FOURIER TRANSFORM Periodic signals which extend over the interval (a, «t) can be effectively represented with the help of fourier series. A periodic signals which are strictly time limited can also be represented by fourier series. A time limited signal means it has zero value outside the specified interval. And asymptotically time limited means as time approaches to infinity (), the value of signal becomes zero i¢.x(!) ->0.as|t| ><} Such time limited signals can be more conveniently represented by fourier tnmsform in frequency domain. These signals are aperiodic because their period Ty > : Fourier transform can also be found for periodic signals. It provides effective reversible transformation link between frequency domain and time domain representation of the signal. We have seen previously that for nonpetiodic signals Ty ->o0, As the period of the signal Tj, fo =0. Therefore the spacing between the spectral components. becomes infinitesimal and hence the frequency spectrum appears to be continuous. Whereas periodic signals has fixed period Tp. Therefore their frequency spectrum is discontinvous as we have seen in the examples in the last section DEFINITION Let x(t) be the signal which is function of time t. The fourier transform of x(#) is given Fourier Transform : X (a) or XG) =f x@eri27' dt since o=2xf Similarly x (f) can be recovered from its fourier transform X(f) by using inverse fourier transform, 1 f xe do Inverse Fourier Transform : x(t) = J xielet afThe functions x() andX(f) form a fourier transform. pair is written by a shorthand symbol as shown below, xo X(f) Other shorthand notation for fourier transform is as shown below, X(f) = FEO) and x) = FUR Fourie transform thus ean be considered as a linear operator as shown in Fig, effete ={fips {o) Fourir transformation and (b) Inverse fourier transformation PROPERTIES OF FOURIER TRANSFORM Linearity (Superposition) Let x1 ()X:(f) represent a fourier transform pair and x2 (i) Xa (f) represent another fourier transform pair. Then for all constants like C, andC» we have, GHO+O mM OOM )+C Xl) Proof : By definition of Fourier transform, FE nO+GuO)= f CnWscu@eim at = C1 fm eit! dacs feels dt = OM (+r Xe (f) By definition of FT. Linearity : C111 4 C22 6 CH ()+O% (/)Time Scaling Let x(@) and X(f) be a fourier transform pair and ‘a’ is some constant. Then by time scaling property, a) x(at) Solution : By definition of fourier transform Fix(a)l = f x(at) e-/28f! at Let t= at then at=tas a Here two cases For a Fix(at)] For a F[x(at)) Combining equation x(at) are possible ; a>0 and a <0; > = Sat =; f xigiet GO)” a 1 £ 4 <0 1 £ «(5 above two equations, ) © ax (E a a Thus the time scaling property is proved.Duality or Symmetry Property Duality property of fourier trahsfotm states that if x) XCF) then x® @ xCs} Proof : By definition of inverse| fourier transform in equation zy = f xe a for t'= -t xD = f Xeni af Interchanging t and f we get, x(-f) = J XH eo 2 at =F IX(9) xf) X@) Time Shifting If x() @ X(f), then Time shifting : x(t—to) <> X(f) e-/2"%0 Proof ; By definition of fourier transform J x(@-to) edt f {x(t-to)} Let t-tp =Then b= tptt odtedt Feito} = fey cPrinr? de ~ jaf X By definition of FT- a ectrum but there is a phase ‘Tiuas a time shift “Yp’ has no‘change on the amplitude SP shift of -2 to Frequency Shifting it x@oX(p then Frequency shifting e)?20¢ x( XE-PO Here f: is real constant. This property is also ‘called modulation theorem. Proof : By definition of FT. J craer xe) emt at = J xo omer at F[elfe x(0] XU-f) Maiptication of the function 2(P), by e/%/"! results im shifting. of fourier spectrum X(f) in positive side by fe. Area Under x (t) Tex) @ X(f), then i Area under x(t): J x() dt = X() ‘That is area under x() is equal to its fourier transform rt Proof : By definition of FT, at zero frequency. XU) = J xQeP* at Let f=0 : x@ = J Oat‘Area under X (f) If x(t) @X(f), then Area under X(f0: f X(f) df = x(0) That is the area under fourier spectrum of a signal is equal to its value at t=0 Proof : By definition of IFT, x) = f xen y Let t= 0, x = f xG) wf Differentiation in Time Nomain If x(t) <> X(f) and first derivative Of x(f) is fourier transformable, then ax Oo (20x) -|- Differentiation of function x(t) in time domain is equivalent to multiplying its fourier transform by (j2/). Proof : By definition of FT, [a0] = la x) em dt Integrating by parts, a Fr [é x 0] eP8 LOT, - , x((j2nf)eri de = j2nf . x() cP dt = j2afxySimilarly it can be easily shown that {ERO} - Gaan?xe (a x= XV) Hence {exo = (j2nf)"x~) = Go)"XH) Integration in Time Domain If v(¢) @X(f), and provided that X(0) = 0, then, xn) (canes 1 2nf oe ffect of Assuming that X(Q)=0, the integration of x(t) in time domain has fhe’ @ dividing its fourier transform by (j2 =f). Proof : Let x(t) be expressed as, x= li x@ a We know that, (ther X(f) Fie = [é{ x0 t - ian de x) «} By differentiation property ie m= PY id x) | fi 1 ‘Lx «| = Frag XW)Conjugate Functions If x(t) >X(f), then tor complex valued time function x(t) we h: Wwe have xO OX CS) Proof : By definition of IFT, xt) = J XY) ci af By taking complex conjugates of both sides x(t) = j X? fF) ent af Now by replacing f wit’ - f gives, x= f XC om af Fo [x*(-f)] xt) oX Cf)Multiplication in Time Domain (Multiplication Theorem) Let the two fourier transform pairs be x (!) 4 X(f) and x(t) 4 Xp(f), then ntaltye [x (Xa (fA) da That is multiplication of two signals in time domain is transformed into convolution of their fourier transforms in frequency domain. The short hand notation for this property is nn oX(f)'%if) Proof: Let us write the RHS of equation 2.2.49 as follows, nl)n() o Xn (f) ie Flu) (t)] = Xa(f) = J X(t) x2 ()e/8 dt by definition of FT -0Let 2=f—f", then by arranging above equation, xa) = | eg-a aa J mi (perie IM! at " i Xo f-0 dd j xi (Qe s2eet dt The second integral above is X; (A) from definition of FT Xn (f) = j X2(f-2) dX or = J %@) maya aa x (t) 2) j % (%) Xalf-%) dd, ‘This property is some times called as multiplication theorem. Since convolution is commutative, equation (2.2.50) can also be written as, x (t)x2 (t) @ Xi (fF) *X2 (f)- Convolution in Time Domain (Convolution Theorem) If x(t) 2% (f) and 22 (0) OX (F) then, J a@ n@-9 dt Om (Rf) ‘This property states that convolution of two signals in time domain is transformed in ‘multiplication of their individual fourier transforms in frequency domain. The short hand notation of ati follows, of convolution can be used to represent this property * ie. At) B®) © &f)%(/) Proof : Convolution of x (¢) and x2 (t) as given equation 2.2.52 j is, p ®+n® + Jaw neleaFfuwen to) = it (on tae} a By definition of FT =f n@elte def aQ-nei citi a = Tanctnn [ae Semin Let—r=a in the second integral. SFE @ +O] = fae tae fm (ayeite da From definition of FT applied to RHS F[n(¢)*22(()] = Xi (F) X (Ff) ie fu@ul-od OX () % Find the fourier transform of the decaying exponential as shown in 0 Va Fig. 1: Decaying ExponentialNormally to show time delays in the function and sign of time, use of unit step function u{(0)] is made. The value of unit step function is always unity i.e. uf) = 1 fort> 0 ‘The exponential pulse in Fig. 2.2.2 is represented as, x@Q = eu Here u(t)=1 By definition of fourier transform we have, X(F) f xf) cP?! dt = J ent (the 28F! dt > = f ertrs2ene de 2 ,fort<0. And u(f)=1 for 20 The lower limit is taken °0' since x(!) = 7 1 xf) “sje nyeaaa, - atj2nf Thus the fourier transform pair becomes, Decaying exponential pulse :e-*" u() ¢ 5 To calculate magnitude and phase spectrum : The function X(f) is expressed as, X(f) = ACA +iBY) Here A(f) is real part of X(f) and B(f) is imaginary part of X (f). Therefore magnitude spectrum of X(f) is given as, IXCA)| = VAP) +B) And phase spectrum is given as, = tan BW of) = tan TyConsider the equation 1 *O © aapaay Multiply and divide RHS by a-j2 xf, Sopa a-j2af MG) = Tejanfova=jlaf _-_a-j2af ~ a +(Qnfy? ie a peep = w+Qnfy: / asQnfy a Here real part A() = Pranfye -2nf and imaginary part B() = raheFrom equation magnitude of X(f) will be, 7 +(2nf)?] = ol ese ~ V at+Qnf)? From equation 2.2.18 phase spectrum will be, 2xf /[a? +(2xf)?] a/ [a2 aot IX) = a OY) = tan { a t 7 1 amen’ [2 KOM TEE 8 (renter (224) -_t -tan-t (xt) [iv ear since @=1 since a o | wed oise = tan-t(02 x) =~ 32.14" ore | pene pot = an 0(f )==tan-t (02) =~ 32.4 oste | x gis eg aos 0 (f )==tan-t (n) =~ 7234" 4.0 Hz 4 ‘7 O(f )=- tan" (2x) =- 8095" WON Taae O87 40.0 Hz 0(f )=—tan- (20 x) = - 8908 1 - = 00159 DEOL T+ (20m)?ee KO ae = pase | lf )= tant (02) = 3214" -05 Hz Ole Tee = 09088 O(f )== tan} (- x) = 72.34" ithe PO rege = 0167 0 (1 )==tanr (-2n) = 8095" |, 7 10.00 He ix Ope aE = gorsa | 9(F)=-tan (-20x)= 8908" mo 4 5 of 7 Amplitude spectrum of decaying exponential pulse of Fig. Here a=1 (assumed). It is even function of frequency. a Phase spectrum. It is odd function of frequency.Obtain the Fourier transform of Raising Exponential as shown in the figure Rising x@) = et u(t) Sign of t in u(-8) is negative and it represents negative time of the pulse. It is zero for positive time (t>0). Therefore integration can be performed from -c to 0 instead of = tote. :-Fourier transform of x() becomes XU) =f xmeiest at ° ° =f et une lest de = f eleeni2een at This fourier transform pair for rising exponential pulse_.can be represented by following equation,Px¢h) (a) Amplitude spectrum and ing exponential pulseobtain the Fourier transform of double exponential signal as shown in the figure Solution : The double exponential pulse of above figure can be represented as, x@) = en; fo a1; t=0 a ett 2 t<0 Fourier transform : X(f) = J x(je-PM dé by equation o- o te ; ct! enitpt ats J Lee Jt dt + f enat e-itat gy i. oe + oe i [eee dt = 1 fei at m1 [eo ]=10-1=0 o- é OF in other words integration at single point with upper and lower limits same is zero only xf) = +04 © aajaey avjanh St ame @reaP Equation can be written in short hand as e-‘h(?). [Here when 1<0 ; e-Na(-t) e*ik-1)] Fourier transform pair is represented as shown below, sypnmnetric double Here Ix = and or) = 0joo" < Fig. 2: Magnitude response ey Obtain the Fourier transform of antisymmetric exponential pulse as shown in the figure Solution : This pulse can be very easly represented with the help of equat ‘ « have used in lst problem. Here overall value of the pulse ie negative ior neg a x)= em; t>0 cove tie ell: t=0 os t<0.. Fourier transform will be, X(f) = i x() e-/?*/! dt from equation 7108 oy = J nett en P dts fljeni dts f eet en ian dt oe a-j2ef atjanp janf + (Qafe Signum function can be used to represent equation compactly, A signum function is defined as, sen) = 1; 0 = <1 <0 :-Equation can be written as, x()-=> e-#ltlsgn(t) Fourier transform pair becomes, 4nf ena sen(t) f(t) = e°%! u(t) Using scaling property find the Fourier transform of 1, (f x(ath o Ta * 5; s, Now consider the given example. FT can be obtained directly as 1 XO) = a5yyaaf ap We know that Fle-'] = Tsj2ef Now by scaling property and a=0.5 ¥ 1 1 F [e-05t] = Bs = aL) 1+j2nl og v2 1 ~ 05+j2nf Obtain the fourier transform of rectangular pulse of duration T and amplitude “Aas shown itt Fig: petor, Fig. 3: Rectangular pulse Solution : This réctangular pulse is defined as, T T a(t) = [4 Sat z (0 elsewhere x)= A wea F)FT of x(t) i x) fx) ec at 2 [ - Ac Pe ae ee zaps "Van A jaf ei /T —e-i8fT 2h sin (x fT) By Euler's theorem. [et ~eisT] Sl> S[> . sin(n fT) fT > By rearranging the equation. sin (nx) AT sine (fT) Since sine x = = 1x aT f ar Negative amplitude of this pulse is made positive by phase shift of 180° Amplitude spectrumThe fourier Phase shi of 180" _4goe make Nl postive Phase spectrum spectrum transform pair of sinc and rectangular function is, A reer) «© AT sinc (fT) Obtain the Rectangular pulse <»sinc pulse. fourier transform of the impulse function shown in below. x(t) =3(t) Impulse FunctionSolution : By definition of FT, AY) fo x@ eitere at j BU) e/28St de The sifting property of impulse function is given as, J £© 8¢-t0) = F (to) Here f@) = eit*f and ty =0 c | X(f) = J enitet § (t-0) dt By rearranging equation = en2nf-0 By applying sifting property. z Delta Function : 8() 1 . Fig shows the amplitude spectrum of delta function. It shows that delta function or unit impulse contains all the frequencies with same amplitude in its spectrum.Obtain the fourier transform of the unit step function shown in pip x(t) 3 0 t Solution ; The unit step signal is defined as, ut) = 1 #20 =0 t<0 By definition of FT, X(f) = J x(t) ets! de =f deitst de 0 1 year = pap leeObtain the fourier transform of DC signal shown in Fig. exit) a at) el Here xy) =.1 ie at) XY) By applying duality property of equation 22.39 x) @ af) Here X() = 1 and 8(-/)=8(f) since 8(f) is even function Therefore, | 1 5(/) de signal [x(f)=1] is transformed into delta function yt yency domain for? Above equation states that in frequency domain. Thus de signal and delta function in frequ fourier transform pair.ie, xy = f xWevet at Here if x()=1, then X(f) =6(f) from equation me *» J te lve dt = af) j eit" dt = 5) This is the important relation and gives result of integration of exponential function, If 1 <> 5(t) form a fourier transform pair of de signal in time domain and delta function in frequency domain, thenj cos(2nft) dt = &(f) Solution : By definition of FT, xy) = f mieten ai Now if x() = 1,X(f)=8(f) _ ftom the given condition in example. J eves! dts 8(f) orf estst dt = ag) ef?nft 4 en i2eft cos (2a ft) = SSS ana : el2aft — e-itast sin(2n ft) = zy a From above two equations we have, anft ei 2afl —e-i2 erst +e 005 (2 nf t)—jSin(2nft) = ta ec s2nst Now substitute value of e~/?"/! from above equation in equation J (eos (2 wf t)—j sin (2 ef D] dt = 8) Now since 8(/) is real valued, complex function will have zero value m above expression, f enn = 3) Hence the result is proved.Obtain the fourier transform of y(t) = ei2Rfet Solution : By definition of FT, J x@er ites dt X(f) = J el@nfet e-faeft dt =f eit -fo! de 8 f-f) Find out the fourier transform of cosine wave shown in Fig. petow,= getty — Dm + eTteht x() = case fet) = By Euler's theorem. By definition of FT, Fee evra x J [etter + ed2efet] ees! dt J fereuctory erusiony at : 7 J creer are F eitusiee a 2 fe G=fo+5 +f] from equation 2261 Fy -f +o +f] ‘Thus, Cosine wave : cos (2 Find out the fourier transform of sine function sin(2 fet) shown in Fig. x(t) = sin(2n fet) Solution : Here since V,, = 1; x@ = sin Qnfet) ei2efet — g-i2nfet sin(2mfet) = 3 By Euler's theoremBy definition of FT, X(f) = { x(t) ecs2ft dt = f sin (2m fet) e~J2"f' dt © gjtnfel — e-/2*fet jin ft yj e dt Ee AS i [ed2ey fot = nit +f0"] ae ‘Thus,