Scribd
Upload
14 views32 pages

Digital Minor 2comm

The document discusses the advantages of digital communication systems over analog systems, highlighting features such as noise resistance, flexible operation, and information integration. It explains the concepts of quantization in pulse code modulation (PCM), detailing uniform and non-uniform quantizers, quantization error, and the significance of signaling rates. Additionally, it addresses the necessity of non-uniform quantization for signals like speech and music due to their high crest factors and varying signal amplitudes.

Uploaded by

Payal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
14 views32 pages

Digital Minor 2comm

The document discusses the advantages of digital communication systems over analog systems, highlighting features such as noise resistance, flexible operation, and information integration. It explains the concepts of quantization in pulse code modulation (PCM), detailing uniform and non-uniform quantizers, quantization error, and the significance of signaling rates. Additionally, it addresses the necessity of non-uniform quantization for signals like speech and music due to their high crest factors and varying signal amplitudes.

Uploaded by

Payal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 32
over a communication ¢: discuSsed earlier in chapter 1, the use of digi important sdvanayes a conpared Oanalogsommctions In particula: digital cominunication system offers the following highty attractive featu: (i) Ruggedness to channel noise and external interference unmatch any analog communication system. (ii)' Flexible operation of the system. { (iii) Tntegration of diverse sources of information into a common formaty ecurity of information in the course ofits tras ihe destination. é and Amplitude may be class fied under two heads ulse analog meditation, Hence, in pension: is accomplished in an anal nodulation, the time and the pul eur in discr*te form and digita ion is therefcre basically a scheme vi esponding dig: tal form, It is for this reas etimes know 2s pulse digital medulati modulation 4 called.pulse code modu he message sig; al is first sampled and th ded off to the nearest one of a finite set x(t) Sampled signal Regenerated PCM wave ‘6. Quantizer Lompors Arab lie? tree tPes As discussed in article 10.5, aq-I uantizer compares thi crete-ting inputz(nT,) with its fixed digit It assigns any one of the x(nT,) any one of the digital level @ Uniform quantizer 4G) Non-uniform quantizer oa t thos! orm gunner Af 3 : i F Therefore, the maximum quantizatit® {p’ maximum quanti s : : acteristic which passes fe for characteris! Berio et 0 Seal ac a 10.4(2).07 the other hand it val to -8, The dotted line show he rou nein aa characterise cof Fig 10-5(a). Hence the digital levels Pgoes not passes through ar an st orrect quantization characteristi¢ 3 8p aro taken at £5/ 2,425 ete, It provides © as and reduces quantization error. eats in a PCM System — ee tienen Bandwidth i eer tenet or pC Mayeten In this section, we shall evaluate the transm! ny 2 ene Let us assume that the quantizer use ‘y’ number of inary digits to 56° ‘digits ‘each level. Then the number of levels { that may | be represented by — ¥ (10.3) — =2 Here ‘q’ represents total number of digital levels of ag-level quantizer. For example, ifv = 4 bits, the total number of levels will be, ~q = 24= 16 levels Each sample is converted nary bits. i.e. Number of bits per sample =v. We know that, Number of samples per second =/, Therefore, Number of bits per second 1s expressed a5 (Number of bits per second) = (Number of bits per samples) x (Number of samples per second) = v bits per sample xf, samples per second ARs (10.4) ‘The number of bits per second is known as signaling rate of PCM and is jlenoted by ‘r’i.e., | Signaling rate -(10.5) for PCM. transmission is given by half of the signaling Vey tee ol be ta. sf Ac (10.6) vevt, -(10.7) +(10.8) \10110. A PCM Receiver In this section we shall discuss a PCM receiver from ical paar Fig 10.6(c) shows the block diagram of PCM receiver and ti gn gan of view. and Fig 10.6 reconstructed signal. The regenerator at the st: pce x,(0 |Low pase] y() | filter 3 fain the pulse and removes the noise. This sign ital words for each sample. = ‘circuit. This signal, at the output of s ough a lowpass reconstruction fil'er to get the [denoted as y(t). As shown in riconstructed quantization at the an. be reduced by increasisg ‘ie binary @ ; to increasing binary digits (bits) per sample. But singnaling rate as well'as transiuission band article. Therefore the choice of ches» parameters * rive an expression for quantization nose (i.e., error) i quantization or uniform quantization. Because of | ors are introduced in the signal. This error is called lefined earlier, the quantization erro’ is given as = x, (nT) -x(nT,) Let us assume that the input x(n7,) to a linear or unif i F continuous amplitude in the range ~ 2pqq 10+ me Se ' From fig. 0. a(a), it may be observed that the total excursion of inputa(nT,) is mapped into ‘g' levels on vertical axis. This means that when input is 45, re : aay! news fm output is > 8 and when input is — 48, output is as 8. Thus, + Xmax Tepresents Sige 1 £8 and —%yox Fepresents —5 3, Therefore the total amplitude range becomes, (23 — : ‘ Total amplitude range = %j,92- (-*max) = 2x, tal amplitude range is divided into ‘g’ ls of quantizer, =(=x q 2px (10.10) a aes (10. Jhormalized to minimum.and maximum valugs ———— Again, now if signal x(t) is equal to 1, then we have eles! Si -(10.1) (10,12) (for normatized signal) dered as. sufficiently small, then it x may be ip will be an uniformly distributed andom jmum quantization error js given (equation (20,18) we (10.14) } quantization error may be assumed as an ww = bution (b) A Uniform distrubution for quantization error Then PDF of uniformly distributed random variable ‘2°18 Rive ae akan = hows an uni Se av ecml i Eigyre (20.72) shows an uniformly distrubuted random va! 7 0 for x b (10, in! Thus with the help of equation (10.15), the probability density function for: quantization error ‘e’ may be defined as 0 for est 1 3 8 fe =42 for Ses 5 aa 0 for eer = --(10.16) Also, from fig 10.7(6), it may be observed that quantization error‘: as Zero average value. In other words, the mean ‘m,’ of the quantization ervor is zero. We know that ce signal to quantization noise ratio of the 1) antizer is defined as, S_ _ Signal power (normalized) N ~ Noise power (normalized) x(t) is known, then it i is possible ts calculate (10.17) Pee resasd.co, v7 gsi Jygrerts|s + Naw (If % =a (10.18) as the mean square value of noise volts e, Since, here é, therefore, its mean square Value = ble?) = LD \ able = fx%felxdz (10,207 = ferrterde (10.21) , above equation may be written as, i exdde se ae x(t) as a voltage signal, is =} al wa = ae or Ble’ = 5 aff eT ee Efe’) = 8. 8 8 (10.22) or Ele = 35 ‘ Now, using equation (10.19), the mean square value of noise voltage woul be Vizie = mean square value 8 12 Also, ifload resistance, R = 1 ohm, then the noise power is normalized ie. rn Noise power (normalized) = Yo (putting R = lin equation 3.18) _ tia _ st ef 12 Hence, finally, we write Normalized noise power st “2 for linear quantization. err) n Noise Ratio for Linear Quantization r linear quantization the signal to quantization .(10.24) its ‘v’ and quartization levels are related as, (10.25) ) to a linear quantizer has continuous am- A Therefore, total amplitude range se = Wimaw Now, the step size will be 2x, be aa! Here, putting the value of q from equation (10.25) in equation (10.26), w bs 2x max . oe Now putting this value in equation (10.24) we get, S _ Normalized signal power can (AR) eS Let normalized signal power be denoted as ‘P. that signal to noise p ower ith increasing bits per sai x(t) is normalized, i.e., N noise ratio will be, Bee ax Po ng mal power ‘P’ is normalized, i.e., atio will be given as ax” s She Sas = 10 loge J 28 s (Fes $ 10 logy, (3 = 274) § baer Gy) rele sane Thus, Signal to Quantization noise ratio for normalized values of power ‘P” amplitude of input x(t) . ; iss | (S)aa 4.8+6v)dB (?* a = mean square value of signal voltage cree), will be, ‘ x*(t) | ara (with R = 1) 0) i P= xXt) Aaa) .35) crest factor is given as, input signal asa voltage signal Crest factor = eee rms value : x, ; : () ow Crest factor = ne since P = x%(t) L When we normalize the signal x(¢), then (@. nerel a ( = ‘max bs “| putting above value ofz,g5 in equatio (10.40), we get . AtOr82) A, Crest factor = TP . For a large crest factor of voice (speech) and music signals P should be very ery less than one in above equation. ees, Pe« io rema be leis . ‘ 5¢ \“1évels, the only way to have a uniform sign! ; de signals and larg | avsp Adjust the step size in such a manner that the ratio } ¢ eaaenaen the step size must be small for small amplitu for large amplitude signals . * The effect of an adaptive step size may al: : 4 /way by distorting the signal before the quantization process »** ‘tion has to be introduced at the receiver to make the overa™ distortionless. we ; ‘Therefore the signal is amplified at low si, nal Jevels- ang 2 etate equivalent to more step Size at Tow signal levels and small step Size at hig {Signal levels. At the receiver a reverse process is done. That is signic is attenifj ~ ated at low signal levels and amplified at high signal levels to «et origi signal. Thus the co i ignal at ittes and ¢xpansion receiver is called combinely as companding Fig 10.9 shows compression ay a ant curves. 4 sy 2 pass Compression Expansion 1 1 r ‘ ‘Linear characteris: ca m Fig 10.9, at the receiver, the signal is expa; ded exactl on curve at transmitter to get original sigal. + dotted liné s uniform quantization. The compression an« expansion det us. think about the desired prin of no 2 iscussing companding, we can + form quantizer can be achieved by using quantizer. Now by cascading this combjiation with * ,own in figure 10.10. This figure depicts the transfer chara h ¥ : aratky, ympressor, ‘uniform quantizer and expander. Te oe re combination of a compression and an expander is called acom, * storally, in an actual PCM system, the combinatioe of compressor ana form quantizer is located in the transmitter whereas the expander is located in i ‘odel of a non-uniform quantizer. ‘igure 10.10 shows the m Fig. 10.10, Model of a non-uniform quantizer In addition to above figure 10.11 illustrates the transfer characteristics of compressor, uniform quantizer and expander. i Quantizer Input Comp EYP EEE Interval |, Compressor input Quantizer output Expander input 1 for positive amplitudes only) saw Companding law companding and A-law companding. +h). ed input and output voltages, and p isa e have plotted thep-law for varying. The antization corresponds to 1 = 0. For a gi 7 ‘he compression curve, which defines the quastan, oth tive of |m| with respect to |v|; that is, Skps, Y dim| _ log(1+p) : 4 dul “ok eae, We see therefore that thept-law is neither stri i rithmic, but it is approxim: ‘ar at low input levels tt loge Wr] <<1, and approximately logarithmic at high input | po S. , levels correspindin, “law companding Another compression law that is used in practice is the so. ° ealled A-lay, In: the A-law companding, the compressor characteristics is Piecewise, i: be a linear segment for low-level inputs and a logarithmic “seman high eat inputs. It is described as i Alm| ig i t ee a i 1+log A’ gsi sa 2 \ 3 lol = L+log(Alm) 1 cine which is shown plotted in-figure 10.12. Practical values of A (as thep ti the 2, in the vienity of 100, The case of uniform quantization cyrre- reciprocal slope of this compression curve is given bythe espect to |v|, as shown by ‘ 1+log A a. ia Os|misz (1+log Alm, slm|s1 am steps over the central linear segment, which hav-tie nall signals, are diminished by the factor A/(1 + la). dB in practice, as compared with uniform quantization, 0.02 04 06 08 MW Normalized input, mi (b) npression laws (a) j-law (b) A-law a Normally for speech and music signal, ap - law compression is used. Fig 10.13 shows the variation of signal to noise ratio wi 5 , 8 tio i evel without companding and with companding. rite Tetpeet ta signal With companding Ei: companding -40 -30 -20 -10 0 Signal level dB —.> Fig. 10.13. PCM performance with p-law companding. _ It can be observed from above figure that signal to noise ratio of PCM remains almost constant with companding. -15. Delta Modulation We have observed 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. | od transmits only one bit per sample. Here, the present h the previous sample value and this result whether ‘or decreased is transmitted. Input signal x(t) is J by the delta modulator. This step'size is kept fixed. input signal x(t) and staircase approximated sig- Sand - 8. Now, ifthe difference is positive, creased by one step, i.e., ‘d'. If the difference is lis reduced by ‘8’. When the step is reduced, is increased, ‘1’ is transmitted. Thus for each ansmitted. Fig 10.14 shows the analog signal ted signal by the delta modulator. fa modulation can be explained by the following impled value ofx(t) and last approximated sample (10.45) le approximation of the staircase waveform Amplitude —+ ' e = totytot 0 OPO Oy Re! : t Binary one —+'0'1 ul iititilolato bit sequence ' i ties. d Tutte Fig 10.14, Delta modulation waveform. : If we assume u(nT,) as the present sample approximation of staircase out- put, then, ul(n-DT,} = 2x7.) : =(10.46) = Last sample apptoximation of staircase waveform Let us define a quantity b(n7,) in such a way that, sis B(nT,) = dsgn fenT,)) his means that depending on the sign of errore(nT,), the sign of step size ed. In other words, we can write bat = +5, ifx(nT,)2 (nT) .(10.47) si 5 if xnTJ< A(nT,) (10.48) T= +5; a binary ‘1’ is transmitted = — 8; a binary ‘0’ is transmitted. __T, = Sampling interval. s the transmitter (ie., generation of Delta Modulated sig- ms (10.47) to (10.49). accumulator adds quantizer output (+8) with the previ- ation. This gives present sample approximation. 1c. ) = u(nT, -T,) + [+5) T,) = ul(n-1)T,) + (nT) (10.49) al approximation u[(n -)T, is restored by delaying one sampled input sign alx(nT,) and staircase approximated tracted to get error signale(nT,). the sign ofe(n7,), one bit quantizer generates an out- » step size is +5, then binary ‘Vis transmitted.and ifit is ransmitted. in Fig 10.15(b), the accumulator and low-pass filter or generates the staircase approximated signal output npling period T,,. It is then added to the input signal. If ' | i ¢ input is binary hen it adds + 6 step to the previous output (which 18 ce- eyed): [f input is binary ‘0’ then one step ‘8’ is subtracted from the delayed © gignal. The low-pass filter has the cutoff frequency equal to highest frequency _ jn x(t) This filter moothens the staircase signal to reconstruct x(t). i f Output (demodulated Input signal) n transmitter Fig. 10.15. (a) A Delta modulatio (b) A Delia modulation receiver jelta Modulation dulation has certain advantages over PCM as under: only one bit for one sample, therefore Channel bandwidth is quite small, dwidth is quite sma T implementation is very much simple for ere is no analog to digital converter required in on. lta Modulation : lation has two major drawba\ oad distortion, or Idle noise these two drawbacks in detail. oad Distortion. This distortion arises because © e input signal. ae 5 ed from Fig 10.16, the rate of ri cks as under: f large ef SCBA yu case approximated sig) vleUU Communication oyst€l™ distor: d . e over loa! Lx input signal s(t), This error or noise is known 88S10PC°T Tope of IB SIMs” “To reduce this error, the step size must beincreé . i high Granular noise Slope-overload distortion Staircase approximation u(t) ' Fig, 10.16. Quantization errors in delta modulation. i i imum oF inini- Sincethe step size of delta modulator remains fixed, i s maximur sini- therefore this modulator is also known i) Granular Noise. Granular or Idle noise occurs when the step 3 te is too large compared to small Variations in the input signal. This very small vartations in the input signal, the staircase signal is chang.d by | large amount (6) because of large step size. Fi, ‘Input signal is almiost flat, the stai alu(t) keeps on oseiltating ty Srpand thi signal ‘The error between t. ut and approximated 81,7 called granular noise. The solution to this proble: ces eréfore, large step js required to accommodate wide dynami the input signal (to reduce slope overload distortion) and small ster: Q luce granular noise. Infact, Adaptive delta modulation is the :.odi- overcome these errors. ual. yn! ofthe signal x(¢), the stép size is increased, Ajo, if e Step size is reduced. Then this mathud is odulators can take continuous changes in step sie ¢ the transmitter and 10,17(5) shows receiver of a or step size control-is added in the diazran greases according to a speci one an example, if it qua is followedasthatintransn’ * FP the same process is followe: as that in trans? Bit, Bae snput decides the step size. It is then applic, jnput ant present Inpul les the : aa Pr aN lator which builds up stall ase waveform. The low-pass filter then _ ‘out the staircase waveform to reconstruct the original signal. ee ere wat ‘oy %. Output & (Variable size) Accumulator wes \s for adaptive delta modulation Jelta Modulation has certain advantages over delta modulation Ordinary q, ~ Pe overload distortion aan S ise, is wider Because of the variable step size, than simple DM. (iii) Utilization of bandwidth is better t] Plus other advantages of delta modul required and simplicity of implementatio han delta Modulation, ation are, only one bit ’ n of transmitter and Toco is ® Example 10,12, Given a sine wave of: frequencyf, and ampli ‘ a m pli 5 \ a delta modulator having step size 8. Show that the ine overlay applied to will occur if listortion &) a here 7, is the sampling period. Solution : Let us consider that the sine wave is represented as, x(t) = A, sin(2nf,,t) It may be noted that the slope of (¢) will be maximum whe t) to ‘t’ will be maximum. The maximum slope o Aa >. Step size = Sampling period @ ill take place if slope of sine wave § maxlA,, 27/m 008 27/nt)1 7. 3 : : | An2 fn > pe ~ : 6 - ee ANS, BS Qn f, Ts SATS 4 ta onerate at five Hm" mple does not differ la rry the same inform Lbvas system, the result- i Fig. 3.19 illustrates time signal x(t) by dotted line. This signal is _ it intervals 7, 27,, 37, ... n' nvq a ‘quantized to the nearest digital e figure 10.19. The encoded binary value of of the samples. in observe from fig. , ST, and 67, are encoded to same value of carried only by one sample. But three samples n_ means it is redundant. Consider another as SE SO Dlog ‘Eh... = tween these 82%) example of samples taken at 97, and 107’, The difference bel they donot chinge, Foner risteTect bit and first tivo bitsare redundant, since t x0) bits (levels) icra) ees xo 6(10) 7, 21, 31, 47, 67, GT, TT, &T, OT, 107, xaT,) Fig 10.19. llustration of redundant information in PCM Ifthis redun dancy is reduced, the rall bit rate will decrease and our of bits require imple will also be reduced. This type of digic! Pulse modulation scheme is know: wn as Differential Pulse Code Modulati The differential pulse code modulation works on the principle of predic: The value of ent Sample is predicted 1 from the past samples. The E i tion may not be exact but ¥ close to the actual sample vlaue. Fig 16.5) shows the tr: smi it ode Modulation (DPOM) syston ted by x(nT,) and the predicted Signal is denoted ty ie difference between the actua B ated sample value is is known as Predict; signal erential pulse code modulation transmitter rence between unquantiz; € predicted value a ‘nT,) and previous predictigy output signale,( att filter. The se prediction filter. This signal is calledx,(nT,). given a es e ore. and more alas she is calledx,(n7,). Thi e that the quantized error signal e,(nT, a ceaad by ual fbits, Thus numb bald fn We er ft @ al in DPC! ample are reduced in Os The quantizer output can be Ere ee =Gnie} = e(nT,) +q(n- J at Hereq(nT,) is eae at ero r. AS shown in figure 10.20, ee i a ot d by sum £ d quantizer outputi.e., tion filter inputz,(n7,) is obt ined by sum <(n7, ) and q need x(nT,) = XnT,) +e(nT,) A utting the value ofe (nT’) from equation 3.51 in the above equation we Putting q' 5 q' get, 207.) = 2(nT,) +e(nT,) +9(nT,) ce) Equation (10.50) is written as, enT,) = x(nT,) — 2(nT,) e(nT,) + i(nT,) = x(nT,) -(10.54) -. Putting the value ofe(nT,) + #(nT,) from above equation into equation (10.53) we get, x (nT) = x(nT,) +q(n7,) (10:55) Thus the quantized versicn of the signal x(n.) is the sum of ori ‘ins sample value and quantization error q(nT,). ‘the quantization error can be positive ‘or negative. Thus equation (10.55) does not depend on the predic acteristics. Reconstruction of DPCM Signal : Fig. 10.21 shows the block diagram of DPCM receiver. Th its the quantized error signal from inc filter output and quantized error sigals are su med Version of the orginal signal. Thus the signal at the ion error (nT), which is intro- J Outpue/ avg t. 10.21. DPCM receiver Digital Pulse Modulation Methods tal pulse modulation methods in details let us now from different aspects, iparision of PCM, Differential PCM, Delta Modula- lulation. This comparison is carried out on the

You might also like