UNIT SAMPLING THEOREM AND ap CORRELATION sia GROUP PART-A SHORT QUESTIONS WITH SOLUTIONS gt. Define and write the conditions of sampling theorem. (Model Papers, 1) | Nov,Dec.-1, (R18), (a) (or) State sampling theorem for band limited signals. ‘Ans: According to sampling theorem, a continuous-time signal can be completely represented in its samples and recovered tack ifthe sampling frequency ‘fis greater than (or) equal to the twice of highest frequency component of the message signal fies - 22h @. Write sampling property of impulse function. Ans: Ifx(¢) be continuous time signal, and 8(0) be a unit impulse function, then the sampling property of unit impulse function isgiven by, 1() 8(=1) = (4) 8004) =x) 03. Define nyquist rate. ‘Ans: Model Papers, a1(i) The minimum sampling rate at which both sampling and reconstruction ofa signal (from its samples) can be performed without any distortion is called nyquist rate. It is denoted by ‘f,’, and is given as, L.=2Sn Where, f,,~ maximum frequency component. (4. Define Nyquist interval. Ans: Nyquist interval is defined as the maximum time interval between the equally spaced samples of the signal during which ‘Sampling rate is equal to Nyquist rate. Nyquist interval is equal to reciprocal of Nyquist rate and is given by, 1 2 Saux jum frequency component, eROM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS SIA GROUP {5List the different types of sampling techniq The different types of sampling techniques are, 1 Ideal or impulse sampling 2. Natural sampling and 3._ Flat top sampling. Q6. Define imputse, natural and flat top sampling. Ans: Impulse Sampling Impulse sampling is defined as a sampling technique Which samples incoming message signal ’) and provides impulses. It is also known as ideal or instantaneous sampling. Itis the simplest sampling technique. Natural Sampling Natural sampling is defined as a sampling technique in ‘which tops of the sample pulses retain their natural shape. Its ‘a complex type of sampling technique. Flat Top Sampling Flat top sampling is defined as a sampling technique in ‘which top of the pulse remains constant and appears to be flat. This technique uses a sample and hold circuit with flat top samples. Itis one of the most popular and simplest sampling technique. This technique has constant samples at the top. Q7. Define aliasing and give its effects. Briefly explain how to eliminate aliasing effect. (or) What are the effects of aliasing and how can you minimize the aliasing error? (Refer excluding Aliasing) (Mode! Paper, 21) | Nov/Dec.-18, (R16), Q(c) Ans: Aliasing ' Aliasing can be defined as a phenomenon in which high frequency components of the signal appears as low frequency components during sampling. It is also known as under sampling. Effects of Aliasing 1. Theeffects of aliasing are due to the interference of high frequency components with low frequency components, distortion occurs in the signal. 2 ‘The information carried by the signal is lost and cannot be retrieved. Look for the SIA GROUP Loco SIGNALS AND SYSTEMS [INTUHYDER ans, Eliminating All Effect I} 1. xo) must be band-limited 10a suitable Fequengy ® 2, Sampling Frequency J, MUSt be greater than tye » band-limiting frequency (f, ). ie, £7 2S ———_esina fitter? Q8. What is anti-allasing filter? Ans: (Model Paper, QM, ayy othe itis defined as a filter used 0 avoid (or) eliminate he errors in the input voltage ie., having frequency greater thay half of sampling frequency. Q9, State sampling theorem for bandpass signa), Ans: The sampling theorem of bandpass signal states tha 1a signal m(f) whose bandwidth is completely recovered and represented from its samples x(n7), if it is sampled at the ‘minimum rate of twice the bandwidth ic., Sampling rate, =2* 2f, = f,-Af, samples /sec ce 1 —— sec. fm Q10. Define correlation of a signal. Ans: Correlation is the process of determining the degree of similarity betweep two signals. The correlation coefficient is | for identical signal and zero for completely different signals. The correlation coefficient is ~ I for identical signal having @ phaSe shift of 180°, * QI1, Define auto correlation. An: A function that gives the measure of similarity between signal and its delayed version is referred to as auto-correlation function, Let x(¢) be.a signal. Then, the auto correlation function of x(0 is given by, Rele)= fxtoxe-o n the TITLE COVER before you buywiT-5 (Sampling Theorem and Correlation) co List the properties of cross correlation function. anst ‘Model Paper-tt, 21(j) The properties of cross correlation function are FIROL=XN) XQ) R= RQ), R(t) = Ry(-2) 4 RAR, CD). 13. State roperties of auto correlation function. Ans: The properties of autocorrelation function are, L R= fro 2 R,O)> Ra) 3 RAD=RC) 4 |F, JRie™ de= 5,0) 5 R@=f(0* £4) @ Fo). Fo) & RA) Q14. List the properties of Energy Spectral Det (ESD). Ans: The properties of ESD are, : 13 Total energy = fy(f)d'= 5- fylardo 2 V(A)= HOE V1) (01) v,(@)= Ho vo) FT FT 3 R(t) <> Wf) (01) R(t) <——> Vo). Q15. Write the Parseval’s relation for continuous time periodic signals. Ans: Model Paper-Ill, Q14) Parseval’s Theorem for Energy Signals The Parseval’s theorem states that the energy of a signal ‘defined interms of Fourier transform. ie B= fixer dr ‘ECTRUM ALLAN-ONE JOURNAL FOR ENGINEERING STUDENTS 5.3 Parseval's Theorem for Power Signals The Parseval’s theorem states that the power of a signal is defined interms of Fourier series coefficients, ie, P Dice a Q16. List the properties of Power Spectral Density (Psp), Ans: The properties of PSD are, Ae . Lo pee Jscordo = Jona 2. $(o)=|Hlo)? S(o) 3. RQ) OS(o) EO Q17. Give the relation between autocorrelation and PSD. Ans: The autocorrelation function for a periodic signal is given as, x,(0) ~ Periodic power signal. The power spectral density fora signal x(t) is given by, Q) It can be observed from equations (1) and (2) that, FUROL=S) (or) fF RQ) <>) Therefore, for a signal, auto correlation function and power spectrum form a Fourier transform pair. SIA GROUP {ESSAY QUESTIONS WITH SOLUTIONS ‘ 5.1 SAMPLING THEOREM 5:14. Graphical and Analytical Proof of Band Limited Signals Q18. Explain the sampling theorem for band limited signals with graphical proof. Ans: (Model Papa 0 Now, 204 ayy Sampling Theorem According to sampling theorem, a continuous-time signal can be completely represented in its samples and recover back, ifthe sampling frequency 4” is greater than (dt) equal to the twice of highest frequency component ofthe message signal Sy ie, Lh, Proof . Consider a continuous time signal m(t) whose spectrum is band limited to f, Hz. i.e. the signal m(t) has no frequency components beyond f, Hz, Then the corresponding Fourier transform of the message signal m(t) is defined as, + Mo) =0 for|@|> 0, The spectrums of m (t) and M(a) are as shown in figures (1) and (2) respectively. tm Figure (1) M@) 2 A Figura (2) The continuous time signal and «) be the Fourier transform of the given input signal m(. ‘The sampling of m() atthe rate Hz can be obtained by multiplying m() by the impulse train 8, (1). The impulse tin 8,, (1) consists of unit impulse repeating periodically atthe rae of ‘7, 7" = + Which is shown in figure (3). in the TITLE COVER before you buyuNIT-6_ (Sampling Theorem and Correlation) yao Theorem and Correlation) 5,00) AT TO Tay fi re (3) The impulse train is multi . , a iplied with the input signal m(t) using multiplier which is as shown in figure (4). Multip result in the sampled signal g(1) which is shown in figure (8). . . m(t) Mukiplier | —>&(!) —_— 5,0 J Figure (4) at) Figure (8) The discrete signal g(*) consists of samples at the rate T, (sampling time) and can be written as, B= m(N8, 0 ( But, the impulse train 8, () is a signal with period ‘7,’ which can be represented in Fourier series as. 1 50) [1 + 2cos wf + 2cos 20,1 + 2eo0s 30.) @ Where, , - Angular sampling frequency = me =2nf, On substituting equation (2) in equation (1), we get, 1 att) mol ioe 2cos wyt +2008 0,f +. | = : [m(t) + 2m(1) cos w,¢ + 2m(*) cos 20,'+.-] @ Applying Fourier transform on both sides of equation (3), we get G(w)= 2 [wy + Mw-0) + Moro) + Mw-20)+ Mo +20) +.) 4) [ Fourier transform of m(t) =M(@) Fourier trsivan of 2n(1)cosu,t = M(o-0,)+M(@+.,) Fourier transom of 2m(t) cosw,t = M(@-20,)+M (@+20,) < oS ‘ Gea) FY Meo-na) © SPECTROM ALLAN-ONE JOURNAL FOR ENGINEERING STUDENTS SIA GROUP {5consi tng of M(a), which is repeating with period, w, = aa (01)f,= 5 8s shown in igure (6), ‘ Go) tot fami | ip, w dp,” Figure (6) The signal m(t) can be reconstructed from g(t) by Tecovering signal frequency spectrum M(w) from G(«). This is possible when no overlapping occurs between the successive cycles of Ga). ‘The condition that must be satisfied to avoid overlapping is, L.2 2S Where, ‘ 1 Sampling frequency, f= T The respective frequency spectrum obtained by the sampling theorem with different conditions is shown in figure (7). ao AT . 22h eh Oh ye Gtf) AT | fy fh oe ci) laity =, - 9 2, ae Figure (7) Look for the SIA GROUP Loco SIGNALS AND SYSTEMS [JNTI 'YDE} RAI Statement For answer refer Unit-V, QI. band limited signal isa signal whose Fourier tan is non-zero or finite interval of the frequency axis, 4. , : Potty number B, exists such that, X() is non-zero only in fe (2,5) Where, Bis also known as bandwidth ofthe signa Consider a band limited signal x(t) having Fourie transform X(f) with non-zero interval. Let the interval be, -BSSSB. ‘Then, the signal x(t) converges as oa = = [XN™ ’ On differentiating above equation with respect to‘? for. ‘n' number of times, we get, 2 a(t) _ 4 ant 0 afew Je 2 = foamxine™™ a a Similarly, 2 we foo Pat a= J Campy x(Ne™ af Thus, the band limited signals infinitely differentiable. ‘Therefore, it reconstructs a smooth curve from its samples. Reconstruction of Band limited ‘Signals from its Samples Consider a bandlimited signal x (+) in [-B, B] whichis given by, 2 x= frnear od) 2 on the TITLE COVER before you buyay Xi) \ x(t) \ —| irr \ Bop “Ta —\__ TA>} The Fourier transform y a (n2B), (DN) ofa bandlimited Signal x) in [-B, B) can be Perfectly re ‘Constructed only by using samy samples (2) _f 2B} = Je (A) ePe207 ay Where, XP) — Fourier inverse of X(f) Figure (2) illustrates the inverse Fourier transform, Xo) 1 2p "125) ~ ss LET | | allt itt -3B -B [8B 3B =a t S Figure (2) Where, 2B = f-fundamental period 2B 3) —n® Fourier series coefficient. The inverse Fourier transform of ¢™" is 5(t~,). Therefore, the inverse Fourier ¥(t) transform of X(/) is given by, (= -it ; {a5 \{-aa) ‘The original band limited signal is multiplied with a periodic train of impulses, we geta signal whose Fourier transform e original bani p \s a petiodic extension of the original spectrum is obtained. The original signal x(t) is retrieved from the spectrum ¥(f) by multiplying ¥(¢) with spectrum. rigi . DECRUM ALLIN-ONE JOURNAL FOR ENGINEERING STUDENTS SIA GROUP J}SIGNALS AND SYSTEMS LINTULHVOEe, 220. Define sampling theorem for time limited slgnal Maximum frequency ofthe signal x jg eA d find the Nyq aS signals, ui Nyquist rate = 2f =2x 1500 1. rect 300t 2. 10008 300nt = 3000 Hz Ans: Nov/Dec.18, (R16), 4(b) Sampling Theorem ~ Nyquist sampling rate = 3 kHz For answer refer Unit-V, QL. 1, reet(3009 If x() denote the given signal then, (0) = rect(3001) Since, the given signal is a rectangular pulse rect(t/t). eee 00r= — ‘The Fourier transform of rect(t/t) is tsinc| ($) or ie, reet('t) se $) ‘The Fourier transform of rect(3001)is — sinc( : rie transform sw (a0 Strictly speaking the bandwidth of rectangular pulse is infinite, As the sine function extends from 0 to «. Hence the Nyquist rate is also infinite. 2. 10.c0s 300nz 10 Ifx(¢) denote the given signal then, (0) = 10 c0s 300nt From cos 300m, we have, cor = 300m anf = 3000n1 000 ny f= 1500 f= 15 kHe Look for the PlINg rate ang Nyquist Interval ofthe folowing function, ty = 8in(200 ~nt) + sin(100 nt), Ans: Dec.-14, int, Qe Given thet, Ad sin(200 nt) + sin(100 ney Highest frequency component in sin(w) is ‘o,.By using this principle, we can calculate the sampling rate, follows, Highest fequency component in sn(200 ni, ive, ©, = 2000 Similarly the highest frequency component in sn(100n) is, 0, ie, 0,=100n So the maximum frequency component in f) is equal to highest frequency component among all (1, and.) ie, 2m, = 0, = 2nf,=200n = f,=100 &: Sy =100 Hz Nyquist rate, f, = 2f, =2 100 = 200 Hz [Nyquist rate = 200 Fa Nyquist interval= In 1 = Leos 309 "005 Se =50 ms +: Nyquist interval = 50 ms IA GROUP Loco Jf on the TITLE COVER before you buyyNIT-5 (Sampling Theorem and Correlation) me come the Nyquist Sampling rate and Nyquist sampling interval for the signals, 4. sine(100 nt) 2. sine*(100 nt) 3. sine(100 nt) + sine(50 nt) 4. sinc(100 mt) +3 sinc(60 tt). Ans: 1. sinc«toonty ‘The given signal is, x(t) = sin (100 nr) = v= Sa (smn et) From the above equation, = or =100m => 2nft = 100m o=2nf) = f=50Hz Nyquist sampling rate, f= 2*f._, =2"50 (Lad = 100 Hz 00 Hz | Nyquist sampling interval, 1 r -———_—_ ~ Nyquist sampling rate 5.9 2. sine*(100 nt) The given signal is, = > a(t) xt) = x)= sine’(100nt) sin’ 100nr (100n1)2 1 (loon)? ( 2 I seo) 41 2x10'y + x(0 contains a D.C component and a sinusoidal component, From cos 200nt, we have, cor = 200m => 2nft = 200m = f= 100Hz Nyquist sampling rate, f= 2 f, 2x 100 200 Hz 100 Hz Nyquist sampling interval, 1 Nyquist sampling rate 3, Sine(100 nt) + sine(S0 nt) 100 : The given signal is, = onds cons x(0) = sin e(100 ne) + sin e(S0n1) 7 1Omes sin(L0Ont) , sin(SOnr) = O° Toon S0nt —— SIA GROUP {3 NTS: SPECTROM ALLIN-ONE JOURNAL FOR ENGINEERING STUDE!=> Oni © SOn = fa 28H f,, The maximum frequency in x(t) is f, = f,= SO Hz Then, Nyquist sampling rate, f= 2f,,. = 50) = 100 Hz Nyquist sampling interval, eS 7, ~ Nyquist sampling rate a 100 = 0.01 seconds = 10 msec. [T= Womseq 4. sinc(100 xt) + 3 sinc'(60 nt) The given signal is, (0) ~ sin (100 ma) + 3 sin (60 m2) Consider, (0) = sin (100 x2) a 100 => Inf = 100m = f=S0Hz SIGNALS AND SYSTEMS LINTU-HYDERARya, ‘Consider, x(f) = 3 sin °(60 mt) sin? 60m (60n0)" 3 {reze] 2 e005? 3 3cos120n¢ ~ 7200n77?— 7200n777 oy = 120nt 2Qafy = 120m f= 60 Hz hye SI a() = 1+ cos 2000nr + sin 4000n1 Consider, ¥,(7) = cos 2000nt 1 = 2000n1 2m ft = 2000nr 2 f= y= 1000 He Consider, x, = sin 4000 a1 wt = 400077 2n fr = 4000n1 ~ = 2000 Hz Therefore, maximum frequency in signal x(¢) is, ox =f, = 2000 Hz, Nyquist rate, = 2 foe =2* 2000 SPECTRUM ALLIN-ONE JOURNAL FOR ENGINEERING STUDEN' J, = 4000 Hz Nyquist interval, Q24. Asignalx( sampled at f, = 500 Hz. If the sampled signal is passed through an ideal low pass filter with a cut-off frequency of 400 Hz, what frequency components will appear in the output? Ans: The given signal is, x(t) = 2 cos (400n1) + 6 cos (6407) Sampling frequency, f = 500 Hz Cut-off Frequency, f, = 400 Hz Consider x,(2) = ¢08 400 nr > of 4000 => Inf, = 4000 100 Hz | Consider x, (2) = eos 640nt O40 ne On => o,f = sia erour 235. 2 The different frequency components in the’ sampled signal are f, +f. f,—Sy 2f, +f, and 2f -f, etc, f+ f;= $00 + 200 = 700 Hz SL,-f, = 500-200 = 300 Hz J,+f, = 1000 +200 = 1200 Hz 2f, -J,= 1000 -200 = 800 Hz J, +f, = 500 + 320 = 820 Hz S,-f, = 500-320 = 180 Hz Since, cut-off frequency f, = 400 Hz The frequency components which appear in the output of LPF are 300 Hz, 180 Hzie., f,-f,and ff, 5.1.2 Impulse Sampling, Natural and Flat Top ; ‘Sampling Q25. Explain briefly impulse sampling. Ans: Impulse or instantaneous sampling is also know as ideal sampling. Ituses a switching sampler, which samples the incoming message signal f() and provides impulses. Figure (1) shows the basic schematic of impulse sampling. f £0) Figure (1) ‘The pulses of the sampling signal width approaches zero " the switch of the sampler approaches zero, when the switch has closing time ‘1’.Then, the output, f(1) gives only instantaneous values which is shown in figure (2). 40) -21,-T, 0 T, 2s Figure (2k: Impulse Sampling SIGNALS AND SYSTEMS. [JNTU-HYDERABA9, The instantaneous value of f) is equal tothe simpy and the area of each impulse. version: ‘The impulse train is, 50> Yae-nTs) Where, 4, (t) Sampling function. ‘The sampled signal is represented as, Kl) =f, =fit). Yse-n7s) = Ler, 6-n1;) Impulse sampling is not possible practically because we never get a pulse with zero pulse width. Herice, this sampling is possible only theoretically Q26. Explain the term natural sampling. Ans: ° Model Paper-t, 10 ‘Natural sampling is a process in which tops ofthe sample pulses retain their natural shape. It is a complex process and hence is not used often. The amplitude of the frequency components in natural sampling reduces for higher harmonics in "* manner. : x The process of the natural sampling is shown in the! figure. m(t) q —t ott) OOO O0E at we. Figure: Natural Sampling Look for the SIA GROUP LOGO {{ on the TITLE COVER before you buy “”uNiT-6_(Sempling Theorem and Correlation) ‘The sampling waveform 5(0) consiste ofa train of pulses paving duration Yand separated by the sampling 7, The base band signal is m(t) and the sampled signal sis shown in figure. ‘The sampled signal consists of Sequence of pulses of varying amplitude whose tops are not fat but follows the waveform of the signal m(f), Using natural sampling, a vith instantaneous sampling is signal sampled at the Nyquist rate may be reconstructed exactly by passing through an ideal lowpass filter with cut- off frequency f,. Where, /, is the highest frequency spectral component of s(t). Natural sampling is more advantages than ideal sampling, the strength of pulses in natural sampling are strong compare to ideal sampling. When they are passed through the channel, there is a very less chance for the signal to be distorted. Main disadvantage of natural sampling is the cross-talk distortion. To suppress the cross-talk, the samples are limited to the duration less than 7/N. Where, » is the number of samples snd 7, is the sampling period. Which result in a large guard band between the end of one sample and starting of next sample. 27. Explain flat top sampling. Ans: The flat top sampling technique employs a sample and hold circuit with flat top samples. It is simplest sampling technique which has constant samples at the top. ‘The flat top sample sequence is given by, ID =P) OKO =p) ® [~ Sew] (1) wf (0 = FOYE-MT,) Applying Fourier transform, we get, Su-m| =) SPECTRUM ALL-IN-ONE JOURNAL FOR ENGI INEERING STUDENTS: Where, p(t) ~ Rectangular pulse with unit amplitude A(t) ~ Sampled signal P(f) ~ sine function, An attenuation of high frequency components is known as aperture effect which is the primary effect of flat top sampling. An equalizing filter with a transfer function, 1 H(f) PF) compensates the aperture effect. The value of P(f) will be constant over the baseband and no equalization will be required, when the time between samples is greater than the pulse width. rip) = 2D ¥ ipa) So feme Where, PU)=Tsine( fe) At f =0, P(f)=tsine(0) 0 aq (oe ayer [SPs Atf = W, p(f) = tsinc(¥t) Lett = —— then, low L fy= c| W.— nn vn) sl = t sine (0.1) _ tsin (0.1) On =0.98t SIA GROUP 2bSIGNALS AND SYSTEMS LUNTU-HYDERARay Time g ithe Kt Tr Discharge A switch ‘Sampling switch T oe = Figure: Flat Top Sampling Q28. Distinguish natural and flat-top samplings. Ans: The differences between natural and flat-top sampling are mentioned in table below. Process is complex and is not used commonly. It is difficult for an ADC to convert the samples. Natural sampling introduces more aperture error and requires a faster ADC convertor. Sampling is done at once without any periodic sampling. frequency components decreases for higher harmonics in (sinx)/x manner. In natural sampling, the amplitude of the 6 Natural Sampling Flat-top Sampling Natural sampling is a sampling technique in 1, | Infiat-top sampling amplitude of the pulse which tops of the sample pulses retain theit is same as the input signal and the tops of natural shape. the pulses are fat. Its easy to generate fiat-top sampling and is the ‘most commonly used sampling technique. Its easy for an ADC to convert the samples to PCM code, Flat-top sampling introduces very less aperture error and can work with a slower ADC converter. Ituses sample-and-hold circuit to periodically sample pulses. In Flat-top sampling, input voltage remains constant after sampling until the next sample is taken, : Look for the SIA GROUP LOGO {Jf on the TITLECOVER before you buy "**UNIT-6 (Sampling Theorem 8nd Cor q29. Explain the differences between v Ans! The differences by 1,| The principle involved in thi i technique is multiplication ty an impulse function » 2,|Iti8 also called as impuis mpling, 3. | The circuit ent for ideal in figure (1), a x(t) K Figure (1) sampling is shown ji a Interference of noise is maximum. The output of ideal sampling is shown in figure (4). Figure (4) |. | Time domain function, x= S)X0T, nT) 8,| Frequency domain function, SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS Sampling rate approaches to infinity, 3 4, tween various sampling techniques are mentioned in table below. ee The principle im technique is chopping principle. Itis also called as chopper sampling The circuit arrangement for natural sampling shown in figure (2), et) SUUL -—p—3 A 4 x(t) s(t) Figure (2) ‘Sampling rate satisfies nyquist criteria, .| Interference of noise in minimum. .| The output of natural sampling is shown in figure (5) Figure (5) .| Time domain function st)= 2 FS sdsinctatyo ge Frequency domain function vy (== ¥ sino(nf, X(t nf). 80-7 Dein i Table 5.15 Flat top Sampling 1. ]The principle involved in this technique is sample and hold circuit, Its also called as rectangular pulse sampling, The circuit arrangement for flat top sampling shown in figure (3), Sampling Discharge switch switch oF ree tt fl xt) | s(t Go c ) Figure (3) 4.) Sampling rate satisfies nyquist criteria 5. Interference of noise is maximum. 6.| The output of flat top sampling is shown in figure (6). | Time domain function, Yxet heat, Frequency domain function,Q30, Discuss the process of reconstructing the signal from its samples. Mayldunest9, (RH), (0) The spectrum of a band limited signal m(‘) can be completely recovered from its samples 8 7(/), by passing the sampled signal is passed through an ideal low pass filter of cutoff frequency f, Hz. The expression for sampled signal g(t) is given by, we (1) Ane att) =m) 8,(0 Where, 8,= Ll +2c0sas+2¢05 20+.) @) T Here, ©, ~ Angular sampling frequency. On substituting the value of () in equation (1), we ge, = E (o +2) cosa + 2m) cos 204+ -] + Q) ‘The transfer function for interpolation filter is given by, Hf) = T,rect (& } w= (4) Taking inverse Fourier transform on both sides of ‘equation (4), we get, ie FH) = m(arel sf } = A) =2f, T,sine (,!) = (8) Equation (5) represents the impulse response h() of the ow pass filter. When the sampling is performed at Nyquist rate, 1 ar > ral (6) On substituting equation (6) in equation (5), we get, A(t) = sin(w,!) Look for the SIA GROUP Losi AND SYSTEMS [JNTU- PHYDERABAD, SIGNALS The recovered waveform at al] Nam f Therefore, sampling rates is, m= 0,0" zZ and is a8 shown in figure LO} at Yn ihe Figure: Recovered Signal 31. A rectangular pulse waveform shown in figure is sampled once every T, seconds and reconstructed using an ideal LPF with a cut. off frequency of f,/2. Sketch the reconstructed waveform for, i = Lsec and T, = psec. kK— 1 + F x(t) 1 \ - T Figure Ans: Given that, For a rectangular pulse waveform shown in figure (1). k— 1 +—___ x(t) + 1 L >t Ty Figure (1) f 2 Cut-off frequency, 5 on the TITLE COVER before you buyNiT-5_(Sampling Theorem and Correlation) al see 1 oT The sampling period ofa signal is given by, Where, J, ~ Maximum frequency. 1 => = =a a3 Hz 2x 6 Fy Hz| ‘Therefore, the respective, rectangular pulse function can be represented as shown in figure (2), xD A >t -1 5 i 3 3 Figure (2) Reconstructed Waveform As the cut-off frequency of ideal LPF filter is, The reconstructed waveform after passing through the tiven ideal LPF is as shown in figure (3) LPF Output reconstructed 5.17 > f= “ =6 Hz ax 12 Therefore the respective rectangular pulse function can be represented as shown in figure (4), x(t), 1 >t een a i 6 6 Figure (4) Reconstructed Waveform As the cut-off frequency of ideal LPF filter is, + The reconstructed waveform after passing through the given ideal LPF is shown in figure (5), Ax) t 7 y ¥ 3 G \. roy ye x, Figure (3) ECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS SIA GROUP gD ae5.18 32. Asignal with Fourier transform X(ju) undergoes impulse train sampling to generate x,(t Yxinnst-ay where, T= 10+, For each of — : the following sets of constraints on x(t) andlor Xo), does the sampling theorem guarantee that x(t) can be recovered exactly from x,(07? 4. Xo) = 0 for foo] > 5000r 2. x(t) real and Xqu) = 0 for Jol > 5000n. Ans: Given that, {A signal undergoes impulse train sampling to ‘generate, a4) = _E nt) 80-1 Time period, 1. XGu)=0 for | o, |> 5000r Here, @, = 5000n 2a, = 10000 On comparing sampling frequency «, with the above modulating frequency 20, we get, ©, >20, Therefore, the signal x(0) can be recovered from x,(1). 2. x(6) Real and Xoo) = 0 for {o,| > $0007 As x(1) is real, Ayo)= XC 70) XG) = X(-J@)=0 for w, > 5000n ...(1) (¥ XG@)=0) Taking modulus on both sides of equation (1), we get, [XGe) |= [7 F0)| = 0 for w > 5000n => Xie) = Ofor w<- 5000n Look for the SIA GROUP Loco {¥ on the TITLE COVER before you buy SIGNALS AND. SYSTEMS [JNTU-HYDERap, Therefore, xy) = Ofor || > 5000n Here, @,7 20, Therefore, the signal x(/) can be recovered from x (1) The signal y(t) Is generated by convolving band limited signal x,(t) with another ba limited signal x,(t) that Ie, y(t) = x,(t) * x,(9 Where, x,(u) #0 f0r fo] > 1000, x,040) = 0 for fol > Impulse train sampling is performed on y(t) obtain, y,(t)= YODA. Specify the range of values for sampling T which ensures that y(t) is recoverable y,(t). Ans: Given that, WO =x) #20 ‘The spectra of the given two band limited signals « and x,() are X,(jw) and X,(f) respectively and are defined X,({) = 0 for jo} > 100% X,{fjw) = 0 for \o| > 2000n From the time convolution property of Fourier trans ie, F.T{x() * x0} =4,00)4,(o) = FT Ot) = YYw) = X(i0)-X 0) = [0] c00e * (Oa » 0008 = (0)... sc00e ss -¥Gjw) = 0 for oo] > 1000 => ©, = 1000n The impulse train sampling of (*) is given by, yf = Ler ye-n7) Where, T ~ Sampling period.5.19 1 = rst z 2x50 7 F909 T <1 msce The range of sampling period values which ensures the recover of (4) from y(1)is, OSTE 1 msec, 5.1.3. Effects of Under Sampling-Aliasing Q34. What is aliasing? Explain its effect on sampling. Ans: (Model Papers, Q10 | Nov.-15, R13}, Q4c)) Aliasing can be defined as a phenomenon in which hi igh frequency components of the signal appears as low frequency components during sampling. It is also known as under sampl ling. Figure below illustrates the spectrum of original signal and its modified spectrum for f,< 20 \ \ \\ A, Effects of Aliasing |. Theeffeets of aliasing are due tothe interference of high frequency components with low frequency components distortion occurs in the signal. 2. The information carried by the signal is lost and cannot be retrieved. SPECTRUM ALLIN-ONE JOURNAL FOR ENGINEERING STUDENTS iA GROUP =5.20 Q35. The signal x(t) = u(t + T,)-u(t-T,) can undergo Impulse train sampling without allasing, provided that the sampling period T< 2T,. Justify. Ans: The given signal is, x(t) = u(t + 7,)— u(t 7,) Sampling period, 7< 27, a (l) ‘The above signal can be represented as shown in figure. x0, Aliasing effect occur's when continuous band limited signal is sampled at a frequency less than the Nyquist rate. ie f<2h, . Where, f-Nyquist rate ‘f,,~ Band limited signal frequency. From expression (1), 1 T2h 8, () = Lent) .:. The sampled version of x(0) given by, a= D007, BCT) On applying the Fourier transform we get, a= Yx0Te ‘When the signal x) is passed through lowpass filter, the higher frequency components gets reduced. The Nyquist rate must be less than the band limited signal. Thus the signal 'x() with sampling period 7<27, will not suffer from aliasing. Trence Justified. Look for the SIA GROUP Loco ‘AND. SYSTEMS [JNTU:HYDERABAD) ea SIGNALS to be sampled with.a sampling frequency of 9, = 150n to obtain a signal g(t) with Fourier transform G(jw). Determine the maximum value of o, for which It is guaranteed that Glo) = 75 Xo) for Ja} S %- Where Xo) Is the Fourier 36. Consider the signal x(t) transform of x(t). Ans: The given signal is, ooo 22) mt = Sampling frequency, w, = 150% Ga) 75 XG) = (l) 0,27 From Fourier transform, xo) = froemar J jo(@2y edt = 50m dy = 50m f=25 ‘Nyquist rate, Sa, = oon = 2x25 Sy, = 50 2n fy, = 2m 50 Sampling frequency, ..(2) ©, = 100n (* @=2nf) From equation (2), “fi [o|st00n XU) = fi |o|2 100% XGa) @ > -100r 0 = 100% Figure on the TITLE COVER before you buyFor Ga) = 75 XY) (ie, < 100m) Thus, the maximum valu. where the signal G(jia) is perfec At, = 757, the bandwidth of the si is Ato, ignal X(jw) is 75 Hz which is exactly equal to the Sampling frequency, f= 75 Hz. Thus, the signal G(ja) is perfectly recovered from Xa), @p = 75m rad/sec i of ©, is obtained at a point tly extracted from XV/in) ie., x(t) has. Spectrum X(f) If|<200 1-1 |/200; x(= 8 { 0; elsewhere (t) is ideally sampled at f, = 300 'e spectrum of x,(t) for |f|<200. Assume that x Hz. Sketch thi Given that, The spectrum, X(/) of the lowpass signal x(t) is, 1-/f)/200; |f|<200 a= Il luis + else where Sampling frequency of 0), f= 300 Hz Then, The spectrum of the given lowpass signal, x(/) can be represented as shown in figure (1). xf) t SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS 5.21 From the definition, the given lowpass signal x(?) is sampled using an impulse train function, a fm pt) = S80-mr,) = Yar-2 | Then, the respective sampling process is represented by x(1) p(t) 1 x) a) = 440) = x) D5 m7,) = Lenn, y8¢—mr,) Where, x,(1) ~ Sampled signal of x(). Applying Fourier transform to equation (1), we get. FT (40) = FT [x(t) p(y} XN XS)* PLY) Where, A), PUP) and Xf) are the spectra of x(t), rf) and «(0 respectively. Then, X() =X) * Ss, DBF Thus, Xf) isa periodic function with replica repeating at period, f= 1. These replica by a factor ‘f” T, Since, the highest trequency component in t (0) is, = 200 Hz and sampling frequency, f = 300 Hz Ks2h, signal The successive cycles of the sampled spectrum sutfers from aliasing ie, then, Here, the original spectrum X(/) cannot be extmeted out of the spectrum X,(/), which is shown in figure Q). wo ff We 500-300-200 100° 109 20 500 ® ang Figure (2) GROUP GD38. Explain briefly bandpass ‘sampling. Ans: Consider a bandpass signal M(1), whose bandwidth is completely recovered and represented from its samples if it is sampled at the minimum rate of twice the bandwidth, The spectrum of bandpass signal is as shown in the figure. Mf) f fff itt, if, he eK > Figure: Spectrum of Bandpass Signal “The maximum frequency present in the bandpass signal is denoted by /, and bandwidth by 2, From the above spectrum, the range of frequencies lies in bandpass signal ie; from .—J, (to) f. tS ‘Where, ‘+f, Highest frequency component S.-J,~ Lowest frequency component. ‘The bandpass signal M(t) is represented in terms of quadrature and inphase components. Assuming, x,{0) ~Inphase component (0) ~ Quadrature component, m(t) = x,(t) cos 2n ft -x/(t) sin 2nft The output of lowpass filter is, ~-StabeHel-a) Hence, for bandpass signal, the minimurh sampling rate must be 4, samples per second in order to completely recover from its samples. (ND SYSTEMS (JNTU-HYDERABAD) “$2 CORRELATION = 5.2. er properties of Coron Farting a39, Explain Cross correlation and Auto correlation and their properties. Ans: Crosscorrelation ‘The eross correlation function for periodic signals x () and x,() is defined as, r 1 RO = 4. OF Jao xf (t- tat or 1t RO = He ap fuOx" (oe S Autocorrelation For answer refer Unit-V QU. Crosscorrelation Properties 1... The Fourier transform of cross correlation function of two signals is equal to the product of Fourier transform of one signal and the complex conjugate Fourier transform of the other signal. ie, FIR] =X) X20) Proof Consider K(f) = alee (l) X(f) = Jaoema »Q) On substituting, ¢= ¢— in equation (2), we get, dt=-dt Then, X(f)= fx (De? at Applying conjugate on both sides, X= Jat-vePMe™ar G8) from itssamples, Look for the SIA GROUP LOGO {ff on the TITLE COVER before you buy’UNIT-8 (Sampling Theorem Correlation = On multiplying equations (1) and (3), we get oh de ~ Fon x (thxy(t- dt} be] = FARA) FIR) = XX) Hence proved 2. The cross correlation function is zero at origin ie. at t =0.R,(0)=0 Proof Consider. 7 froxe-oar ak frcostina RO) ap funn 7 =0 (¥ x(0) and x(1) are periodic) Hence proved 3. The cross correlation function satisfies conjugate symmetry property i. R(t) = Riy(-%) Proof Consider 1 . dt Ree ge run tee ” “ot Vt : (0 afd 6) noo wb Put t=-1 SS eu fr (oxi + de RO pT Taking conjugate on both sides, we get, | eee beh fein iret | ® tsp fainniend | Putrer—t i | | R(-)= Rt) Hence proved 4. The cross correlation function does not satisfy commuta- tive property i.e. R,,(t) * R, (2) Proof Consider, fxj(t-tdt and tit As the above equations are not same, Thus, R(t) # (2) Hence proved 5. Cross Correlation Function for Energy Signals The cross correlation function for energy given by, Rp = Jaw tut (or) Ray frst Vp (t= tht Autocorrelation Properties 1. The auto correlation function at origin is equal to the energy of signal RAO) = ff (oat . T ~ . UM ALL-IN FOR ENGINEERING STUDENTS SIA GROUP a SPECTRUM ALL-IN-ONE JOURNALSIGNALS AND SYSTEMS [JNTU-HYDERABAD) Proof From the definition of auto correlation function, RAD J Hlohe-Tar Putr=0. = R= Jroa Whee, J sviste energy ofthe signa 7 (0)>1R,, (D). 2. Maximum value of auto correlation function occurs at origin i.e., R,, Proof Consider, JliozncenPar The intégral is positive non-zero quantity for # 0. Thus, f (AO#Ate+ Dae > 0 = [ror] eens fionern>o =) First two integrals in above equation are equal. Jc is energy ofthe signal J 1700 is energy of same signal shifted by an amount equal er} ‘Also from property (1) J 4200 d= 2, (0) Thus R, (0)= Jeo- rete Thus equation a becomes, R,(0) + R, (0) £2R,(7)> 0 Third term comes from definition of auto correlation, 2R,,(0) +2R,, (1)>0 => R,(0)> £R,(7) Hence proved. Look for the SIA GROUP LoGo {{5 on the TITLE COVER before you buyUNIT-5 (Sampling 7 hebéei 3. Auto correlatio mand Corrslation) cd " function isan even finetion of 7 RuD=R, ny Proof From the i definition of auto-correlation function, AOAG-T)at = JAne-rar RCD = [AORO-Crye = fronerna Let, 14+T=x at= de > RCD JHe-NAOe Changing the variable x to 1, we get, RCD = [Aone-nar =RD Thus, 8, =2,- Hence proved. ‘4 Fourier transform of auto correlation yields m times the energy density spectrum off. ie, F@P= [RM =n 5(0) Proof ~ From the definition of Fourier transform, FIR, (Dis, Jaunemrar =f fiwse-nema ar Jaumer*tar = J Ae Ma | Ken! Par Putting t- 7: f rate*tar= Fw) [five ma x, =3 dT = de =F (0) Fo) = FOF Ja (DeP aT = 2S) Hence proved. ; SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS5.26 5. Fourier transform of (cross correl Proof Consider, Fourier transforms of f() and f-(0), A) Fo) LO Flo) Similarly, Ait) OF (0) S{-) OF (-0) ‘The cross correlation function is given as, RAT) =F * £1) (* represents symbol for convolution) From definition of convolution, Fourier transform of f(0) * £(-1) is, IM) * £0) P F(0).F(-0) Thus, R(T) =O * 5) © F,(0)F(-0) Hence proved. Graphically R,,(7) is same as R,,(T) where itis folded back about the vertical axis at T= 0. R,,(7) = R,,(-T) Proof Q40. Find the cross correlation between [u(t) + u(t —1)] and 6 u(t), Ans: From the definition of cross correlation function, RAD =f O.fle+Tar ’ RD) = JOA -Tae RecN = [AOKO-C-T)ae . Jao.orra = R,(T) {given above] Thus, R(T) = Ry(-T) Hence proved. Given that, XO = [u() + ut 1)] And x,(0) =e" u(t) Look for the SIA GROUP Loco 45. SIGNALS AND SYSTEMS LINTUHYDERARAD, n function) is F,(0) F(a) ie, FO FCO, on the TITLE COVER before you buyUNIT-5_ (Sampling Theorem and Correlation) 5.27 The cross Correlation (R,.(t)) of two signals x,(f) and x.(1) is given as. Ry Jor ney Ndty QB) ad Ray = flue) suet, ayet Mut, 1) dt, flurnye ute, i) dey flay =) eMule, 1) de, a ‘The limits of the second term in the cross correlation function equation (4) depends on two constants *t° and */* On assuming both + and 1 to be greater than zero, we get, Case (i) t>t When t> 1, Equation (4) can be written as, afew dys feo I, feoranefeorna “etter =e&(-@-e)]+e0-e9] a=1te™ | Case (ii) T x1x1=0.5 — Between interval to 1 | o 2 aor? J r 7 7 | —— | | | ] | Common area=0 | Tt a ao 1 3 t ed ++ The correlation function R,,(t) as obtained from the table is, Ri2(t) J -2-1 9 12 Figura Hence, the cross correlation between triangular and gate function results in a triangular function, but with doubl time period. Look for the SIA GROUP LoGo {{f on the TITLE COVER before you buypling Tt em and Correlation) 2 Energy Density Spectrum, Parsevals , Power Density Spectrum 43, Derive the expression for energy density spectrum function of a energy signal f(t) from fundamentals and interpret why It is called energy density spectrum. Ans: The energy ‘E” of a signal f(t) is defined as the energy of the signal f(1) in frequency domain, ie, E=JPWar W” If F(@) is the Fourier transform of f(#), then Ly joe fo “3y [Foe deo me) (On substituting equation (2) in equation (1), we get, “fae a jor, ef rod [roe aol Changing the order of integration, we get, 1a. fe ob je E Lj rej ‘fide af 1 pL FOFCode 17 2 E=— ||F@/ do = 3 andl | ® Equation (3) states that the energy of a signal is given by the area under the |F(w)/* curve. Consider the signal f() applied at the input of an ideal band-pass filter, whose transfer function is shown in figure (1). ‘This filter suppresses all frequencies except a narrow band Aw (0+ 0) centered at frequency «,, The output response of the filter r(1) = A(0) * (0. Where, #) - Output of filter h(t) — Impulse response of filter. SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS ‘And energy *E," of the output signal is given by get 2 , 75 JFon) de Since H() = 0 every where except over a narrow band Ao. Fo,) = ta dw =2| Flw,) AF +H) do — _L -0, a, Figure (1): Transfer Function of H(o) The plot of signal /{?) and spectrum of its density functions is shown in figure (2), (3) and (4). sft) Figure (4): The Energy Density Spectrum SIA GROUP & TTAND SYSTEMS [JNTU-HYDERAI LS. sian transmitted through the 8.32 are “The energy ofthe output signals 2. | Fa P fen 2} FU) components off Hence 2| F(a) P. dfs the energy per unit bandwidth (Hz). ity spectrum is shown in fig ‘The units of Energy Density Spectrum (ESD) are Joules Per He, The energy density | Flo) P=| Fl-o) 5 i itive frequency components is From equation (4), it is evident that the energy contribution from negative and posi buted by +ve frequency components and other p, half is contri Dee the Energy Density Spectrum (ESD). From the energy of output signal, £ = 2 | F(@) | Flo) Pis called contributed by -ve frequency components. For this reason “The Energy Density Spectrum (ESD) () is defined as, ¥/(@) =| Fo)? uation (4) it is evident that ESD function is a real and even find out the total energy contalt From ined in the frequency band Iflsw Q44. For the signal x(t) = e** u(t), We=al2n. ‘Ans: Given that, Input signal, x(2) = €* u() Energy contairted in the band, |f| < W=? a We For W = > (On applying Fourier transform to equation (1), we get, F(x) = Fle“u() 1 WO we Ir Fle™u(n} =, whew any] “The expression for energy spectral density (ESD) is given as, v= HOF pt -(r| [riston Pay] 1 a? +(2nf)* 5.33 given by the area under the ESD curve. Therefore, the energy coritained in the band UisWis curve from ~ Wto + W and is given by, " Eneray in(— Ww) = fycpyap " = Ja lf da? «Qnty WW 1 1 aff a [teed tanf E le Gn” &IL, [: laeea™ @] a2 [ag tf 20) tag if 28 aan |" (t= ( a } “sheer pe na a G 1 [ (22). “Z atts) = | tan FOE) tant a a a v SIA GROUP &3qB frequency, S, 1 f, = 2f,. Find the output power Ans: The given signal is, MO = cos 01+ 2 sin 30+ 0.5 sin 4oye When the above signal v() is passed through RC low pass filter, sit Sey and sin 400, become zero, i., sin 304 = 0, sin 4ayt= 0. “+ The output of low pass filter is, M1) =cos wr Then, the power spectral density of output power is, 2 Jim zi v(t)P dt it 2 - dingp Js gt dt — tim Li(esee ies? 4 T= 2T 6052 ot 4 = tim 1] flare eal ft fe 4] if Ly ire im feos 2agt at ee i 7 20) 1 L awd = —(27) = gpl tho = Eenio SIGNALS AND SYSTEMS [JNTU-HY De, RABAD, G46. State and explain Parseval’s theorem, Ans: ‘The Parseval’s theorem states that “energy aleulated in the ine domain same asthe necting domain”. ie, fo wia- poma () (or) Jot afro Where, XY), YU) ae the Fourier transforms of, respectively. Derivation From th definition of fequency convoltion theorem, fam mena [xNHNe™ 7 0) Also, we) $25 Ep and ET, YH ye) 1, YH) Taking convolution of x() and (—) using convolution theorem, we get, frost - POWs a On substituting ¢= 0 in equation (4), we get, fart = frrntaar ~S) Replacing ‘A’ by #, we get, Joona 5 Jere XN) fxs a on the TITLE COVER before you buyUNIT-5_(Sampling Theorem and Correlation) a7. State and prove Parseval’: and energy type aneval's theorem for power Ans: Parseval’s Theorem for Energy Signals Statement The Parseval’s theorem, States that the ene is defined interns of Foon al the energy of a signal ‘transform, = fi XN af Proof The energy of continuous time signal x1) is given by, E= fi x(0)P dt For energy signal, the average power is zero. Then, the Parseval’s theorem for energy signals is given by, Ly 2 E=— || x(@P ral (oP doo Where, o=2nf = do=2ndf - eek zl 3g JIXCor an af iE fix@ra Li Parseval’s Theorem for Power Signals Statement The Parseval’s theorem states that the power of a signal is defined interms of Fourier series coefficients, ie, P= DCP Proof ~ Let (represents a function, Then, bP = x(x", Where, (I) x*(f) — Conjugate of x(1). SPECTRUM ALLAN-ONE JOURNAL FOR ENGINEERING STUDENTS 5.35 The average power (continuous time signal) of x/¢) is. 1 1 2 poz Jie a 2) r, On substituting equation (1) in equation (2), we get, 1 1% ot feoeroa B) 4 According to exponential Fourier series, w= YC,e"™ (4) On substituting equation (4) in equation (3), we get, Q48, Using Parsevals theorem for power signals, evaluate fe-*u(tat Ans: The given integral is, fe 2(0) dt From the Parseval’s theorem for power signals, p 1 2 2 = bt Pr path Jixot a DG SIA GROUP >SIGNALS AND SYSTEMS [JNTU-HYDERAE rine dt j y(t) dt = Jevtuna = DIG? f m =ICP Where, C= Je u(t) dt ie, . C,=FTle*w(o) = fetune ar Let = O- feteta [Fu = 1 fort> 0] a = ferra 0 - feta [¥ u( = 1 for t> 0] 0 Tung * at “o . +l Then, ; 2 2 t 1 1 , = = feta ar = rol Lal 2 15.37 UNIT-B_(Sampling Theorem and Correlation) UMTS (Semping Thecrem and Gorin) BF S (t) 49. A signal y(t) given by Yit)=C,+ S.C,coa(nugt+0,). Find the auto correlation and PSD of yt). ™ Model Papert, 11 Ans: The given signal Is, HO =C,+ LC cos(noy+8,) nt Auto correlation of y(1), R¢r) = PSD of (1) =? Auto correlation of (1) = 7 Auto correlation of function {is given by, Toe R¢)= Lim L t froseesa Then, auto correlation function of y(#) is given by, 1 i (1) y(t+t)dt = R@=tim ps y+). m2 » 8,) | Co+ S.C, cos(naye + nays + Je = agefimt 7 J [er¥e. stro + »| A x fn n= Lim 1 7 [aroc$ cos(ntagt + 091+ 8, +d CoC, cos(neayt +8, ore > “in +c? FS cos(nayt +6,)cos(ndgt nas] nl = Te 1 Lim 1 ai Chat + Lim ne CC Lesinay+noy+9, dt T in e Lim LFF C2 cos(no 148, )cos( npr + net +8, dt shim? J Soc, cos(nyt +O,)dt + Fame = i YLcF cos(ney tml fine 72 cr 6,,)c0s(noye +nayt+8, )dt rim Jrcos(nay+ = RDS +404 5 Tr JE jdt Cs ci S Stim “Footanoy +02, ) + 608(n0091)]. . OT TO in SIA GROUP x NTS SPECTRUM ALLIN-ONE JOURNAL FOR ENGINEERING STUDE!nm Joos mop! cy ie & DL bm aye c} + sk Due J? Sch ff) =Ch+ Lark sono 2 Power Spectral Density (PSD) of 1) is given Py> PSD = FIR) = esp=e(c Scams] m= = PsD- ci} Seeeve-my ser ( Fleomu,t] = 2n{8(o—n0,)] + 60 +n). Q50. For the signal ott) = 2a(t? + mine the ‘essential band width B Hz of g(t) such that the ‘energy contained In the spectral components of g(t) of frequencies below B hz Is 99% of signal energy E,. Ans: ‘The given signal is, 2a On Te ‘Then, the energy of the signal g(/) is given by, z,- fish a V[_2 7 See) - 4a jotta o=n2 LL. - oo=atan® 2 enn 12 1 2 . ty asec? od8 = 6,740 | Ghat ora) #12 =40 ae a I + x sec? 8 a' sec’ @ 2 = Toast 08 Fan 4% [1-2 a ain 2 8)? 48 : a2] a2. 4 Jw =2m a 2n n ET The energy contained in the spectral components &(t) of frequencies below B Hz is given by, 1 £, = 7, 1GUa)P? (Aa) = £,=|Gja)PB ‘Where, G(jw) - Fourier transform of g(#). Look for the SIA GROUP LOGO {5 on the TITLE COVER before you bu lyuNIT5 (Sampling Theorem and Correlation) Then, the Fourier transform of the given function g(?) can be calculated using duality property ie, 2a wet Gt MOm ee ADs Xj) = = wo +a Then, the synthesis equation for this Fourier transform pair is, oO 4a "do Replacing by ~1 in the above equation, ie, Ga) =2ne Then, . E, = |2ne** ? B =4r? Bem 1. (2) But, given that, £, = 99% (E,) ie, £, = 0.99(E) From equations (1) and (2), 2n 4m Bew =099% = we p= 299-200 2na [pod 0158 mire | “a SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS Qs: Ans: 5.39 Find (or power) of tht output voltage y(t) of the system shown In figure. I the input voltage PSD. 8,(.0) = rect(w/2). Calculate the power (mean square value) of input signal x(t). 19 x) IF Ya) Figure The given system is, 19 ww Input power spectral density, 5,() = male ) Power (mean square value), P= ? From the above figure, 1 jo wt jo 2()= x) 20 (0) 0) “ia ™ d and output voltage, (1) = Grol d O= ribs ("From equation (1)] | 4m 0" Teja dt Sia GRouP 4FOO be JjoXo) im io) Yo) = 1 (1+ jo) Y(a) X@) fo (1+ jo) Ho) = w= (2) oO I+ jo The relation between output-and input power spectral density is given by, S(@) - |H(@)? So) »@) ‘Where, H()- Transfer function $,(@) - Output PSD S(w)~ Input PSD ‘On substicuring equation (2) in equation (3), we get, wf 53(@) ay ) S@)= S(o)= w= (4) On substituting the corresponding value of S,@) in equation (4), we get, - Orel (ta) (2 The power or mean square value of x() is given by, So) 6) On substituting equation (5) in equation (6), we get, Power, (P) = 37 [sto SIGNALS AND SYSTEMS [JNTU. HYDERA, Bap Q52. Explain the relation between auto correlation function and energy/power spectral density function. (or) Prove that autocorrelation function and energy ‘spectral density function forms a Fourler transforms pair, ‘The Fourier transform of auto correlation ism times the energy density spectrum of f (1), Mathematically, it is. expressed as, Foyt four ar an so) Look for the SIA GROUP LOGO {5 on the TITLE COVER before you buyUNIT-5 (Sampling Theore mM and Correlation) Proof From the definition of Fourier transform, Joutrermrar . JROKO—r a ar J Ade arf from Nar Fo) J Ae—T ear Put rr e x > 0-ar=_ = aT =a ad fo Wear = Fe) Jac xed =F) Fo) =IF@)F + Joulte "Tar = 5 Si) ( IF @)P=7 S(@)) Hence Proved, Q53. Determine the autocorrelation function and Snergy spectral density function of xit) = e-# u(t). Ans: The given signal is, (1) =e u(t) Auto correlation fufction, R (1) =? ESD of x(1) =? Autocorrelation Function The expression for autocorrelation function of ral signal AU) is given as, R(t)= J x(x(t=9) dt. (1) Here, x(1) = €* u(t) is real ‘VECTRUM ALLAN-ONE JOURNAL FOR EN 5.41 Therefore, equation (I) can be wrtien aw ROO) = Feuer yy = a= fetewn (501 fort 0 and (tt) =| for >) = RE femern = noe [etee dt, for 120 R= er ferry uy = RO=e| 0-e2* = RQ! RQ) =e] "|e o Ronee et RQ) = —— forall values of t 2a The waveform of abi ove autocorrelation function is as shown in figure, RO,5.42 Energy Spectral Density Function The expression for Energy Spectral Density (ESD) function of v(2) is given by, Wo) > FIR(T)) =f Rime ae ‘. - =H f etemtacs fener at . 2a} 4 jecmasfemmra ° eet T ~(at jo) Lt Q54. Prove that auto correlation function and energy! power spectral density function forms Fourier Transform:pair. Ans: The auto corelaton funtion fora periodic signal i givenas, ae RO F J x0 xa .() SIGNALS AND SYSTEMS (UNTU-HYDERARA, ‘Wher x, (is the periodic power signal and i given " Y, x= zx x(t= mT,) “2% On substituting equation (2) in equation (1), we get J x) y x(t-t mT,) dt As the limits are from (~<°, -), the signal x(1 is said tp bbe non-periodic signal. I< = )-7 4 x a(t) x*[t— (t+ mT,)] dr ims --Q) The equation of auto correlation function for energy signals is given by, e ra-f x(t) x*(t- 1) dt Equation (3) becomes, z z R(t+m4T,) (4) RO= ‘Taking Fourier transform on both sides of equation (4), we get, Le FRO= 7 Y Arctnry 0 wn “72 wen , CROASWD) By implementing the shifting property of Fourier transform, we get, WA) =IXP ~~ 6 * On substituting y(/) value in equation (5), we get, FAR) died eer FA Look for the SIA GROUP LOGO { on the TITLE COVER before you buy5.43 1 Te Le Nm Pam) fem) le 7 DY owe (80) 1) he. FIR) = sy) or Rt) A> sip Therefore. fora signal, auto correlation function and power spectrum form a Fourier transform pair Q55. Show that the auto correlation function of g(t) = C cos(w,t + 6,) is given by, R,(2) = (67/2) cos «,t, and the corresponding PSD is, S,(o) = (C*r/2) [d(w - 0,) + dl + 0). Ans: The given function is, (1) = Ccos(w,t +.) It is a periodic function. The auto correlation function of g(7) is given by, r = R= 7 Jetoe-o fe c0s( gf + 8p ).C COSCO ~ OyT+ Oy dl + r 2 R= ; Je OS( Wof +0) COS(CWyf ~ OT + Oy dt or aI . SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS. SIA GROUP x&5.44 SIGNALS AND SYSTEMS [JNTI r DERABAD, £08(2004 — ~ DoT +209) + C08( (04) ) 9 ; xe dt (#20084 cosB = c09(4 +B) + cosa gy ce DF J (608(2e09t ~ (ot 285) + cos(ngt)dt r ay 2 ce r z j [c0s(29t) cos(«o,t-28,) +sin(2091)sin(«yt-28,)+ cos(yt)] de = 2 r ce cos(@,t-26,)} esse +sin(ayt-! ea +005(o, 02, ? or 2 19. 22m) - c08((09t 28, Xsin| 2a,{ 7 2)" 204 2 +sin((09t-20,) {ost as = reocovn( $5) gees 209)[sin(2n)+sin(2n)] ° OF ‘4 $iM(@pt= 285-008 2+ 08 20) Ifo ay + 008(@yt}(T) | co =Gl0+ 0+ cos wx] R@= (S ost a oR) = (So 097 Look for the SIA GROUP Loco ‘on the TITLE COVER before you buy “"unit-6 (Sampling Theorem and Correlat n) Then, the power spectral density of gi is given by, S(@)= FTIR) oo) Cc = SE Tleos (0,9) 2 7 (IS (@-a,) +5 (@+0,))} 56. Explain the relation between convolution and correlation. Ans: ‘The expression for correlation function of periodic signal (i) is given by, 2 Re) = Lt = | xtext—ar .@ in Ea eT Where, T ~ Period of the signal, x() The convolution between a signal x(¢) and its delayed version is given by, p= f x(f).x((-Dat = Q) Generally the correlation between the two functions x,() ‘sd(0)is same as the convolution between the functions x1) dx), Hence, equation (1) becomes, 2 : J x(t)x(-t-1)dt Rqy= Lt DECTRGM @LLAN-ONE JOURNAL POR ENGINEERING STUDENTS _ 5.45 ft = R= WE) x1 ar 1 RO = Lt FEO") [> From definition) n= Lt AL(exeny| Late TO | Hence proved. 5.2.5 Detection of Periodic Signals in the Presence of Noise by Correlation, Extraction of : Signal from Noise by Filtering Q57. Explain the method of detection of periodic signals in the presence of noise by correlation. Ans: A noise signal is defined as a signal with random amplitude variations. Considering p(t) as a periodic signal and n(/) as a noise signal, then, from correldtion property, is Spn(t) = ed P(t)n(t-7dt = 0, for all t t Detection by Auto Correlation Consider, . AO= pl + n(t) (ld Where, (0) ~Periodie signal (0) ~Noise signal Ad) Received signal Auto correlation function of /{¢) is given by, r 3c) = Et ; j SOflt-Ydt—.Q) A GROUPS546 r wont Lf it i n Fd a r L r P().p(t=1) dt+ from» dt + fro p(t-1) de+ prone» dt 7 z 9,0 = 9 + Bye 5,048, 0 Where, ,,(t) ~ Auto correlation function of p() 9 (0) ~ Auto correlation function of n(t) (0) — %,, (®)~Cross correlation of p(t) and n() . As p(t) and n(¢) are uncorrelated then, $= §,(=0 * — ( pltyandn(t are uncorrelated) 9,0 = 9,04 9,0) An auto correlation function of a periodic signal covered by noise is as shown in figure below, Figure The periodic component 6,,(t) and the nonperiodic component ® ,(t) are separated when the nature of 8 (2 is periodic Hence, auto correlation function of ¢) cannot be calculated over a infinite interval. From the equation (2), the auto correlatio function is evaluated over a finite interval 7. When the cross correlation functions are evaluated over (~ °°, °°) then @ ,(t) an 5 (2) are zero. The cross correlation function become smaller for large value of 7. Look for the SIA GROUP Loco 4B on the TITLE COVER before you buyUNIT-6_(Sampling Theorem and Correlation) Detection by Cros ‘orrelation ‘The cross correlated received signal is given by, Ad) = Pl) + n(n) r 2 » Fy = Lt 1 flo +molea—nar Fa eT [re m2 =A) fonmew-nar+ Jmnreu-rar T Léa Te +O) Where, (1) ~ Locally generated signal whose frequency is same as that of p(1). The generated signal c(t) is a periodic function and is uncorrelated with the random noise signal n(t). + O@=8@ (6 Fy=0) In the above equations p(0), c(t) and 6,0) are periodic functions hence their frequencies are same. The cross correlation between ft) and c(t) gives 6, (2), does not give any noise terms. Q58. Explain briefly the Process of extraction of a signal from noise by filtering. Ans: Model Paper-iil, 11 Filtering technique is the extraction of a given signal in the frequency domain. If the given signal is extracted in the time domain then it is known as correlation technique. Relationship between Correlation and Filtering Consider f(0) and f(A) as two input signals, whose cross correlation function is 6,,(2). £0 °F) £0 © F,@) Then, f,, () © F, (w) F, (0) Multiplication of the spectra F,(«) and F, (- «) in the frequency domain is equivalent to cross correlation of (0) and A{0 in the time domain. The cross correlation in time and frequency domain is as shown in figures (1) and (2) respectively. £0) £0 8:0 Crosscorrelation Figure (1): Cross Correlation in Time Domain Fo FORGO) — 7 RCO PS £0) o. Figure (2): Cross Corre tion in Frequency Domain SPECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING STUDENTS SIA GROUP &5.48 i SIGNALS AND SYSTEMS (JNTU. HYDERARa, he gee gun FAt) —Input signal F () ~ Transfer function 6,0) - Output From figure (2), iti noted that when the signal (1) is applied to F,(-0), the output will be 9,(¢). Ths signal application effects the cross correlation between signals f(/) and fi) which represents filtering. The impulse response of a system is given by, A) = FTF, (-@)] . A(QO F(a) + MOKA (Hones) If s() is the desired periodic signal component ind n(t) is the random noise component thei the received signa ig given by, Aa) = s(t) + n() The cross correlating signal f(¢) with another periodic signal c(t) of the same period as that of s(, detects the desired periodic signal components s(#) which is present in /(*). This means that the cross correlation function is performed by a system which has a unit impulse response o(-#) or a transfer function C(-) ie, CW) © Clo) 1) € c(-0) Where, C(#) = Periodic signal with period 7,, * (0) = Fourier transform of periodic signal c(t) Therefore, C(a) and C(- ) consists of impulses located at w= 0, + Wy + 200, + 34... mm, The magnitudes of impulses located at different frequencies is equal to 27 times the corresponding coefficients of the exponential Fourier series for oC). ‘Therefore, the Fourier series representation of c() is given by, a= Dee ‘Where, ont “ % Then FT. [e()] = Clo) = 2n E ¢, 5(0-no,) And C(-) =2n E ¢,* 8(0-n0,) [¥ CCW) =C*(W)] | Look for the SIA GROUP LOGO {J} on the TITLE COVER before you buy