Scribd
Upload
100 views47 pages

Convolution and Correlation

Uploaded by

Kishan Adhikari
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
100 views47 pages

Convolution and Correlation

Uploaded by

Kishan Adhikari
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
= Convolution and Correlation of Signals 4, INTRODUCTION gr ottion 1S mathematical Way of combining two signals to form a third signal. It is the a ost important atu in Digital Signal Processing. Using he strategy of impulse a son, systems ae ia ied by a signa called the impulse response Convolution is te este three signals of intrest the int signal, the output signal, and al operation tats very smart convolution ation is a mathe ce a third signal. This third signal paras with convolution: re meation uses two signals 1 prod ead the oross-correlation af the two input signals. Ifa signal is correlated with itself, the vetead called the autocorrelation. ‘ems, its properties, correlation, enerBY and power scaling signal isi This chapter presents the ast specta, Rayleigh’s enerey 12, CONCEPT OF CONVOLUTION which is used convolution theor’ theorem and Parseval’s theorem. to express the input-output relationship symoluton is mathematical operation 3fan LTI system. Mathematically, the convolution z(t) of two time functions 21) and y(0) is defined #8 ai) =x) x YO. (OR) 2) = x(t) * 90) = f xeyyt-n)de ‘An encircled multiplication notation ® or # i used to denote convolution. "Thus, zt) = x0) © y(t) i8 read as x(?) convolved with y(- Bi wae 201 292 + signals and Systems 73 CONVOLUTION THEOREMS .e convolution theorems associated with Fourier sey transgy. forms, h Here, we will discuss th convolution theorems are: Time Convolution Theorem 7 i «Frequency Convolution Theorem 7.3.1. Time Convolution Theorem The time convolution theorem states that convolution in time domaj In ig multiplication of their spectra in Frequency domain. Mathematically, “aia, if x(t) 2 X(@) and y(t) <——> Ya), : then x) * () 3 XY) Proof: The Fourier transform of x() is given by X(@) = Fix) =f x) e/at Fla) * yO] = f (a+ y(] ear -« ATI B ° roe ut, a) * (0 = f x(0) t-2) dr 2 Fix) + yOL= J [i x(t) (t-7) ‘] el at Interchanging the order of integration, we get, Fix) * yO) = f « Jxt= 2H" ar | a ed Using the time-shiftin, a + ig Property, we get i 100, J x02)" at = HQ) eter ila me re ne . sth FIC) *O1= f x(2)¥(@) e-H0" gp = Y(@) ij = I AO) de = HeX(@) 2) 1) 3 X(@)Ko) 7; This is the time convolution theorem. 4 Convolution and Cotrelation of Signals * 203 . ey Convolution Theorem n theorem states that the multiplication of two functions in time domain io ve of their spectra in frequency domain. olution x) 2 X(@) and y() — Ko) Inx(t) Wt) —— X(@) * ¥(@) x(t) ¥() 9 X(f) * HCA) the inverse Fourier transform are given by pe one asform and Xo) = FIX) =f xt eI dt x) = F'IX@)] = = fh: X(o) ei do Fix(t) yO) = { x(t) y(t) eI dt =f ifzi J xa e™ abocima invechanging the order of integration, we get FE) Ol= 55 Lj jaa y(n) ee ala da ‘ wel x0] yneto™ ala 1 eyes Add ar J XY@-Ad. Fit) yo = 3 X10) Y(@) x) YO) — xo Y(@) 2m xt) YQ) <> X(@) Y() This is the frequency convolution theorem, hi ai 294 + Signals and Systems 7,4 GRAPHICAL CONVOLUTION Normally, analytical computation of convolution becomes complex and gop. functions. However, the convolution can be estimated by inspection, ang ce iat with regard to convolution can be made without actually computing the convoy Ni In many appiaions of communication systems, the convolution by ing tO.*y information needed without complicated calculations. This convolution by inspection ide hy is called the graphical convolution. ‘The procedure for graphical convolution between two functions, 2) and ce " ed, x) #0 = f x(t) ye-2) de %, is as follows: - (2) isthe first function, where an independent variable ‘ris replaced by ty re variable ‘7’. ity ¥C1) is the mirror image of y(2), i.e., (7) is lipped around the negative Janis, y(t — 1) represents the function y(-t) shified to right by t seconds. Fora particular value of =a, integration ofthe product x2) y(a~ 2) represents the gy under the prodet curse (common area). Ths common sea represen the como’ x(0 and y(t) for a shift of r = a. ie, J 2 ya-1) dr =[x()* Ona The procedure is repeated for different values of t to evaluate the convolution, For a negative value of t, the function y(~2) is shifted left by t seconds. The function 1) may be considered as a rigid frame progressed along t-axis. The value of convolution obtained at different values of t (both positive and negative values) may be plotted on a graph to get the combined convolution. Bey a 7.5 ENERGY DENSITY SPECTRUM. Spectral density: It is the distribution of power or energy of a signal per unit bandwidth as a function of frequency. Energy and power signals: Signals with finite energy, ie. 0 < E < co and P = 0 are called energy signals, e.g. aperiodic signals like pulse. Signals with finite averge power, ic. 0 S(@) of S(f) (172) 5, The autocorrelation is periodic with the same period as the periodic signal itself, ie. R(@) = R(t + nih), n= 1, 2,3, .. an) 7.41 RELATION BETWEEN AUTOCORRELATION AND SPECTRAL DENSITIES The following are the relations between the autocorrelation R(2) and ESD yi f) and PSD S({). L. The autocorrelation function R(t) of an energy signal (pulse) and its energy spectral density (ESD), (A) forms 2 Fourier transform pai, ia RO A) 306 rp signals and Systems -correlati n of two energy signi (0) and 0 7 Ven Ry (= J xy (fem df a Proo re same, then the autocorrelation is given by If both the functions a xx (nel df= I XCDP ed ap R= i =F (FOP) But ireph= v) R= Flly(f)) > FRI = YW) Re) 9 Wh ‘The autocorrelation function R(#) and the power spectral density SD), 5 ) rm a Fourier transform pair. ofa R@ — SF) signal fo tn ie. Proof: The autocorrelation function of a power (periodic) signal x(t) in = in mtg series coefficients is given as: Riny= YS X,X.,e0"0" where X,, X., are the exponential Fourier series coefficients. R= z Ix,P elt? Taking Fourier transform on both sides, we have FIR@= | | y een | ie le Interchanging the order of integration and summation, we get, AR@I= S| XP J ern) " =n S| 1X,P 6(@— nay) = 5 IX,P 3g mf) REL ‘The RLS. is the PSD S(a) or S(f) of the periodic function x(1). c FIR] = S(f) , R@ = FUS(f)] Rt) — Sf) v Cenwvolution ang Correlation og ignals peTWe CONVOLUTION AND - / aattON : \ CORRELATION and correlation are closely related With the following Minor di aio * sical time ‘plays the role of a g . ifferences: vation, phys! Nmmy Variable and i disappears afte gc e integral Sa : Mu convolution, delay parameter, ‘7’ plays ne g(t) is a function of the delay tion May" init ¢ fu ‘n arame Parameter 7, Whereas, Convolution is a come” of time f- ® the role of a dummy Variable, jon can be obtained by convolving x(-1) and y*¢y, A does not depend on which function is be juti¢ ing shifted, whereas, correla i i * ', COrre] ‘ conval! convolution is commutative. lation 7 ie oes y£CTION OF PERIODIC SIGNALS IN PRESE| B coRRELATION siodic signals in presence of random noise/AWGN el ce of ran is of great importance in eres various applications in RADAR, SONAR detection, detection of ve Cel in brain waves, the detection of cyclic components in ocean wave analysis, * gc comtpO NCE OF NOISE BY etc. . . ‘ . 1 Gon techniques provide a powerful tool in the solution ¢ of the above problems, al encountered in practice is a signal with random amy , plitude variations. Such spenise sign i iodic signal. related with any peri . is une eriodic signal, n(t) represent the noise. [et s() represent the p 1m = lim — | s(t)n(t-r)dt=0 Ral) fim 7 J et =m 184 Detection by Autocorrelation las{) be the periodic signal mixed with noise n(t). Signal will be Tie eceived signal wil Si leenteen tice the received signal y(t) is also periodic. = 1 F = lim —fy() t-7) dt f R@ jin pO ae tin Jls(e) + n(e)] [se —2) + 00-2) dt T30T T RO = RO +R +RO+RO 908 + Signals and ‘Systems As s(f) and n(f) are uncorrelated, We have RO=RO +R) he autocorrelation function of a periodic signal is peridic ang that of for large values of 7. ‘As s(f) is periodic and n(@) is non-periodic, R,(@) is periodic and Ry (2) is arbitrarily small for large values of ¢, Therefore, for sufficiently large values of 7, the autocorrelation ofa siven 5 calculated by the numerical techniques used for convolution; even on digital compa ea, We know that th function tends to zero 7.13.2 Detection by Cross-correlation 4 The detection of a periodic signal can be carried out by cross correlating the receive « another periodic signal of the same frequency. Detection by cross-correlation ig “Seta yg effective than by the autocorrelation, The disadvantage is that it is necessary ae Oi frequency ofthe signal to be detected, beforehand. In many cases such’ as RADAR, 0 frequency is known beforehand. Let the received signal be re yt) = 5) +n) Let the locally generated signal of same frequency as that of s(t) be 1(0) Seer Ry = jim = Jo +n(t)] I(t = 7) dt a eee eed Ry) = Ry () + Ry (2) If IG) and n(t) are uncorrelated, Ry) = 0 oipatad : : Ry(O)= Ry (2) «a Itis clear that if the cross-correlatio eee 1 of the corrupted signal y(t) with it) yields a periodic sgn, (0) must contain a periodic component of the same frequency as that of 1(), Rome oo oh y(t) = a Hence, we can conclude th mt BelO)+ Ra) . clude that : cae hae whether a periodic component is present or not, for any vale RO=R@ IMPORTANT FORMULAE wy 1. Convolution of x(t) and O=10 89) = Taerye- a Convolution and Correlation Of Signals * 309 ion 0 an x(t) * y() con olution theorem: y X(@) Yo) 4 i 2 convolution theorem: 2nx(t) y(t) ——> X(@) * Xo) 0 f rl energy . . B= J Iatof ar= f 1xX(pP at snrmalized POWET ane OL * P= x (= lim — J x(n? at T3“T in P= EGP oe «_parseval's theorems payeigh’s Energy Theorem: B= J IX(A)P df paseval’s Power Theorem: P=) [X,° 6, Energy spectral density (ESD) vif) = KP W4(A) = HAP wc) 1. Power spectral density (PSD) s= EL w,P OU =n) S(@) = 5 1x, 7 5(@—na) i Sf) = LHCPP SA) 8, For energy signals, R() <> WW) For power signals, R() <> Sf) 9. Cross-correlation a 1 wg eaydt= lim & f xte+ayy"Qat Rat) = lim {ny ana jmz J Ry) = Rul : we x(t) and yt) are orthogonal if and only if Ry) is and §) 310° signal 40, Autocorrelation RO= paoseona= fatto (oa Ree RO ace) ¢ RO) For all & (0) = P for power signals. 1 signals. 0) = E for onere SOLVED PROBLEMS Problem 7.1. Show that the convolution of a function x(¢) with a unit i the function itself Unit impulse fue Solution: Mo x 5) = FFE) FBON But F(x()} = X@) 7 F(S) = | x(t) x 6) = F“(X(o)] = wo. Problem 7.2 Calculate y(0 if and Solution Given a) ee, —wetce KM) =t, o0 ‘ins Given x) =" u(), a> 0 Mt, E= j |x(oP? at = feat = fer at ‘Therefore Now, according to Parseval’s theorem, 4 E= x J Ix@P ao X(@) = F [x] j x()e i dt = fete lar 0 = fe @tiorar 0 eat ioye ° ~ | (a+ jot ev -e X(@) = — at+jo Now, we write, X(@) = Th erefore, Ko = c onvolution and Correlation Of Signals + 319 tc Sell IX(o)P dw 1 ae 2na 1 E=— ; for 2a : E. ¢e from Eqs- (@) and Gi), the energy is same. Hence, the Parseval’s theorem is verified. . ¢ 141. Find the power for the following signals. EH aaa w Acos ot . 7 : ; (at. fi, ais a constant and f(#) is a power signal with zero ‘Solution’ fp Given 0) =A COS yf tial Fourier series of the cosine function exists only at +0, |The exponent ‘Te amplitude of the two components located at i is A/2. . A ie, yet Fixe ‘Therefore, P= > ix, I nom AY, (AY ={—] +/= 2) +(4) Therefore, Oe P= Mean square value of x(t) pre = = f A®cos* apt dt Tin Therefore, 2 x) =atf(t) The power corresponding to a constant a is &. The power of f() is its mean square value f2(1). Therefore, Total power, P= a? +f°(0- This can be proved as follows: we have, , 1 72 2 at P= lim— f (a+ fat fing J (ii) Given 1 1/2 F aid 2 2 lim if fader J P@d+2a J roa} Toe! lr id 12 sa +f(y+0 Poa +f) Problem 7.12 Find the autocorrelation, power, RMS value and sketch the PSD, fOr the g x(0) = (A + sin 1001) cos 2002. meh seb [INTU:’May 204 Solution: Given x() = (A + sin 1001) cos 200¢ =A cos 200 + sin 100r cos 2007 = Acos 2001+ $ {sin 3001 - sin 1004] Let x() = x(t) + x2(8) +350) : : where x4( = A cos 200t A x(t) = sin 300r 4) = Asin 100r Autocorrelation of x(?) is RO) = Rus(2) + Rant) + Ryg(t) + Ria(t) + Rix(t) + Roy(t) + Rog(t) + Ry) + Rul) Ry) = Rai) Rix) = R(t) Rat) = -Rop() (#0) RM) = Ryo) + Rot) + Ry) y | But Convolution and Correlation of Signal B21. , sit) dt im wr Rio= fm = tim & fA? 60s 2006 co tam T am s(C— 2) dt = lim a a 20s 2008 cos 2 Toe 27 ara 200 (tr) de ame = phe OT } J feos 4001 — 2007) + cos 2007] dt = Jim A» cos 2007 [i], rs Seint ‘ ° 2T ry * Am. = 2 uincsoor — 20071", __ A? cos 2008 pes R= Rylt) = jim nF alt Xp(#) x(t- 7) dt yor -m W 1 TI2 1 im — J sin 300r si . lim 1a jn 300t sin 300(t— 7) dt 172 im — J 2 sin 300¢ sin 300(t- An. er J sin 300(t— 7) dt w 1 12 tim -L f [cos 300 ¢ — cos (6001 ~ 3002) dt ny To 8T _tp [a @00r - 3007)" = tim 2 cos 3007{f sin (600t ~ 3001) | te a Wr TI [00 un 600 12 1 | Threfore, — Rpp(1) = oe 3007 Re()= lim L% xg(t) x(t - 7) dt oT a TI2 = lim ui L sin 100t sin 100(t-7) dt t=T In 322. + Signals and Systems lim L 7 2 sin 100r sin 100(t ~ z) dt Toe 8T nn " 72 tim 4 [ {c0s1007 ~cos(2001 ~ 100) Toe 8! ng 1 ni2 sin(200t ~ 1092) 72 = lim— ~ lim —| 1} = im ir cos 100t [1] 775 i” er 7 | S220 00 Tr 1 Therefore, Ry) = gos 100r Therefore, RD = Ry + Ra + R332) 2 = cos 2007 +4 cos 3007 +4 aos 1007 Now, power Therefore Now, RMS value = JP = PSD S(@) = FIR(1)] FI ax cos 2007 + t cos 3007 +t cos 1007 2 8 8 So) = Xs 150+ 200) + 5(w - 200)} +2 150+ 300) + 5(a - 300)] +2 150+ 100) + 5(@ 100) -n0 So) : a. ae 822+ Signals and Systems = tim 7 2sin 1001 sin 1001 ~ zr) dy T= 8T in = lim — L? [cos 1007 — cos-(200¢ ~ 1002) a T+ 8T 5p | sin(200r =100e) yr = di gp 008 1008 7 say “Tr 1 Therefore, Rela) = 008 100 Therefore, R(1) = Ryi(D) + Root) + R332) 2 = cos 2t0e +2 cos 300F + cos l00r Now, power P= R(0) apy BS +o45 278 8 Therefore tala 4 Now, i RMS value = /P = PSD S(0) = FIR(Q)] a = is cos 2007 + 4 0083007 +t zoos | S(a) = Frise-am ee 200)1+ = [5(e + 300) + 50-300) +£16(@ +100) +5(@~100) “mR So) Convolution and Corteaton of Signal + 823 ine the enerBy spectral density (ESD) of a Gate function of width ¢ and es 4x) “m0 wm + |<— 7 —>| Figure 7.9 Problem 7.13. pove figures the Gate function of width + and amplitude A is given by: the a From t a A, Ith<= ya x= 2 0, elsewhere x(a) = f xe! at cai 12 f Act at -t/2 " A feior?2 _ gloe!2y jo A pelo _ ny jo 2A sin =) @ 2 sin(wt/2) (at/2) Trefore, X(@) = ArSa (=) Teor, the Fourier transform of a Gate function is a sampling function Tare, ESD y(o) = (oP on -e2se(*2) 824. + Signals and Systems Problem 7.14 A signa TL rad/s. off (i) Test whether the input is an energy signal, Te, (ii) Find the input and output energy. 4 1) is passed through am ideal LPF ign Cut. Solution (i) Given input signal x(t) =e u(t) Energy of the input signal, = fixoPar = fear 0 = feat 0 Therefore, E; = 0.167 units = 6 Now input power, lim ai E To» 2T eee (e lim —|—|= T= 27 (3) Therefore, input energy is fi = inite and power is zero, x(t) i is (ii) Now, ESD of the output ¥( is given by , x(t) is an energy signal. ar WO) = |H(a)?? y,(o) qe 4 ods. nai ACO) = XC)? “ ; sano Now X@) = Fle* u(y) = 1 3+ jo rmlet) Va) a +9 a LPF, the square of the transfer function is given by: Hoe = fr lols1 0, otherwise for Vy) = IH)? y,(w) 1 =lea9 lo|<1 0, otherwise ‘Therefore, the total energy of the output signal is given by 1s y= 3 )mode 1} ==] 7 do Ta +9 sant G) a 3 Tere, Fo = 232 = 0.0341 units 3a " Trblem 7.15 Find the PSD of a periodi¢ function f(#) with period 7. ion: Let f(t) be the periodic function with period 7. . _ tft) be the truncated portion of periodic funetion f(t) over a period of 7 Srl) = f(t) x Gate function Gr(t) Fr =fO GrO -1 oF) Fi F,(@) = drsa( z s (@) [Using Frequency, convolution theorem] its Fourier transform is given by As f(O) ¥ F,5(@- nay) 2 T Fo =T > F,84{ Fx 500 —na) rene property of impulse function, we have, = T F,(@)=T Y #,52[(@-n) 2] Using sampling The PSD is given as: _ @P = = lim Y IAL Tote IF the limit 7 —> =, the sampling square function is reduced to delta function, 2x5 = (ng Therefore S(@) = 22 IF, 6(@- nay) i Problem 7.16 A function f(0) has a PSD of S(a). Find the PSD of (i) integral of faa (ii) time derivative of f(). Solution: Given function is f() with PSD S(@). (i) Let £0 <— Fo) Then for the integral of f(1), we have J rod Go+re -o jo ‘Therefore, if f(0) is the in iputs with Fourier transform F(q), then the Fourier transom of the output, ¥(@) = — F(a) jo Therefore, Transfer function, Ho = XO) F@) = Mia) Fo) F(o) Therefore, Ho = jo of Signala +327 Now. (ay? = Now, the output PSD is given by SK) = MHC? Scan 5 ne ‘Therefore, S\(@) = Gr 5t@) ) Let £0 — Fa) “ ‘Then for derivative of f(), we have Gy dt a —— jo F(a) Now, the transfer function, jo F@) H(@) = ——— = j Fa)! |H(@)P = @? ae ut PSD is given by 5,(@) = |H(o)P S(o) S,(@) = @PS(o) ‘A power signal g() has a PSD S,(«) = NIA?, -2nB < @< 2nB, shown in figure 1a probe 7 d N are constants. We ie ne PSD rand the mean square value of its derivative ded pel! [JNTU: May 2003; Nov. 2006, 2007; Feb. 2008) 45,(@) = NIA* -2nB 0 2mB o Figure 7.10 = a(t) ion: Given input signal Input PSD, S,(@) = 7a output be the derivative of g(t). gl) > 6) “0 eS joGo) systems, 328 + signals ond Transfer Output H(@) = ~~ Input _ jaG@) “~~ G(a) H(o) = jo ‘Therefore, \ ) i Therefore, =a (ao = © ., Output PSD, Therefore, Outp' sa) = war S(0) 2 aN Therefore, S(O) = ee Mean square value of output = Area under output PSD 2B gy 1 fH aw 20 inp A a N [2 I" a 2nd? | 3 LoaB Problem 7.18 The periodic signal g(t) is passed through a filter with transfer function Hla) fn ignal g(t) i gnal g(7) is i i ; 8 peti shown in figure below. Determine the PSD and RMS value of the input signal. Assume te petiod is 2. [GATE: 2001, JNTU: Nov. 204 a(t) “HO He A : | a -72 0 mm = ae ib Garegnl.” Figure 7.11 Convolution ang Conelaten og as ven figures, We observe, g(t) ig : = Petiodie with 0 al Fourier series, we have ail’ gon! jo" pom : whos" a= LY Reta , a 1 mid Fea | nem a a2 : m2 ti ae Lama aid T=25 2am, 5 Po mga yy Fre 2 a. [, cA pee ee 2jnt 0 Sl> a Therefore, arn= evel 20d ‘fans f 4eea a! re 2a an berg 30 a) = 30 +_ signals and SS nal is pSD of a periodic St i an ¥ Mal 5(@- neo) psD= A Y en 2a 2 y(t) and transfer function, HO) = U1 4 A $a, an input x(f) = € put. x) =e u(t) Problem 7.19 A filter has the energy spectral density of the out Solution: Given 1+jo 1 FP 1+ jo aia (1+ ja)(1— ja)| 1 Hey= —— lH = Therefore, Hoy= l+0° Let x(t) — X(@) Therefore, lt) 2+ jo Therefore, oo (@) = Input ESD 2+ jo Y(@) = IX(@)P 1 Therefore, Now, output ESD, VCO) = HCO? vile) ¥@) = (a) \ 4 1+? ote é lation functi ‘ ff 7 ie autocorrel ion of an aperiodic Th ‘ yi " R(t) = et 20? omer Signal is the normalized aver; elt psp and ‘ABE Power content of the signal, ou 1 r Ria) = eto? {UPSC; tes: 1959) __, Given yan RO — Sw) we S(@)= | R(t) ergy ' 4 =f eR? jor dt S(@) = 2x02 ¢-(oo} 2 je somolzed average Powers P= R() = T unit ral 7.21 Determine the autocorrelation and PSD of the following signal: a(t) =A sin (@t + 9) \oution: Given. x(t) = A sin (apt + 9) 2 a = = = Period = Ty Tt ix) is a periodic signal with period To, the autocorrelation is given by’ 1”. RQ= lim — J x(x +r) dt Ty To yn 1 W2 = lim — J Asin (@pt +9) Asin (opt +r + 9) dt Th ia Ae Be lim — a +2g+ apt)idt fim 2) fost cos (2apt +29 + Wp " 2 Tyl2 2 Tyl2 Jim A cos age at im AQF cos (ant +26 +008) “ty " Ty 2Ty Ty 21 75/2 932. + Signals and Systems 332_+_Sinas oe Simplifying, we get 4, aa A R(t) = = COS MyT Az PSD, S(@) = FLR()]= |& cos a S(@) = #8 st 40y)+ 80 = a4) Therefore, aol Problem 7.22 The autocorrelation of a random process is given by R(z) = eh aa the PSD, . Solution: Given R(t) = eal PSD, S(@) = FIR()] = Jett ci de ° 7 : Jt eI de + fe eit ap ee 0 (2a-joye Pf ,-Qa+ joyr = fee jo | | 2a+jo - | 4 2a- jo 2a+ jo 4g? + @ 4 Therefore, s@= 4 4a + 0 Problem 7.23 Find the relation between input and Output PSDs. of a. differentiation characterized by y= 7 XO, Solution: Given v()=T dy;(t) dt Assume v(t) = ela" Volt) = feat v0) Assume VAD = HCE) Vas) = ail a Lonivotition ang : UN Sata + 999 Nf) = jor ta ef SUD = HCP S\(f) = (oT sf) e =OPS(f) it ‘The energy signal X() =A sine 2Wt is applied t : oe Fa bain aPPFOXIMALeexPESson Tor Rn vey PF with bandwidth 1S : pee Given x(1) = A sine 2Wp ot Pa that, " FLA sine xi] = anf) xk ~Ay(t 2. sf = ine =|“) nf . N= KDE (4) nf Now S\A) = HHCMP = 5,0) form RC ZPF, we have, 1 joa 7 14+ (FB) 2 Therefore, (p= Ae : -n{ 5) 1eqiay QW When WS B, Sf) = SA) afar 2 A R ‘Therefore, RO = i) a Le df = Syne Wr aCe Therefore, Ry(o) = sine 20 Problem 7.25 Verify Parseval’s theorem for the energy signal x(t) = & u(), a > 0. Solution: The energy of a signal x() is given by B= Jimnpar= fet war at . ° em \° = fletPar= fem «|= 0 0 @ ystems have +s theorem, We fing 10 Parseval’s theo! i. seayP », accordil ee : Now, a0 a5 il je 5 a JO" ay 7 tye im de= J eM ult)e 2 fxne FLX) je eo X(@) = ie ~ Tati) dp : jee are fe 0 0 2 1 1 oso] = F —_f[e 0 atjo = Tat jo) * eae cea an io|, 6-18 24 lem arjo)\a- jo) aor a tar a tay 1 Kol = > Therefore, ee +0? Therefore, Lia _(_z\j_1 “male sl 2a i Thus, from Eqs. () and (ii), we see that the energy is same in both cases, Hence, Parseval’s theorem is verified. OBJECTIVE TYPE QUESTIONS '. The advantage of convolution integral in the analysis of linear network is (a) It can be used for all types of functions (b) It gives a direct method to obtain output from input (©) Both (a) and (b) @ None S ~ The output PSD SN) of SA) is Siven by ® SM=HNE sin © SM=IHIS,(H a) @ None a linear network of transfer function Hf) and int PSD Se Ind Correa of Sign a J W) pv ction (b) N func! ever a (@) None Pte faction of CoP unctio® (te axiom function is a measure of rela t se of a random process my spre ence of two random process ca M : y-depe™ w ie a ll “in of the random process he me @ mn variables X and Y have zero mean, th ye Or = rw? @ 100 e* for orthogonality of 1() and x jon a(t) in term: : 6 me: 20 ) Rx) =1 of comelation is “ it?) st @ Rift) = Ue | given bY sp is sive” 1 BP pi pe | UME E= fixe ag jal? (@ None Correlation isa tool for ¢) Signal analysis (b) Signal comparison @ Frequency response (d). None 9, It the cross-correlation between two functions x)(2), x2(¢) vanishes, then they are (a) Conelated (0) Uncorrelated (¢) Non- periodic @ Equal 10. The eo transform of autocorrelation is @ E (b) PSD 2) fe of the (a) or (b) (d) None 11, The auto-correlation is maximum at : (@ t=0 (:) T=° (d) None (¢) Depends on function 12. The autocorrelation at origin is equal to (a) Power (b) Energy (©) Any of the above (a) or’ (b) depending upon the signal @ None 336 * Signals and Systems r even signal convolution and correlation ay re ignals, the convol . For even signals, i (b) Zero () Ba @) Difficult to fing (©) Unequa ¢ autocorrelation function is ce a a function of T e Olt feo iPr: (c) Both (a) and (b) mn = 15, Parsevals theorem for non-periodic signals (nergy signals jg oa @ [Pod= frie ‘ &) | Pod= firorar © [rode Jieuear 3 Q (@ None 16. Parseval’s theorem for energy signals is also called } (a) Parseval’s power theorem P (b) Bessel theorem (c) Rayleigh energy theorem ei (@ Schwarz theorem. 17. The autocorrelation at t= 0 is (@) Maximum (b) Zero (©) (d) Minimum . The autocorrelation and PSD form pair (@) Fourier transform (b) Laplace transform 3 (©) Z-transform (d) Fourier series | @ sampling function is a : . The autocorrelation of (@) Triangular function 12 iy (©) Signum function. | m (©) Square of sampling function (q) Ge function ik 20. Highest value of the autoconelaton of a function 10 sin 10m is @ 50 () 100 io fi 0 (@ depends on time ¢ Answers to Objective Type Questions 1. (b) 2. (@) 3. (by a 1.0 Oe aa ni) SoCo) SA) 4a eo ey in e QO 2 Be 18. (@) 19. @ 20. (a)

You might also like