CORRECTED SOLUTIONS 10 @ xli}all + Cu 2g Oe Flénree \y \\\ .\ 1.21) 4!) =x(i[ 5lt+2)-0lt-2)) wy Neste ot 4543 ght) -2G@)=-2E gh B)=xL-8)= “oF Tokobo no — FIGUREChapter 1 Answers 1.1. Converting fom polar to Cartesian coordinates: jor te ‘cos(—n) = ib os) + isin) =f a Vie" eal in) 3) =145 Vie =1-j5 1.2. Converting from Cartesian to polar coordinates: B= beh, 20, 1-38 18. (a) Boo = [renee =}, Poo =0,because Bog < 00 fo (0) 22(9 = HD, onl] = 1, Therefore, Bae =f leatPat = at = 00, Poo = ue 2, a ie li fim ah [leo at = in [a= Jim = (6) 23) = cost). Therefore, Bae = J ale [toa = ce, rom fn df cottt= in [" (ESED) a mt 4)” wn, fen}? = (§)P ule}. Therefore, Bas = 3 fel? = (4) mln] = because Bay < 00. $8+8), [agfn|? = 1. Therefore, Boo = oI teal (e) zaln] = x 1 inl? Pam Jim sae Dll = i awreT 2, a (£) zal] = cos(n). Therefore, Bas = s test x 1+ cos( Fn) Poe aw eT a owt Go) = im aye (ae) 1.4. (a) The signal 2[n] is shifted by 3 to the right. The shifted signal will be zero for n <1 and n> 7. (b) The signal {nis shifted by 4 to the left. The shifted signal will be zero for n < ~6 and n> 0.1.5. 1.6. ur. (c) The signal 2{n] is flipped. The flipped signal will be zero for n < —4 and n > 2. (a) The signal 2{n] is flipped and the fipped signal is shifted by 2 to the right, ‘This new signal will be zero for n < —2and n> 4. (e) The signal 2[n] is flipped and the flipped signal is shifted by 2 to the left. This new signal will be zero for n < ~6 and n > 0. (a) 2(1 — #) is obtained by flipping 2(¢) and shifting the flipped signal by 1 to the right, ‘Therefore, 2(1 — t) will be zero for t > -2. (b) From (a), we know that 2(1—t) is zero for t > ~2. Similarly, 2(2—1) is zero for t > —1 ‘Therefore, 2(1 — t) + 2(2 ~ t) will be zero for ¢ > ~2. (c) (34) is obtained by linearly compressing (t) by a factor of 3. Therefore, 2(3t) will be ero for t <1 (a) 2(¢/3) is obtained by linearly stretching 2(¢) by a factor of 3. Therefore, 2(¢/3) will be ero for t <9. (a) 21(¢) is not periodic because itis zero for t <0. (b) 2a{n] = 1 for all n, ‘Therefore, it is periodic with a fundamental period of 1 (c) za[n] is as shown in the Figure S1.6. Figure $1.6 ‘Therefore, it is periodic with a fundamental period of 4. (a) Eofarinl} = Hern) + aif-nl) = $n] ~ ul ~ 4) + ul-n) ‘Therefore, Ev{xr{n]} is zero for {nl > 3. (b) Since 29(t) is an odd signal, Ev{2(t)} is zero for all values of t (c) Evzsnt) = Has) + aan) = HIG) aln — 8) = Ga ‘Therefore, Ev{zs[n]} is zero when |n| <3 and when |n| -+ oo. (d) Fle) +a(-1)) = Evfzald}= y(t 2) — ea(—t+ 2) 1 2 ‘Therefore, Ev{24(t)} is zero only when |t] +a 1.8. (a) Re{er(t)} new (b) Re{z2(t)} 22 cos(0t +) Vicos($) c05(8t + 2x) = cos(3t) = e% cos(3t + 0) new (6) Rezalt)} = et sin(3t + 2) = et cos(3t + §) (d) Ree{za(t)} = ~e7% sin(100¢) = e-% sin(100¢ + n) = e°*cos(100¢ + §) right 1.9. (a) 21(¢) is a periodic complex exponential. y(t) = jel = 0D) oe ‘The fundamental period of z(t) is 3§ = §. ill be (b) z(t) is a complex exponential multiplied by a decaying exponential. Therefore, z(t) is not periodic. ‘ll be (©) es{n] is a periodic signal. a(n] = e'7*" = z2s[n] is a complex exponential with a fundamental period of 2F = 2. (a) saln] is a periodic signal. The fundamental period is given by N = m(s2%) = m(f). By choosing m = 3, we obtain the fundamental period to be 10. (e) zsin} is not periodic. zs[n] is a complex exponential with wy = 3/5. We cannot find ‘any integer m such that m(22) is also an integer. Therefore, zs(n] is not periodic. 1.10. (t) = 2cos(10¢ + 1) ~ sin(4t ~ 1) Period of frst term in RHS = 3 = § Period of second term in RUS "Thorefore, the overall signa is periodic with a period which isthe least common multiple of the periods of the fist and second terms. ‘This is equal to 1. qa. Period of the first term in the RH! Period of the second term in the RHS = m( 2%) = 7 (when m = 2) Period of the third term in the RHS = m( 2%) = 5 (when m = 1) ‘Therefore, the overall signal 2[n] is periodic with a period which is the least common multiple of the periods of the three terms in z[n). This is equal to 35, 1.12, ‘The signal z{n] is as shown in Figure $1.12. z[n} can be obtained by flipping u(n] and then shifting the flipped signal by 3 to the right. Therefore, z[n] = u{~n +3]. This implies that M = —Land m9 = ~3.sro 12s Figure $1.12 1.18. : 0, t<-2 ara = [ cars9—ae-aya={ 1 -22 2 Ba [tas La 1.14. ‘The signal 2(t) and its derivative 9(¢) are shown in Figure $1.14. 3 3 we ‘Therefore, xt) 1 1 ~ appa | 3 J ae . > 2 Figure $1.14 ‘Therefore, SS e243 J se-28—1) 3, and ty = 1. ‘This implies that Ay 1.15, (a) The signal z2[n), which is the input to Sp, is the same as yi[n]- ‘Therefore, vale) = rn —2)+ 5aain—3] = nle-2)+ jul -3) = dayfn—2) + 4nyfn—3] + 5(2naln — 3} + 40i(n — 4) = xy[n— 2] +5y[n— 3) + 2n[n - 4) "The input-output relationship for is fn) = 2z[n — 2] + San — 3] + 22[n — 4)i (b) The input-output relationship does not change if the order in which $; and Sp are connected in series is reversed. We can easily prove this by assuming that; follows Sy. In this case, the signal x[n], which is the input to S;, is the same as yon]. Therefor, nln) = 2xy[n] + 4x[n - 1) = 2afn] +4ynin— 1) 1 2(aafn—2] = 2xy[n—2] +5: alr ~ 3) + A(zaln~ 3] + 5zan— 4) in ~ 3] + 2xa{n— 4] "The input-output relationship for $ is once again ln = 2e{n — 2] + 52{n ~ 3] + 22m — 4) 1.16. (a) The system is not memoryless because y[n] depends on past values of z{n]- (b) The output of the system will be yfn] = é[n}é[n ~ 2] = 0. (c) From the result of part (b), we may conclude that the system output is always zero for inputs of the form d[n— k], k € Z. Therefore, the system is not invertible. 1.17. (a) The system is not causal because the output y(t) at some time may depend on future values of x(). For instance, (—n) = 2(0). (b) Consider two arbitrary inputs 2(t) and za(t). 21(t) + n(t) = 2: (sin(e)) t2(t) — walt) = 22 (sin(t)) Let 29(t) be a linear combination of 2,(t) and z(t). ‘That is, a(t) = ary (t) + baa(t) where a and b are arbitrary scalars. If23(t) is the input to the given system, then the corresponding output y(t) is 2 (sin(t)) ax; (sin(t)) + bare (sin(t)) = anit) + b(t) Therefore, the system is linear. us(t) 1.18, (a) Consider two arbitrary inputs 2i[n] and xan) nine alr] ulnl= > alk)¥ all fn] — voln} ‘That is, where a and b are arbitraty scalars. If x(n} is the input to the given system, then the corresponding output yan] is Let zal] be a linear combination of x3[n] and za[n -za{n] = az[n)} + bee nino win] = > asl ey nine nine = ¥ @alti+emlk) =2 YO milk] +o Yo malt enn a keno = oyile) + byl] ‘Therefore, the system is linear. (b) Consider an arbitrary input 2; [n). Let Y alt mln) be the corresponding output. Consider a second input xan) obtained by shifting =n] in time: zy{n) = 21[n — ra} ‘The output corresponding to this input is winl= SS all= SO alk-ml= Dall toning Bem aes Also note that nenitne uin—mj= > lal. enomi—no ‘Therefore, valn] = wil — mi) ‘This implies that the system is time-invariant, (©) I le{nl| < B, then vin] < (2no + 1)B ‘Therefore, C < (2n9 + 1)B.1.19. (a) (i) the ifn) (b) (@) Consider two arbitrary inputs 2:(t) and 22(t) a(t) lt) = a(t ~1) 22(t) + yn(t) = Paa(t — 1) Let 23(t) be a linear combination of z(t) and z(t). That is, s(t) = axs(t) + boa(t) where a and b are arbitrary scalars. If <3(t) is the input to the given system, then the corresponding output ys(¢) is us(t) Pas(t 1) P(az(t— 1) + bra(t —1)) ay; (t) + bya(t) ‘Therefore, the system is Hinear. Consider an arbitrary input 2;(t). Let w(t) = P(t - 1) be the corresponding output. Consider a second input «9(t) obtained by shifting ,(t) in time: 9(t) = a(t to) ‘The output corresponding to this input is walt) = fxa(t - 1) = #xy(t— 1 to) Also note that a(t ~ to) = (t~to)?xi(t ~ 1 ~ to) # vnlt) ‘Therefore, the system is not time-invariant, Consider two arbitrary inputs z,[n} and z2{n). 23{n—2] ain] — valn] = 23[n — 2] Let 23{n] be a linear combination of x;(n] and za[n]. That is, alr] — mln) as{n] = a23{n) + bala) where a and b are arbitrary scalars. If 23[n] is the input to the given system, then the corresponding output ys{n] is wl] = 23{n-2] (azy[n ~ 2) + beofn — 2)? a2ziin — 2] + Ben — 2] + 2abey[n~ 2}xain — 2] # ays{n| + bysln] ‘Therefore, the system is not linear. 7(i) Consider an arbitrary input 2,[n}. Let ann] = 23[n - 2) be the corresponding output, Consider a second input 2a[n] obtained by shifting y(n] in time: a{n] = eiln — no} "The output corresponding to this input is alr] = 23fn = 2} = aff —2— no) ‘so note that yal — no] = x{[n — 2 — no] ‘Therefore, valn] = viln — mo} ‘This implies that the system is time-invariant. (c) (i) Consider two arbitrary inputs zy{n] and 22[n) fr ain] — wah ziln] + viln] = ain] -2iln— 1] a(n + 1] — xa{n— 1} That is, Let an} be a linear combination of 2,[n] and ean. axg{n] = a2; [n] + bx{n] where a and b are arbitrary scalars. If za[n] is the input to the given system, then the corresponding output ysin] is za{n +1) — asm — 1} axy{n + 1] + bz,{n +1] - axy[n — 1] — bza[n - 1] a(zy[n + 1] — mn - 1]) + (za[n + 1] ~ z2[n - 1) cay;[n) + byaln] ‘Therefore, the system is linear. (Gi) Consider an arbitrary input 2[n]. Let & =z woud al) =eiln+1)-nfr-1] be the corresponding output. Consider a second input 2a[n] obtained by shifting 22y(n] in time: zn] = z[n — no} ‘The output corresponding to this input is = aaln-+ 1} — 29{n— 1) = zifn+1— ng] -2iln— 1-0) wal 8@) @ (ii) then ting Also note that valn — no] = 21h 1 = no] —a1[n — 1 — no] ‘Therefore, vel] = n(n — n9] ‘This implies that the system is time-invariant. Consider two arbitrary inputs x(t) and 22). z(t) + w(t) = Od{x(t)} 22(0) + walt) = Odfen(0)} Let 2a(¢) be a linear combination of 24(t) and z(t). That is, a(t) + bara(t) ‘where a and 6 are arbitrary scalars. If z(t) is the input to the given system, then ‘the corresponding output ys(¢) is us(t) Od{as(t)} Odfaz,(t) + bxa(t)} = aOd{x(t)} + bOd{z2(t)} = ays(t) + byalt) ‘Therefore, the system is linear. Consider an arbitrary input 2(t). Let nit) = Odf2,(9) = HOH be the corresponding output. Consider a second input z(t) obtained by shifting 2,[n] in time: ‘t2(t) = 21(t — to) ‘The output corresponding to this input is w(t) = Odfey(o} = BOSD _ uta ) Also note that y(t — to) ~ 21(-t + to) ~~ 2 wilt —to) = # wit) ‘Therefore, the system is not time-invariant.1.20. (a) Given a(t) = — vit a(t) =F — yt) = Since the te isa, a= Horse) nya Hor re) ‘Therefore, 4(t) = cos(2t) —+ yr(t) = cos(3t) (b) We know that Using the linearity property, we may once again write 1 e-del8 + cbeI8!) = cos(3t — 1) a(t) = FoF tej) ml) = ‘Therefore, -2y(t) = c0s(2(t — 1/2)) —+ sn(t) = cos(3t ~ 1) The signals are sketched in Figure $1.21. “40-0 2 xQt) rt ox(et+) 7 2 3 aor 6 ort i oI + ae oat Figure $1.21 (Can) re sig ae sete in igure $1.22 THB. ‘The even and odd parts are shld in Figure $1.2. 10x{n-4) rn xbr aoe % he |e f TT ae Ter “6 lo fa mle @ ©) @ ” o Ln rt 2) = % ©) Figure $1.23Figure $1.24 1.24, ‘The even and odd parts are sketched in Figure S1.24 1.25. (a) Periodic, period = 2n/(4) = #/2. (b) Periodic, period = 2x/(n) =2. (c) a(t) = [1 + cos(4t ~ 2n/3)}/2. Periodic, period = 2n/(4) = */2. (A) 2(¢) = cos(4nt)/2. Periodic, period = 2x/(4n) = 1/2. (c) x(t) = [sin(4nt)u(t) — sin(4xt)u(—1)}/2. Not periodic. (£) Not periodic. 1.26. (a) Periodic, perio (b) Not periodic. (c) Periodic, period = 8. (a) z{n| = (1/2)fc08(3en/4) + c0s(nn/4)]- Periodic, period = 8. (e) Periodic, period = 16. (2.27, (a) Linear, stable. \—~ (bp) Memoryless, linear, causal, stable. (e) Linear (a) Linear, causal, stable. (e) Time invariant, linear, causal, stable. (®) Linear, stable, (g) Time invariant, linear, causal1,80. (a) Invertible. Inverse system: y(t) = 2(t+4). (b) Non invertible. The signals z(t) and z(t) = 2(t) +2 give the same output. (c) Non invertible. d[n] and 26(n] give the same output. (4) Invertible. Inverse system: y(t) = de(t)/dt. (e) Invertible. Inverse system: y[n] = 2[n + 1] for n > 0 and y[n] = 2{n] for n <0. (f) Non invertible, zn] and ~[n} give the same result. (g) Invertible. Inverse system: yfn] = 2[1 —n]. a 7 f tNis periodic. ie. let 2{n] = g{n] + bln] where (2) False. vii] periodic does no imply 1, neven _{ neven { oO, nodd | 4 Aled = { (Gy, nod “Then y(n] = 2[2n] is periodic but z[n] is clearly not periodic (8) Tue. aft +N} = afnls yal + No] = valrl where No = 27 (4) ‘True. yoln-+ N] = yal; n+ No] = alo] where No = N/2 alr 138) (@) Consider - DY afr] = 210] +) tated + at-nl} If 2{n] is odd, 2[n) +2[-n] = 0. Therefore, the given summation evaluates to zero. (b) Let yfe] = 2i{nlzein). ‘Then y{-n] = 21[-n]zal— ‘This implies that yn] is odd. (c) Consider 5 x ate) + Using the result of part (b), we know that 2¢[ Jel isan od sig, Therefor, wing the result of part (a) we may conclude that a > ze[nja ‘Therefore, 2?[n] == - 5 atin] + = a(n] (A) Consider [lew [2 sous [2 a(tyde + Again, since ¢(t)z0(t) is odd, elthzalthl [Leranae=o. iz (ide = f. * (eat + f a(t)dt 16 ‘Therefore,41.35. We want to find the smallest No such that m(27/N).No = 2rk or No = EN/m, where k is an integer. If No has to be an integer, then V must be a multiple of m/k and m/k must be an integer. This implies that m/& is a divisor of both m and NV. Also, if we want the smallest possible No, then m/k should be the GCD of m and N. Therefore, No = N/ged(rm, NV). 1.86. (a) If z{n] is periodic efo("+")T = eT, where wy = 2n/Tp. This implies that Teh Fy = 78 rational umber 2m, NT = ork = (b) 167/75 = p/q then {n] = &%%0/0), ‘The fundamental peviod is a/ged(p,g) and the fundamental frequency is 2n 2m » oy wo 2h ged(p,a) = =P acd(p,a) = Pged(p, 4) = “geal eed, a) = Fecdlnna) = Frsedla) = “sed P (©) p/ged(p.a) periods of 2(t) are needed. 1,87. (a) From the definition of day(¢), we have dat) =f 200+ s)uiner [lvcerneteidr oya(-t). (b) Note from part (a) that ¢co(0) = dze(—f)- This implies that g-o(t) is even. Therefore, the odd part of dzz(t) is zero. (c) Here, doy(t) = dze(t — 7) and dyy(t) = dz2(t). (£38) Wes a 2609 69, Tans 1 Jim da(2t) = im, s8ayalt) ‘This implies that ' S(at) = 582) (b) The plots are as shown in Figure S1.38. 1.89, We have Fim vat) = Fim ua (0)6(t) = 0. Also, 1 Jim wa(t)5a(6) = A(t)Figure $1.38 We have ce - oft) -[ u(r)6(¢ = 1) f u(e)é(t — x)ar. ‘Therefore, , <0 at-7) | w-{ 1 1>0 -ulz)a(¢ ~ 7) = 6(¢~7) undefined for t= 0 1.40, (a) Ifa system is additive, then 0 = a(t) — 2(t) — v(t) — u(t) = 0. Also, if a system is homogeneous, then 0=0.2(t) > y(t).0=0. (b) y(é) = 2?(t) is such a system. (c) No. For example consider )= x(r)dr with 2(¢) = u(t)—u(t—1). Then 2(2) = 0 for t> 1, but y(t) =1fort>1.(g) Since r) > O,ra > 0 and ~1 < c0s(61 ~ 63) <1, (al-lal? = sf+r3-2nre FB-+ 1 + aria coa( ~ 0) ar +220? wad and (lea + lea)? = Ff +08 + rare 2 [an + a2P 1.54, (a) For a= 1, itis fairly obvious that For a # 1, we may write a-a)S @ = ‘Therefore, | (b) For jal <1, im aN = Wit ‘Therefore, from the result of the previous part, ete ool 1 (a) 1 (d) We may write 1.55. (a) The desired sum is, PaChapter 2 Answers 2.1. (a) We know that (821-1) ‘The signals «{n] and h{n] are as shown in Figure S2.1. pF a) { ie Figure $2.1 From this figure, we can easily see that the above convolution sum reduces to h{-Yz{n + 1] + hfd}2{n — 1] 2e[n + 1] + 22[n - 1] maln] This gives 25 + 1] + 4é{n] + 26[n — 1] + 26[n ~ 2] — 24[n — 4] nl (b) We know that DY dlklein + 2-4) nl +2] + h[n] Comparing with eq. (S2.1-1), we see that nln + 2} (c) We may rewrite eq. (S2.1-1) as DY zlklaln —k] k Similarly, we may write usin) Comparing this with eq. ($2.1), we see that usln] = viln + 2] 302.5. The signal y(n] is ln] = [rn] « A In this case, this summation reduces to 9 a vin] = So allhin - k] = alm — &] 0 From this it is clear that y{n] is a summation of shifted replicas of h[n]. Since the last replica will begin at n = 9 and An] is zero for n > N, y[n] is zero for n > N +9. Using this and the fact that y[14] = 0, we may conclude that V can at most be 4. Furthermore, since y[4] = 5, we can conclude that hin] has at least 5 non-zero points. The only value of N which satifies both these conditions is 4. 2.6. From the given information, we have: D alelain - 4] = > Gtalk — te vlr x hln] k-1] = DG ytun +k-1] r= Replacing k by p—1, (26-1)‘| [| qi gna | | | ala) ote 234s ase Figure $2.7 2.8. Using the convolution integral, a(t) » A(t) f 7 Given that h(t) = 6(¢ +2) +26(¢ + 1), the above integral reduces to a(t) +y(t) = x(t +2) +22(¢+1) ‘The signals z(t +2) and 2e(t + 1) are plotted in Figure $2.8. 2 os x42) ' o let) é 0 . ae 82.8 Using these plots, we can easily show that a(r)h(t —r)dr = ie A(r)a(t — r)dr. t+3, —2t-4 0, (t-5) <7r<(t-4) 34‘Therefore, A=t-5, B 2.10. From the given information, we may sketch z(t) and h(t) as shown in Figure $2.10. (a) With the aid of the plots in Figure $2.10, we can show that y(t) = 2(¢) « A(t) is as shown in Figure $2.10. ze) * * he) 0 t + ow + ow 1 te t Figure $2.10 Therefore, t O 5, the integral is 6)_~8t-5) 3Therefore, the result of this convolution may be expressed as 0, oo UhfAll = 0 Al cos(F8)| a co = ‘This sum does not have a finite value because the function k| cos($)| increases as the value of k increases. Therefore, f(r] cannot be the impulse response of a stable LTT system. (b) We determine if h cos(2t)\dr is absolutely summable as follows > [hall = 3 gle 00 Therefore, ha[n] is the impulse response of a stable LTT system. 2.16. (a) True. This may be easily argued by noting that convolution may be viewed as the process of carrying out the superposition of a number of echos of h{n]. The first such echo will occur at the location of the first non zero sample of z{n]. In this case, the first echo will occur at Nj. The echo of h{n] which occurs at n = N; will have its first non zero sample at the time location N; + Nz. Therefore, for all values of n which are lesser that Nj + Np, the output y(n] is zero. (b) False. Consider ln] = fn) + hf = DY akain-4 From this, yin-1) = > aff}h{n — 1-8) b=00 = 2{n]* hin —1) ‘This shows that the given statement is false. 37(c) True. Consider u(t) = ato ene) = [a(n nar From this, v(t) = [xem =n)dr [ * a(—n)h(-t-+r)dr a(—t) #h(—t) ‘This shows that the given statement is true. (d) True. This may be argued by considering * a(a)h(t — r)dr. u(t) = 2(t) «h(e) In Figure $2.16, we plot z(r) and h(t — r) under the assumptions that (1) x(t) = 0 for t > Ty and (2) h(t) = 0 for t > Ty. Clearly, the product x(r)h(t — r) is zero if a(t) h(-T) T % 4-T, Tv Figure $2.16 t-_>T. Therefore, y(t) = 0 fort > Ty +7. 2.17. (a) We know that y(t) is the sum of the particular and homogeneous solutions to the given differential equation, We first determine the particular solution yp(t) by using the method specified in Example 2.14. Since we are given that the input is <(t) = e(-1+3))tu(t) for t > 0, we hypothesize that for t > 0 p(t) = Kei, Substituting for z(t) and y() in the given differential equation, (<1 43)) Kel H3 4 ac eH80 = (1481 38This gives (-14+3)K+4K=1, 3K ‘Therefore, = (49mg wl = sae 1 -t>0 In order to determine the homogeneous solution, we hypothesize that : walt) = Ae"2.18. Since the system is causal, y[n] = 0 for n < 1. Now, vil) = 4ylo}+2ft]=04+1= wal = git pte vi] = jubl+aisi=t+0-4 yaa Therefore, a ie) = yt 3 2.19. (a) Consider the difference equation relating y(n] and w(n] for S: ln] = ayln — 1} + Buln) From this we may write and = jul 1) ~Guln — 2) ‘Weighting the previous equation by 1/2 and subtracting from the one before, we obtain gure -y- wen ~ty ypu Substituting this in the difference equation relating w[n] and z{n] for 1, dy) — Sun — 1) - avin — 11+ Svln -2 B 26 26° ‘That is, 1 ale = (a+ 5)uln— 3] ~ Sule — 2] + Ban] Comparing with the given equation relating y[n] and z[n], we obtain B=1a) sO! = f uui(1— 1) cos(2er)dr which is the desired integral. We now evaluate the value of the integral as dealt) = sin(2nt)|,21 = 0. @ (ios sin(2rt)|a1 2.21. (a) The desired convolution is ale] = alr hin) = DY alélhfn-4] k5=00 = Bao" forn>0 - att! = (FS hi orate (b) From (a), | ufn] = (n+ 1)o"uln. (c) For n <6, ofS dye She utr] = 4") Sg) gp = = For n > 6, in Sh yk S ke un] = 4") o- ah Ogle fo eo ‘Therefore, _ J /M-1/ayar, nse vel={ iaycugye noe (d) The desired convolution is vin] = SO afklain- A] ey vot An] + h{n — 1] + hf — 2] + Alm — 3] + Afn — 4] a[0}hn] + x[t}hfn — 1] + 2[2]hfn — 2] + 2[3]h[n — 3] + 2[4]Aln — 4)(c) The desired convolution is ut) = [stone nar 5 f sin(wr)h(t — r)dr. lo ‘This gives us 0, t<1 (y=) C/m[-cos(ntt— Dj, 1 -2 yinJ= 9 1, n=-3 0, otherwise ‘Therefore, y{n] (c) We have 1/2yuln +3). waln] -aln] + xa| ufn +3] — un + 2] = d[n +3}. (d) From the result of part (c), we get vuln} afr + 3] = (1/2)"*4ufn + 3). valn] + 21] The proof is as follows. 4, = fwd = [i f.xone-nerae = [2 [= raed = a 2(r)Andr AzAn " (a) Causal because h[n] = 0 for n <0. Stable because }>(})" nao 5/4 < 00. (b) Not causal because h{n] # 0 for n <0. Stable because } > (0.8)" =5 <0. (c) Anti-causal because hn] ° 0 for n> 0. Unstable because > (1/2)" = 00 a (€) Not causal because h{n} # 0 for n <0. Stable because > 5" = 7 < 00 =0 for n <0. Unstable because the second term becomes infinite (e) Causal because hf as n+ 00. (£) Not causal because h{n] # 0 for n <0. Stable because > |A{n]| = 305/3 < oo a7(g) Causal because h{n] = 0 for n <0. Stable because > |h{n]| = 1 < ov. 2.29. (a) Causal because h(t) =0 for t <0. Stable because f[ In(Q)|dt = e°8/4 < 00. (b) Not cansal because A(t) # 0 for t <0. Unstable because [ a(t)| = 00. (c) Not causal because A(t) # 0 for t < 0. a Stable because f [A(t)|dt = €19/2 < 00. (A) Not causal because A(t) £0 for t <0. Stable because f[ [n(t)|dt = €2/2 < 00. (e) Not cansal because h(t) #0 for t <0. Stable because i Ih()|at = 1/3 < 00. (E£) Causal because h(t) = 0 for t < 0. Stable because f |a(t)|dt = 1 <0. (g) Causal because h(:) = 0 for t < 0. Unstable because [ JA(2)|at = 00. 2.30. We need to find the output of the system when the input is | to assume initial rest, we may conclude that y[n] = 0 for n < [rn]. Since we are asked ). Now, fn] = 2[n] - 2y{n — 1). ‘Therefore, xy(0] = 2{0] — 2u{-1] > ull] = [2] — 2y[0) and so on. In closed form, vuln] = (-2)"ul ‘This is the impulse response of the system. 2.31. Initial rest implies that y(n] =0 for n < -2. Now ln) = 2{n] + 22[n — 2} — 2y[n — 1] ‘Therefore, w-2] = 1, uf-l) ld] = 56,y{5] = -110, yn] = -110(-2)""* forn > 5. 2.32. (a) If ya{n] = A(1/2)", then we need to verify Clearly this is true. 48(b) We now require that for n > 0 ( ‘Therefore, B= —2. (c) From eq. (P2.32-1), we know that y{0] = z{0] + (1/2)y{-1] = x{0] = 1. Now we also have yOJ=A+B 4 A=1-B=3. 2.33, (a) (i) From Example 2.14, we know that n= [5 (ii) We solve this along the lines of Example 2.14. First assume that yp(t) is of the form Ke* for t > 0. Then using eq. (P2.33-1), we get for t > 0 2» Kal, 2Ke™ + 2K 7 We now know that yp(t) = }e% for t > 0. We may hypothesize the homogeneous solution to be of the form — sult2.37. Lot us consider two inputs n(t)=0, for allt and a(t) = effu(t) — u(t - 1)} Since the system is linear, the response yi(t) = 0 for all t. Now let us find the output yo(¢) when the input is #2(t). ‘The particular solution is of the form yplt)=Ye', O 1, the only unknown Fourier series coefficients are a; and a_;. Using Parseval’s relation3.10, Since the Fourier series coeffiecients repeat every IV, we have a, = 415, a= 16 and a3 = 7 Furthermore; since the signal is real and odd, the Fourier series coefficients ay will be purely imaginary and odd. Therefore, ay = 0 and ay =a, m=-a-2 a =-O-5 Finally, 4.11, Since the Fourier series coefficients repeat every V = 10, we have a; = ani = 5. Further- more, since a(n} is real and even, ay is also real and even. Therefore, ay = a-1 = 5. We are also given that ® 5 Deb? LO la{nll? = 50. 10 ‘Using Parseval’s relation, SX lee? = 50 wich ‘ ¥ lai? = 50 1 * Jal? + la? +05 + Sola? = 50 = 8 B+ dolar = 0 i ‘Therefore, ay = 0 for k= 2,--- ,8. Now using the synthesis eq.(3.94), we have 8 i= > ayelvkn Dd weit ase> é = belli" 4 5S" = 10c08(En) 8.12. Using the multiplication property (see Table 3.2), we have ESS by + Dba + Wn + Des 893.13. 3.14. Since by is 1 for all values of k, it is clear that by + 2b.-1 + 264s + 20,3 will be 6 for all values of k. Therefore, , for all k. Let us first evaluate the Fourier series coefficients of z(t). Clearly, since z(t) is real and odd, ax is purely imaginary and odd. Therefore, ay = 0. Now, oc 5 f° -(t)erI2H/ tgp ve = Effet 8Jo er HORI) Ge yuo Clearly, the above expression evaluates to zero for all even values of k. Therefore, _ { 0, k a={4 if When (2) is passed through an LTI system with frequency response H(jw), the output y(t) is given by (see Section 3.8) ult) = So ag jkup)elrt kere where wy = 3 = §. Since a, is non zero only for odd values of k, we need to evaluate the above summation only for odd k. Furthermore, note that opr _ sin(kn) H (shave) = HRI 4)) = Foray is always zero for odd values of k. Therefore, u(t) is periodic with period N = 4. Its Fourier series coefficients are 0. ‘The signal z| jee = 5, forallk From the results presented in Section 3.8, we know that the output y[n] is given by 3 Sau (cir!) itn = TH e0 + LAH TH (SOA CIEE) + LT(eH)) 3 (s3.14-1) 90From the given information, we know that yin] is Ca yin) = cos(en+ F , = coaline® = cos(fn +9) eet a ilies 5 it uartl 50 +5¢ = Latntn 4 LetieD = 30 See Comparing this with eq, ($3-14-1), we have H(e®) = H(e) =0 and H(é#) =2e%, and H(e4) = 208% 3.15. From the results of Section 3.8, ut) = oH Ghul where wo = %f = 12. Since H(jw) is zero for || > 100, the largest value of |k| for which ay is nonzero should be such that [lwo < 100 ‘This implies that |k| <8. Therefore, for |k| > 8, ax is guaranteed to be zero. 8.16. (a) The given signal 21[n] is aife] = (yh = = er ‘Therefore, 2{n] is periodic with period N = 2 and it's Fourier series coefficients in the range 0 ellm/2)t§(t — 2k). 100 3.24. (a) We have =f vats [@—oae= 1 (b) The signal g(t) = dz(t) at is as shown in Figure $8. 24. a Figure $3.24 ‘The FS coefficients by of g(t) may be found as follows: it if. =i fa-5faro and be = aekdt ert = zat eo, (c) Note that a(t) = 20 ES oy = fron ‘Therefore, 3.25, (a) The nonzero PS coefficients of z(t) ae =P) =1/2 (b) The nonzero FS coeficents of =() are hay 7(©) Using the multiplication property, we know that (ul) es 2) Therefore, 1 jl ee = a4 be = lk 2) — FrOlk +2}. This implies that the nonzero Fourier series coefficients of z(t) are co (4) We have 2(t) ‘Therefore, the nonzero Fourier series coefficients of z(t) are e = c-2 = (1/44). in(At) cos(4t) = Fin) 3.28. (a) If x(¢) is real, then 2(t) = =*(t). This implies that for x(t) real ax not true in this case problem, z(t) is not real. (b) If 2(¢) is even, then 2(t) = (1) and ay = ap. Since this is true for this case, x(t) is 1» Since this is even. (c) We have _ Galt) Fs, 3p 2 a(¢) = Se = jk oe Therefore, nal k=0 k= —k(1/2)!*\(2n/To), otherwise Since by is not even, g(t) is not even. 3.27. Using the Fourier series synthesis eq. (3.38), afr] = ap + age OXINOM 4g g¢°H2IR/NI 4 gyeiA(2HININ 4 g_ gem H2H/¥In 2+ 2el*/6eH(45/5)" 4. ne F8/5Q-HAR/SIN 4 git/Sei(BR/5)" 4 -IR/S—-a18H/5)N ast(c) Here, T = 1, wy = 2m and 1/2, k=0 a, = 0, keven,k #0 Sale) odd Therefore, the FS coefficients of the output are 1/4, k=0 by = apH (ku) = ¢ 0: Keven, k #0 site sce tee + he], odd 3.35. We know that the Fourier series coefficient of y(t) are b, = H(jkwo)ay, where wy is the fundamental frequency of x(t) and ag are the FS coefficients of z(t). If y(t) is identical to 2(t), then by = a, for all k. Noting that H(jw) = 0 for |u| > 250, we know that H(jkw) = 0 for |k| > 18 (because wp = 14). Therefore, a must be zero for [el > 18. 3.36. We will first, evaluate the freamency recnanse af the evetam Caneidar an innit lanl af tha8.37. The frequency response of the system may be easily shown to be Se Joe 1-2 H(e*) (a) The Fourier series coofficients of z{n} are 1 or= forall k. ‘Also, N = 4. Therefore, the Fourier series coefficients of y{n] are bg = aptt eI") =F | (b) In this case, the Fourier series coefficients of z{n] are 1 ag = z[1 +2cos(kn/3)], for alk. ‘Also, N= 6. Therefore, the Fourier series coefficients of y[n] are by = aH (oo fi 1 gut 2c08(kx/3)} =e Teor 8.38. The frequency response of the system may be evaluated as H(e¥) = 8 — i 1 +e, For a[n], N= 4 and wo /2. The FS coeliicients of the input z[n} are 1 ay = 3) forall. r se conte af tha nntnnt are(a) 2(t — te) is also periodic with period T. The Fourier series coefficients by of x(t — to) are ba = Bf alt ~ taper nOrae TIr _eé eace [atnenerrar Ir =e HHr/Moq, Similarly, the Fourier series coefficients of x(t + to) are oy = OH/T0gy, Finally, the Fourier series coefficients of x(t ~ to) +.2(t + to) are ty = by op = CH P/M0gg 4 eHHOAIT0gy = 2e08(KIato/T) ae. (b) Note that Eo(2(t)} [x(t) +2x(-#)]/2. The FS coefficients of 2(—t) are = bf (tester, b= R fi: a(-t)e dt eo Mae [Tyr =; HE (re dr os ‘Therefore, the FS coefficients of Ev{x(t)} are Oe tbe OK+ Ow 2 2° (c) Note that Re{z(t)} = [o(t) + 2*(t)]/2. The FS coefficients of 2*(¢) are 1 #4) SEORIT I fe o . oe be Conjugating both sides, we get OE KOn/T tgp = z fi: (te! ‘dt = 0g ‘Therefore, the FS coeflicients of Re{a(t)} are e+ be _ Oe tak, ar rr) (a) The Fourier series synthesis equation gives a(t) = D> ager, 104Differentiating both sides wrt t twice, we get a e 2 Bol) Fearn By inspection, we know that the Fourier series coefficients of d?x(t)/d!? are —b4fr-a.. (c) The period of 2(3%) is a third of the period of z(t). Therefore, the signal «(3t ~ 1) is periodic with period 7/3. The Fourier series coefficents of 2(3t) are still a,. Using the analysis of part (a), we know that the Fourier series coefficients of 2(3t — 1) is eo HOTT gy, 3.41. Since ay = a_p, we require that x(t) = 2(~t). Also, note that since a, = a,42, we require that a(t) = (eH, ‘This in turn implies that (t) may have nonzero values only for t = 0, 41.5, +3, 4.5, 1, we may conclude that 2(t) = 6(t) for -0.5 < t < 0.5. Also, since ‘2(t)dt = 2, we may conclude that x(t) = 26(¢ ~ 3/2) in the range 0.5 < t < 3/2. os ‘Therefore, x(t) may be written as a(t) > (t—h3) +2 7 6(t — 3k — 3/2) 3.42. (a) From Problem 3.40 (and Table 3.1), we know that FS coefficients of x*(t) are a”. aN AEN me at Nato thotFigure $3.43 (A) (1) Ha; or a is nonzero, then a(t) = aset9"/7 4. and a(t + to) = ane FH) +. ‘The smallest value of |to| (other than |to| = 0 for which ti = 1 is the funda- mental period. Only then is a(t + to) = asre®"? + -- ‘Therefore, to has to be the fundamental period. (2) The period of 2(t) is the least common multiple of the periods of ih2/T and eil@x/T\t, ‘The period of ef*(2*/7) is T/k and the period of eM"*/7" and T/1. Since k andl have no common factors, the least common multiple of T/k and T/! is T. and = a(t). Since 2(t) is real, a1 8.44, ‘The only unknown FS coefficients are a1, 0-1, a2, and a; . Since ay is real, a, = a1. Now, z(t) is of the form x(t) a; = 1 cos(wot) + Aa cos(2uot + 8), we.3.47, Considering 2(t) to be periodic with period 1, the nonzero FS coefficients of z(t) are ay = a_y = 1/2. If we now consider 2(t) to be periodic with period 3, then the the nonzero FS coefficients of x(t) are by = b_3 = 1/2 3.48. (a) ‘The FS coefficients of «{n — no] are wa 1 ae = Dale —nolevnel™ = wot Saige enh n=O =e (b) Using the results of part (a), the FS coefficients of 2{n} — 2{n ~ 1] are given by ay = 0p — €I** ag = [1 — eH Ma, (c) Using the results of part (a), the FS coefficients of 7 0, k & =a,[l-€ = {o ead . (A) Note that z{n]-+2[n+NN/2] has a period of N/2. The FS coefficients of z{n]-+2[n—N/2] are given by ~ z{n — N/2] are given by ea a = 25 N, ay xy [aidan i for 0 {ne P* N N = 0, for k even. a ete speach similar to part (a). we may show thatFigure $3.62 (c) The de component of the input is 0. The de component of the output is 2/n. 3.68. The average energy per period is i 2 ee 1a] __ La? 7 e(0Pat = Sout = Soot = TA Ta We want Nv such that wa 2_pgite D lee? = 0933 “wh This implies that 1-20?" +20? _ +a? 1-a? Solving, we bal 5a? + = Blogga” — N (w -1): a =1)n TG 3.64. (a) Due to linearity, we have ult) = Sree Acdu(t. r (b) Let z(t) > lt) and a(t) — volt). Also, let a(t) = ax (t) + bro(t) — a(t). ‘Then, P[axy"(t) + bx2"(t)] + taxy (t) + bx2'(t)] = ay(t) + by2(t) a(t) 122I 4. pericclic, cn TA 2 2 lt] + ye) yl= 1 (Ck) £- pe) ” (dk)sr xtt) — (Gx) T26 = urs = \e) \ we “Tt ae & O-ztyR =H 2 a) ] () | @ |” CT e Hl + z oI | a * yi (OQ is He perodic Spore wave giver in table 4, with. & chifld oy & = % LB b=-S (ite=-1) a for Yr 2), with Th = he fooner cect cuenks olerved in bk | Y2,) & aj Tel a Rib, PS - ne OE & dee oe e Dosen. on Ue) cet oe Z 50. sin KB) doy won eet Oe =e Hi | 7 sin (Ar) sin (4°)(0) Te 3 = r= AL O- tl, S=4 3 le) yh) “@ (ce) (m=) si (ue) + Using, formate, for fooner coefGcints of periodic Squne wave given in fade 2 & shithing prpety in tabh 3.1. (agile «yet : _ 3a Cn= Sin ( BL] eS & deeeZsin 1) e ve AT kar bee Cet de (Urvanty Property) be, eC Tategtetn property) Ok = OR = SR Jj Kwe JAE on jar om X= —3i [sin Zale) oe _2snt™e | 2a, far ar vat jee = 2 [2éin (te? . sin (2 )e” > zakT= 2 => Wo=fT qet sl 6U)-24(41)] o Z2 Sy. a” Nip xt) = Al) 2slE-1) (SE 42) | Using fourier coetfiventa for SL) given in fable UL Bw snitting propery 1 we get : doe t -2ebet. 4 Bee eee onl ae T Zz uw) 2 7 al. 7 xy pe egvala yt) = ot yee => Ase De (park &)) = Bote)(a) Te3 > We = Zr \P 1 a OH = ty3=4 Ta Ss ne nN = T (ee) (Ce) a 4 Pr aaa ee a | (=v red ole) & ) Gan be obbinud using fourier coeficunt for pericdic Salam wave gn in tabi Y\2 & Shithing Property in tebe O41. bn = sin (24) eo HE kar © Ke sin ES) es Kr Oc = Det Ce (Lintanity propeily ) = sin 78) SF sin) @ OS AIT AMYChapter 4 Answers 43, (a) Let a(t) =e-%-Yu(t— 1), Then the Fourier transform X (ju) of 24) is X(ju) = [letenae- nema = Pretvesra eH /(2-+50) |X (jw)| is as shown in Figure $4.1. (b) Let 2(t) = el. Then the Fourier transform X (js) of x(t) is X(je) = £ ettledetge 34 /(2 + ju) +e /(2 - jw) sen (4+ 0?) |X Gw)| is as shown in Figure $4.1. Ixiest ixGun 2. 3 a 3 a ad Figure $4.1 » 4.2. (a) Let 21(t) = 6(t +1) + 6(¢- 1). Then the Fourier transform X1(jw) of 2(¢) is Xan) = [ ace + 1) + 6(t ~ yje "at = & +e =2e08w |X; (ju)| is as sketched in Figure $4.2. (b) The signal x(t) = w(-2 — #) + u(t ~2) is as shown in the figure below. Clearly, Stu(-2 = 1) + u(t -2)} = 6(¢ 2) — 6 +2) 129‘Therefore, Xaljw) = [ * file 2) — 6(¢ + 2)]e-Hat = ete = 25 sin(20) |X (ju)] is as sketched in Figure $4.2. =, SH, IT Figure $4.2 “3 Me 4.3. (a) The signal 2:(t) = sin(2nt + */4) is periodic with a fundamental period of T = 1. ‘This translates to a fundamental frequency of wy = 2x. The nonzero Fourier series coefficients of this signal may be found by writing it in the form j 0 = 3 ) 3 3° | a ‘Therefore, the nonzero Fourier series coefficients of 2;(t) are 1 sgt i _. 1 minis gy SSS——————————_. 45. ‘Therefore, the nonzero Fourier series coefficients of a(t) are 1 n/t pat, = sere, a ‘From Section 4.2, we know that for periodic signals, the Fourier transform consists of fa train of impulses occurring at ku. Furthermore, the area under each impulte is 2 times the Fourier series coefficient aj. Therefore, for z(t), the corresponding Fourier transform X2(jw) is given by Xalju) = 2ragé(u) + 2xay6lw ~ up) + 2xa-16(w + wo) = niles) + 1o*!*5(0 — Gr) + ne F*/5(w + 6x) (a) The inverse Fourier transform is a(t) = (0/2) f° tand(u) + nblw — An) + x8lu + 4rd 1/2n)faned + nett + xe") = 14 (1/2) + (1/2)e9 = 1+ cos(4zt) (b) The inverse Fourier transform is nit) = 4/2m) [” Xaljn)etdo = (1/29) [ze ann [’ {aetae (o%— lait) — LPN Ui8) — (4s sin? 1)/(r8) From the given information, a(t) = (1/2) f © xGu)eaa " (apn) ff “Tx GueAor oman (jen) f ° ye terres 2 = EH 77 in{3(t — 3/2)] ‘The signal 2() is zero when 3(t — 3/2) is a nonzero integer multiple of x. ‘This gives for KET, and k #0.ee ‘Therefore, the desired result is FT (Odd part ofe(t)) = S2g - 2 4.10. (a) We know from Table 4.2 that SHE &T, Rectangular fonction ¥ (ju) [See Figure $4.10] "Therefore cay £%, (1/2n) (Rectangular function Y (jw) + Rectangular function ¥ (i) ‘This is a triangular function ¥; (jw) as shown in the Figure $4.10. Geo Lilie) Figure Using ‘Table 4.1, we may write sint\? er, ad «(SREY EB xia) = agave) ‘his is as shown in the figure above. X (jw) may be expressed mathematically as jim — -2S0<0 X(Gw) =) —j/2m, OS w2 0. otherwise (b) Using Parseval’s relation, CeGiyerel IXGu) Pao = 5 13541. 4.12. We know that 20) B IxG2), nay &S EHG2) ‘Therefor, Glin) FT (x(3%) « h(3t)} = Now note that Y (jus) = FT (z(t) + h(t)} = X (jue) HG) From this, we may write vo =x69) (0) Using this in eq. (**), we have and att) = Fula0) Therefore, A= } and B (a) From Example 4.2 we know that ewes 2 Tea Using the differentiation in frequency property, we have vn £8, jf 2 Aju bed do iF wy (b) The duality property states that if aft) Gia) then 7 Gt) ES 2xg(ju) Now, since 7 o-¥ &8, Siw -— ‘we may use duality to write 4jt_ rr, 4 — ge Tare t Multiplying both sides by j, we obtain i wen ll Trap A steve 136— ne 4.18. (a) Taking the inverse Fourier transform of X(jw), we obtain A dey Le Hap t ett ge “The signal 2() is therefore a constant summed with two complex exponentials whose fundamental frequencies are 21/8 rad/sec and 2 rad/sec. These two complex expo- nentials are not harmonically related. ‘That is, the fundamental frequencies of these complex conaplex exponentials can never be integral multiples of a common fundamen tal frequency. Therefore, the signal is not periodic. (b) Consider the signal y(t) = 2(t) + h(t). From the convolution property, we know that ¥ (iw) = X(ju)H Gu). Also, from h(t), we know that wy 2imes i “he function H (ju) is zero when w = kx, where k is a nonzero integer. ‘Therefore, ¥ (ju) = X Gu) H Gu) = Sle) + lw -8) Aju) ‘This gives ‘ 1 ade det WO = Fe Be ‘Therefore, y(t) is a complex exponential summed with a constant. We know that a _complex exponential is periodic, Adding a constant to 9 complex exponential does nos fost its periodicity. Therefore, y(t) wil be a signal with a fundamental frequency of 2/5. (c) From the results of parts (a) and (b), we see that the answer is yes. 4.14, Taking the Fourier transform of both sides of the equation FAU(1 + ju)X (juo)} = AZ*ult), ve obtain a | XU) = TFG) at ‘aking the inverse Fourier transform ofthe above equation a(t) = Aetu(t) — Aem*u(t) Using Parseval’s relation, we have | [LixGartay = 2 [icra ‘Using the fact that [ 1X (ju) du = 2x, we have [Leora 137Substituting the previously obtained expression for <(t) in the above equation, we have [ [ater + Aen! — 2AM = 1 [tet ate atte tide = 1 A A/2=1 = A=VvB We choose A to be VIZ instead of V2 because we know that 2(¢) is non negative 4.18, Since 2(t) is real, z(t) + ra =) er, En{a(t)} = Ref X(ju)} We are given that TFT (Re{X(ju)}} = [tle ‘Therefore, evfa(p} = +4 ‘We also know that 2(t) = 0 for t < 0. This implies that <(—t) is zero for ¢ > 0. We may conclude that = Iter. a(t) =e" fort > 0 ‘Therefore, 2(0) = 2te-tu(t) 4.16, (a) We may write a) = Yo a — bn A) = at > ‘xilt — kn/4) Therefore, oft x me ~kx/4), | | | its Fourier transform G(iw) is also an impulse train,= a Le We see that G(ju) is periodic with a period of 8. Using the multiplication property we low that ae abr} If we denote FT {884} by A(jw), then X(ju) X(jw) = (1/2m)[AGw) +82 J) 4 (wo - 8k) aes “x (ju) may thus be viewed asa replication of 4A(ju) every 8 rad/sec. This s obviously periodic. Using Table 4.2, we obtain Alia) te lw] <1 0, otherwise ‘Therefore, we may specify X ju) over one period as te fol st X=, Ics 4.17, (a) From Table 41, we know that a real and odd signal signal s(t) has a purely imaginary sey odd Fourier transform X(ju). Let us now consider the purely imaginary and gad signal j2(2). Using linearity, we obtain the Fourier transform of this signs) to te jXGw). The function jX(j.) will clearly be real and odd. Therefore, the given statement is false, (b) An odd Fourier transform corresponds to an odd signal, while an even Fourier transform coresponds to an even signal. ‘The convolution of an even Fourier transform with an sett Eourier may be viewed in the time domain as a multiplication of an even and Sad signal, Such a multiplication wil always result in an odd time signal, ‘The Fourier transform of this odd signal will always.be odd. Therefore, the given statement is true- 4.18. Using Table 42, we see that the rectangular pulse x(t) shown in Figure $4.18 bas 9 Fenrir hanorm yu) = sin(Sw)/w. Using the convolution property of the Fourier transform, we may write | alt) =2i(t) * 21(t) © Xow) = XiGiw) Xi (iw) (22) ‘The signal a(t is shown in Figure $4.18. Using the shifting property, we also note that ; Sage) £5 doe (22022) | 1391 oT, 1 ju (sin(3w)\? Jot) EB for (221) Adding the two above equations, we obtain ae) = Bas(e +1) + Joate—1) cote) (# ‘The signal h(é) is as shown in Figure $4.18. We note that h(t) has the given Fourier transform H (ju). Xk) a 4346 54 t Figure $4.18 Mathematically A(t) may be expressed as i <1 _| “Hag, sings Mom) NE = s 6(¢—28), see then w(t) = 2x (t) + 21(t — 1). ‘Therefore, X (Jw) = Xi(Gw)2+e™) Sse ~ beyl2-+ (194SM (i) Using the Fourier transform analysis eq. (4.9) we obtain 1 et X(ju)=—+ Gw= 55 (j) 2(t) is periodie with period 2. ‘Therefore, X(ju) 1 ibe)slo— kr), where X (jw) is the Fourier transform of one period of x(t). That is, 1 [: Lettie eft 21146) T= 1450) je Xu) ft 4.22. (a) ao={5 ee 7 (b) 2(6) = JorF*/5(e — 4) + JePrPA(t +4). (c) The Fourier transform synthesis eq. (4.8) may be written as a(t) EL Gane oman From the given figure we have a=} [ames cos(t— 3) ~ | t=3 7 G3 (a) 2(t) = Asint + 2 cos(2nt) (e) Using the Fourier transform synthesis equation (4.8), cos3t | sint —sin2t Gat ie? 4.28, For the given signal o(), we use the Fourier transform analysis eq. (4.8) to evaluate the corresponding Fourier transform = mete) T+ia Xoljw) (i) We know that y(t) = z0(t) + 20(-t)- Using the linearity and time reversal properties of the Fourier transform we have 2-207 cosw — we“! sinw iGo) = ta 143(i) We know that ‘t2(t) = to(t) — t0(—t). Using the linearity and time reversal properties of the Fourier transform we have Kal) = Koln) ~ Kl —du) = j [27 (ii) We know that a(t) = xo{t) + xo(t + 1). Using the linearity and time shifting properties of the Fourier transform we have Xoljiu) +e! Xo(—jw) = 2 Xs(jw) (iv) We know that 24(t) = trol) Using the differentiation in frequency property Xa(ju) = ‘ol5w). ‘Therefore, 1=20rteH — jwe™ (+ ju? 4.24, (a) (i) For Re{X(je}} to be 0, the signal 2(¢) must be real and odd. ‘Therefore, signals in figures (a) and (c) have this property. (ii) For Zm{X (jw)} to be 0, the signal z(t) must be real and even. Therefore, signals in figures (¢) and (f) have this property. (Gi) For there to exist a real a such that eJ%X jw) is real, we require that 2(t-+a) be a real and even signal, Therefore, signals in figures (a), (b), (e), and (f) have this Xa(jw) property. (iv) For this condition to be true, 2(0) = 0. Therefore, signals in figures (a), (b), (c)s (@), and (f) have this property. - (v) For this condition to be true the derivative of x(t) has to be zero at: t = 0. Therefore, signals in figures (b), (c), (e), and (f) have this property. (vi) For this to be true, the signal 2(t) has to be periodic. Only the signal in figure (a) has this property. (b) For a signal to satisfy only properties (i), (iv), and (v), it must be real and odd, and a(t) = ‘The signal shown below is an example of that. 2'(0) =0.ne 4.25. «(t) : eo, iG (a) Note that y(t) = 2(t-+1) is real and even signal. Therefore, ¥ (jw) i also real and shen, This implies that <¥ (Ju) = 0. Also, since ¥ (ju) = eX (jis), we know that 4X (jw) =u. (b) We have _ X(j0) = f a(t)dt = 7. (c) We have 7 i X (judas = 2nax(0) = An. (a) Let ¥ (jw) = 282#e%"., The corresponding signal y(t) is w0={3, ceria ‘Then the given integral is [Lx G00¥ Ge) = auf + VO} (e) We have 16. [ * xjuyPaa = anf” teQPat = (f) ‘The inverse Fourier transform of Re{X ju)} isthe Ev{a(®)} which i (2(0 + a(-1))/2. This is as shown in the figure below. pp, ¢yceif (a) (i) We have - 24 _ OM) (1/2) T4jo 2450” Q+juP 145‘Taking the inverse Fourier transform we obtain desta — Lert + beet ult) = Fouls) ~ Fe-*u(e) + 5te- ule) (ii) We have 1 1 vou) = xoaos)~ [zap] [asap] a), 0/4) a/4) 0/4) 24 jo" Bt juP? 4+ jw” 4+ jw? ‘Taking the inverse Fourier transform we obtain u(t) = her Ma(e) + fte-Mule) — Ferttutty + hte, ¥(jv) = X(jw)H(Gw) fel k=) 1 2, 1p T+j0 "T= jo ‘Taking the inverse Fourier transform, we obtain " u(t) = pet (b) By direct convolution of z(t) with A(t) we obtain tet we) = less t>s ‘Taking the Fourier transform of y(t), 2e~7® sin(2w) 60) = OF) 7 [ e™ ) eH 2sin(2w) ~ [T+jw, w = X(jw)H iw) 4.27, (a) The Fourier transform X (jw) is X(ju) = [lanema fowe [ome a MOLD Gy grinyge sora 6Po (0) The Fourier series coofcents 0, ate 1 yea on = FL He Ly [7-53 eae 20 [ = Sil — irye oor Paty Comparing the answers to parts (a) and (b), it is clear that Ly ah: a= FXG) where T= 2. 4.28. (a) From Table 4.2 we know that pit) = D> once EB Plju) =2e Y) anblwe — hen) From this, . ¥ Gu) = L(x) «HGw)) = Yo aeX(ile~ hw) (b) The spectra are sketched in Figure $4.28. 4.20, (i) We have Kalu) = [XGuller HII = X Gade From the time shifting property we know that za(t) = a(t). (i) We have Xj(ju) = LX Gu|OAOHIM = XGuIO™ From the time shifting property we know that y(t) = 2(0-+ 8). (iii) We have Xelju) = |X Gullo FO = X"CGu). From the conjugation and time reversal properties we know that elt) = 2°(-t). Since x(t) is real, c(t) = 2(—t).(iv) We have Xljus) = |X (ju)]o IGP = x Gae™ From the conjugation, time reversal, and time shifting properties, we know that a(t) = 2"(et~ A). Since x(t) is real, z4(t) = =(-t—d) 4.30. (a) We know that w(t) = cost © W (jw) = a[6(w ~ 1) + 5w +1) | and att) = 2(¢ cost #3 GlGu) = 2 {XGw) + WGe)} ‘Therefore, 1 ' | Gu) = 5XG(e - 1) + 5XG(w + 0). 4sa re asst since G (ju) is as shown in Figure $4.0, itis clea from the above equation that X (0) is as shown in the Figure $4.30. nati) 3G" = t we ey cra Figure $4.30 ‘Therefore, _ 2sint a) = (b) Xi(ju) is as shown in Figure $4.30. 4.81. (a) We have a(t) = cost #5 X (jw) = a[dlw +1) +5 U} (i) We have y(t) = u(t) 3 H(i) = ie + n8(w). ‘Therefore, . Y¥ jw) = X (jw) Gi) = F162 +0) — sw -1))- ‘Taking the inverse Fourier transform, we obtain u(t in(t). (a) We have hal = 261) +5e-%u(t) €B tala) = 24 Therefore, ¥ (juo) = XG) EG) = Flu + 1)-(w-1)} ‘Taking the inverse Fourier transform, we obtain v(t) = sin(t). 9(iii) We have ha(t) = 2e-tu(t) & He(jw) = (1+ ju? ‘Therefore, ¥ (ua) = XG) Hh (Gu) = Fate +1) - 4-1) ‘Taking the inverse Fourier transform, we obtain u(t) (t). (b) An LITT system with impulse response hal) = J ate) + hal] will have the same response to x(t) = eos(t). We can find other such impulse responses by suitably scaling and linearly combining y(t), ha(t), and h(t) 4.32, Note that h(t) = fiu(t—1), where y(t) = Sit at ‘The Fourier transform H) (jw) of y(t) is as shown in Figure $4.32. From the above figure itis clear that /y(t) is the impulse response of an ideal lowpass filter whose passband is in the range |u| < 4. Therefore, h(¢) is the impulse response of an ideal lowpass filter shifted by one to the right. Using the shift property, feos lcd moe = {Ghar * (a) We have Xilju) = xoitslw —6) + neBSlu +6). Its clear that Yi (iw) = Xi Gw) A iw) = 0 > n(t) = 0. ‘This result is equivalent to saying that X,(jw) is zero in the passband of H (jw). (b) We have Kali [Eprwe = 3k) - w+] . ‘Therefore, ~ Yin) = XaGW) AEG) = 2 [A/MMHlw 9) —Hlw 4 I}. ‘his implies that twit) = Fsin(ae — ). ‘We may have obtained the same result by noting that only the sinusoid with frequency 3 in Xa(jia) lies in the passband of (jw) 150. ee (c) We have rao {5 Shee Yo(ju) = Xa(iu)H (a) = XaGwwle "This implies that sin(st) aa alt) = za(t— 1) We may have obtained the same result by noting that Xa(jw) lies entirely in the passband of (jw) (@) Xa(jw) is as shown in Figure $4.32, pai \ =O € we Eqy er 9 + oo Figure $4.32 ‘Therefore, Ya(jeo) = Xq(juo)H (ju) = XalGw)e* ‘This implies that t) = a(t wd = 240 Meni ‘We may have obtained the same result “by noting that X4(jw) lies entir passband of H(ju). (a) Taking the Fourier transform of both sides of the given differential equation, — YG) 2 HG) = X Gu) ~ =F S (cate 2). rely in the 4.33. ‘we obtain, Using partial fraction expansion, we obtain 1 jot ford (jw) = 151‘Taking the inverse Fourier transform, ate) u(t) —eu(t). (b) For the given signal 2(t), we have X(jw) = G+ ier ‘Therefore, 2 1 Ca Bots) Oia Using partial fraction expansion, we obtain 12 ¥ (jw) = X (ju) HG) ee ts GorOP jur4 “Taking the inverse Fourier transform, a rC——_U ult) = eMule) — Fee Mule) + He“ Me) — Fertu(e (c) Taking the Fourier transform of both sides of the given differential equation, we obtain oe a) +2041 Using partial fraction expansion, we obtain V2 — 225 V2 + 225 HG) = 24 AON, ve jo = * 5 - aE ‘Taking the inverse Fourier transform, A(t) = 26(t) — VBC + 2j)e CIty — VALI — ayer 4.84. (a) We have ~_ +4 © 0= ob Bj CCross-multiplying and taking the inverse Fourier transform, we obtain ult) , gdult) = az(t) get ge + ult) = + 4a(t). (b) We have H(jw) = = - Tio Tie" ‘Taking the inverse Fourier transform we obtain, A(t) = 2e-*u(t) 152(6) We have \ : XU) = Tia” Oi “Therefore, YG) = XGWOY) = ETI Finding the partial fraction expansion of Y(jw) and taking the inverse Fourier trans- form, ult) = dealt) — Feral). 4.35. (a) From the given information, = Vere Go) = YES = 1 ‘Aso, = cos(t = > + cos(V3t — >). 4.36. (a) The frequency response is YG 3(3-+5u) X(jw) 4+ ju) 2+ jw) H(ju) = (b) Finding the partial fraction expansion of answer in part (a) and taking its inverse Fourier transform, we obtain: - A(t) = 5 le +o] ut (c) We have YGw) (9+ 3jw) XGw) 8+6jw-u?” CCross-multiplying and taking the inverse Fourier transform, we obtain @y(t) | edult) _ de(t) Set OG +t 8ult) = 3-FE + 9200). 153(a) Note that a(t) = xi(f) + x(t), _fh wi SGlo- FM = Ge) YO uw — 4D) eo peas ‘This may also be written as Gu) = FO XGak/DAGtw ~ KE) =F YY SCsnk/2)5(5(w — FF). Clearly, this is possible only if Glixk/2) = X(Gnk/2).‘Therefore, an LTT system with impulse response h(t) = }4(t) may be used to obtain g(t) from a(t), (a) Taking the Fourier transform of both sides ofthe given differential equation, we have ¥ (jw)[10 + ju} = X(5w)[Z (4) — 1). Since, Z(ju) = dg +3, we obtain from the above equation () 34 2ju HG) = X Ga) = TF jo) IO (b) Finding the partial fraction expansion of H(jw) and then taking its inverse Fourier transform we obtain 1 agi) = betas) + Zen. We have We) =2() +n) Gu) = X(u) HG) From Parseval's relation the total energy in y(t) is a= [wore=z = Ef GorrnGeyPae IY (iw) Pio 1 On, pata wi i ce han ali we Rix iePas zelX Gun For real x(t), |X(—jwo)|? = |X(jwo)|?. Therefore, B= Lx(iun)PA. Let g(t) be the response of Hi (jw) to a(t) coswet, Let g(t) be the response of Ha(12) to a(t) sinwet. ‘Then, with reference to Figure.4.30, y(t) = a(t)e™t = a(t) coswet + ja(t) sinwet, and w(t) = gilt) + joa(t)- Also, f(t) = e-*4+w(t) = [coset ~ j sinuset]lar(t) + joa} ‘Therefore, Re{f (t)} = git) cos wet + go(t) sinwet. ‘This is exactly what Figure P4.46 implements. 1594.47. (a) We have nal = MO EMED Since h(t) is causal, the non-zero portions of h(t) and h(—t) overlap only at t = 0 Therefore, 6, 1<0 A=) Ae), t= 0 (s4a7—1) De(t), 4 >0 Also, from ‘Table 4.1 we have helt) © Re(H(G)} Given Re{H(ju), we can obtain h(t). From hg(t), we can recover h(t) (and conse- quently (ju) by using oq. (8447-1). ‘Therefore, H(i) is completely specified by Re{ul ju). (it Re{H(ju)} = cost = 3" them then, hel) = Falt-+ 1) + Z6tt—D). ‘Therefore from eq, (S4.47-1), A(t) = 6(¢~ 2). (©) We have nay = MOHD, Since h(t) is causal, the non-zero portions of A(t) and A(—t) overlap only at t = 0 and h(t) will be zero at t= 0: Therefore, 0, r<0 A(Q)= 4 unknown, t= 0. (447-2) Dhelt),“"—-t> 0 ‘Also, from ‘Table 4.1 we have hot) 5 Tm{H(i)}- Given Tm{H(ju), we can obtain hot). From h(t), we can recover A(t) except: for + =0 by using eq. (S4.47-1). If there are no singularities in A(t) at t = 0, then H (jw) can be recovered from h(t) even if h(0) is unknown. ‘Therefore H(jw) is completely specified by Zm{H(ju) in this case.