, 16 Marks] An $ Solution : (i) To obtain signal power : In example 2.3.1 we have derived that slope overload distortion will not occur if é An § Ta, Here A,, is peak amplitude of sinusoided signal 3 is the step size fy, is the signal frequency and T, is the sampling period. From above equation, the maximum signal amplitude will be, - 6 An > ae a ww (23.6) Signal power is given as, 2 Peg Here V is the rms value of the signal. Here V = + Hence above equation becomes, A, p= (4n)7r (4 Normalized signal power is obtained by taking R = 1. Hence, pe ab 2Digital Communications 2+13 Delta Modulation Putting for A,,from equation 2.3.6, 32 > ae = (23.7) This is an expression for signal power in delta modulation. (ii) To obtain noise power We know that the maximum quantization error in delta modulation is equal to step size B. Let the quantization error be uniformly distributed over an interval [~8,5] This is shown in Fig. 234 From this figure the PDE of quantization error can be expressed as, Fig. 2.3.4 Uniform distribution of quantization error f(0) 0 for e<8 fe) = x ir ~8d The noise power is given as, i V poise Noise power = —A Here V2, is the mean square value of noise voltage. Since noise is defined by random variable ‘&' and PDF /, (¢), its mean square value is given as, mean square value = E[e?] =e mean square value is given as, E[e?] = fer (edeDigital Communications 2-14 Delta Modulation From equation 23.8, F[e?] = u so % | = 1f83 83 = als +5] =» (23.9) Hence noise power will be, A 82 noise power = | —- “RK Normalized noise power can be obtained with R = 1. Hence, . at noise power = (2.3.10) 3 This noise power is uniformly distributed over -f, to f, range. This is illustrated in Fig. 23.5. At the output of delta modulator receiver there is lowpass reconstruction filter whose cutoff frequency is ‘W'. This cutoff frequency is equal to highest signal t frequency. The reconstruction filter passes part of the noise power at the output as Fig. 2.3.5. From the geometry of Fig. 23.5, output noise power will 2 Output noise power= ™ x noise power = 5 & & We know that f= + hence above equation becomes, fs Fig. 2.3.5 PSD of noise 2 Output noise power= ms 23.11)Digital Communications 2-15 Delta Modulation (lil) To obtain signal to noise power ratio Signal to noise power ratio at the output of delta modulation receiver is given as, S$ _ Normalized signal power N ~ Wormalized noise power From equation 2.3.7 and equation 2.3.11, 8 8n2 p27? N° Wr,8? 3 3 * Seat = (2.3.12) Lo Zn This is an expression for signal to noise power ratio in delta modulation. Example 2.3.6 : A DM system is designed to operate at 3 times the nyquist rate for a signal with a 3 kHz bandwidth. The quantizing step size is 250 mV. (i) Determine the maximum anaplitude of a 1 kHz input sinusoid for which the delta modulator docs not show slope overload. (i) Determine the postfiltered output SNR for the signal of part (i). (May/June-2004, 8 Marks} Solution : Given data Bandwidth, W = 3 kHz Signal frequency, f,,= 1 kHz Nyquist rate = 2f,, = 2x1 kHz = 2 kHz Sampling frequency, f, = 3x nyquist rate = 3x2kHz = 6 kHz Step size, 6 = 250 mV 1 Tet ‘oRDigital Communications 2-16 Delta Modulation {i) To obtain signal amplitude Slope overload distortion does not occur if, é An S aT Putting values in above equation, 250x109 2n*1000 Ag < 6000 S 0.2387 V. Thus maximum amplitude is 238.7 mV. {ii) To obtain signal to noise ratio Signal to noise ratio of delta modulator is given by equation 2.3.12 as, Ss 3 NN” ertwpars 1 (6000)° 4.5610 = -33.41 dB 8n? x 3000x1000) x Here note that signal to noise ratio is very poor since sampling frequency is low. ium Example 2.3.7 ; A DM system is tested with a 10 kHz sinusoidal signal with 1V peak to peak at the input, It is sampled at 10 times the Nyquist rate, (i) What is the step size required to prevent slope overload 7 (ii) What is the corresponding SNR ? Solution : Given data Signal frequency, fn = 10 kHz Signal amplitude, An = ge 05V Nyquist rate = 2% fy = 2x 10kHz = 20 kHz Sampling frequency, fe = 10x Nyquist rate = 10x 20kHz = 200kH2Digital Communications 2-17 Delta Modulation (i) To obtain step size From equation 2.3.4 we have, 6 fn Safad Under this condition slope overload will not occur. From above equation step size will be, 5 = fT Am Putting values in above equation, 8 2 2nx10,000x 0.5 1» 200x103 2 0.187 V Thus the step size greater than 157 mV will prevent the slope overload. (ii) To obtain signal to noise ratio Signal to noise ratio of delta modulation system is given by equation 2.3.12 as, S23 N ” eewyr> This is post filtered signal to noise ratio. In this example value of "W" is not given. Hence we will calculate signal to noise ratio from equation 2.3.7 and equation 2.3.10 as, §2 Ss _ 82272 N §2 T = 3 © BRpaT? Putting values in above equation. Ss. N x? x (10,000)? x_—_L_. (200x103)? 15.2 = 11.8 dB.Digital Communications 2-18 Delta Modulation nme Example 2.3.8: Consider the test signal m(t) defined by a hyperbolic tangent function m(t) = A tank (Bt), where A and B are constants. Determine the minimum step size 8 for della modulation (DM) of this signal, which is required to avoid slope overload. Solution : The given signal is, m(t) = A tanh (Bt) " 4 mit) = a-—b - ame = 4 cosht*(Bt) To avoid slope overload, 3 a = = max| sm(y| & B oF we . tr mA cos hpi Ts > ABmax| cos kB Maximum value of is 1. Hence above equation becomes, 1 cosh?(Bt) & 7 = 4B or 5 ABT, Thus the minimum value of step size is, 8min = ABT, to avoid slope overload. ‘> Example 2.3.9: In a single integration DM scheme the voice signal is sampted at a rate of 64 kHz. The maximum signal amplitude is 2 volts. Voice signal bandwidth is 35 kHz, Determine the minimum value of step size to avoid slope over load and calculate granular noise power. Solution : The given data is, f= 64x10?Hz 64x109 An fn = 2V W = 35 kHz = 3500 HzDigital Communications 2+19 Delta Modulation i) To determine step size (3) Slope overload distortion will not occur if, é An SORA Putting values, 2 < 5 7 2nx 3500x ——__ 64x103 o 6 2 0.687 volts. Il) To calculate noise power Noise power is given by equation 2.3.11 as, WT,82 3 = 86 mW. 3500x, Noise power = ium) Example 2.3.10 : In a single integration DM scheme, the voice signal is sampled at a rate of 64 KHz. The maximum signal amplitude is 1 volt, voice signal bandwidth is 3.5 Kitz. i) Determine the minimum value of step size to avoid slope overload. if) Determine granular noise N,, iii) Assuming signal to be sinusoidal, calculate signal power and signal to noise ratio. iv) Assuming that noise signal amplitude is uniformly distributed in the range (1, 1) determine the signal power and signal to noise ratio. [Nov.-2005, 16 Marks] Solution : The given data is, f= 64K, A,,=1V, W=35 kHz. |) To obtain step size Slope overload will not occur if, 3 ne Im < Putting values in above equation with 1 T= aoage and fn = 35 KHDigital Communications 2-20 Delta Modulation 1s 4 2nx3500x—__ 64x109 8 > 0.3436 volts. ii) To obtain granular noise power Noise power is given by equation 2.3.11 as, _ WT8? _ Ws? No = 3. Oh 3.5x103 x (0.3436)? 3x 64x 109 = 2.15 mw iil) To obtain S/N ratio Amplitude of the signal is 1V. For sinusoidal signal, the signal power will be = “nil Pebe aw Hence signal to noise ratio will be, s 1/2 2s = = 232: N 15x 103 3 S _ S_ «(Fl = log 19 2 = 23:66 dB iv) To calculate $ ratio if signal is uniformly distributed over the range (1, 4) fy) Fig, 23.6 shows the pdf of the signal. It can be easily calculated that fx) = 3 if it is distributed over (-1, 1). * Hence mean square value of the signal can be calculated as, Fig. 2.3.6 Uniformly distributed signalDigital Communications 2-21 Delta Modulation 1 l 1 ifs? 1 [Pm ton al 3 x gos Normalized signal power = — = x? with R=1 a1 = iw Hence signal to noise ratio becomes, S _ Signalpower_ 3 N Noise power 2.15x103 = 185 or (&) = 1Olog 1p 155 = 21.9 dB NJap Theory Questions 1. Explain delta modutation in detail suitable diagram. Expluin ADM and compare its performance with DM. 2. What is slope overload distortion and granular noise in delta modulation and how it is removed in ADM ? Unsolved Example 1, What is the maximum power that may be transmitted without slope overload distortion ? TAns. + a wae 2.4 Comparison of Digital Pulse Modulation Methods. Table 2.4.1 shows the comparison of PCM, Differential PCM, Delta Modulation and Adaptive Delta Modulation. The comparison is done on the basis of various parameters like transmission bandwidth, quantization error, number of transmitter bits per sample etc.Passband Data Transmission 3.1 Introduction There are basically two types of transmission of digital signals : 1) Baseband data transmission : The digital data is transmitted over the channel directly. There is no carrier or any modulation. This is suitable for transmission over shart distances. 2) Passband data transmission : The digital data modulates high frequency sinusoidal carrier. Hence it is also called digital CW modulation. It is suitable for transmission over long distances. 3.1.1 Types of Passband Modulation The digital data can modulate phase, frequency or amplitude of carrier. This gives rise to three basic techniques : 1) Phase shift keying (PSK) ; In this technique, the digital data modulates phase of the carrier. 2) Frequency shift keying (FSK) : In this technique, the digital data modulates frequency of the carrier. 3) Amplitude shift keying (ASK) : In this technique, the digital data modulates amplitude of the carrier. 3.1.2 Types of Reception for Passband Transmission ‘There are two types of methods for detection of passband signals. 1, Coherent (Synchronous) detection : In this method, the local carrier generated at the receiver is phase locked with the carrier at the transmitter. Hence it is also called synchronous detection. 2. Noncoherent (Envelope) detection : In this method, the receiver carrier need not be phase locked with transmitter carrier. Hence it is also called envelope detection. Noncoherent detection is simple but it has higher probability of error. {3 - 1)Digital Communication 3-2 Passband Data Transmission 3.1.3 Requirements of Passband Transmission Scheme Any passband transmission scheme should satisfy following requirements - 1. Maximum data transmission rate. 2. Minimum probability of symbol error. 3. Minimum transmitted pawer. * 4, Minimum channel bandwidth. 5. Maximum resistance to interfering signals. 6. Minimum circuit complexity. 3.1.4 Advantages of Passband Transmission over Baseband Transmission . Long distance transmission. . Analog channels, can be used for transmission, . Multiplexing techniques can be used for bandwidth conservation. Problems such as ISI and crosstalk are absent. newepe . Passband transmission can take place over wireless channels also. a . Large number of modulation techniques are available. Drawbacks of Passband Modulation 1. Modulation and demodulation equipments, transmitting/receiving antennas, interference problems make the system complex. 2. It is not suitable for short distance communication. 3.1.5 Passband Transmission Model Fig. 3.1.1 shows the model of passband data transmission system Transmitter Receiver Fig. 3.1.1 Model of passband data transmission systemDigital Communication 3-3 Passband Data Transmission ‘1. Message source : It emits the symbol at the rate of T seconds, pv Encoder : It is signal transmission encoder. It produces the vector s, made up of 'N’ real elements. The vector §; is unique for each set of M' symbols. » Modulator ; It constructs the modulated carrier signal s(t) of duration 'T’ seconds for every symbol m;. The signal s(t) is energy signal. 4, Channel ; The modulated signal s,t) is transmitted over the communication channel. + The channel is assumed to be linear and of enough bandwidth to accommodate the signal s(t). N, 2 Detector : It demodulates the received signal and obtains an estimate of the signal vector. 6. Decoder : The decoder obtains the estimate of symbol back from the signal vector. Here note that the detector and decoder combinely perform the reception of the transmitted signal. The effect of channel noise is minimized and correct estimate of symbol ii is obtained, * The channel noise is white Gaussian of zero mean and psd of ea 3.2 Binary Phase Shift Keying (BPSK) 3.2.1 Principle of BPSK In binary phase shift keying (BPSK), binary symbol ‘I’ and ‘0 modulate the phase of the carrier. Let the carrier be, st) = A cos (2m fy!) -- B21) ‘A’ represents peak value of sinusoidal carrier, In the standard 10 load register, the power dissipated will be, 1 P= 34? A = ¥2P we (3.2.2) « When the symbol is changed, then the phase of the carrier is changed by 180 degrees (x radians) © Consider for example, Symbol ‘l= s,() = V2P cos (2nfy t) w= (3.2.3)Digital Communication 3-4 Passband Data Transmission if next symbol is '0" then, Symbol 0" => 39 () = V2P cos (2nfy t+ 1 vw B24) Since cos (0+ x) = -cos®, we can write above equation as, 89 (0) = -V2P cos(2n fy -» (3.2.5) With the above equation we can define BPSK signal combinely as, b(t) V2P cos (2x fy 8) - (3.2.6) Here b() = +1 when binary 'I' is to be transmitted = -1 when binary '0' is to be transmitted 3.2.2 Graphical Representation of BPSK Signal Fig. 3.2.1 shows inary signal and its equivalent signal b(f). Fig. 3.2.4 (a) Binary sequen (b) Its equivalent bipolar signal b(t) () BPSK signal As can be seen from Fig. 3.2.1 (b), the signal b(t) is NRZ bipolar signal. This signal directly modulates carrier cos (2n ff).Digital Communication 3-5 Passband Data Transmission 3.2.3 Generation and Reception of BPSK Signal Nov./Dec.- 2005 3.2.3.1 Generator of BPSK Signal Bipolar NRZ BPSK evel signal Carrier signal Fig. 3.2.2 BPSK generation scheme + The BPSK signal can be generated by applying carrier signal to the balanced modulator. © The baseband signal (0) is applied as a modulating signal to the balanced modulator. Fig- 3.2.2 shows the block diagram of BPSK signal generator. * The NRZ level encoder converts the binary data sequence into bipolar NRZ signal. 3232 Reception of BPSK Signal Fig. 3.2.3 shows the block diagram of the scheme to recover baseband BPSK signal. The transmitted BPSK signal is, s() = b(t) V2P cos (Qn fy nal from BPSK signal cos?(2rtt +0) Bandpass from filter channel cos(2nfat +0) 1 tt) V2P cos*(2ntyteo) we C Integrator ‘Synchronous demodulator (muttipiier) SkkTp) lof bit) V2P cos(2atgt+0) Si Fig. 3.2.3 Reception BPSK schemeDigital Communication 3-6 Passband Data Transmission Operation of the receiver 1) Phase shift in received signal : This signal undergoes the phase change depending upon the time delay from transmitter to receiver. This phase change is normally fixed phase shift in the transmitted signal. Let the phase shift be 0. ‘Therefore the signal at the input of the receiver is, SQ) = b()V2P cos (2n fy t +0) we B27) 2) Square law device : Now from this received signal, a carrier is separated since this is coherent detection. As shown in the figure, the received signal is passed through a square law device. At the output of the square law device the signal will be, cos? (27 fo +0) Note here that we have neglected the amplitude, because we are only interested in the carrier of the signal. We know that, 1+ cas 20 SS cos? @ = cos? (Qn fy +8) = Tt s2Grfot+0) dah eos2tarfyt +0) Here 4 represents a DC level. 4 or 3) Bandpass filter : This signal is then passed through a bandpass filter whose passband is centered around 2,f,. Bandpass filter removes the DC level of 5 and at its output we get, cos 2(2n fy t-+8) This signal has frequency of 2fy. 4) Frequency divider : The above signal is passed through a frequency divider by two. Therefore at the output of frequency divider we get a carrier signal whose frequency is fy ie. cos (2r fy t +8). 5) Synchronous demodulator : The synchronous (coherent) demodulator multiplies the input signal and the recovered carrier. Therefore at the output of multiplier we get, b(t) V3P cos (2n fy +0) cos (2x fg t+0) = bi) V2P cos? (2n fy t +0) = 6) ARB «Ep + cos 2(25 fy +0)Digital Communication 3-7 Passband Data Transmission [P = (yz I + cos 202 fy +0) B28) 6) Bit synchronizer and integrator : The above signal is then applied to the bit synchronizer and integrator. The integrator integrates the signal over one bit period. The bit synchronizer takes care of starting and ending times of a bit, * At the end of bit duration T,, the bit synchronizer closes switch Sq temperorily. This connects the output of an integrator to the decision device. Itis equivalent to sampling the output of integrator. © The synchronizer then opens switch $, and switch $, is closed temperorily. This resets the integrator voltage to zero. The integrator then integrates next bit. * Let us assume that one bit period ’T,’ contains integral number of cycles of the carrier. That is the phase change occurs in the carrier only al zero crossing, This is shown in Fig. 3.2.1 (c). Thus BPSK waveform has full cycles of sinusoidal carrier, To show that output of integrator depends upon transmitted bit + In the k” bit interval we can write output signal as, p kt So (KTy) = oar fe J Wr cos 2 (re fy t+ Oy) at (k-1) Tp from equation 3.2.8 The above equation gives the output of an interval for i!" bit. Therefore integration is performed from (k~1)T,, to KT;,. Here T;, is the one bit period. * We can write the above equation as, pl Te AT 1 59 (KT)) = oury {e J ldt+ J cos 22K fat 40) | DT, DT, 3 kT Here [cos 2(2xfy +0) dt =0, because average value of sinusoidal waveform is (KD Tp. zero if integration is performed over full cycles. Therefore we can write above equation as, 7 tl 89 (KT) = very fe fla (k-1) TpDigital Communication 3-8 Passband Data Transmission ip = HOTS thy, = wary fE er, -@-0} S 5, (Tp) = veT YS Ts - 3.29) Le This equation shows that the output of the receiver depends on input i.e. sy (kT,) a b(kT,) Depending upon the value of b(kT;), the output so (kT;) is generated in the rece This signal is then given to a decision device (not shown in Fig. 3.2.3), which decides whether transmitted symbol was zero oF one. 3.2.4 Spectrum of BPSK Signals Step 1: Fourier transform of basic NRZ pulse. We know that the waveform b(t) is NRZ bipolar waveform. In this waveform there are rectangular pulses of amplitude £V,. If we say that each T. pulse is ap around its center as shown in Fig. 3.24. then it becomes easy to find fourier transform of such pulse. The fourier transform of this Fig. 3.2.4 NRZ pulse type of pulse is given as, sin (nf T)) (fT) Step 2 : PSD of NRZ pulse. For large number of such positive and negative pulses the power spectral density S(f) is given as X(D = V,T, By standard relations 3.2.10) XP T, s(f) = B21) Here X(@) denotes average value of X(f) due to all the pulses in b(f), And T, is symbol duration. Putting value of X (f) from equation 3.2.10 in equation 3.2.11 we get,Digital Communication 3-9 Passband Data Transmission 80 = “FE xi Tp 5 ‘Step 3 : PSD of baseband signal b(t) For BPSK since only one bit is transmitted at a time, symbol and bit durations are v2 Te er ? same ie. T), =T,. Then above equation becomes, 2 yap, [sitet Td si) = V3 1 vw (3.2.12) The above equation gives the power spectral density of baseband signal b (0. Step 4: PSD of BPSK signal. The BPSK signal is generated by modulating a carrier by the baseband signal b(). Because of modulation of the carrier of frequency fy, the ‘spectral components are translated from f to fy +f and fy ~f. The magnitude of those components is divided by half. Therefore from equation 3.2.12 we can write the power spectral density of BPSK signal as, 1[ sin ef - NT, , 1f sin fa + /ATs s = VOT sie a Afsin (fo + ATs ose = VET aT aGy-Dty | 2, 0 eDT ‘The above equation is composed of two half magnitude spectral components of same frequency 'f above and below fg. Let us say that the value of + V, = +P, That is the NRZ signal is having amplitudes of + /Pand-P. Then above equation becomes, _ PT, {| [sinx(f=fo) TF , Uf sin ef +7, 1 Sere = YL Sion | aE amenme| fe The above equation gives power spectral density of BPSK signal for modulating signal b(f) having amplitudes of +P. We know that modulated signal is given by equation 3.2.3 and equation 3.2.5 as, s() = 4V2P cos (2x fy!) since A=¥2P If b(t) =+ VP, then the carrier becomes, 6) = J2cos2nfy ww» (3.2.14)Digital Communication 3-10 Passband Data Transmission Plot of PSD * Equation 3.2.12 gives power spectral density of the NRZ waveform. For one rectangular pulse, the shape of $(f) will be a sinc pulse as given by equation 3.2.12. Fig. 3.2.5 shows the plot of magnitude of S(f). Fig, 3.2.5 Plot of power spectral density of NRZ baseband signal Above figure shows that the main lobe ranges from ~ fy to + fy. Here fy =~ > Since we have taken + V;, lobe is PT). * Now let us consider the power spectral density of BPSK signal given by equation 3.2.13, Fig. 3.2.6 shows the plot of this equation. The figure thus clearly shows that there are two lobes ; one at fg and other at — fy. The same spectrum of Fig. 3.2.5 is placed at +f) and= fo. But the amplitudes of main lobes are 27> in Fig, VP in equation 3.2.12, the peak value of the main z - . fb Sees) tt tar ampttude of 1 mainiobe = _ | att OO dt t r T 2 [to 2h =! Zp “fotfy lye 2h a Ea | | ! Fig. 3.2.6 Plot of power spectral density of BPSK signal Thus they are reduced to half. The spectrums of $(/) as well as Sgpcx (/) extends over all the frequencies.Digital Communication 3.11 Passband Data Transmission Interchannet Interference and ISI : * Let's assume that BPSK signals are multiplexed with the help of different carrier frequencies for different baseband signals. Then at any frequency, the spectral components due to all the multiplexed channels will be present. This is because S(f) as well a5 Sppcx (f) of every channel extends over all the frequency range. © Therefore a BPSK receiver tuned to a particular carrier frequency will also receive frequency components due to other channels. This will make interference with the required channel signals and error probability will increase. This result is called Interchannel Interference. * To avoid interchannel interference, the BPSK signal is passed through a filter.This filter attenuates the side lobes and passes only main lobe. Since side lobes are attenuated to high level, the interference is reduced. Because of this filtering the phase distortion takes place in the bipolar NRZ signal, ie. b(t). Therefore the individual bits (symbols) mix with adjacent bits (symbols) in the same channel. This effect is called intersymbol interference or ISI. + The effect of ISI can be reduced to some extent by using equalizers at the receiver. Those equalizers have the reverse effect to that filter's adverse effects. Normally equalizers are also filter structures. 3.2.5 Geometrical Representation of BPSK Signals We know that BPSK signal carries the information about two symbols. Those are symbol ‘I’ and symbol '0", We can represent BPSK signal geometrically to show those two symbols. (i) From equation 3.2.6 we know that BPSK signal is given as, s(t) = b(t) ¥2P cos (2m fy!) ++ (3.2.15) (ji) Let's rearrange the above equation as, s() = b(t) {PT “fe =C2Hf 9 w= (3.2.16) b (iii) Let 6 0 = Ae cos (2m fy #) represents an orthonormal carrier signal. Equation & 3.2.14 also gives equation for cartier. It is slightly different than $, (t) defined here. Then: we can write equation 3.2.16 as, s(t) = b® YPT, 00 ww (3.2.17)Digital Communication 3-12 Passband Data Transmission (iv) The bit energy E,, is defined in terms of power ‘P’ and bit duration T,, as, E, = PT vw» (3.2.18) ©. Equation 3.2.17 becomes, s@) = +E, ww (3.2.19) Here b(t) is simply £1.- (v)Thus on the single axis of 4, (t) there will be two points. One point will be located at +./E, and other point will be located at -J/E,. This is shown in Fig. 3.2.7. Represents Represents symbol °0" symbol "1" ca mE wea -}-——e.n§ ——| Fig. 3.2.7 Geometrical representation of BPSK signal At the receiver the point at + JE, on 6, (t) represents symbol ‘I’ and point at -/E, represents symbol ‘0’. The separation between these two points represent the isolation and ‘0’ in BPSK signal. This separation is normally called distance ‘a’. 32.7 it is clear that the distance between the two points is, d = +E, -(-/E) d = 2JE, w= (3.2.20) As this distance ‘d’ increases, the isolation between the symbols in BPSK signal is more. Therefore probability of error reduces. 3.2.6 Bandwidth of BPSK Signal The spectrum of the BPSK signal is centered around the carrier frequency fy. lef, 1 then for BPSK the maximum frequency in the baseband signal will be fy see Fig. 3.2.6. In this figure the main lobe is centered around carrier frequency fo and extends from fy ~ fi, to fo +f,» Therefore Bandwidth of BPSK signal is, BW = Highest frequency ~ Lowest frequency in the main lobe = fo + fy -Yo-fr) BW = 2f, «32.21Digital Communication 3-13 Passband Data Transmission Thus the minimum bandwidth of BPSK signal is equal to twice of the highest frequency contained in baseband signal. 3.2.7 Drawbacks of BPSK : Ambiguity in Output Signal Fig. 3.2.3 shows the block diagram of BPSK receiver. To regenerate the carrier in the receiver, we start by squaring b(t) V2P cos(2nfyt+0). If the received signal is -b(t) J2P cos (2n fy t +8) then the squared signal remains same as before. Therefore the recovered carrier is unchanged even if the input signal has changed its sign. Therefore it is not possible to determine whether the received signal is equal to b(l) or -b4). This result in ambiguity in the output signal. This problem can be removed if we use differential phase shift keying. But Differential Phase Shift Keying (DPSK) also has some other problems. DPSK is given in detail in the next section. Other problems of BPSK are ISI and Interchannel interference. These problems are reduced to some extent by use of filters. ump Example 3.2.1: Determine the minimum bandwidth for a BPSK modulator with a carrier frequency of 40 MHz and an input bit rate of 500 kbps. Solttion : The input bit rate indicates highest frequency of the baseband signal. Hence, = 500 kbps = 500 kHz. > From equation 3.2.21, the bandwidth of the BPSK system is given as, BW = 2f, = 2x 500 kHz = 1 MHz Review Questions 1. Explain BPSK system with the help of transmitter and receiver, and state its advantages/disadvantages over other system. 2. Derive an expression for spectrum of BPSK system and hence calculate the bandwidth required.Digital Communication 3-14 Passband Data Transmission 3.3 Differential Phase Shift Keying (DPSK) Principle : Differential phase shift. keying. (DPSK). is. differentially coherent modulation method. DPSK does not need a synchronous (coherent) carrier at the demodulator. The input sequence of binary bits is modified such that the next bit depends upon the previous bit. Therefore in the receiver the previous received bits are used to detect the present bit. 3.3.1 DPSK Transmitter and Receiver 3.3.4.1 Transmitter | Generator of DPSK Signal Fig. 3.3.1 shows the scheme to generate DPSK signal. Input sequence s(t) = b(t) V2P cos(2nlgt) = £V2P cos(2nte!) DPSK signal Fig. 3.3.1 Block diagram of DPSK generate or transmitter Operation and waveform of transmitter The input sequence is d(). Output sequence is b(t) and b(t~T,) is the previous output delayed by one bit period. Depending upon values of d (t)andd (tT), exclusive OR gate generates the output sequence b(!). Table 3.3.1 shows the truth table of this operation. di bt -Th) by o(-1V) Ov) o(-1v) O(-1v) 1) 10V) atv) oy) avy 1Vv) 1(v) o(-tv) Table 3.3.1 Truth table of exclusive OR gate An arbitrary sequence d (t) is taken. Depending on this sequence, b(t) andb(t-T}) are found. These waveforms are shown in Fig. 3.3.2. The above table 3.3.1 is used to derive the levels of these waveforms.Digital Communication 3-15 Passband Data Transmission eps le Fig. 3.3.2 DPSK waveforms From the waveforms of Fig. 3.3.2 it is clear that b(t-T,) is the delayed version of b(® by one bit period T,. The exclusive OR operation is satisfied in any interval ie. in any interval b(t) is given as, bi) = d() @b¢-T,) -- 83.1) While drawing the waveforms the value of b(t-T;) is not known initially in interval no.1. Therefore it is assumed to be zero and then waveforms are drawn. Important conclusions from the waveforms 1. Output sequence b(t) changes level at the beginning of each interval in which d(f)=1 and it does not changes level when d(!)=0. Observe that d(3)=1, hence level of b (3) is changed at the beginning of interval 3. Similarly in intervals 10, UL, 12 and 13 d()=1 Hence b(t) is changed at the starting of these intervals. In interval 8 and 9 d(#)=0. Hence b(t) is not changed in these intervals. 2. When d()=0, b()=b@-T,) and When d (t)=1, b()=bE-T,) 3, In interval no.1. we has assumed b(t-T,)=0 and we obtained the waveform as shown in Fig. 3.32. If we assume b(t-T,)=1 in interval no. 1, then the waveform of 6(¢) will be inverted. But still b(t) changes the level at the beginning each interval in which d({)=1 4. The sequence b(t) modulates sinusoidal carrier.Digital Communication: 3-16 Passband Data Transmission 5. When f(t) changes the level, phase of the carrier is changed. Since b() changes its level only if u(t) =1 ; It shows that phase of the carrier is changed only if d(jak In BPSK phase of te careier changes om both the symbol “L’ and ‘0’. Whereas in DPSK phase of the carrier changes only on symbol 'I'. This is the main difference between BPSK and DPSK. J 6. Always two successive bits of d(#) are checked for any change of level. Hence ‘one symbol has two bits, Symbol duration (T) = Duration of two bits (2T;,) ie, T = 2M, = (3.3.2) As shown in Fig. 3.3.1, the sequence (t) is applied to a balanced modulator. The balanced modulator is also supplied with a carrier J2P cos (2m fy #). The modulator output is, si) = b(t) V2P cos (2n fy 8) = (3.3.3) cos (2r fot) w+ (3.3.4) = The above equation gives DPSK signal. Fig. 3.3.2 shows this DPSK waveforms. As shown in the waveforms the phase changes only when d({)=1. 3.3.1.2 DPSK Receiver Fig. 3.3.3 shows the method to recover the binary sequence from DPSK signal. Fig. 3.3.3 {a) and (b) are equivalent to each other. Fig, 3.3.3(b) represents DPSK receiver using correlator. Fig. 3.3.3(a) shows multiplier and integrators separately Operation of Receiver 1. Phase shift in received signal : During the transmission, the DPSK signal undergoes some phase shift @. Therefore the signal received at the input of the receiver i Received signal = 8 (t) ¥v2P cos (2n fy t +8) w» (3.3.5) 2. Multiplier output : This signal is multiplied with its delayed version by one bit. Therefore the output of the multiplier is, Multiplier output = b()b( -T,) (2P) cos (2r fy +0) cos [2nfy 1-T)) +0]. (3.3.6) We know that, cos(A) cos (B) = feos (A ~B) + cos (A +B)]Digital Communication 3-47 Passband Data Transmission bit} b(t-T,,) (2P) cos(2nfot+ 0) cos[2nfy(t-T,,)+ 0] pC Integrator if bit) VRP casiaefgt+ 0) _| Synchronous ° or Mutipior forty D(L-T,) V2P cos[2nfglt-T,)+ 0] a {a) (b) Fig. 3.3.3 (a) DPSK receiver (b) Equivalent diagram of DPSK receiver using correlator Here A= Infyt+@ and B= 2nfy(t-T,)+0 -. Multiplier output = b(t) b(t-T,) P few 2m fy Ty +00 [+= (:-E) +} B37) fo is the carrier frequency and T, is one bit period. T;, contains integral number of cycles of fy. We know that, 1 fo a IfT,, contains 'n’ cycles of fy then we can write, n fo" = fo=z, fol, = 0 w= (3.3.8)i Communication 3-18 Passband Data Transmission Putting fy T =n in first cosine term in equation 3.3.7 we get T, Multiplier output = Ob e-TopP aan (- ye =} Since cos2xn=1, the above equation will be, Multiplier output = b (i) b(t -T,) P+b@b(t -T,) P cos [ext (+-% +a] 83.9) 3. Integrator : The above signal is given to the integrator, In the k" bit interval, the integrator output can be written as, kTy sy (kT,) = b(kT,)b[(k-1)T,]P fat (k=1) Ty kT T + bKT,)O[K-HT,]P J cl anso(*—] 29] (k-D Ty The integration of the second term will be zero since it is integration of carrier over one bit duration. The carrier has integral number of cycles over one bit period hence integration is zero. Therefore we can write, 5 (kT,) = b(kT,b[k-1)T,] P[KT, —(k-1)T,] = b(kT,)b[(k-1)T,] PT, oe (3.3.10) Here know that PT, =E, ; ie. energy of one bit. The product b(kTs)b[(k-1) Tp] decides the sign of PT,,. The transmitted data bit d() can be verified easily from product b(kT,)b[(k-1)T,], We know from Fig. 3.3.2 when b()=b(t-T,),d (= That is if both are +1V or =1V then b (t)b(¢=T,)=1. Alternately we can write, If b@b(t-T,) = 1V then d(t) = 0 We know that b()=b(=T,) then d(f)=1. That is b()=-1V,b(t-T,)=#1V and vice versa. Therefore b(t)b(t-T,) =—1. Alternately we can write, If b()yb(t-T,) = -1V, then d(t) = 1 4. Decision device : The decision device is shown in Fig, 3.3.3 (b). We know that, 5) (KT}) = W(KT,Yb [KT] PT, from equation 3.3.10Digital Communication 3-19 Passband Data Transmission PT,,, th d(§=1 and te stk Tp) = 4 PTarten d@ <1 an +PT),,then d(t) =0 3.3.2 Bandwidth of DPSK Signal We know that one previous bit is used to decide the phase shift of next bit. Change in b() occurs only if input bit is at level ‘I’. No change occurs if input bit is at level ‘0. Since one previous bit is always used to define the phase shift in next bit, the symbol can be said lo have two bits. Therefore one symbol duration (T) is equivalent to tro bits duration (2T,,). ie. Symbol duration T = 2, 3.11) Bandwidth is given as, 2 Bw = 2 ei Ty or BW = fy o B3.12) Thus the minimum bandwidth in DPSK is equal to f, ; te. maximum baseband signal frequency. 3.3.3 Advantages and Disadvantages of DPSK DPSK has some advantages over BPSK, but at the same time it has some drawbacks. Advantag: 1) DPSK does not need carrier at its receiver. Hence the, complicated circuitry for generation of local carrier is avoided. 2) The bandwidth requirement of DPSK is reduced compared to that of BPSK. Disadvantages : 1) The probability of error or bit error rate of DPSK is higher than that of BPSK. 2) Since DPSK uses two successive bits for its reception, error in the first bit creates error in the second bit. Hence error propagation in DPSK is more Whereas in PSK single bit can go in error since detection of each bit is independent.Digital Communication 3-20 Passband Data Transmission 3) Noise interference in DPSK is more. In DPSK, previous bit is used to detect next bit. Therefore if error is present in previous bit, detection of next be can also go wrong. Thus error is created in next bit also. Thus there is tendency of appearing errors in pairs in DPSK. ina Example 3.3.1: The bit stream 1011100012 is to be transmitted using DPSK. Determine the encoded sequence and transinittd phase sequence. Solution : Fig. 3.3.4 shows the encoded bit stream b(t) and the transmitted phase. The input bit stream is represented as dit). The encoding waveforms are shown as per the DPSK generator of Fig. 3.3.1 and Table 3.3.1 bt -Tp) is the encoded sequence delayed by one bit period. DPSK encoded sequence bit} 0. C t | Fig. 3.3.4 DPSK waveforms In the above figure observe that the delayed output sequence b(t-T,) is assumed ‘0’ intially. The encoded sequence b(t) is given as, WH) = dQ) @ WE-T,) From equation 3.3.1band Data Transmission Digital Communication 3-21 In the Fig. 3.3.1 observe that the transmitted signal is given as, s(t) = b(t)}V2P cos(2rf,t) +J2P cos(2nf,t) { VIP cos(2nf,{) when Wi)=1 ie. s(t) = “" ~ VIP cos(2f,!) when bth =0 The above equations can also be written as, st) = | V2P cos(2nf,t+-0) when B()=1 ~ VIP cos(2xf,t+n) when b(t) =0 The transmitted phase sequence is shown in Fig. 3.3.4 as per the above equation. Review Questions 1. With the help of block diagram, waveforms and expressions explain the operation of DPSK transmitter and receiver. 2. What ave the advantages and disadvantages of DPSK ? What is the bandzvidth requirement of DPSK ? 3.4 Quadrature Phase Shift Keying May/lune-2006 iple * In communication systems we know that there are two main resources, ise. transmission power and the channel bandwidth. The channel bandwidth depends upon the bit rate or signalling rate f,. In digital bandpass transmission, a carrier is used for transmission. This carrier is transmitted over a channel. * If two or more bits are combined in some symbols, then the signalling rate is reduced. Therefore the frequency of the carrier required is also reduced. This reduces the transmission channel bandwidth. Thus because of grouping of bits in symbols, the transmission channel bandwidth is reduced. = In quadrature phase shift keying, two successive bits in the data sequence are grouped together. This reduces the bits rate of signalling rate (ie. f,) and hence reduces the bandwidth of the channel. * In BPSK we know that when symbol changes the level, the phase of the carrier is changed by 180°, Since there were only two symbols in BPSK, the phase shift occurs in two levels only. * In QPSK two successive bits are combined. This combination of two bits forms four distinct symbols. When the symbol is changed to next symbol theDigital Communication 3-22 Passband Data Transmission phase of the carrier is changed by 45° (x / 4 radians). Table 3.4.1 shows these symbols and their phase shifts. Sr.No. Input successive bits Symbol | Phase shift in carrier fat qv) O(-t¥) Ss ald | jo2 (tv) Oivy S) 3n/4 | i<3 Orv) av) Sy 5x14 | ied WV) 1(1V) Ss 714 Table 3.4.1 Symbol and corresponding phase shifts in QPSK Thus as shown in above table, there are 4 symbols and the phase is shifted by x/ 4 for each symbol 3.4.1 QPSK Transmitter and Receiver 3.4.4.1 Offset QPSK (OOPSK) or Staggered QPSK Transmitter Operation and waveforms. [April/May-2004 Step 1: Input Sequence Converted to NRZ type : Fig, 34.1 shows the block diagram of OQPSK transmitter, The input binary sequence is first converted to a bipolar NRZ type of signal. This signal is called b(). It represents binary ‘1’ by +1V and binary '0' by -1V. This signal is shown in Fig.3.4.2(a) AP; sint2ntgt) be sett) Binary Bipoier | yyy OPSK signal data NRZ level Demutiplexer Adder sequence | encoder att es sl) VP, cos (2af,t) Fig. 3.4.1 An offset QPSK transmitterDigital Communication 3-23 Passband Data Trans! Step 2 : Demultiplexing into odd and even numbered sequences The demultiplexer divides b(t) into two separate bit streams of the odd numbered and even numbered bits. b, (t) represents even numbered sequence and 0, (8) represents odd numbered sequence. The symbol duration of both of these odd and even numbered sequences is 27),. Thus every symbol contains two bits. 3.4.2 (b) and (c) shows the waveforms of b, () andb, Observe that the first even bit occurs after the first odd bit. Therefore even numbered bit sequence 6, (¢) starts with the delay of one bit period due to first odd bil, Thus first symbol of b, () is delayed by one bit period 'T;,’ with respect to first symbol of b, (t). This delay of T,, is called offset. Hence the name offset QPSK is given. This shows that the change in levels of b, (t)andb, (t) cannot occur at the same time because of offset or staggering. Step 3 : Modulation of quadrature carriers The bit stream 0,(t) modulates carrier 4{P, cos(2z fy t) and b, () modulates VP, sin(2xfyt). These modulators are balanced modulator. The two carriers AP, cos (2n fy Nand JP, sin (2 fy #) are shown in Fig. 3.4.2 (d) and (e). These carriers are also called quadrature carriers. The two modulated signals are, 5.) = b, YP, sin 2xfy 0 ws (34.1) and 5) () = b(t) yP, cos(2zfy 8) .. (3.4.2) Thus s, (f) ands, (t) are basically BPSK signals and they are similar to equation 32.3 and equation 3.2.5. The only difference is that T=27, here. The value of b, (t)andb, (t) will be +1V or -1V. Fig. 3.4.2 (f) and (g) shows the waveforms of s, ands, (t. Step'4 : Addition of modulated carriers The adder of Fig. 3.4.1 adds these two signals b, ())andb, (t). The output of the adder is OQPSK signal and it is given as, st) = sy (+s, () = by ()afP, cas (Zr fy +b, U)a/Py ste (2x fo) we (84.3) ‘Step 5 : QPSK signal and phase shift Fig. 3.42 (h) shows the QPSK signal iepresented by above equation. In QPSK signal of Fig. 3.4.2 (h), if there is any phase change, it occurs at minimum duration of T,,. This is because the two signals s, (f) ands, () have an offset of 'T,,”. Because of this offset, the phase shift in QPSK signal is 5 tt is clear from the waveforms of Fig, 3.42 that &, (Q)andd, () cannot change at the same time because of offset between them. Fig. 3.4.3 shows the phasor diagram of QPSK signal of equation 3.4.2.Digital Communication 3+24 Passband Data Transmission Soot { t 5 | 6 ‘odd—}-event ods. — Tt0 bit sequence 9 0 oy hod j= First symbol of b,(ty starts hare Le peinp Evenmumbered 4 | [bit sequence t { + tt fet belt) 8 t 7 oe 4 | | | Vr t { i Jf 4 [Tt Lh |~ First symbol of byt) stants here — btodap ee bn hh +—t+44 $B, cosl2nfot) © + NP alt Hai} i iW 1 tr Oz, i igs; e082 i 5 RE. Lo NT t BitNo: 0 . Phase Phase Phase 4 ‘shittot shit of snwft of | | Me ppp PRR EEE Fig. 3.6.2 QPSK waveforms (a) Input sequence and its NRZ waveform (b) Odd numbered bit sequence and its NRZ waveform (c) Even numbered bit sequence and its NRZ waveform (d) Basis function @4(¢) {e) Basis function 4, (t) (f) Binary PSK waveform for odd numbered channel (g) Binary PSK waveform for even numbered channel (h) Final QPSK waveform representing equationDigital Communication 3-25 Passband Data Transmission Since b, (t) andb, () cannot change at the same time, the phase change in QPSK signal will be maximum z/ 2. This is clear from Fig. 3.4.3. Amplitude = |stt)i ~ V2P, = 2,8) = VP, cos (2afgt) VP, sin (2nfgt) ie.blt)=4 bylt)=-4 VP, 60s (2nfgty i ' ! sin (2nfaly ait) = -VP 0s (rtgt) + VP, sin (xf) | 8(t) =P, c08 (2nfgt) + belt) = 4 4#L-----¥.-----2 bal = : balt)= 1 elt) = Fig. 3.4.3 Phasor diagram of QPSK signal 3.4.1.2 Non-Offset QPSK * We known that there is an offset of "T,’ between b, () andb, (). If we delay 6, @ by 'T,' then there will be no offset. Then the sequences b, (i) andb, (é) will change at the same time. This change will occur after minimum of ‘21, © Asa result, the signals 5, ())ands, () will have phase shifts at the same time. The individual phase shifts of s, (t) ands, (€) are 180% Because of this the amplitude variations in the waveform will occur at the same time in So (ands, (). Therefore these variations will be more pronounced in non offset QPSK than OQPSK. «Filters are used to suppress side bands in QPSK. Since phase changes by 180° in non offset QPSK, amplitude changes are more. Hence filtering affects the amplitude of non-offset QPSK. In OQPSK, the phase changes by 90°, hence amplitude changes during filtering are less. ¢ amplitude variations are more in non-offset QPSK, the signal is affected communication takes place through repeators. These repeators highly affect the amplitude and phase of the QPSK signal.Digital Communication 3-26 Passband Data Transmission 3.6.1.3 The QPSK Receiver ‘April/May-2004; April/May-2005 ‘Received signal a s(t) sin(2afgt) bn, VP, bWTe Py correlator - 1 sty, correlator - 2 \ cos 4121) s(t) C08 (2xf,t) 605(2fgt) sint2afgt) Fig. 3.4.4 QPSK receiver Fig. 3.4.4 shows the QPSK receiver. This is synchronous reception. Therefore coherent carrier is to be recovered from the received signal 3 (2). Operation Step 1: Isolation of carrier The received signal s(() is first raised to its 4" power, ie. s4 (lj. Then it is passed through a bandpass filter centered around 4fo. The output of the bandpass filter is a coherent carrier of frequency 4fy. This is divided by 4 and it gives two coherent quadrature carriers cos (2x fy 1) and sin (2x fy (0. Step 2: Synchronous detection These coherent carriers are applied to two synchronous demodulators. These synchronous demodulators consist of multiplier and an integrator. Step 3 : Integration over two bits interval The incoming signal is applied to both the multipliers. The integrator integrates the product signal over two bit interval (ie. T, = 27). Step 4: Sampling and multiplexing odd and even bit sequences At the end of this period, the output of integrator is sampled. The outputs of the two integrators are sampled at the offset of one bit period, T,. Hence the output ofDigital Communication 3-27 Passband Data Transmission multiplexer is the signal b(t). That is, the odd and even sequences are combined by multiplexer. To show that output of integrator depends upon respective bit sequence. * Let's consider the product signal at the output of upper multiplier. s(t) sin Qn fy ) = b, (t) YP, cos 2m fy t) sin (2 fot) +b, () YP, sin? (2nfyt) —... Gad) + This signal is integrated by the upper integrator in Fig. 3.4.4. (2k VT, (2k+1) Tp J s@sinQrfyndt=b,(QJR feos (2x fot) sinQQnfy dt (2k-1) Th (2k-1) Ty (2k+1) Thy +b. (OYPR sin? Qxfyt)dt ODT, Since Esin(2x) = sinx-cos.x and sin? (x) = 4 U1 ~ cos (20) * Using the above two trigonometric identities in the above equation, (21) Ty b(t) fe 2+ bf NT J s@sinQnfy Oat — ol J sindnfy dts 20 fooide (2k-1) Tp (2k-1)Ty (2k-D) Ty (2k+1) Tp (2k-1) Ty * In the above equation, the first and ‘third integration terms involves integration of sinusoidal cartiers over two bit period. They have full (integral number of) eycles over two bit period and hence integration will be zero. ke Th b. OSB _ f s(t) sin(2nfy dt = Pe OSF ate 42k-1) Th b. VP, 1 OE a, = be {PTs a (34.5) + Thus the upper integrator responds to even sequence only. Similarly we can obtain the output of lower integrator as b, (1) JP, Ty.| Communication 3-28 Eventhough bit synchronizer is not shown in Fig. 3.4.4, it is assumed to be present with the integrator to locate starting and ending times of integration. The multiplexer is also operated by bit synchronizer. The amplitudes of signals marked in Fig. 3.4.4 are arbitrary. They can change depending upon the gains of integrator. Ambiguity in the output : In Fig. 3.4.4 observe that even if the received signal is negative, the recovered carrier remains unaffected because of the 4! power conversion of the signal. Therefore it will not be possible to determine whether the transmitted signals were positive or negative fie. +b, (Q) or —b, () and +b, (or —b, (0 This is phase ambiguity in output similar to BPSK. This problem can be recovered by employing differential encoding and decoding of b(t). 3.4.1.4 Carrier Synchronization in QPSK Both the carriers are to be synchronized properly in coherent detection in QPSK. Fig. 3.4.5 shows the PLL system for carrier synchronization in QPSK. ‘Quadrature: carrier Phase shift N cos (2nfo!) of 90° mm oN “a iaw filter PLL Input |_Sevice ay signal {__________» Inphase carrier cos (2nfot Nx/2) Fig. 3.4.5 PLL system for carrier synchronization The fourth power of the input signal contains discrete frequency component at 4fy. We know that, cos4 (2m fot) = cos(Snfyt+2nN) Here ‘N’ is the number of cycles over the bit period. It is always integer value. When the frequency division by four takes place, the RHS of above equation becomes cos 2n fot +f). This shows that the output has a fixed phase error of SS Differential encoding may be used to nullify the phase error events. The PLL remains locked with the phase of '4fg' and then output of PLL is divided by 4. This gives a coherent carrier, A 90° phase shift is added to this carrier to generate a quadrature carrier.Digital Communication. 3-29 Passband Data Transmission 3.4.2 Signal Space Representation of QPSK Signals ()Fig. 3.43 shows the phasor diagram of QPSK signal. Depending upon the combination of two successive bits, the phase shift occurs in carrier (see table 3.4.1). That is the QPSK signal of equation 3.4.3 can be written as, s() = ¥2P, cos [ear f4(2m+1) 3] m=0,1,2,3 ww (34.6) Here, the above equation takes four values arid they represent the phasors of Fig.3.4.3. (2)The above equation can be expanded as, s(t) = J2P, cos (2 fy done [cams »4] - J2P, sin 2x fy t) sin [c2msn 3] (3)Let's rearrange the above equation as, s() = { PT ew[omen ell LF cos ashy ) - [er T; sin (cam +1) Fil fe sin (Qn fy) G47) ar 4) Let (0) = ie 0s (27 fy 8) B48) - and $) (0) = e sin (2r fy #) B49) The above two signals are called orthogonal signals and they are used as carriers in QPSK modulator. 6) Let b, (t) = View [amen 5] .w 34.10) and b, () ~V2 sin [com +04] ow (3411) (6) With the use of equation 3.48 to equation 34.9 we can write equation 3.4.7 as, 1 1 s(t) = ¥PT; “Fpbo Oh O+VRT yt (428 = fe Br,04,0+ 2 26.000 (T, = symbol duration and T, = 27,Digital Communication - 3-30 Passband Data Transmission or Th = wh on (3.4.12) Then the above equation becomes, S(t) = PT, by WO, + YP. Th be M42 ~~ (3.4.13) Since bit energy E, = P.T; (3.4.14) $(t) = fy By (td) C+ By by (B42 CH ww (34.15) Comments The above equation gives signal space representation of QPSK signal. The two orthogonal signals, $ (t) andy (0) form the two axes of the signal space. Fig. 3.4.6 shows the signal space representation. b,(0 = b(t) ~6,(t)-<- b= 4 byt) 1 A Dg{t)= bylt)= 1 oatd= fFsincant Fig. 3.4.6 Signal space representation of QPSK signals + The possible 4 nal points are shown by small circles on 4, $2 plane. From cach signal point, we obtain two bits, For example from point ‘A’, we obtain two bits as (1, 1) and from “B’ we obtain bits, as (-1, 1). The distance of any signal point from origin ‘0’ given as, Distance ‘OB’ = JT, +21, (227; =T) .. (3.4.16) (PT, =B) (34.17)Digital Communication 3-H Passband Data Transmission Thus the length of each signal point from origin is /E,- * We know that d, (!) andb, (!) represent two successive bits. There is an offset of 'T;’ between b, ()andb, (t). Therefore &, (Pand>,,(t) both cannot change their levels simultaneously. Therefore either b, () orb, (t) can change at a time. «Let's say that b, (f)=b, (#) =1 representing signal point ‘A’ in Fig. 3.4.6. In the next bit interval if b, ()=~1, then signal point will be 'D’. Otherwise if 8, (0) changes its level (ie. 6, (2) ==1), then next signal point will be 'B.. Thus from signal point ‘A’, then next signal points will be either ‘D'or 'B’. Distance between signal points : Normally the ability to determine a bit without error is measured by the distance between two nearest possible signal points in the signal space. Such points difiered in a single bit. For example signal points ‘A’ and 'B' are two nearest points since they differ by a ‘tb, (t). As ‘A’ and 'B' becomes closer to each other, the possibility of error increases. Hence this distance should be as large as possible. This distance is denoted by ‘d’. In Fig, 3.4.6, the distance between signal points ‘A’ and 'B' is given as, a = (VE)? d= PE we (3.4.18) or d= 2/PT, =2 JE, on (BA.19) Compare this distance with the distance of BPSK signals given by equation 3.2.20. This shows that the distance for QPSK is the same as that for BPSK. Since this distance represents noise immunity of the system, it shows that noise immunities of BPSK and QPSK are same. 3.4.3 Spectrum of QPSK Signal Step 1: PSD of NRZ waveform The input sequence b(t) is of bit duration T,. It is NRZ bipolar waveform. In section 3.2.4 we have obtained the power spectral density of such waveform as, 42 SW) = Vz T, ae from equation 3 /P,, then above equation becomes, w= (3.4.20) The above equation gives power spectral density of signal b(t).Digital Communication 3-32 Passband Data Transmission Step 2 : PSDs of even and odd numberd sequence This signal is divided into b, () andb, (i) each of bit period 2T,,. If we consider that symbols 1 and 0 are equally likely, then we can write power spectral densities of b, (and, (as, 2 5.) = RT (oe) 4.21) . 2 and 5.) = PT, payee we (34.22) In the above two equations we have just replaced 7;, by T, andT, is the period of bit in b, ()andb, (0). Step 3 : PSD of QPSK signal Since inphase and quadrature components [b, (t)andb, (f)] are statistically independent, the baseband power spectral density of QPSK signal equals the sum of the individual power spectral densities of b, (i) andb, (i) ie., Self) = Se(N +5.) - sin(nf 7.) = 2P.T, (S| ww (34.23) This equation gives baseband power spectral density of QPSK signal. Upon modulation of cartier of frequency fy, the spectral density given by above equation is shifted at +f). Thus plots of power spectral density of QPSK will be similar to that BPSK given in Fig. 3.25 and Fig. 3.26. 3.4.4 Bandwidth of QPSK Signal We have seen that the bandwidth of BPSK signal is equal of 2f,. Here T, a is {a the one bit period. In QPSK the two waveforms 6, (#) andd, (t) from the baseband signals. One bit period for both of these signals is equal to 27. Therefore bandwidth of QPSK signal is, or BW = fy -» GA24) Thus the bandwidth of QPSK signal is half of the bandwidth of BPSK signal. Earlier we have seen that noise immunity of QPSK and BPSK is same. This shows that inspite of the reduction in bandwidth in QPSK, the noise immunity remains same asDigital Communication 3-33 Passband Data Transmission compared to BPSK. BW of QPSK can also be obtained by plotting equation 3.4.20 as shown in Fig, 3.4.7 below. Fig. 3.4.7 Plot of power spectral density of QPSK signal BW = Highest frequency - Lowest frequency in main lobe -z¢-(-z) which is same as we obtained in equation 3.4.24, 3.4.5 Advantages of QPSK QPSK has some definite advantages and disadvantages as compared to BPSK and DPSK. Advantages : 1) For the same bit error rate, the bandwidth required by QPSK is reduced to halt as compared to BPSK. 2) Because of reduced bandwidth, the information transmission rate of QPSK is higher. 3) Variation in OQPSK amplitude is not much. Hence carrier power almost remains constant.Digital Communication 3-34 Passband Data Transmission ym Example 3.4.1: Write the waveforms for a binary sequence 101100 modulated under QPSK. Solution ; Let us assume the offset QPSK or OQPSK system. The waveforms are drawn similar to those in Fig. 3.42. Fig. 34.8 shows the waveforms for given sequence. The input sequence is written at the top alongwith bit numbers Fig.(a) shows NRZ waveform for odd numbered bits. Fig.(b) shows the NRZ waveform for even numbered bits. Observe that the signal is assumed ‘O’ at the beginning for even numbered bit sequence i. b,(t). The two quadrature carriers are shown in Fig. (¢) and (d). Fig. (e) shows the binary PSK signal generated due to b, (i), ie. so(2). Fig. (f) shows the binary PSK signal generated due to b,(f), ie. s,(f). The signals s,(i) and s,4!) are added to get the final QPSK waveform s(t). It is shown in Fig. 3.4.8(g). Observe that the phase shifts of F occur in this waveform. The individual binary PSK waveforms are staggered due to offset QPSK. (See i ig. 3.4.8 on next page) ‘wm Example 3.4.2: In a QPSK sustem, the bit rate of NRZ stream is 10 Mbps and carrier frequency is 1 GHz, Find the symbol rate of transmission and bandiwidth requirement of the channel. Sketch the power spectral density of the QPSK signal. Solution : (i) To obtain symbol rate and bandwidth Bit rate f, = 10 Mbps Bandwidth of the QPSK system is equal to Lit rate Hence BW = f, = 10 MHz Symbol duration and bit duration are related as, T, = 2, Lilith Symbol rate = fe=ap- = HE = 5 MHz (ii) To obtain power spectral density ; Carrier frequency f, = 1 GHz psd of NRZ signal is given as, _ sin(n fT) Spl) = 2RT, “ge ] Upon modulation of the carrier of frequency f,, the spectral density given by above equation is shifted at + f,.Digital Communication 3-35 Passband Data Transmission _| sequence [Pot Even numbered o bit sequence I pope -t Lo i WP le ti Peosi2st) 0 1 ~F, w Lanett) g Isat = HBC, cos 2att) j= “a a= VE 0; sins i. || OPSK Signal (jst 5,0 5500 Fig. 3.4.8 QPSK waveformDigital Communication 3-36 Passband Data Transmission Fig. 3.4.9 shows the psd plot. fot hy Fig. 3.4.9 Spectral density plot of QPSK In above figure observe that the two lobes are placed at +1 GHz. Width of the main lobe is (1 + 0.005) GHz to (1 - 0.005) GHz. This is equal to bandwidth requirement of the system. That is, BW = 1.005 - 0.995 = 0.01 GHz = 10 MHz im Example 3.4.3; For the input binary sequence {yet ao a 1, 1}. Find the transmitted phase sequence and sketch ik ‘hanoiled waveform for QPSK Solution : The given sequence {b,} is in NRZ form. Fig. 3.4.10 shows the waveforms of QPSK. (See on next page) Fig, (a) and (b) shows the NRZ waveforms of even numbered and odd numbered samples. Fig. (c) and (d) shows the quadrature carriers. Fig, (e) and (f) shows the PSK modulated quadrature carriers. Fig. (g) shows fhe QPSK waveform. The transmitted phase is shown at the bottom. Review Questions 1. With the help of block diagram and relevent expressionsfwnvejorms explain QPSK transmitter and receiver. 2. Compare psd and bandwidth requirements of QPSK with that of BPSK, 3. Represent QPSK signals in the signal space and find distance between them, What is the significance of this distance ?Digital Communication “Tit number [input sequence :| (Odd numbered. sequence bt) E Fan xt) [seve 2 (0B, cos (2 SQlt) = {t)9P, sin (2x) PSK signat sit) 980 #56) i t — | ' if ‘ - | ranemitos | | phase sna 4 1 fot c tof pp os O— 0-0 —--F - 1a $ inca i Ee + Fig. 3.4.10 Waveforms of QPSK 3.5 M-ary PSK BPSK transmits one bit at a time and it has only two symbols. Hence whenever the symbol is changed, the phase shift is, “tt a . Phase shift in QPSK = ‘umber of symbols aye Tor 180°Digital Communication 3-38 Passband Data Transmission In QPSK two successive bits are combined to form 4 distinct symbols. Hence whenever symbol is changed, the phase shift is, 2 2n = Fe n/2or 90° Phase shift in QPSK= Sana = This can be extended further for “N* bits. IF we combine N successive bits, then there will be 2 =M possible symbols. Whenever the symbol is changed the phase shift is, thi 2n ig 5 Phase Mea =———— ti a G5. Phave shift in Mary PSK = ae = iy 5.1) The duration of each symbol will be NT, thus NT, = Ty -- B52) Since there are M-symbols, this method is called M-ary PSK, The transmitted waveform is represented in M-ary PSK as, S() = J2P, cos (2x fy t+6y,) + B53) m= OL, Breen M =H The symbol phase angle is given as, Om = Ome + G54) 3.5.1 Signal Space Diagram Equation 3.53 can be expanded as, > C05 Gy, COS (2 fy )— J2P. sind,, sin(2nfy Let's rearrange the above equation as, 7 cos by cos (2% fo 0 - YP Te sin, sin (27 fy 8) = PLT, 6054), 9) O-yP.T, sind, by -- 85.6) H = casi 3.5.7} Jere 1 = 7 eos nfo w= (35.7) D and tO =F sneha -- G58)Digital Communication 3-39 Passband Data Transmission The above two equations are orthonormal waveforms. Fig. 3.5.1 shows the signal space diagram based on equation 3.5.6. The orthonormal carriers , (t) and, (#) form two axes. The signal points $y.5;, m1 are placed on the circumference of the . The signal points are equispaced with the phase shift of or The distance of cire each signal point from the origin is ¥P. T.. 3 0B cosretgn Fig, 3.5.1 Signal space diagram or geometrical representation of M-ary PSK signals Here PT, = E, (Symbol energy) G59) Thus we can say that QPSK is the special case of M-ary PSK with M =4. Then the signal space diagram of QPSK and 4-ary PSK will be similar 3.5.2 Power Spectral Density of M-ary PSK PSK and QPSK are the special cases of M-ary PSK. The symbol duration for M-ary PSK is given by equation 3.5.2 as, T, = NT, wo» 3.5.10) Here N is the number of input succe spectral density of QPSK is given as, sin(af TP fT, ive bits combined. The baseband power from equation 3.4.23 Sargrsey ) = 2FT, If we put PSK ie, NT), in above equation we will get power spectral density of M-ary sine NT) epnTot 5.11) afNTy,Digital Communication 3-40 Passband Data Transmission The above equation gives the power spectral density of baseband M-ary PSK. ‘T,’ is the duration of one bit. Fig. 3.5.2 shows the plot of Sp (f) for M-ary PSK. Fig. 3.5.2 Plot of power spectral density of baseband M-ary PSK signal When the baseband M-ary PSK signal modulates the carriers of frequency fy, the spectrum shown in Fig. 3.5.2 above, is centered around + fy. 3.5.3 Bandwidth of M-ary PSK The spectrum of Fig. 35.2 is placed at carrier frequency after modulation. Therefore the bandwidth required by the system is equal to the width of the main lobe ie, BW = f.-Cf) = 2% = 2 uped 1 Ty “ds 7. * Symbol period = 2 WT = "5h (7, =NT) ws (3.5.12) = we ap el “N- (4 =i) w (35.13) Above equations shows that, as the number of successive bits (N) per symbol are increased, the bandwidth reduces. 3.5.4 Distance between Signal Points (Euclidean Distance) In BPSK and QPSK signal space diagrams, the determination of distance between two neighbouring points is quite straight forward. This distance is required to calculate probability of error. Therefore its calculation is more important. Fig. 3.5.3 shows the signal space diagram of M-ary PSK. In this diagram M=8 In figure, thePassband Data Transmission This value is same as that we have obtained earlier in equation 3.4.18. 3.5.5 Transmitter and Receiver of M-ary PSK 3.55.4 Mary PSK Transmitter Fig. 3.54 shows the simplified M-ary PSK transmitter. The serial to parallel converter forms a symbol of 'N' successive bits. That is the output of serial to parallel converter is 'N’ bit word, Senat to Meary PSK parallel oa converter sigeal Fig. 3.5.4 Transmitter block diagram The digital to analog converter output remains unchanged till last N‘ bit is received. Then depending upon the input ‘N' bits, the output of D/A converter is defined. This output is m(0). Again the serial to parallel converter starts taking bits for next 'N' word. The output of D/A converter remain unchanged till last bit is received. Thus 1 (f) is hold for the period of NT,,. m(t) takes 2‘ = M different values, depending upon the input bits. The vollage m(l) is applied to modulator. This modulator modulates the phase of sinusoidal carrier depending upon the amplitude of the symbol 1m (1). 3.552 Neary PSK Receiver Fig. 35.5 shows the receiver of M-ary PSK. It is similar to QPSK receiver. The input signal s(d) is raised to M! power. The bandpass filter extracts the frequency component Mf. This frequency is divided by 'M’ to obtain carrier frequency fy. The coherent carriers are thus generated and applied to the two multipliers. The outputs of the multipliers are given to the integrators. The integrators integrate over the period of T, =NT,. The outputs of the integrators are sampled after the period T, in every cycle and applied to the analog to digital converter. The integrator outputs are proportional to T, P, andT, P,. These voltages are applied toDigital Communication 3-43 Passband Data Transmission o_M-ary PSK signal st) Parallel to serial s(t) sin(2afgt) cost2ntet) sin(2rfgt) Fig. 3.5.5 M-ary PSK receiver A/D converter, which reconstructs ‘N’ bit symbol. This ‘N’ bit symbol is given to the parallel to serial converter. It then generates the bit sequence b(t). ‘w= Example 3.5.1: A 4ary PSK has the transmitted wavefornts, in 5) = se(2nf, 3) ww» (3.5.15) i = 0,1,2,3and 0. Thus the minimum distance will be distance between S, and S;. We know that distance between S, and S. will be, diy = JE, +5 = 26 Since $, is at the centre of S and S2, the distance between S, and S; will be, ays = B= 20.707 JE In standard 8-level PSK, all the 8 symbols will lie on the circle of radius JE. Equation 3.5.14 gives the distance as, dy = 2YE sin F = VE, sin 5 =0.765 JE From above result, it is clear that, distance is more in case of standard PSK. As the distance between the signal points becomes more, the probability of error reduces. Hence it is clear from above results that probability of error is less in case of standard B-level PSK. imp Example 3.5.5 Derive an expression for the spectral spread of 16-ary PSK system. Solution : Power spectral density of M-ary PSK is given by equation 3.5.11 as,Digital! Communication 3-48 Passband Data Transmission agra i T b S For M = 16, N = logy M=log, 16=4. Putting this value of N in above equation, . 2 Sy(f) = 2P,* ver | 4 sin(4nfT),) }* on I) | Review Questions 1. Explain M-ary PSK system with the help of transmitter and receiver. 2. What is the bandwidth requirement of Meary PSK ? 3.6 Quadrature Amplitude Shift Keying (QASK) [or Quadrature Amplitude Modulation (QAM)} We have seen in the preceding sections that the correct detection of the signal depends upon the separation between the signal points in the signal space. In case of PSK systems all points lie on the circumference of the circle. This is because PSK signal has constant amplitude throughout. If amplitude of the signal is also varied, then the points will lie inside the circle also on the signal space diagram. This further increases the noise immunity of the system. Such system involves phase as well as amplitude shift keying, It is called quadrature amplitude phase shift keying or simply QASK. It is also called quadrature amplitude modulation or QAM. 3.6.1 Geometrical Representation and Euclidean Distance of QASK Signals (or Signal Space Representation or Signal Space Constellation) Let us consider the case of 4 bit symbol. Then there will be 24 =16 possible symbols. In the QASK system, such 16 symbols are represented geometrically as shown in Fig. 3.6.1. [It shows Geometrical representation of 16 QASK signals. The distance from the neighbouring points is d=2a. Let the signals be equally likely. Then the average energy associated with the signal can be obtained as (Considering first quadrant), E, =} [fat +02) + (002 +02) +(a2 +902) + (90? + 902)]Digital Communication 3-49 Passband Data Transmission ot) . : . ° 56 sts Bie Sag Fig. 3.6.1 Geometrical representation of 16 signals in QASK system = 10a? a= fO1E, «» (3.6.1) Since d=2a we have, d= 2/0i, = J04E, -» 3.6.2) This gives the distance between two signal points in 16 QASK. In each symbol there are 4 bits. Hence bit energy and symbol energy are related as, E = 4B, d = f04«4E, Jf16E, ++ (3.6.3) ‘The distance for QPSK is given from equation 3.4.19 as dopsx = 21Ey = /4E, + G64) and the distance for 16-ary PSK is given from equation. W digpsk = 2E sin 2p sine B= AE, = 2/0I5E,Digital Communication 3-50 Passband Data Transmission = /06E, s+ (3.6.5) Thus the distance of 16-QASK is greater than 16-ary PSK where as it is less than QPSK. 3.6.2 Transmitter and Receiver of QASK 3.6.2.1 Transmitter of QASK Signal for 4-bit Symbol The signal in Fig. 3.6.1 is represented as, s() = ky ab, +k, 00,0 (3.6.6) Here k, and ky will take values of #1 or £3. $; (!)and}y (f) are orthogonal carriers having the values as follows oi () = w- (3.6.7) and 6 () -» (3.6.8) " From equation 3.6.1 we know that, a= (OIE, o» (3.6.9) -. We can write equation 3.6.6 as, s() = ky 02 costxfy Nk, oats sin(2n fo 0 (3.6.10) 3 = FT, We know that then the above equation becomes, s(t) = ky (OLE cos (2x fy +k (OZP, sin (2n fy t) e» 3.6.11) This equation gives the QASK signal. Here k, andk, defined the amplitude of the modulated signal, Fig 3.6.2 shows the transmitter for 4 bit QASK (or 16-QASK) system. The input bit steam is applied to a serial to parallel converter. Four successive bits are applied to the digital to analog converters. These bits are applied after every T, second. T, is the symbol period and T, = 4T,. Bits b, andb, , are applied to upper digital to analog converter and by, andb;, 3 are applied to lower digital to analog converter. Depending upon two input bits, the output of digital to analog converter takes four output levels. Thus A,(QandA, ( takes 4 levels depending uponPassband Data Transmission YP ,cos(2rtgt) QASK signal sit) Clock at every T, Fig. 3.6.2 Generation of QASK signal combination of two inputs bits. A, (f) modulates the carrier ./P, cos (2n fy Hand A ,(f) modulates ,/P, sin (2x fy #). The adder combines two signals to give QASK signal. It is given as, sO = A, YR cosQnfy +A, fF sin(Qn fot)... (3.6.11 (a) If we compare the above equation with equation 3.6.11 We can write A,@andA,(t) = +¥02 or +302 w+ (3.6.12) (depending upon input to D/A converter) 3.6.2.2 Receiver of QASK Signal Fig. 3.6.3 shows the receiver of 16-QASK (4 bits QASK) system. The put signal s(t) is raised to 4!" power. It then passed through a bandpass filter centered around the frequency 4f, the signal is then divided in frequency by four. It gives a coherent carrier cos (2n fy t). Quadrature carrier sin (2x fo) is produced by phase shifting of 90°. The inphase and quadrature coherent carriers are multiplied with QASK signal s(t). Since the amplitudes of A,(f) and A, (#) are bit constant and equal, let us check whether we can really recover the carrier correctly. The 4” power QASK signal is, s4() = P2[A, (0) cos (2 fo) +Ag(D sin (rf OT 3.6.13) This signal is passed through a bandpass filter of 4fy. Therefore we will consider only the frequencies of 4fy ie,Digital Communication 3-52 Passband Data Transmission SUNEAS(IVP, cosd(Qalgth+ A,(ONP, sin(nfgt ‘Analog 2 to digital Bit conwertery sequence a bit) 4 | I é Des | sin(2nfgt) Fig. 3.6.3 4-bit QASK receiver block diagram si) = 2 faso + AN) ~ 6 AZ) AZO] cos 4 (2x fo f) B +3 [Acl) Ag {A20- AZ O}] sin 42 fy (3.6.14) The average value of second term will be zero, hence only first term is passed through a bandpass filter centered at 4f. This happens because all power of A,(t) and A,(t) in the first term are even. The integrators integrate the- multiplied signals over one symbol period. The output of integrators at sampling period give A,(t)andA,(). The analog to digital converters gives the four _ bits bye, 1/b442 Ande, 3. The parallel to serial converter then generates the bit sequence b(t). 3.6.3 Power Spectral Density and Bandwidth of QASK Signal The QASK equation given by equation 3.6.11 is similar to that of M-ary PSK given by equation 3.5.5. Therefore power spectral density of baseband QASK signal will be, ‘sin (n fT.) aft, The above equation gives power spectral density of A,(i)andA,(0). When they modulate the carriez, the main lobe given by above equation is shifted at carrier frequency fo: si) = ar o» (3.6.15) _ PT, [sin nf fo) T, P Bt [rareealt : so = ae re | [ear | 69Digital Communication 3-53 Passband Data Transmission This equation gives power spectral density of QASK signal. Bandwidth of QASK Signal : We can plot the main lobe of QASK signal given by equation 3.6.15. This plot is exactly similar to that shown in Fig. 35.2. Therefore bandwidth will be, BW = Ch) =f since f, = (since T, + 8.6.17) = (since f, w+ (3.6.18) ‘Thus the bandwidth and power spectral density of QASK is similar to that of M-ary PSK. 3.6.4 Comparison between QASK and QPSK QASK and QPSK are both quadrature modulation techniques. They have certain advantages and disadvantages over each other. Table 3.6.1 shows the comparison between QASK and QPSK. Modulation Quadrature phase Quadrature amplitude and phase 2 | Location of signal | All signal poinis placed on| Signal points are replaced points circumference of circle symmetrically about origin 2YOTBE, tor 16 symbols | 2VO%E, for 16 symbols Complexity Relatively simpler Better than QASK Relatively complex Noise immunity Poor than QPSK. But better than M-ary PSK. 8 | Error probability Less than QASK Higher than QPSK. Lower than M-ary PSK, 7 | Type ot Coherent Coherent demodulation Table 3.6.1 Gomparison of QPSK and QASKDigital Communication 3-54 Passband Data Transmission Review Questions 1. Explain the differences between QASK and QPSK systems giving corresponding expressions and signal space representations. 2. Explain QASK system with its transrtitter, receiver and signal space representation. 3. Winat is the bandwidth of the transmitter in terms of input bit duration i.e., input signal bandwidth? Explain the meckanism by which the bandwidth reduction is made possible in QASK system ? 3.7 Binary Frequency Shift Keying (BFSK) ‘(May/iune’= 2006) In binary frequency shift keying, the frequency of the cartier is shifted according to the binary symbol. The phase of the carrier is unaffected. That is we have two different frequency signals according to binary symbols. Let there be a frequency shift by Q. Then we can write following equations. If BQ) = 1; Sy (t) = J2P, cos (2m fy +Q)E @71) If b(t) = 0; 5 (t) = «[2P, cos (2n fy -2)t (37.2) ‘Thus there is increase or decrease in frequency by 9. Let us use the following conversion table to-combine above two FSK equations. b(t) Input Table 3.7.1 Conversion table for BPSK representation ‘We can write equation 3.7.1 and equation 3.7.2 combinely as, s(t) = 2B, cos [(2nfy +4.) f] ~- G73) Thus when symbol '1' is to be transmitted, the carrier frequency will be fo+(2} i If symbol ‘0’ is to be transmitted, the earrier frequeney will be fy -(3) 7 ta =to +2 for symbol ‘1’ w= G74) fh =h-& for symbol ‘0’ B75)Digital Communication 3-57 Passband Data Transmission _ iP _ . P.Ty {sin(nfy Tp))? , PT, {sin (why, T,)|? | sur FE pe farsa ese Re | ELT 5 Be (ees Th v3.7.1) Fig. 3.7.3 shows the plot of power spectral density of BFSK signal given by above equation. fiy and fj, are selected. Such that, fu-fi = fp w- (3.7.12) With such selection, it is clear from the spectrums in the above figure that, the two Power spectral density Sesk(th Fig. 3.7.3 Power spectral density of BFSK signal frequencies fyy and f, can be identified properly. The interference between the spectrums is not much with the above assumption. Bandwidth of BFSK Signal : From Fig, 3.7.3 it is clear that the width of one lobe is 2f,. The two main lobes due to f,, and, are place such that the total width due to both main lobes is 4f,. ie., Bandwidth of BFSK = 2f, +2f, or BW = 4f, wes (3.7.13) If we compare this bandwidth with that of BPSK given by equation 3.2.21, we observe that, BW (BFSK) = 2xBW (BPSK)Digital Communication 3-58 Passband Data Transmission 3.7.3 Coherent BFSK Receiver Fig. 3.7.4 shows the block diagram of coherent BFSK receiver. There are two correlators for two frequencies of FSK signal. These correlators are supplied with locally generated carriers 6,(t) and 2(t). If the transmitted frequency is f,,, then output s(t) will be higher than s2(t). Hence y(t) will be greater than zero. r 5,0 Signal oe | Decision Choose 1 if yt) > 0 o i device Choose 0 if yit) <0 Ty, vee fra 0 ® 4 sa vane |Z cos (2nf,t) Fig. 3.7.4 Coherent BFSK reciever The decision device then decides in favour of binary ‘1’. If s4(f)>s)(#), then y(<0 and decision device decides in favour of 0. The coherent carriers are generated using, similar methods discussed earlier. 3.7.4 Noncoherent BFSK Receiver Fig. 3.7.5 shows the block diagram of BFSK receiver. The receiver consists of two bandpass filters ; one with centre frequency f, and other with centre frequency f, Since fy —f, = 2fp- the outputs of filters do not overlap. The bandpass filters pass their corresponding main lobes without much distortion. ~Digital Communication 3+59 Passband Data Transmission Bandpass filter at Envel { detector i = 3 ° aK Comparat BC) signal Bandpass filer at Envelope |__} f. detector Fig. 3.7.5 Block diagram of BFSK receiver The outputs of filters are applied to envelop detectors. The outputs of detectors are compared by the comparator. If unipolar comparator is used, then the output of comparator is the bit sequence 4 (1). 3.7.5 Geometrical Representation of Orthogonal BFSK or Signal Space Representation of Orthogonal BFSK ‘Orthogonal carriers are used for M-ary PSK and QASK. The different signal points are represented geometrically in 6,6) plane. For geometrical representation of BFSK signals such orthogonal carriers are required. From Fig. 3.7.1, we know that, two carriers 9; (f) and $y (@) of two different frequencies fy and f, are used for modulation. To make 6, (#) and, (t) orthogonal, the frequencies f,; and f, should be some integer multiple of base band frequency ' f,’ ie. fa = mfr ». B7.14) and tL = "hy vs GF15) Here f, = 2, then the carriers will be o0) = iE cos (2mm fy, t) +: (87.16) 7? and be) = | cosanm fi, 0) w+ (87.17) TyDigital Communication 3-60 Passband Data Transmission The carriers 4, (H)andé, (t) are orthogonal over the period T,,. We can write equation 3.7.1 and equation 3.7.2 as, sy@® = JP, [Fewenty DY and sp = {2T, Resorts 5 ® a a Here fu = fotze and fi =fo- 35 Using the relations of equation 3.7.14 to equation 37.17 we can write above equations as, su) = ¥PTy 410 ws (3.7.18) and si) = YET 420 vue (R719) Based on the above two equations we can draw the signal space diagram as shown in Fig. 3.7.6. alt) a= VeP,Ty= V2E, ot) * Decision boundery Fig. 3.7.6 Signal space representation of orthogonal BFSK Distance between signal points : ‘There are two signal points in the signal space. The distance between these two points can be obtained as, d? = (/PT,)? +(JP, T,)? = 2P, Th d= 2PT, sos (87.20) ‘y We can write above relation as, d= \2y so G72)Digital Communication 3-61 Passband Data Transmission As compared to the distance of BPSK, we observe that this distance is smaller. 3.7.6 Geometrical Representation of Non Orthogonal BFSK Signals Whenever the carriers 6, ().and$, (#) eat are non orthogonal, then the signal point $;, (or $, ( will not lie exactly on the axes 6; (Nand@2(t). Such a representation is shown in Fig. 3.7.7. 4o(t) The distance ‘d’ for non orthogonal signal shown in Fig. 3.7.6 is given approximately as, 2 40) * Ev) Fig. 3.7.7 Geometrical representation of we. (3.7.22) non-orthogonal BFSK signals 3.7.7 Advantages and Disadvantages of BFSK Even though the generation of BFSK is easier it has many disadvantages compared to BPSK, First thing is that its bandwidth is greater than 4f,,, which is almost double the bandwidth of BPSK. The distance between the signal points is less in BFSK. Hence the error rate of BFSK is more compared to BPSK. From equation 3.7.3 we can write, sit) = Jf2P, cos {a(t) 28} cos (2a fy t)— f2P, sins {(1) O HY sin (2x fo Since dQ) = 41 cos 0.1} = cos (Qt) and sin f2OQ = +5in(Qh=dysinQy By standard trigonometric relations s(t) = Jf2P, cos (Ot) cos (2n fy t) ~ J2P. d (ty sin (021) sin (2 fy 1) o-- (3.7.23) In the above relation the first term, /2P, cos(Q) cos (2 fy #) carries no information. The second term, 2P, d (l) sin (01) sin (2x fy t) carries the information signal d (#). Thus only half of the transmitted energy carries the information signal.Digital Communication 3-62 Passband Data Transmission Review Questions 1. Drmw the block diagrams and explain the operation of BFSK transmitter and receiver. 2. Compare BFSK and BPSK. 3.8 M-ary FSK In the last section we studied BESK for two symbols. This principle can be extended further to ‘N’ successive bits. These ‘N’ bits from 2 -M different symbols. Every symbol uses separate frequency for transmission. Such system is called M-ary FSK system, The principle of transmission and reception of M-ary FSK is different than BFSK. 3.8.1 Transmitter and Receiver of FSK 3.8.4.1 Transmitter Fig. 3.8.1 shows the M-ary FSK transmitter. The ‘N’ successive bits are presented in parallel to digital to analog converter. These “N’ bits forms a symbol at the output of digital to analog converter. There will be total 2 =M possible symbols. The symbol is presented every T, = NJ, period. The output of digital to analog converter is given to a frequency modulator. Thus depending upon the value of symbol, the frequency modulator generates the output frequency. For every symbol, the frequency modulator produces different frequency output. This particular frequency signal remains at the output for one symbol duration. Thus for ‘M' symbols, there are ‘M! frequency signals at the output of modulator. Thus the transmitted frequencies are fa far-fy-1 depending upon the input symbol to the modulator. Input bit at sequence eMart PSK Bit) signal convertor Clock at every T, Fig. 3.8.1 M-ary FSK transmitterDigital Communication 3-63 Passband Data Transmission 3.8.1.2 Receiver Fig. 3.8.2 shows block diagram of M-ary FSK receiver. It is the extension of BFSK receiver of Fig. 3.8.1. The M-ary FSM signal is given to the set of ‘M’ bandpass filters. The center frequencies of those filters are fy, fy. fo/-~fy-1- These fillers pass their particular frequency and alternate others. The envelope detectors outputs are applied to a decision device, The decision device produces its output depending upon the highest input. Depending upon the particular symbol, only one envelope detector will have higher output. The outputs of other detectors will be very low. The output of the decision device is given to ‘N’ bit analog to digital converter. The analog to digital converter output is the ‘N’ bit symbol in parallel. These bits are then converted to serial bit stream by parallel to serial converter. In some cases the bits appear in parallel. Then there is no need to use serial to parallel and parallel to serial converters. Envelope detector Fig. 3.8.2 Block diagram of M-ary FSK system 3.8.2 Power Spectral Density and Bandwidth of M-ary FSK We know that for M symbol fy, f,.fy ---fy.1 frequencies are used for transmission. The probability of error is minimized by selecting those frequencies such that transmitted signals are mutually orthogonal, If those frequencies are selected as successive even harmonics of symbol frequency f,, then transmitted signals will be orthogonal. Let's say that the lowest carrier frequency fy is the k” harmonic of symbol frequency ie., fo = ¥ - G81) Then the other frequencies will be, Ay = (k+2) fifa =k 44) ... ete w+ GB.2)Digital Communication 3-64 Passband Data Transmission Thus every carrier frequency is separated by 2f, from its nearest carriers. Fig.3.7.2 shows the power spectral density of BESK (for two symbol FSK). In this plot the two symbol frequencies f,, and fy are separated by 2f, (Here f, =f, for BFSK). The same principle of BFSK is extended to M-ary FSK, That is M-carriers are added with separation of 2f, between the carrriers (Note here that f, is symbol frequency and not f;): Therefore power spectral density for M-ary FSK will be simply extension of BFSK. Fig, 3.8.3 shows the power spectral density of M-ary FSK. ale? ah, y= ON, t a Fig. 3.8.3 Power spectral density M-ary FSK Observe in the above figure that, the separation between the two nearest main lobes is 2f,. Bandwidth of M-ary FSK : From Fig. 3.8.3 it is clear that the width of one main lobe is 2f,. If there are M-symbols, then power spectral density spectrum will have M_ lobes, Therefore bandwidth of the system for M-symbols will be BW = Mx(2f) = IME +++ (3.8.3) We know that 2% =M and f, fe we can write the above equations, = 2.2N fo BW = 2-2! N «+ (3.8.4) Nea = Nh .- (3.8.8) On comparison of above equation with M-ary PSK bandwidth of equation 3.5.12, it is clear that M-ary PSK needs comparatively large bandwidth.Digital Communication 3-65 Passband Data Transmission 3.8.3 Geometrical Representation of M-ary FSK or Signal Space Representation We know that in M-ary FSK, mutually orthogonal signals are used for transmission. Equation 3.7.18 and 3.7.19 give the two mutually orthogonal signals for BFSK. Similarly we can write the equation for M-ary FSK ie. s0= JBI 4% | 3,0 = JR o,0 5) = JET 40) 688) Sa (0 = VETS Oy ol Here s9 (#), 5; (0,52 (t). Sj) are mutually orthogonal signals for ‘M’ symbols. The orthogonal carriers $9 (8,0; (0/42 (0 -..44-10 ete. can be represented as follows (ie. extension of equation 3.5.16 and equation 3.5.17) do = te cos (27 fy #) $2 () =f cos nfs 0) » B87) Oye O E cos (2r fxg. 9 ett) NS (/Pate)” d= /2P,T, = (2B, oC) Galt) Fig. 3.8.4 Signal space (Geometrical) representation of M-ary FSK for M= 3