DIGITAL SIGNAL PROCESSING A Computer-Based Approach Sanjit K. Mitra Text Errata List 1. Page xii, Line 17: Replace “Short-term” with “Short-time”. 2. Page ain, Live 4, Replace “eeeealuedu” wil “ipbet wece-ucsb edu” Page xvi, Line 6: Replace “5.5” with "8.5". Page 13, Line 2: Replace “Asin (Q,)” with “Asin (Qgi:”. 5. Page 25, Interchange Figure 1.22 (a) and Figure 1.22 (). 6, Page 125, Eq. (3.10): Replace“ "with = y”, a sa of b 1. Page 127, Table 3.2, Fist line: Replace“ Length-N sequence" with“ Sequence "and replace “N-point DFT” with “ DEFT" 8, Page 127, Table 32, Line Som op: Replce “ea 50) * wit “eal 2) 4. Page 131, Ba. (2.13) Replace eo” yi me, 10. Page 148, Eq. (3.75) and Eq, (3.76): Replace “L ” with “L—1 11. Page 155, Eq. (3.88) Replace“ (N-M) " with (Naf 1)". 12. Page 175, Line 13 from bottom: Replace “2!/( 1+ 427)» with 3 13.Page 175, Line 12 from bottom: Replace “-3n( 2! Ju " with “3 n—1)(2 14.Page 175, Line 10 from bottom: Replace the equation with . .036(} =)" eta) = [024(=t)'+036(2) Juin +03600-0 (Jute 15. Page 183, Eq. (3.149): Replace “i {k]" with “h(n] 16.Page 184, Replace “Figure P3.2 shows four" with “Figure P3.2 shows two" 17.Page 184, Line 3 from bottom: Replace “length-sequencs” with “length-N sequence”. 18. Page 197, Problem M3.3: In the numerator of expression in (a), replace “0.1915¢" with “0.19157, and “o191se"™ with “0.1915<%"" 19, Page 198, Problem M3.17: Par (ii) - Replace “ld > 0.4" with “Id > 0.948683", 20.Page 198, Problem M3.17: Part (iv) - Replace “ld >0.4" with “lel>05", 21. Page 211, Line below Eq. (4.48) Replace “/™**! wigh # /2/M 22, Page 211, Line 2 from bottom: Replace “3.20” with “3.21”, 23 Page 219, Line 8 from bottom: Replace “H,,{n] " with “hz (nl ". 24. Page 220, Line 8 from bottom: Replace "lg jo, (| "with “logo (2)|". 25. Page 222, Eq. (4.68): Insert aj” in front of *<7°/*sia(2 erste")26.Page 230, Figure 4.19: Replace “v [4m] ” with“ [-n} "and replace “w (£n) " with w [=a) 21. Page 247, Eq. (4.1390): Replace “Ay (2) +A, (2)" with “Ay(2) ~A, (2) " 28. Page 252, Line 5 from the begining of section 4.8: Replace “zero locations” with “pole locations” 29, Page 254 After Line 4: Replace“ [T2," with“ 1)" TT 2," iat is 30.Page 271 Problem 4.61 (b) line 3: Replace * (4n+0,)/" with" (2e-+ 0) /M" 31, Page 274 Problem 4.67 (c) Replace “G (@) ” with “F (w) ”. 32.Page 275 Problem 4.76 (a), (b) and (c): Replace “Hg” with “Hy, (2) ”. 33.Page 276 Problem 4.82 (d) Replace “H," with “H, (2) " 34, Page 277, Problem 4.88: Replace “1,” with “H, (2°, and “H," with “H, (2)” 35.Page 277, Problem 4.88 (6): Replace “(1+ 2°) in the numerator with “3154 652" 465 36.Page 280, Line 5 from bottom: Replace “4.84” with "4.83". 37. Page 280, Line 2 from bottom: Replace “4.97” with “4.96”. 38. Page 290, Last Line of Figure caption: Replace “cos (2x:)” with “cos (262)". 39. Page 340, Eq. (6.36): Replace “a” with “D”, 40. Page 314, Line 12 from bottom: Replace “9.4” with "6.4" 41. Page 327, Eq. (5.65): Replace “2R, + R” in the denominator by “2(R, + R) ” 42, Page 374, Line 3 from bottom: Replace “A, (2) " with “Ay (2) 43, Page 393, Line 1: Replace “0.39" by "03x", 44, Page 415, Problem M 6.2 (b) and Problem M 6.3 (b) : Replace “four” by “two”, 45.Page 45, Poder M68 Replae mente of th “oowsaial tae" sate ean eve 46, Page 547, Example 8.20: Replace “A = -0.325,5" with “A = -0.375,,”. 47. Page 557, Line 5 from bottom: Replace “Bq, (8.84)" with “Eq.(8.88)". > * 48. Page 568, Problem 8.44: Replace “ 3° o_,27" with * a2” 49. Page 570, Figure P8.5(a): Replace “142 with 142 8 8 ‘50. Page 577, Line 2 from top: Replace S’a_,2' with Y a2” im i 51. Page 579, Last Line: Replace “[Mit74a]” with “(Mit74c]’ 52. Page 659, Figure 10.4: Replace the figure here with the figure shown below: xin] +} | M | —syin} '53.Page 669, Figure 10.15(b): Replace “L” with “M”, 54,Page 678, Example 10.7 line 5: Replace “f, = 031” with “fy = 031",‘55. Page 695, Lines 1-3 below Eq, (10.61): Replace this sentence with “As aresult, y{Ln+k) = axx{n) . ie. the ‘input samples appear atthe output without any distortion at all values of n, whereas the in-between (L 1) sam- ples are determined by interpolation.” '56,Page 707, Equation 10.109 and the equation two lines below: Replace “(1-[(e*)}*-[in¢*)°)» with tele ae} 57.Page 731, Problem 10.14 (b),(c): In the denominator. replace “0. * with "0.8?" 58,Page 732, Problem 10.19: Replace“ $: (/®n4), > with Shley> cnt ht 0 ay salen -neoyred uns -cne 60.Page 755, Section 11.5 Tie: Replace “Short-Term” vith “Short-Time". 61. Page 755, Line 10 from bottom: Replace “togn” with “2ayn” 62. Page 798, Lines 11,13, and 35: Replace “2°” with “*™, and “[-3]" with “-1) 63,Page 829, Lines 4 and 5: Replace “f, =0.68" with *f=0.18” and “f,=0.8" with “f, 20.3" 64. Page 829, Lines 7 and 8: Replace “f, =0.68" with “Yj, =0.18" and “f,=0.8,0.77,0.74 and 0.71" with *fy03,027/0.24 and 0.21" 65. Page 829, Line 11; Replace “f, =0.68" with “/, =0.18" and "fy=0.71" with “fy =0.21" 66. Page 829, Problem M11.) lin 2: Replace “0.46” with “0.36” (67.Page 830, Problem M1113, line 2: Replace “Program 11_4” with “Program 11_S”. (68. Page 830, Problem MII. 14, line 2: Replace “Program 11_5” with “Program 11_6" 69, Page 830, Problem MII.15, line 2: Replace “Progranm 116” with “Program 11_7"7 Solutions Manual to accompany Digital Signal Processing A Computer-Based Approach Sanjit K. Mitra Department of Electrical and Computer Engineering University of California, Santa Barbara Prepared by Rajeev Gandhi ‘wcB vay tt McGraw-Hill Boston DUET Ridge, IL DubuyseyIA Madison, WE New York San Francisco St. Louis Bangkok Bogoti Caracas Lisbon London Mat Mexico City Milan New Delhi Seoul Singapore Sydney Taipei Toronto eeChapter 2 24 (@) xnl={3 -4 206 39 3} Thus weain{s 93602 -4 t dnJ+xi-n) fs 9 3 9 _ 9395 Hence, sy inte [8 2227233 3. 9393 a cel rr in i 3 (b) uiny={-2 12 5 0 ot 6 3} . Thus, u{-n] $ 6 on 0521 = 1 tee Piasoragr gy Hence u,,tnl= and uggind={-1 bri 3934-41 7 1} (©) v{n]= Acos(«»gn)+Bsin(gn). Thus, v{-n] = Acos( n)—Bsin(«99n) Hence v,{n]=Acos(@yn) and vggla]=Bsin((on) 2 | (a) s{n]=n™ ‘Thus, ef-n] Hence g,y{n]=n* and g,glnl=0 (e) blnj=n? Thus, h-n]= | Hence h,,{nl=0 and hgglt} 2.2 (a) {x{aI}=[Aa® }. Oy)} = F{ac" sara) "} where A* and a, represent the complex conjugates of A and ct, respectively U Lag ni Xpealt gfttal-x<-2 >I} = fae’ -As(an™ } (b) {h{n}} fon -4-j5 24j3 5+) -seith t Hence net-a= [6-1 Sj 1B 4488 2-2} cea 3 Thus, host B-}+ i} B 2 =124j3 -2- een and nie {e PP BPR zea} 2.3 (a) {x{n}} {aa"} where A and 0. are compiex numbers, with [al <1.24 2.5 2.6 27 Since for n <0, [al can become arbitrarily large hence (x{n]} is not a bounded sequence. (b) {y{nJ}=Aa*y[n] where A and o: are complex numbers, with fal <1 In this case |y[nJ S|A] Vn hence {y(n} is a bounded sequence. (ec) {h{n}}=CB"y{n] where C and B are complex numbers, with J3|>1. Since [B|? becomes arbitrarily large as n increases hence {h{n}} is not a bounded sequence. (a) {e{n}} =4sin2(@,n). Since [g[n] $4 Wn hence {g{n]} is a bounded sequence, (e) {v{n]}=3cos(@,n”). Since |v[n]] $3 Wn hence (v{n}} is a bounded sequence. Recall, x,,{n]= +(x{n]+x{-n)). z Since x{n] is a causal sequence, thus x[-n] = 0 Vn >0. Hence, xin] =x,,(01+x,,[-n] = 2x,,{n], Va >0. For n= 0, x[0] = xey{0]. ‘Thus x{n] can be completely recovered from its even part. tx{n}, n>0, z 0 n=O. Likewise, x(n] =3 ‘Thus x{n} can be recovered from its odd part Wn except n = 0. 2y,qln] = yln]—y *{-a]. Since yfn} is a causal sequence y(n] = 2y,,{n] Vn >0. For n=0, Im{y[0]}=y,q{0]- Hence real part of y{0] cannot be fully recovered from y.,{0] ‘Therefore y(n} cannot be fully recovered from y,,[n]- 2y,,{n]= yln]+y*[-a}. Hence, y(n] =2y,,{m] Vn >0. For n=0, Refy{0]} = ‘Therefore y(n} cannot be fully recovered from y,,{n]. _s(0]. Hence imaginary part of y[0} cannot be recovered from y.,{0] 1 Kyla) = 2 (xin]+xl-n)). ‘This implies, x,,{-n]= + (x(-n]+ xfn}) =x,,{0]- Hence even part of a real sequence is even. x,glnl=4(x{n]=x{-n}). This implies, x,gf-nl= 5 (xE-n)- x{n}) =—x,qln]- Hence the odd part of a real sequence is odd. RUS of Eq. (2:170a) is xs{0] +x,yln—N]~ Hala] +x4{-n)) +5 (ln -N}+x*{N—n)) Since x[n] = 0 Vn <0, Hence Keglt]+ Xogl—N]=4 (x{n]+x*(N~a) RHS of Eq. (2.170b) is in], Oy]) for OSnSN-1, Since, x[<—n >y]=x[N-n], it follows that Xgl) F(x{n}+x*0N—n}), IsmSN-1. For n=0, Xpes(0}=4(x(01+x*(0)) = Re(x101} Similarly xpcq{a)=4{abel—a*CS—a>y]) =p (ale) x#IN aly 1eMeN 1. Hence, for =0, Xpeq{O}= £(x{0]~x*10]) = Im(x{C}} 2.9 — x[n]=cos(2mkn/N), Osn 1. Hence, Nel Net x E, = Ycos?(2nkn/N) =$) (1 +eos(4rkn/N) = Yoeos(4rkn/N). m0 a0 0 x Nel Let C= J’ cos(4nkn/N), and S= Dsintankn/N). 250 Na Therefore C+jS= pelt — =0. Thus, C=Re{C+jS}=0 Re{C+jS}=0, it follows that E, 2.10 (a) Average power of sequence x{n} = jn] is teM tim LS yt = tim 2 Pay = lim gap DUH = tim aap M+ a (b) The sequence x{n]= Acos(2t +4) is periodic with a period M. Hence el Met Py nae cost 4g) ALY (coo 4g) = A 2.11 (a) Consider a sequence defined by x{n]= >) 8{k). ee If n <0 then k = 0 is not included in the sum and hence x(n} = 0, whereas for n 20, 5 1 020, included in the sum hence x[n] = 1 ¥n20. Thus x(n]= daw-{f neo, Kewe 1 nt 0, nso. 1 0, 120. (b) Since {a} = 9, it follows that win —1 Hence, p{n]—p{n—1212 Now x(n] = Asin(aon +4) (a) Given x{n]={0 1.5 0 -1.S}, it follows that the fundamental period of x{n] is N = 4. Hence @, = 2/4 = n/2, Solving x{0]= Asin() =0 we get @=0, and solving x{l] = Asin(r/2)=15, we get A= 15. (&) Given x{n}= {V2 V2 -V2 -V2}. Again the fundamental period is N = 4, hence y= 2nl4 = W/2. Next from x{0]= Asin(6) = V2 and x{I]=Asin(n/2+9)= Acos(9)= 2 it can be seen that A = 2 and 4 = 1/4 is one solution saticfying the abave equations (©) x[n]={3. 3}. Here the fundamental period is N = 2, hence @y =. Next from x[0] = ‘Asin(g)=3 an a{I}—Asin( | x)= Azin(})~—3 observe that A = and = 1/2 that A= 3 and = 1/2 is one solution satisfying these two equations. @ xtnjJ=(2 V2 0 -V2 -2 -VZ 0 2}. Thus the fundamental period is N= 8, hence @, = 1/4. From x[0]= Asin($) = 2 and x{l] = Asin(g + 1/4) = V2 we note that A=2 and $ = 7/2 is one solution satisfying these two equations. 23 (a) AS Kf} e129, 5, [0-48] = 08075478! = X,[n]. Hence N = 8 is the fundamental period of the sequence X,[n] (b) Xj{n]=cos(0.2mn). Let N be the fundamental period of the sequence. This implies that Kq[a+N]=cos(0.2nn +0.2RN). For Xy[n] tobe periodic , we must have O.2nN = 2x. Hence N = 10 is the fundamental period of the sequence %,[n} (c) %,[n]=2c0s(0.1nn)+2sin(0.2mn). Let N be the fundamental period of the sequence. This implies that X,[n-+N]=2cos(0.Inn-+0.tnN)+2sin(0.2mn +0.2nN) = X,{n). For to be periodic in N we must satisfy 0.1nN = 2nr and 0.2nN = 2k. Therefore N = 20r and N = 10k. Hence N = 20 is the fundamental period of the sequence X,{n} (@) X4{n}=3sin(0.8nn)—4c0s(0-Irmn). Let N be the period of the sequence. Then x4[n+N]= 3sin(0.8nn +0.8nN) —4.0s(0.lnn-+0.1nN)=i,[n]. Hence N = 20 is the fundamental period of the sequence X,[n} (€) Xs{n]=5sin(0.1nn) +4cos(0.91n)—cos(08nn). Let N be the period of the sequence. Then Z,[n +N] = Ssin(0.Inn +0.1nN) + 4c0s(0.9nn +0.92N) —cos(0.8rn + 0.8RN) 20 is the fundamental period of the sequence X,[n]. s{n]. Hence, N= modulo 5. Since X¢{n+5]=(n+5) modulo 5 =n modulo 5 = X¢{n]. is the fundamental period of the sequence X¢{n]2.14 2.18 2.16 (a) x{n] = Acos(@,n) (a) 5 = 0.151 and N = 2nr/ivg. Substituting the value of @y we get N= 401/3. N takes an =3, Hence N 2nrltg. It can be shown here N= 100 is the fundamental period of integer value when 40 is the fundamental period of this sequence (b) @y = 0.18 and the sequence. (6) 4 = 0.225n and N = 2nr/@q. It can be shown here N = 80 is the fundamental period of the sequence (a) c= 0.3m and N= 2reléng. Tt can be shown here N= 20 is the fundamental period of the sequence, (©) ©, = 045m and N = 2nt/io,. It can be shown here N = 40 is the fundamental period of the sequence. (a) x(n) =gln}g{n]. Hence x{-n] = g{-nlg{-n]. Since g[n] is even, hence g{-n] = g{n}. ‘Therefore x{-n] = g{-n)Jg(-n] = glnlg(n] = x(n]. Hence x{n} is even. (b) ufn} = g{nJh{n}. Hence, uf-n] = gf-nJh{-n] = g{n](-h{a}) = ufn] is odd. -g{nJh{n] = -u[n). Hence (©) via} = h{njh{n). Hence, v{-n) = hfn}h{r] = (-h{n})(-h{n]) = h{n]b{a] = vin]. Hence vin} is even. D Bvin]+yv{n—1]. Hence, From figure, vin] = o.v{n-1]+x{n) and y(n] cey{n -1]= 08 v(n-1]+ ry vin=2},and yin] y(n -1]=B(v{n] -av{n—1])+7(vin-1]-cev{n-2)). Thus, y(n] = ory(n ~ 1] + Bx(n] + yefn 1] x(a} (b) va}It follows from the figure, v{n} = x{n] + x{n-3}, un} = x[n-1] + x{n-2], and y[n] = aav{n]+Bufn}. Hence, yo] = o-(x{n]+ x{n—3)) +B( x{n ~1)+x{n -2)). (©) Here vin} = x{n} + dyu(n) aid ute] = afn) - vet] = fn] xfn 1) dyufa-t). Thos, yln]= v{n—1]+4,0(n] =x{n]—vfa]+djufn}. Asa result, dyln—1= 4 xn -1]-dufn-1)+dy00~1) An —1] +x{n]+(d, ~1)xln]-x{n—1]). Hence, y(n} = d,x(n] +x{n—1]-d,yfn— 1). and y(n}+djyn—- (a) Ie follows that v[n} = x(n} - dyw(n}, w(n] = v[n-1] + dgu(n}, and ufn} = v[n-2] + x(n]. Thus, w{n] = x{n—1]—d,w{n 1] +d,xfa} +d5xin—2}-dydwin—2], and win} +d,win 1) +d)d,w(n—2] = x{n=1]+d_s{0] +d,x{n-2]. However, yln) = vin -2]+d,win}, Hence, y{n}= x{n—2]—d,win~2]+d,w{n] from which we get 4,yln~ 1] =4,x{n—3]—dp win —3}+dyw(n—N] and 4 d,yln-2}=4)4,x[n-4]-d7d,win—4] +d7d,{n~2]. Adding the last three equations we arrive at_y[n]-+d,y{n—1] + dyd,yla 2] = xfe—2] + d,x(n—3]-+ did,x[n— 4] ay (win=2}+ dyin 3} + dydywta A) +4,(vdn +4,(0~ 1) +n 2) = did,xin}+4,x{n-1]+ x{0-2]. x{n}=0.58[n-+2)-+1.58{n - 1} dfn — 2] + 8{n —4]+0.758[0—6). Since 8{n]=y{n}-H{n—1], Hence x{n]=0.5p{n +2]—O.5p{n +1] +1.Sufn-1]-2.5¢{n -2}+ H[n-3]+pln—4] pln —5]+0.75p{n- 6] -0.75u{n -7] (a) w{n]=—3{n +3} +8{n) + 48{n-1] - 26[n—3] +360 - 6] (b) Using the fact that 8{0}=y{n]-yln—1], we getwin] = —pln +3]+ fn +2}+ nn] +3p{n— 1] —4ufn—2]- 2nfn— 3} +2p{n—4]+3u{n - 6]-3p[n-7]. 2.19 Let y;{n] and yo{n] be the outputs of the system corresponding to the inputs x;{n) and x2{n}, respectively, Let y3[a] be the output corresfonding to the input xs{n]=ax,{n] + bx2{n] For the discrete-time system to be linear, we need to show that y3{n] = ay,{n]+by,[n] (a) yln}=o,x{n]+0,x(n- 1] +c4,x{n -2}+0,,x{0 3]. Hence, 31) + b(o xn] +4,x,[0 =1]+a3x3[n -2]+04x410-3)) ala} =a (ex in}-+ 05X01 4546021 raga =ay,{nl+bya{n]. Hence the system is linear. (b) yin] = boxin}+byxln—I]+bpxln—2)+a,y(n—1]+a,yin—2].. Hence, yal} = bo {axa} + bx2fn}) +b, (ax,{n—M1+bx,{n 1) +hp(axjfn—2}+ bx{n—21) +a,ysin— M+ agysfn—2] = a(box,{n] +b,x,[n — 1] +bx,[0 -2}) + b(bpx2[n] +b,x2[n — 1] + b,x, — 2]) + ays{n—1]+a,ys{n-2] yala} = aly,{n]-ay,{n—1]—ayln-2.) +b(yafal—a,y,{n -1]-a,yx[n-2)) + ayy, ~ 1) + aays{n 2] ; ~ yala}=ay,{n]+ by,{n}, ie. the system is Tinear for a causal input only if yy{-1} = yil-2I = yol-HI = yol-2) = 0. © ain ft n=0,L,42L, 0, otherwise. fax,[n/L]+bx,[n/L], n=0,41,22L,.... ao) -{ 0, otherwise, =ay;(n]+by,[n]. Hence the system is linear. 1 Met 1) Ml a) @ ya) wees Hence, foley Dale Hegy Dale wey Dale =ay,{n}+ byp{n}. Hence the system is linear. © yol= nylon +5 (xyln—+x, (0-41) yalal=xagl0] +H (tygKn—U+ Xggl0+0) = axyln]+bx24C0] +4 (an ln-M+bs,,{n—H+an,,In +1] +bx2y(0-+ 1) =ay;,{n]+by,{al. Hence the system is linear.2.20 (xyn-21 + xy{0-+1)} aln}+ yatok= ago (agin te Ags 042) 5 (pl 2}+x5y(a+H) = axp,{n] +bx3,{0] 4 (ann M+ bran “H+ anya +2}+bxagln-+2) +Zfoxyt0-2+ begin Aan lott + bea) =ay,{n]-+by,{0) Hence the system is linear. (a) y{a) = nx{n. For an input x;{n] the output is yj{n] = n xy{n}, and for an input x;[n] the output is y2{a] = nn xgin}. Thus, for au input xgfn] = ox, E0] +B xp{n}, the onrput yoin] is given by gta} = u( cxf} + Bxan}) = oy, [n+ By, fn] Hence the system is linear. Since there is no output before the input hence the system is causal. However, ly{nl] being proportional to n, it is possible that a bounded input can result in an ‘unbounded output. Let x(n] = 1 Wn, then y(n] =n. Hence y[n]— 2 as n — o°, hence not BIBO stable. Let y{n] be the output for an input x(n}, and let yj{n] be the output for an input x;(n}. If x,{n]=x{n—no] then y,{n] =nx,{n]=ax{n— ng] However, y{n—ng]=(n—ng)x[n no]. Since y;{n]# y{n— Mo], the system is not time-invariant. (b) y[n]= *°En] For an input x;{n] the output is y;{n} = x 3{nl, and for an input xy{a] the output is ya{n] = Ayu). ‘Thus, for an input x30] = & {9} +f xa{nl. the output ya{n] is civen by yale = (ox (0)+Bxgla)? # 0p fa} +B°x3C0] ‘Hence the system is not linear. Since there is no output before the input hence the system is causal. Here, a bounded input produces bounded output hence the system is BIBO stable too. ‘Also following an analysis similar to that in part (a) it is easy to show that the system is time- invariant. 5 (© y(nj=B+ > )xin- 4) ny Since PB #0 hence the system is not linear. However, here a bounded input produces bounded output. Thus, the system is BIBO stable, Also it is easy to check that the system is causal and time invariant.- s @ yinj=B+ Doxin-4) é ‘The system is non-linear, BIBO stable, non-causal and time-invariant. (©) yln] = ax[-n] ‘The system is linear, stable, non causal, Let y(n] be the output for an input x[nJ and y,{nJ be the output for an input x,{n]. Then yfn] = ox[-n} and y,{n] =ox,{—n] Let x,{n]=x{n—no], then y,{n]=ox,{-n] =ax[-n—ng), whereas y[n~ng]= e reduces to Kai(ks $) which is equivatent to K? =a or in other words, K= Va Itis easy to show that the system is non-linear. Now assume yj{n] be the output for an input {rien 8) xy{n). Then yf Tie i! > 1 xin~n9] If x[n]=x{n—ngJ Then, viet ia-ureS=2a) ‘Thus y,[n]=y{n—ng}. Hence the above system is time invariant. 2.24 As 8[n] = wln]-p{n—1], Te6{n})=Th{n}} - Tafa -1)} => bin] = slaj—sta— 1 For a discrete LTI system ylo}= Srtken—K) = Do olkI-stk-myxin—K] = YsikIxfn—K]- DY sfkUn{n—) oe kere ee fea 2.25 Now y(n] = x[n]@hln] = = Yxn— kabel eo Substituting h{n} = 0.58[n - 2] + 8{n - 2] + 0.58{n - 1], and x(n] = 8{n ~ 3] + B{n - 2] + 8{n— 1] + B[n J, we obtain yfn] = = (0.58(k -21+ 8k 1] + 0.56(K)\G{n-3-K] + Bln -2—K]+Sfn-1—K}+3fn~K). Hence vini= {os S22) is os} 2.26 xol={9 10 Bt ratol= {2 10 (a) yq{n) = xy[01® hyfo -{9 2050 a (db) ya{n] = xgfn]@ hgfn} -{ 2 Tt (© yale} = x1l9]@ hla} ={9 12-44 a} =44 it (@ alo) = x2i1@ nto =f2.27 2.28 2.29 yla} = x{n]@bln} = YY x(n—KIhCk] — Substituting k by n-m in the above expression, we get y[n]= ),x{mJh(n—m] = h{n]@x{n] Hence convolution operation is commutative. Let yfn} = x(n} (b,{0]+hyfn}) = = xin —Ki(hy Ck) + hylkl) kewe 5 Keo > xin—kghylkd +) xln=kIhg lk] = xlo} hyn] + x{n] hota). kewe ke Hence convolution is distributive. x30] Q x2in]@x1[0l = x3la] @ al] @x1l-) As xq{n] @xi{[n] is an unbounded sequence hence the result of this convolution cannot be determined. But xq{n]@x3{0] @xzL0] = x2{n]@ (x3{n]@ xy Ln]) . Now x3{n] @xj n] = 0 for all values of n hence the overall result is zero. Hence for the given sequences xg{n] @ xan] @ xyL0] # xgln] @xsl0] @ xf] w{n] = x{n]@h{a]@ein}. Define yin} = x(n] @hln} =) xCk]hfa—K] and fn] = hin} @ala) = > stkdbln—ky ‘ Consider wy [n] . (xin]@ala})@egtn] = ym] @aln} = Y) elm] Y) xl] h'a-m—K]. : ta Yetetkent—m kom Difference between the expressions for w;{n] and wo{n] is that the order of the summations is Next consider waa] ~ x[a]@(h{a]@sta)) = x{a]@ ln changed. ‘A) Assumptions: fn] and g{n] are causal filters, and x{n] 0, form <0, ynl= {52 aoinn- KI, form20. Thus, win] = 0" gelimlyln—m}= 0" get) Yep} hn —m— Kk) [All sums have only a finite number of terms. Hence, interchange of the order of summations is 0 for n <0. This implies justified and will give correct results. B) Assumptions: hfn] and g{n] are stable filters, and x(n] is a bounded sequence with [kins X. : " Here, ylnd= Soy Mame) = 2, WORD e+e fm] with fess tonfs e,X2.30 2.31 2.32 2.33 In this case all sums have effectively only a finite number of terms and the error can be reduced by choosing ky and ky sufficiently large. Herce in this case the problem is again effectively reduced to that of the one-sided sequences. Here, again an interchange of summations is justified and will give correct results. e Hence for the convolution to be associative, itis sufficient that the sequences be stable and single-sided. yfn]= Yxn- kJh{k]. Since hfk] is of length M and defined for 0Sk x{0] = y[0]/h(O] k=0 i Mi In general, x{n]= sbifio- Drea ash i 2.34 yfnl nfo} where x[n — k], k = 1, 2,.... have been computed before. 2.35 (a) Given (yin]}=2 8 20 40 60 6R 62 40} and (hin})={2 4 6 8}. Therefore length of x{n] = 8-4 +1=5 8 1 . Using c= fin- Sete, es of x[n] for n =0, 1, ...., 4 can be computed and is given by (x{n}}={1 2 3 4 (b) Here x(0] = y(0)/h[0] = 3, x[1] = RO] =z obtain {x{n]}={3 -2 4 -6 7} (© x{0) = an = 1+ j2. Following the procedure outlined above, we arrive at +j =~ 2. Continuing further we {x{nJ}={1+j2 2-J3 3+j}. 2.36 y[n] = ay[n - 1] + bx{n]. Hence, y[0] = ay[-l] + bx(0]. Next, yL1] = ay[0] + bx{1] = a’y{-1]+abx{0]+bx{1]. Continuing further in si(a) (b) ©) 2.37 2.38 (a) f[n} = ln ~ 1] + fn - 2]. Let fl) () Let y,{n] be the output due to an input x,{n]. Then y,{n}=a"*!y[-1)+ If x,[n] = x{n— ng) then 2 ang yjlal=a™ y+ Yat *bxtk—ng) =a™y-+ Ya" 0“ bxfe] Kno = 3 Le However, y[n—ng]=a™*0My(-1}+ Ya" "0'bxlr]. i= Hence y;{n]# yn—no] unless yl-1] =U. For example, st y-1] = 1 then the system is time variant. However if y{-1] = 0 then the system is time -invariant Let yj{n] and yo{n] be the outputs due to the inputs x;{n] and xa{n]. Let y(n} be the output for an input ox,{n] +Bx,{n]. However, ay fn] +Byp{n}= aa"*"y{-1+Ba™ y(t] +0) a"*b x(k] +B 9a" "bx, fk) ko ie whereas yin}=a™ y(t] +a) a" bx tk]+B Ya" *bx,(k]. 0 io Hence the system is not linear if y[-1J = 1 butis linear if y[-1] = 0. Generalising the above result it can be shown that for an N-th order causal discrete time system to be linear and time invariant we require y{-N} = y{-N+1] = -.- = y[-1] = 0. The difference equation representation is yln]=yln—1J—x,{[n}+x,{n] with x,{n]$ ylo] and initial deposit = yf-1) car", then the difference equation reduces to 215 or" ar"! ar"? =0 which reduces to r?-r-1=0 resulting in ee. we, roi=a( ZZ) +08) Since f[0] = 0 hence a +0 =0. Also fl] = pence S178 5G 81=%2 ao 2 LLNS) tial conditions are equal tw zero. (n - 1] + y(n - 2] + 8[n—T1]. Hence, Solving for , and cy, we get ot ylny = yf = 1] + yn 2] + alu 1]. As system is LTH, the Let x(a} = 8{a}. Then, ylnyl0] = yf- 1) + y{-2]=0 and y[I] = 1. Forn> I the corresponding difference equation is y(n] = yln ~ 1] + yln— 2] with initial conditions y{0} = 0 and y[1} = 1, which are the same as those for the solution of Fibonacci’s sequence. Hence the solution for n > 1 is given by se) yta} 5 Thus f{n] denotes the impulse response of a causal LTI system described by the difference equation y(n) = y(n - 1] + y(n - 2] + x{n - 1] 2.39 y[n) = y[n- 1] - x{n] + Ris the difference equation representing the system. y[0] = D and x{n] = P. Hence y{n} = yfn— 1] - (P -R) or y{n] ~ y[n - 1] +P-R=0. Thus, y{1] = y{0] - (P - R), and y(2] = yf1] - (P - R) = y[0] - 2 - R). Continuing the process, we get y[n] = y(0] - n(P - R). Let N be the number of years needed to repay the loan fully then y{N] = 0. Then N = y[0}/ (P-R) = DAP - R), 2.40 y[a] = ay{n—l}+x{nJ. Denoting, y[n} = yrel] + yim{n], and « =a +jb, we get, Ygql 1+ iy gq Ct = (0+ ONY pel 11+ IY} {0 ~ 1) +E]. Fquating the real and the imaginary parts . and noting that x{n} is real, we get Yeelt] = ay, <= 1]—by ql 1)+ xE0}, a Vil] = by gl +Ay ig (0 Therefore 1 b Yigl®— N= 2 yjqlMI- 27,01 Hence a single input, two ontpnt difference equation is 2 by gin y,.fn=I]+x{n] a 2 Yeelh] = ay, gf thus by, ,{n—1) =ay,(n-I]+(a? +b?)y,.[n-2]+ax{n—I]. Substituting the above in Eq. (1) we get Veeltt] = 2ay,-[n~1]-(a? +b*)y,.[n—2]+x[n]—axfa—l] which is a second-order difference equation representing y,¢{a] in terms of x(n} 2.41 The impulse response of the cascade is given by h{n] = hy[n]@hg[n] where hyn] =a"y[n} and hin] =B"pln]. Hence, hfn] (Sater ea K=0 n+l, S03, 2.42 gin} = biol@hin} where niay=["Z 9053.Hence atol={1 4 10 20 25 24 ish, and ator={uv2s 4i25 2S AIS 1 24125 16/25}. T 2.43 For a filter with complex impulse response, the first part of the proof is same as that for a filter with real impulse response. Since, y[n]= h{k}x{n—k], ke bind -| Seer < Svtkin -kh a wo Since the input is bounded hence U S|x[nJ SB,. Iheretore, |yinJ SB, ‘Dy Inlkh Soif S.nlkl=S<-> then |y{n]] < B,S indicating that y[n] is also bounded. ee To proove the converse we need to show that if a bounded output is produced by a bounded input then S <2. Consider the following bounded input defined by x{n} =F =a Then y0)= a YlkI=S. Now since the output is bounded thus $<. ie — ‘Thus for a filter with complex response too is BIBO stable if and only if Ih{kJ=S and S'Ihgtk |<. Hence, > etk] <=. c r « Hence the cascade of two stable LTT systems is also stable. 2.45 Consider a cascade of two passive systems. Let yj{n] be the output of the first system which is the input to the second system in the cascade. Let y{n] be the overall output of the cascade. ‘The first system being passive we have Spanf? < Sorin Likewise the second system being also pasive we have S'btni?< Spink? < Siar,indicating that cascade of two passive systems is also a passive system. Similarly one can prove that cascade of two lossless systems is also a lossless system. 2.46 Now, b{n]=a"u{n]. Therefore yin]= ) h{k]xin—k] =) oxin—k] Kemm io exin)+ Soxin=K] = a(n} os(a=1-k} =x] +eyl0—1 Hence, x[n]=y[a]-ayln—1]. Thus the inverse system is given by yfn] = x[n]—ax{n ~ 1] ‘The impulse response of the inverse system is given by atol={1 -a}. 2.47 y(n] =y[n-1]+y{n—2]+x{n—l]. Hence, x{n~ 1] = yln] - yfn - 1] - y{n - 2}, ie. x[n} = y[n + 1] - yln] - y{n — 1]. Hence the inverse system is characterised by yla} = x[n + 1] - x(n] - x{n ~ 1] with an impulse response given by ato)={t 7 -} ; Ld 2 2.48. yln]=pox{nl+ p,x{n—1]—d,yln—1] which leads to x[n}=—yfn]+—Lyfn—1]-—L xn —1] ‘0 Po Po Hence the inverse system is characterised by the difference equation 1 4, Py y,lo] =x {n}+—xy(n—1]-2Ly fn 1) Po Po Po 2.49 Let the first order causal system be yfn] = a y{n 1] +b x{n] +e x{n~ 1} Let the input to the system be x[n] = 8[n]. Then its impulse response samples are given by h(O] =b, h{1]=ab+c, h(2}=ahf1]. Solving these three equations we get 2M) ye ‘{21a(0] asap DEMO e=h- M N 2.50 Y'pyxin-K}= J. d,yln—K}. mt io . ‘s Let the input to the system be x{n]=8[n}. Then, ‘Y" p,Sin—k]= 1d, h{n—k]. Thus, k=0 ko P, = )/4,hit—K]. Since the system is assumed to be causal, h{r-k] = 0 Vk > ¢. é p, = Ddyhtr—K] =) hikld, é io M 2.51 vinj= Digxta —k]. Define another cause! FIR system with input and output related by N. y(n} = Yaa =k]. The inverse system cf the new system is defined by ket2.52 2.53 vinl N M N Yayyin—K1. This implies Yoytn-K= Dayyin-K. ko fa) i=0 Hence a causal IIR digital filter can be realised as a cascade of an FIR filter and inverse of a causal FIR fliter. Let Yip,xin-K]=yinl+ )'4,yln—K] be the difference equation representing the causal imo TR digital filter. For an input x{a] = 3[n], the corresponding amtput ic then yfa] = hn]. the impulse response of the filter. As there are M-1 {py} coefficients, and N {dj} coefficients, there are a total of N+M+1 unknowns. To determine these coefficients from the impulse response samples, we compute only the first N#M+1 samples, To llstrare the method, without any loss of generality, we assume N = M = 3. Then, from the difference equation reprsentation we arrive at the following N+M+I = 7 equations: HO] = Pg. ALL] +HI0}d, =p), h(2]+billd, +hlO}d, =p, h3]+hl2}d, +h{I]d, + hlO}d, = py, hl4]+h(3}d, +h[2}d, +hll}d, = 0, AIS] +h4]d, +hI3]d, +h[2}d =0, H{6] +H{SI4, +h{4]d, +h{3Id, =0. Writing the last three equations in matrix form we arrive at a ( bn] 4 ( bEs}|1{bC4) {3} Ea] } a, | h{6}} Lbts] bl4} Id, |} 10. 4,| fois nz) amy 'yheay and hence, {dy |=-|hi4} h3}hE2I] | hE] 4{ LMS) 514) ME3}) bf) Substituting these values of (d;) in the first four equations written in matrix form we get Po| fro) o o off Pi |_| hi) ho) Oo | py] {hl} mL) lO] 0 Id p,| LM) hz) nit} Bolas (a) vin] = (hyfn} + h3{n] @hs{o}) xin] and y{n} = ha{n] @ vin] + bln] @hgln)] @xto]. Thus, yfn] = (ha{n} @hy la] + bon] ®hsln]@hsta} + hs{n} @hgin}) @x(n}. Hence the impulse response is given by hh{n} = haf} @hy a] + holm] @hgln] @hsta] + hgla] @ hain](b) v{n} = h4ln]@ xfn} + by fn] @hgln]@x{n} Thus, y{n} = b3{n]@ vin] + by[n] @hs{n]@xin} = hg{n} @hgla] @xln] + by6n] Ohy Ln] @hyln] @xtnj + byln] @hs(n]@xin] Hence the impulse response is given by bln] = bln] @bgln] + hyln]@ hyln} @hgin] + hylan] @hsin] 2.84 ufn] = hy{n]@hpln] + hs{n} Rut, hyln]@holn} =65{nl+28{n—-11438{n—2'+8[n-3]. Hence, hfn]=78{n]-8[n—1}+33[n—2}+ 8[n—3]+7 8[n-4]+6 [1-6]. 2.85 The ctate equations are given by sainl=s,{a—-1], s3{n]=s,{0-1], si{n +1] x{n]-0.85,{n]-0.6s,{n]-04s3{a] yin}=0.4(x(n]~0.8s (n]~0.6s{n}-.4s3(a) + 0.69,[n] +f] +0.8e3C0] sin+ll] f-08 -06 -04)sjnl} [1 ie, |ssin+t]=| 1 0 0 Jsp{nl] +] 0}xtm) sini} LO tO 4sitmy} Lo stn} yin}=[0.28 0.76 0.64} s,{n)|+0.4xn) sala] 2.56 For a second-order system, the difference equation is of order 2, and is of the form poxtn]+ p,xin—1]+ pyx{n-2)= yfn}+d,yln 1] +d,yln—2]. We thus need to determine 5 coefficients, po, P1, P2, dj and dy trom the state-space description. To this end, we first determine the first 5 samples of the impulse response samples from the state-space description of the causal LTI discrete-time system: D, for n=0, wolfearin, terns Os 04 L Hee, A=[%5 03) B-[}], ¢=[2 3}, and D=08 was 3f%5 b3]5 5, Thus, b(0] = 0.4, nil) =[2 af] 1 wai-e 3{.%5 Sart] 1, nd wai=[2 3.5 9a] [s=o23 Next, following the results of Problem 2.52, we arrive at 4) _ fn) ayy yay) pss yp 1.87 ]_y-08" 45)> [hE] h2)) | nea)" 11.87 5.5} [0.231]" [0.23Po| fro) o oft] [o4 0 Of) o4 pi {={Ht] hf] 0 Fd, |=]11 04 0 |-08)=| 10.68 | ~ p,| (nl a] wOlJa,| [55 11 04}0.23] |-3.208 Hence, the difference equation is given by 0.4 x{n] + 10.68 x{n - 1} -3.208 x{n - 2] = y[n] -0.8y{n — 1] +0.23y[n—2]. 2.87 62 =E{(x-m,)"} =E{x? +mz -2xm,}= Ex”) +Elm,)~26(xm,) Since m, is a constant, hence E(m2}=m? andE(xm,)=m,E(x}=m72. Thus, (x2) +m? 2m? = E{x?} -m?. os 268 Ele ty)— [fot covddrdy = [fxty(nrmedy« ff yey eatery =f figisori ee fol [fyCuants)y = [acon ft 94 E{x) + Ely}. 2.59 mean = m, = E(x}= f_ xp¢(x)dx variance =02 = E{(x—m,)4}= J” (x—m,?px(0dx aft ape _xdx x fa) px(x) “ } Now, m, . Since —~—y is odd hence the integral a Px eae? Wore x +0! = is zero. Thus m, =0. 2 apr _x7dx RI ea this density function. 7. Since this integral does not converge hence variance is not defined for a ) p= Fer. Now. m, xeMlix=0. Next, belo lax e“*[" pm 2 oe axl = ofopeeel oh Be | 3 (© p,0d= (Ppl )"“8-9. Now, r=) FSD fSe-0ac=> (8) ott-0)* np a 0 2 2 eHienal een loof =E(x2)-m? = [7x S(ope-0 85 ¢x = ebdx — (np)? -e(0 pet ~p)'"'-n?p?=np(l-p). Sota! @ ade BH 9. Now, m=fx) ESE soot a5. z 10 0? =E(x)-m? ae PSone! © pw=7pe™ Praca), Nom 1 pa Brae? oy tina? — ce woows= <5 f 281200 Gy = al RTD. 2 2 ots? a2 in? ae _(,_®)\2 of = Ex?) - ~2E 2(2-B 2, BBE} my =p fe woods ( 2 2.60 Recall that random variables x and y are lineerly independent if and only if efhs+apf po V aay except when aj =a) =0, Now, efe,P a? }+e{p bi }+E (ey apxy "PE [e(02)*x*y} = bien? } +p.) ebe} > 0 Va, and a, except when ay = a Hence it x.y are statistically independent they are also linearly independent. 261 dgylQFE{xin+ Ay*In]}, Oyyf-OHE(xIn—Ay*Lal}, oyq(=E{yin + 1x0}. Therefore, by, lIHEfy *ln+ ata} =Efxln Aly *[0]}=9,y{-2) Hence $,y{-A=6,,*U) Since Ypy{4]= Oxylé-mg(my)*. Thus, Yyyl4] = yl] — my(m,)*. Hence, Yayl)= OgylEl = melmy)*. AS a result, Yyy El = gy 14) — (m,)*m, Hence, Yqy{—E) = tye) ‘The remaining two properties can be proved in a similar way by letting x = y. 2.62 Efstn}—xin—eP?} 2 0. E{hint"} + Efjsin- a} = E{x(n}x*{n-<]} ~ Ex *Inlx{n J} 20 26,,(0] - 26,14202.63 2.64 2.65 2.66 Sexl0]2 44041 Using Cauchy's inequality Efn?}e[y?} ~.. Thus fim Y,,(4* fim 4,04 [+6. Hence tim 6,,{0=|m,l° on a Be c= Bx -xF}. To find the value of that minimize the mean square error C, we ifferentiate C with respect to K and set it to zero. ac ft Thus $o = BtX-K)] = O which results m X= E(X), and the minimum value Of Cis oF ik VsaX+bY. Therefore, E(V} =a E(X} ~bE(Y} =am, +bm,, and 2 =E((V—m,)*) =El(@X—m,)+¢¥-m,)7). Since X.¥ are statistically independent hence 62 =a2o? + bo? M (@) dal Jim raz DAEMe+ a}. Since xin] is periodic with period N, hence xlneN] = Ao 1 M. M x(a). Now, rn +N]= ate i ae fee] = lim Deletes = Ho). Hence r{n] is also periodic with a period N. (b) sf-n]= Sxterxte-n} = S)x{¢-+n}x(44=r{n}. Hence efa] is an even function of n. (©) Since YY{4\-Af¢+n})* = 0, it follows that 2x{0]~2r{n]= 0. Hence r{0] 2 x{n] for all ron values of n. (@) For an aperiodic sequences r{0] =} x"[4] which is equal to the energy of the sequence. (ane F x 1 im - Ul Mom 2M +1 2° Substituting n = 0 in the autocorrelation sequence of a periodic random sequence we get r{0] is For an periodic sequence, the average power is defined by P,, = lim equal to its average power.(€) Since the autocorrelation sequence of a periodic sequence is also periodic with the same period, hence an way to determine the period of the actual sequence is by measuring the distance between two maxima’s say r[0] and s(N] in the autocorrelation sequence. Note that duc to part (c) r{0] will always have the maximum value 2.67 (a) -x[n}= wn( 22) Since this sequence is periodic with a period M it follows that ae io 2 sl so 22) 5 fa a of sn a Eo) alae) ‘Thus, r{n] is periodic. ‘The period of the sequences r{n} and x[n] is M. x(Ox,(0+4 (b) x,{n]=nmodulo7. Since the above sequence is periodic with period 7, it follows that an] 6 Yatra = {15 wo s778 woh, tn} is periodic with a period 7. =) (©) The sequence x,{n] = (-1)" is periodic with a period 2. Hence 1 ¥{n] = FrslOuln += {h r=) } Thus, r{n} is periodic with a period 2. M21 b= input('Desired length = '); n= i:b; FT = input(‘Sampling frequency = ');T = 1/PT; imp = (1 zeros(1,L-1)];step = ones(1,L); ramp = (n-1).*step; subplot (3,1,1); stem(n-1, imp) ; xlabel(['Time in ',num2str(T), ' sec']);ylabel('Amplitude'); subplot (3,1,2)7 stem(n-1, step); xlabel([{'Time in ',num2str(T), ' sec']);ylabel(‘Amplitude'); subplot (3,1,3)7 stem(n-1, ramp) xlabel({‘Time in ',num2str(T), * sec']);ylabel(‘Amplitude'); M2.2 aAsinput (‘The peak value L=input (‘Length of sequence ='); Nzinput('The period of sequence ="); FT=input('The desired sampling frequency =") ‘T=1/FT; t=0:L-1; MoAtsawtooth (2*pitt/N) ; y=A*square (2*pi* (t/N), 40); subplot (211)stem(t,x)7 ylabel (‘Amplitude xlabel({'Time in ',num2str(T),'sec']); = subplot (212) stem(t,y)i ylabel (‘Amplitude’); xlabel({'Time in ',num2str(T),'sec']); M2.3 (a) The input data entered during the execation of Program 2_1 are Type in real exponent = -1/12 Type in imaginary exponent = pi/6 Type in the gain constent = 1 aype in length of cequence - At (b) ‘The input data entered during the execution of Program 2_1 are Type in real exponent = -0.3 nype in imaginary exponent - pi/3 type in the gain constant = 1.5 type in length of sequence = Si ‘The plots generated are shown below: Real par Imaginary part o 0 2% 3 0 30 ‘Time index n M24 (a) b = input('Desired length = '); A = input (‘Amplitude = '); omega = input('Angular frequency = '); phi= input('Phase = '); n = O:L-1; x = A*cos(omega*n + phi); stem(n, x); xlabel('Time index") ;ylabel('Amplitude');M2.6 M27 M28 (b) ' ' | 35 os rill IL) gts rd i ut + a $ i | LTT] iL) ss rr a Tine nex Tine index (i) cos(0.25mn) (ii) cos(0.2nn) airs 10 a 4 5 2 of i Sieg Pgs dy rill shill, & 12) of J | Ae TET a0 10 20 30 40, io} 10 720 30 40 Tine index Tine index 2cos(0.1mn) + 2sin(0.2nn) (iv) 3 sin(0.8xn) ~ 4cos(0.17tn) t = 0:0.001:2; fo = input('Frequency of sinusoid in Hz FT = input('Samplig frequency in Hz gl = cos(2*pi*fort) ; plot (t,gl,'=") xlabel (‘time’) ;ylabel(‘Amplitude') hold mx Ostsem: cos (2*pi*fotn/FT) ; Plot (n/P?,gs,"0") hold off € = 0:0.001:0.85; gl - coa(S*pite);g2 = coo(1itpite)sg3 ~ coe (26rpite); plot (t/0.85,gl, '~',t/0.85,g2,'--',t/0.85,g3,':") xlabel (‘time’) ;ylabel (‘Amplitude’) hold O:1:8; cos (0.6*pitn); plot (n/8.5.gs,'o') ‘As the length of the moving average filter is increased, the output of the filter gets more smoother. However, the delay between the input and the output sequences also increases {Tis.can be seen from the plots generated by Program 24 for varfous values ofthe iter length.) alpha = input('Alpha = '); yo = liyl = 0.5*(yo + (alpha/yo)); while abs(yl - yo) > 0.00001 y2 = 0.5*(yl + (alpha/yl)); yo = ylr yl = v2: end aisp('Square root of alpha is'); disp(y1)M29 alpha = input('Alpha =‘); yo = ljyl = 0.5*(yo + (alpha/yo)); while abs(yl - yo) > 0.00001 y2 = 0.5*(y1 + (alpha/yl)); yo = yl; yl = y2; end 7 disp('Square root of alpha is’); disp(yl) M2.10 y = input (‘First sequence = h = input(‘Second sequence = [x,x] = deconv(y,h); disp (x) (@ 1 2 3 4 5,(b) 3 -2 4 -6 7, @ 1+ i2 2-43 3+ it Chapter 3 34 x(e!®) = Yi xinJeH". If (x{n]} is absolutely summable then Sita <-. Thus, s ST hunkeo| < Siot X*(e%) (e) Re{x{n}} = F tala} + tn) Now the DTFT of x*{n] is X*(e~!). Hence, using the linearity property of the DTFT we obtain Re(x{n]) eo Pxei) +Xr(eI). L 3B i Im{x{n]} <= FX) X*(e)), (4) Im{x{n}) {x{n] ~x"{n]). Hence, using the linearity property of the DTFT we obtain (©) x.gin]=2(x{n]+x*[-n)). Using the lineasity property and the results of part (b) we get alo o> 3 [x(o! rX#(0!)} - ReGKCH)}. 1 (O) Xcaln] = 7 Xeglt] F(xee™)-xe™) = jlm(X(e™)}. (x{n]-x*(-n]). Using the linearity property and the results of part (b) we get 3.5. (a) xevin] = Feat + x{-n]). Since x(n) is real, x{- n} = x°[- nJ. Thus Xeyin] = ={x{n] + x"[-n]}. Now using the linearity property and the results of Problem 3.4 2 part (b) we get x,,{n] = Ffxei®) +X* (ei = Refx(el™)} L 2 galt 69 4 (X(e!®)—X*(el®)) = jlm(X(e!9)) (b) Xoafn] = > {x{n] ~ x{- n]} =F lato] ~x"[-n}}. Thus () Since x(n] = x"[n}. This implies X(e!)=X*(e7J). Thus X,,(ei®)=X,,(e7#) and X jg (CH) = -X jg (7H). Also [K(!®] = f X70!) 4X7, (61%) = [x2 (eH) 4X2 (eI) im im re in Fe (OF) + XFCE) Xe) Xie) a) aren 50) — tan~! Xin 1 Xigy = jo (eI), Likewise, ang X(@!®)= tan! + oo) Xe 7X), 283.6 x{n] = F frei reirao. Hence, x*{n] = (a) Since x[n] is real and even, hence X(ei®) = +(e), Thus, xt-n} = 3 [x(eiye"Ido, cn Therefore, xin]= 4 (aln}+sl-n)) = [XC cox(onde, Now x{n] heing even. X(ei®) = X(e7I), As a result. the term inside the above integral is even, and hence x{n]= 4 Jx0e!)costomdo 0 (b) Since x(n] is odd hence x(n] ~ xf a. ek ‘Thus x{n] = (stn) xt-n) J xc sinconde. Again, since x{n] = ~ x{-n], X(el) =-x(e). The term inside the integral is even, hence x[n]= 2 J X(e!)sin(onjdor o ciate? 4e-IMo%e-i? 37 rnl=otentayn spun) =aar( Sees ME A ays Aetna! = Nei(aeiits Ae-it(aeis = Fol(aeito "winds eM (ae 1__iA.-ie Geigy 17 Ogee in]. Therefore, A xeimy =A, (el): ae 3.8 I uM (oe) xgfal=4%" ISM: then, x,(ei%)= Sater + 0, otherwise. & Lao He I00M4) ay song Ler Meio 5 rae glee (@) xslnJ=o"y{n+3}, — fol= Fate 5 404-01 s= Do = elBamtk-DIN 1 -ei2an(k-1) 1-1 Ik —1 # EN then S = 7 area TN = |g PRMEDIN = 9 Not N= -¥ N. Ifk—1=:N then S= ” Wy = Yea a0 * Nol ence yy _ {Ns for k=1=0N, ran integer, a 2 N -{5 otherwise Rela Na . Sta] =) X(e}hfn—e]. Then jfn-+KN]= Skits +kv=r}. Since fifn] is periodic inn with i=0 = net > eelhin-1) = Fa), hence jn} a period N, h{n+kN—r] = h{n—r]. Therefore 9[n+KkN] is also periodic inn with a period N. (01 0 2}and h{nj={2 0 1 O} 3 Now s101= 5, xt f(0— 0} — sfoyht01 + RMBEa}+eEAIAL21+ RIAN = 0 Similarly 7 DY Mele) = ROTA + RE FLO} + X(2] HES) + E3}HL2] = = Continuing the process we can show that {2]=0 and j[3]=5. i{n]={2 -1 0 3} and A{nJ={-1 2 1 Of Following a procedure similar to the above problem, we can show that j{nJ={4 8 0 —4} Since H,{n+eN]=i,{n], hence all the terms which are not in the range 0,1,...N-1 can be accumulated to P,ln}, where OSky]. Therefore Xtk] = Yig(y]Wae 20 1 xa = YgiN+n-npwak+ Ygln—no]Wi = aan, Nomgt Nal = Sawer YE atnwert* = wi Y atniwet = we*GIK) net, =o a Re ~ Ce) ula}= Wy" Yutnwek = Sy in wee Fo a0 ifk2 ky, x Tween), itk Nel Nel Net Not (A) hin] = Gln}, Therefore, Hik}= Yh{njwek = Ginywek =D) Yetqwerw mo a0 m0 0 Nel Nad > Serve ‘The second sum is cy zero ‘only ifk=r=Oorelse ifr =N-kandk #0. Hence, Ng(0], if k Hk] eee if k> 0, 7Nel<—k>y]- Nel Not Nat (e) ula) = YelmJhf< n—m >y]. Therefore, UIKI= YD) elmIbl,] Wg = m=0 120 m=0 35Na Nel Nel = ¥ elm) Shl< a—m>y]We = DatmIH(kIWg* = HIKIGE m0 m0 n=O ~ Not Net Ne (© v{n} = ginhfn}. Therefore, V{kI= >” elnlhin] Wet “4m Su b{njGew"* W> = t= a=0 f=0 eo i i ta Hw om ces ry) a=0 3.26 0 =0 Not Nal, (Net 1 NING! Net Tie? = Vee Saw Exo “ xtax"tg yw n=O ao 720 250 2-0 Na ito Since the inner sum is non-zero only if ¢=r, we get. |x{n] =y LIxEI . a0 =0 Na 3.27 Xtk]= > x[njw™. Na (a) X"[] = Sx" [nW-, Replacing k by N - k on both sides we obtain a0 Nol : X"IN-K1= Sot tawr) Sere Thus x*{n] <> X*(N -k] = X"[<-k >n]. ~ a0 Nel (by X*tK] = Sox*inpwo™. Replacing n by N ~n in the summation we get aaa a X*1K) = Yoxt tN nwo = Px [Nn a=0 a0 Thus x*[N a} = x"[<-n >n] © X"EK1. (©) Re(x{n]} = F(xta] + x"[n]}. Now taking DFT of both sides and using results of part (:) we get Re(xin}} <> $(XIK] + X*[<-k>n]} 2 (a) j Im{x{n}} = $6xtn} = x*{a]) this imples j Im{x{n}} <> 4(XK]-X'Ie-k >nl) (©) xpestn] = Foe) 4+ x*[<-n >N]} Using linearity and results of part (b) we get pesln] o> E(X(k) + XK} = Rel XOK)- 1 (9 xpcaln] = 2 {x(n} ~ x°{<—n >yy])- Again using results of part (b) and linearity we get ~ pean] € 4 (XTK] -X"(KI} = jm (XK) 36Not 3.28 X{k} = Re(XCkl) + j ImEXtk]} = Jy xfn}e~!H07N a0 DET {xin]+x{<-n>y]}- From Table 3.6, x*[<-n>y] = X*[K}. Since xin] is real, x{<-n>y]=x*f<-n>y] X*(k}. Ths, KggGk] = {XU + X*1K0} = Re(XOK. x{nJ—x[<—n >y]}. Asa result, X,,{k]= $ext) ~X*[k]} = jlm(X(k]}, (©) Since for a real sequence, x[n] = x*[n], taking DFT of both sides we get X[k] = X*[<- key]. This implies, Re(X{k]} + j Im{X[k]} = Re{X[<- k>y} -j Im(X{<— boy}. ‘Comparing real and imaginary parts we get Re(X{k]} = Re(X[<- ky} and Im{X{k]} =~ lm{X<- on] Also [XEK] = /(Re(X(K]))? + (Im(X(K]))* = {(Retxte-K>y)))° (cleat -k> 1) inca) an-if ~2tXE—K ay) Re(X(k]} RelX{<-k>y)) [xt<-k>y i and axgtxti = tan" } = ~arg(XI<-k >]y) 3.29 (a) For a sequence to have a real DFT, the sequence must be periodic conjugate symmetric. ‘As x4la] is the only periodic conjugate symmetric sequence, it will have real-valued DFTs. a (b) For a sequence to have a imaginary valued DFT, it must be periodic conjugate anti- symmetric, Neither sequence is periodic conjugate anti-symmetric. Nol Nal 3.30 (a) Now, X(N/2] = Yox{nJwgh/? = ¥(-1)"x{n}. Hence if x{n] = x{N ~ 1 - n] and N is m0 m0 Not even, then J’ (—1)"x{n] = 0 or X[N/2] = 0. =O " Nel (b) X{0] = })x{n] so if x{n] = - x[N - 1 - 2], then X{0} = 0. m0 N Net a N= (©) xt20= Sxtnw?" = ¥ xintw?" +” xintw? a0 ad anna 1 Ma MS Saw? 5 nen Swe = 5 cata +xin + pwr’ a0 = a0 Hence if x{n] =~ x[n + 4] =~ x{n+M), then X[24] =0, for £=0,1,....,M=1 3.31 Let x{n] = a (where a is some real constant), for 0SnSN-1 37 de“1 IN, ifk=0, . J2M0KIN Since yen aaK/N [Ne . Now, X[k] = Savy a re s = -{t elsewhere, cA a = eee eve Thus the N-point DET X(k} is real valued Nol Sxinji" = then X{K] = { Nel Next, X(e) Yee". TEN is even then a=0 a0 X(ei) = aenSON-D/2| Yno{ee) Hence in this case X(e!) has linear phase IEN is odd then X(ei®) =aeHOND/] 1+ Yoo = I this case also X(e!® has linear phase. 3.32 (a) (G{k]}={14+j -2.1+j3.2 -12-j24 0 0.9+j3.1 -03+jLI}. hh{n] = gl6], and hence H{k]=Wé*GIK]. k=0.1,....5. Therefore . trek) ={Wecto) won) WicIa) wor) wiScr4) w2%ctsi} ={l+j, -1.7213- 8.4187, -1.4785+ 22292, 0, -3.1347—j0.7706, 1.1026 j0.2902} (b) In this case h[n]=We*"g(n]=(-D"g{n]. Therefore, _ tony) = [5a] WzPett) WESel2) WARE] We!2aL4) We"Sat6I} = {41 -35 12 -5 2 -33} MN-I Nel Net 3.33 Ytkl= Yytnwiks = Sxtaiwek. Thes, vikMi= Y xfalwat = X1kl. a0 0 Y{kM] Hence, Xtk] 9 3.34 (a) X(0] = ))x{n] = 22. 0 3 (b) X(5] = (-D* xin] = - =) ° (e) Xk} =10x{0] = 20. to (2) = Ye D Ih}. The DTET of eK! KI] is xfen - 4219]. Thos 0 S = 10 x{< 0 - 4>49] = 10 x[6] = 0. (e) From Parseval's theorem Spx? =10 ke? Thus Sixt? - to 0 0 383.35. Since x(n] is real, X[k] = X"[< XI] = X*t] =-5+j4, X(10] = X"[2] =3 + j2, X19] =X"BI= ‘X[8] = X*[4] = 2-j5, X{7] = X"[5] =6 +2), ~ iH | (a) x(0) = aaa * in ) x16) = oka = 3 mo (c) ¥.x[n] = X[0] = 10. 0 Thus S 2Ri< 0-4 >12) = XW) = 2-15 wu | (@) S= Ye" Fx{n]. The DFT of e?*™/7x{n] is Xi< k ~ 4 >12) 1 i (©) Sato = 5 Deer = = o 2 | 3.36 Now, yeln] = Dielk]h[4]. Hence, 0 ye] = g(0h[0]+ gf1Jh(3] + ef2]h(2]+ gf3}h{1], Yel xO Jo 1] + gt tTh(O] + s{2]h[3] + gf3}hI21, yel2] = g{0Jh{2] + gf1]b[1] + g(2}h{0] + g{3}h[3], and Yel3] = g(O}b{3] + g(Jh(2] + g{2]h(1] + gf3}h{0). 3 Likewise, y, {n]= 'elk]h[n—k]. Hence, =o {0} H{0}, gfO} h{1] + eft] h(0}, 0] h{2] + gf] ft] + gf2]h(0}, (0) h{3] + aft} h(2] + gf2}n{1] ~ 2f3) (0), y; 4] = of11bC3}+ gf2)5(2]+gf3]000), a(2]h{31+ ef3]b(2], and Comparing ye{n] with yL{n] we observe that 10] + yL{4], LU) + yL{5}, L{2] + yi{6], and LL] + LIT. 3 3.37 (a) yp{a} = >)" efk}h{n—k}. Thus, y, [0] = s(0]h{0]= 5x (~3)=—15, 0 yz C1] = g(O}B(1] + g{ITh[0} = 5x4 +2%(-3)=14, ca y_ (21= ef01h(2] + sl 1} ht] + gf2)h{0] = 5x0+2%4 +-4(-3) =—4, 39y_[01= e(016(3) + gl!) + ef2) b(t} + g{3]h(0] = 5x2 +2044 %44(-1)x(-3) =28, y [4] = afO1b4] + 1) HD} + g(2Ih(2]+ ef3]h{t] = 5x (-1)+2x2+4x0+(-I)x4=-5, y (51 e(01(51+ e(1)M(4) + e(2)HD}+ a3] 2] = 5x2+2x(-1)+4%2+ (1) x0=16, = (6) = elITb15]+ ef21h(4)-+ g{3]b(3] = 22+ 43¢(—1) +(-1) x2 =-2, y_ (7) = ef 2}b(5] + gf3) hl4] = 4x2+(-1)x(-1)=9, yL[81= af 3}h{5] =D x2 Hence, (y_{nJ}={-15 14 4 29 5 16 2 9 2} 5 (b) yeln] = YgelkIn{< a —k >], where wam={5 aio 4} Thus, ko Y¥elO] = gO] h(0} +g [1b{5]+ g_(2}h4] +g,[3)h13] + 2,14)b12] + 8.1/0) (0) h{O] + gf 1} H(S]+ ef2]h[4] + e{3I(3] = 5 x (-3)+2x2+4 x (-1)+(-1)x2=-17, Yell] =s,(O}b{] +s, UIHO]+ g,(2)H{5] + g,[3]b{4]+ g,[4]hI3) + 8,{5)h(2] O}N(L] + BEL} HO] + g(2IHLI] + gI]N[4]=3x4-42.x(-3)+ 402-4 -D ACY -23, gelO}h(7]+ g_(Ih{t] + g_[2}h{0] +2, (3)b(5] + 2 [4] hf4] + 2 (5)h03) = g{0)h{2]+ g{IDh(I] + gf2)h(0}+ ef3]h{5}=5x0+2%4+4x(-3)+(-Dx2=-6, LO} hL3] +g, (h(2}+ g,{2) ht] + g,f3]h(0] +g, [41h{5] +g, (5]h{4] {0} 3] + eft] A(2] + g(2]HE1] + eL3]h[0] = 5x2+2x0-+ 4x4 + (1) x(-3) = 29, ycl4] = g,[0lbI4]+g,(UbI3) + 2, (2)h{2] +2, (3) h1] + 8_(4] h[0]+ 8, {5]h15] {0} h(4] + g{ U1 h{3}+ ef2] b(2] + eC 3]hI1] = 5x (-1)+2x2+4x04(-1)x4 = gs lO}h(5}+g,(UhL4] + g,[2)h[3] +g, (3)h[2] +e, (4)b{1) + 8 {51 hL0] = g{O)h{5] + g{lDhE4) + ef2}h(3] + e(3)h(2]= 5x2 +2x(-1) +4x2-+ (1) x0 =16. Hence, (ycinJ}={-17 23 -6 29 -5 6} ¥el: Yel (©) Using MATLAB command ££¢ we determine the two 6-point DFTs, Glk]= {10 5— 5.1962 14]1.7322 8 1-j1.7322_ 5.5+ 5.1962} and H(k]={4 -1.5—j2.5981 -3.5-j0.866 -12 -3.5+j0.866 -1.5+j2.5981} Thus. G[kJAK]= {40 -21~j5.1962 -2~j6.9282 -96 -2+ j6.9282 -21+ 5.1962} Next, using the MATLAB command if £t we compute the IDFT of Ge{kJH[k] resulting in (cln}}={-17 23 -6 29 -5 16} 3.38 Let ufn] = ge{n] +j h{n], where gc{n] is giver in Part (b) of the above problem with Utk] denoting its 6-point DFT. Now, ufn]={5-)3, 2+i4, 4, -1+j2, -j j2}. Using the MATLAB command ££t we compute its 6-point DFT which is given by UtkI = {10+ j4, 7.5981 j6.6962, 1.866~ 1.7679, 8-j12, 0.134—j5.232, 2.4019+ j3.6962} Thus, Ul< - k>y] = UIN ~ K] = {10+ j4, 2.4019 + j3.6962, 0.134~j5.232, 8-j12, 1.866 —j1.7679, 7.5981 6.6962} Therefore, Gytkl=F(Ulk+U*t<-K> yp = {10 5-§5.1962_ 14j1.7322 8 1-j1.7322 5+ j5.1962}, and (Ulk]-U*{<-k>y]}= {4 -1.5-j2.5981 -3.5-j0866 -12 -3,5+j0.866 ~1.5+ j2.5981} 3.39 We need to show git IO = hla] @®@ete). Let x{n] = g{n]@hfn} = Samten- m>y) > md Nel and y(n} = h{n]@agln] = Yo htmlgty) a Latmigin-rm]+ Y)bmfalV-n—m) ms nes Net Swe mlg(m]+ Fo en mig{m] = Sten m>, etm) = xt ment ao Hence circular convolution is commutative. Nol 3.40 (a) y[n] = g{n]@bln] =} elmIh{,,]. Thus, = Nel Nel Net Dial = Setoionen- m>y] Sen] Sem] = m=O a0 Ameo (b) yn} = gla] ®@alnj = Setomen- m>y]. Thus, Ne ae ze 1)" yn] = Sa Sten mSyI-D® m=0 n= Nel (mot No = dal mf Snsen- my ye Se 1) hfa = n| a=0 =m Replacing n by Nen-m in the first sum and by n-m in the second we obtain Naf Nel m-l Ee 1)" {a} ~ Euof § a. beay—ya-Nom Si rena] - a=0 - (Se “oes Sc wna = cos 228 etn) = MAM (eizeeniN 4 gi2atn/N 3.41 yin) = cos{ 22 tn} = Ae i +e Hence Y[k] = Xlek 4 Coy) E Mek = >y] an] Wl +5 tn Wg" 3.42 Since x[n] is real hence X{k] = X*[<—K >}. Thus, *[<— 1 >g] = X"[8] = 5.5 + 58.0. X"[< -3 >9] = X*[6] = 9.3 - j6.3.7 aLX(5) = xi X*[< -5 >9] =X") *[< ~T >9] = X"Q2] =1.7~}5.2, and 5 - ja ~ 3.43 Since x(n] is real hence X{k] = X"[<~—k >n]. Thus, X[2] ais 2 >9] = X"[7] = -4.1527 - j 0.2645, X(3] "[<-3 >9] = X*[6] = 6.5 ~j 2.5981, X[5] = X*[<- 5 >9] = X*{4] = 63794 - j 4.1212 and X{8] = X"[< - 8 >9] = XL] = 2.2426 +). 3.44 Since X{KIs are purely eal, x{a] = x "(<= >y). Thus "13] = ~ 0.1098 ~ j1.6708, [2] = 0.25 - j0.125, and *[1] = - 0.642 + 50.0795. 3.45 Since ¥{k] = W#*X(k] hence y{n]=x[< 1-4 27]. Thetefure, via = fr x{5} x{6] x{0} xf) x{2] va} = fy 2053 2 =} 3.46 x{0] = 1, x{1] = -L, x{2] = 2, x(3] = 3, x[4] = 0, x{5] = Since G{k] = W*X{k] hence g[n] = x[6]. Thus, aint = fxg x4) x{5} xf0) x0) xa} = (3 001-1 2} N x Not z 3.47 yinJ=x(5n], On st « Therefore, Y(KI= J yin]W2h 5 = ¥ ssa. ~ 0 Let oy eS Now, x[Sn] = xe Dxtmwe™" = ni xx [m]Wy73 - Hence, =0 m0 1 Niset NET Nel N/So1 Y= x, Ly XlmlWays Was = 3p DXlel DY Wes" a=0 m=O m0 a0 Nis-l N NaN pa 3N pg a Since Swi" =] 3" makk+S kA Sk kA: zw eter Thus Yoel = 5 (kd +Xtk + $+ X0k + Bex T+ xtk +). 3. 48 vin] = x{n] + y{n}. Now the DFT of the resl part of the sequence vin] is x(k 3 {vey *(<—k>4}} and the DFT of the imaginary part of the sequence is YtK 1 . 3m V*[<-k>g}}. Therefore, 2 1 1 7_ ts xta={l ain, PHI. $F Zod ~t4+j, ana A A 1 4 vin=[3, 5+), -+id, 345g, § 3-ip. 3.49 v{n} = g{n] + jhfn] = {3+ j2 2+j 1+j 4+j3} 42350 3.51 3.52 Since F! fi Lt 1 pasi27 fossa -j} 2 vow w i dy 24d | B | therefore, batt lp tej 7} 2-5 j]s+ia) | 4-j 1 -j V*{k]=[l0-j7, -j3, -2+j, 4+j] and V*[<4-k>,]=[10-j7, 4+), -2+j, -i3]. Hence, Gik}=4{VIkI+Vfc4—K>y]}=[10, 2+j2, 2, 2-j2} and tak) =5,{VIRI-V lcd Kol} 1+j2, (@) Lat pla] = IDPT of Pik]. Hence poy] pl tot ot 2 pO) tt j=l 3 pRy| aft <1. -1 4 por} Lb = -2+j3] Lo Similarly Ufa] = IDFT of D{k]. Hence, ‘qoy) fl 1 ot iff o7 1 auy|_iji j -1 -jj/04+j09]_|-09 gf2}}"4]i -1 1-17 25 |} 06 f aa} 1 =j -1 jJjlo4=jos} | 0 | Now, P(e!) = p[0]+ pllJe” + pl2]e2” + pi3}e3! and D(ci”) = d[0] + dftJo I” + [2}eF + dp3]eF. (ei) 24307 4 de Pi Die) 1-09 +0679" Therefore X(ei” (b) Here, p{n] = IDFT of Pk] = (0.75 2.25 2.75 3.25} and ‘d{n} = IDFT of D{kj = (1 0.6 0.5 0.4). Peei®) _ 0.754+2.25e7H +2.75e720 43,.25e731 Therefore X(e!*; ee = Dye) 10.6699 +0572 0.409 Nol Nal (eI) = xfnfeH" and Kk) = YY xine” n=O ' Bey . Now {n= ES x{mje22RKm/ My. M M cameo f ES am peaaeom = Sxir-m. mad k=O = 7 ‘Thus &{n] is obtained by shifting x{n] by multiples of M and adding the shifted copies. Since the new sequence is obtained by shifting in multiples of M, hence to recover the original sequence take any M consecutive samples. This would be true only if the shifted copies of x{n] did not overlap with each other, that is, if only if M 2 N. 1 F thus F = NF". Thus ylo) = F( FC FCF txt} =N FAC FIN FF (xln)})}) = NP xf0]. 4323 Fa 19 3.53 yn} = x{n]@n{n} = Yo x{kIbln—k] = Yhlk]x(a—k] = >) bfkIx[n -k}. ko ket Kato _ 23 9 un) = x{n] ata] = YblkIx[< n—k >3¢] = DAKkRL< n—k 39] k=0 Kt Now forn 219, x[<—K>9]=x{n—k]. Thus ufn] = y{n] for 19 vingwee =D) xtanwget = > xiang 0 a=0 a0 Fork 2 N, let k= kg +#N where ky = y. Then, Nel Nel Yk] = YUky +2N} = Sexy) = Yi x{a]Wye = Xiko] = X[< k >]. n=0 a0 (b) Since Y{k] = X[< k >7] fork =0, 1, 2, 20, a sketch of ¥[K] is thus as shown below. 4 3 2 Y(k} 1 012345678 20 x 3.87 xo{n]=x[2n+1]+x{2n}, x(n] =x(2n +1] (20), y,En] = yl20 + 1}+yl2n},and N ~ yolnl = yl2n+1]-yf2n}, OSns> Sirce x[n] and y{n] are real, symmetric sequences, it 44 OE |follows that xo{n] and yo{n] are real, symmetric sequences, and x;{n] and y\{n} are real, anti- symmetric sequences. Now consider, the (N/2)-length sequence ufn] = xo{n]+y,{n]+ j(x,[n]+y{n}). Its conjugate sequence is given by u*[n] = xoln] +y;La]- j(xy{n]+ yoln]). Next we observe that Ul n> yp9]= Xgl ya] Yl —M ya} I(myl<—9>yy2]#¥ol<— Py a]) = ko{n]—y,{n]+j(—x,n]+yo{n}}- Its conjugate sequence is given by uA[<—n>y/91= xolal-y (al j(-xjln}+y,{n))- By adding the last 4 sequences we get ‘4xo{n] = nln} +u*[n]+ul<—n >,,/2)+u*l<—n >y4/2) From Table 3.6, if U{k] = DFT(ufn]}, then U*[<-k>y,2]=DFT{u*(n)}, U*[k]=DFT{u"[<-n>y/9]}. and Ul<-k>y/q]= DFT (ul<—n>y,9]). Thus, Kolk= DE (Rg tal) = (UIKI+ UE yg) FUCK > 9+ UK) Stary, J4x,{n] = ufn]—u* [a] -uf<—n>y 9] +U*[<-n>y/]- Hence, ylUlRI-U Eko yy] Ul -k> 2] UEKI}. Likewise, 4y,{0] = uln}—ul<-n>y,.]+0*{n]-u*[<—2>,,/2]- Thus. Yylk]= DFT ty {0)} = 4 (UlKI-Ul< Kk >y 3) U*E<—K>y/9]- U*E)} J4yoln] = uln]+ul<-n>y,]-u*[n]-u*[<-n>y,2]- Hence, Yo{k]= DFTIy (nl) =z (UIKI + Ul<—K>yq]~U "Lek >y 21 U*tK)} X,(k] = DFT{x,[n]} = nally, 2.58 glol= Ha(2n}+a(0n +1), Mo}={ot2m)-x0n +, 0y 21+ (1-WyDHE yah OSKSN-1, 3.59 g{o}=a,x(2n]+a,x{2n+1] and h{n}= ax(2n]+ayx(2n-+I], with aja, #aa5. Solving for x{2n] anh ~a,e(n) +h and x{2nel),, we get x{2n) = 2220)~82MO) os prone 1p = een taN) rh erefore 384-8283 a4 9295 “1 Nl z X()= Yi xnz-* = Yo xf2n}e- +27 SP xan eye ™ a0 m0 m0Ba mi z z YS stn} +(-2, +427) Y hlnl2-®. Hence, m0 (ay —asWa™)Gle k>y,g]+(-a, +8, Wy™)Gly/2] OSE SN— eee aaa xtkj- 819g 79283, 3.60 G(z) has poles at z = ~1+tj 2, and 2+j 3. Hence there are three possible ROCs. @ Ry: [2] 3 VS. The inverse = trancform gfn] in thie cace ic a left-sided cequence. (ii) Ry: V3 0.3 ne te The ROC of Xj (z) is given by Ry: |z| > 0.3 bi>os 5S (ii) x5[n] =(0.2)"pfn-5]. Thus, X5(2) Soares 5-5 = 0272" 50.2. The ROC of X3(z) is given by Ky: [2] > 0.2. © 1-021" (iv) xg{n] =(-0.2)"H[-n=1]. Thus, X= D xgla2 The ROC of X4(2) ie given by Ky: [2] < 0.2 mal (b) (i) Now, the ROC of X;(2) is given by %: [z] > 0.3 and the ROC of X2(2) is given by Ry: [2] > 0.5. Hence, the ROC of ¥(2) is given by Ry Ry = Rg: |z| > 0.5 (ii) The ROC of ¥2{2) is given by Ry ORs = Rj: [2] > 0.3. j (iii) The ROC of ¥3(2) is given by Ry ORs = BW. Hence, the z-transform of the sequence. ya{a] does not converge anywhere in the z-plane. F (iv) The ROC of ¥4(2) is given by Ky =z: fz] > 0.5. (¥) The ROC of Ys(2) is given by Nyy =. Hence, the z-transform of the sequence = ysln] does not converge anywhere in the z-plane. 463.62 (i) Zistn}) = 8(alz [0] = I, which converges everywhere in the z-plane. ~ Gi) See Example 3.15. Zola) = Leta (iv) See Example 3.29. (¥) x{n} = £* sin(wyn)pfn} = 5 citer eo 0" ufn}. Using the results of (iii) and the linearity property of the z-transform we obtain 1/ 1 7 jr ) fsin(og)2 v fal> ie = 2rcos(Wo yet +z Set = Sera ‘The ROC of X1(2) is [2] >a which includes the unit circle. On the unit circle z=e!. ‘Thus which is the same as the DTFT of x(n). pe Xy(0!)= Xi @racit "gw (b) xp{n]=a"yf-n], fol>1. Thus, X= SY) xgfnke™ . L ye De aig Het ‘The ROC of X(z) is 1, the ROC includes the unit circle. On the unit L a Soru-ne* circle z=. Thus, X,(e!)=X,(2),-.i0 Teele (ce) x3{n]=al"ly{n} = a"u{[n] which is the same sequence as that in part (a). Hence X3(z) = X,(2)= ))xy{ale™ cor Ie [njsM, i on clecuhere, Therefore, X,(2) a M eM (= oe Ma-M. Met, M+) Br Se ae Oe racer) One dere + dete d-o 7 t=az7 n=M neoM ROC of X4(z) is the entire z-plane, Thus, the DTFT of xg{n] is given by 47X,(0!)=X,¢a| hacia = i (©) xg{o]=ap{n+3}, fal<1. Thus, oe ta X= “= Bye" = YyaPuin+3} = Since the ROC of Xs(z) includes the unit circle thus X5(e!)=Xs(2)| ig = (© xgla}=nefn] Now, XQ)= Sy na%utn}e™ d Now differentiating X}(z) (from part (a) ) we get -l_ Thus, AX eo Bact 2" =-X@). d 1 aX, Hence X,(2)= Ro 7 1, -Nsnsn, 3.64 (a) vinl={g elsewhere. Nga z-@NeD) NHL yo= Sota ROC of ¥4(2) is the entire z-plane except z jott_gciotsen sf) 1 gaia = sin(@/2) Since the ROC includes the unit circle, we get ye”) Hl sn (b) yginl=ftoye NSASN 0, elsewhere 1, -N/2 aT in TaN RAN =, ; TIGITAN =a RTIN AT, The ROC of Y3(z) is the entire z-plane except z= 0. The DTFY of y3[n] 1s theretore given (e-i#NHVDVN ja +1/2) _ gjntW41/2)/2Ng—jH1/2)) by Ys(e!)= 502), io = aT RTAN Cg RTA TOT 2 e e e e ) 1 (eSB FH/2)72Nf41/2)_g-j41/2)/2Ng-JO4N41/2)) 2 (EIDIIGIRTAN — <=iRTAN,=j02) 1 80((o-RN+9) a sino+ 2nd sof 357) 2 sin (w- F9] 2 3.65 (a) Y\(2)= bi<+. Consider X,(2)= kel>4, From Table 3.8, the =4z" 4 inverse z-transform x,[n] of X,(z) is given by x{n]=4"y{n]. From the time-reversal property of the z-transform given in Table 3.9, the inverse z-transform y(n] of ¥ (2) is thus given by y lo) =x,[-n] = 4-H fn] (b) Y5(2) 115 thus, from Table 3.8 and the arity property of the z-transform given in Table 3.9, we obtain asf ef fo - Then, Let reos(@)=4 and rsin(®) in yafn} we get yaln] = (rc0s(6)— jrsin(®))(rcos(6) + jrsin(®)" w{n] + (reos(@) + jrsin(@)\r cos(@}— jrsin(8))" W[n] = (cele i084 42#hgi96-i00 en] = 2 (2-(400/2) ge DE ( + atttei8e-i0® in} = 2 (2 eoe( 59m. J Md tan(®) =I, oF, O=F. Substituting these (©) Yx@)= +t 3 |z|>$.Since the ROC is l2l>4. hence y3ln} Fata (2) weot~3a. (a) Y,@)= $0.5. Therefore, ‘We next determine the inverse z-transform xy{n] by the long division approach: -I X (2) = Sa 4 1.25271 40.8125 2? +0.45312 9 +0.238324 + 107527" + 01052 Therefore, {xy{n]}={1 1.25 0.8125 0.4531 0.2388 +. (©) We first determine the inverse z-transform xe{n] by a partial-fraction expansion 2 4 8 X.(2)= =4 » OS< oe (z+0.5)(2+1) “Teel Troset t Since the ROC is an annular strip, xe{n] is a two-sided sequence. The second term in the partial fraction expansion has an ROC given by [2|<1, and has an inverse z-transform which is a left- 50sided sequence —4(-1)"{-n—1], whereas, the last term has an ROC given by [2|>0.5, and has an inverse z-transform —8(~0.5)"1{n] which is a right-sided sequence. Therefore, x_{n]=45{n] - 4(—1)"w{-n -1]-8(-0.5)" fn]. We next determine the inverse z-transform xc{n] by the long division approach. To this end we need to develop a power series expansion of 4(1 + 27!) in a positive power of z, and, a power series expansion of 8/(1 + 0.527!) in a positive power of z“!, Now, Az 1+z 38-42 ' 422 eet tat las lee a es 4 = 421-2422 -23 +24 —--) = 42-42? +423 — 424 +425 —---, whereas, -2 +097 -vzoz §+---. Therefore, fx.{n]}= a 1 8) X4@=— Fore anus dcose cba hasoe @) Xs@= Fa HIG one O05 NM —022 N+ 0.6 We first determine the inverse z-transform xg{n] by a partial-fraction expansion 02281 0.0918 | 0.7576 | OLN X,@= - 07576, ONL >06. 6 Get 1-05e7 U-0521 1-022" 4 xq{n] = 0.2231(-0.6)" fn] + 0.1111(0.2)" pin] -0.0918(0. 5)" y{n] + 0.7576(n +1)(0.5)"p[n +1] We next determine the inverse z-transform »a[n} by the long division approach. af i =— > ——_ -__,,___ _,, _, (2=0.5)P(2-0.2)(2 + 0.6) 21 = 0.272? +0.222 i = 140.627! +0.632-? +0.3227 +0.26012-4 +0.121927 +» Therefore, (xgfn]}=(| 06 0.63 0.32 02601 0.1219 --} X4@) L ot 2jsin() {1- rez 8 sin((0+198) sing) Mh 3.68 Gt2)= Sralnl2™* witha ROC given by Hy (a) Therefore G*(z)= Sig*[n](2*)"" and G*(2*) Thus the z-transform of g*[n] is G*(z*).(b) Replace n by -m in the summation. This leads to G(z)= )’ g{-m]z™. Therefore G/2)= Sigi-m)2-®. Ths the transform of en] is GI/2). Note that since x has heen replaced by 1/z, the ROC of G(1/z) will be 1/ Re. (e) Let yfn]=ag{n]+Bh{n]. Then, ¥¢@)= Y(agin]+Bhfn})2-" =a. gin}z"™ +B), hfn}e™ = aG)+BHE) In this Case Y(z) will converge wherever both G(z) and H(z) converge. Thus the ROC of Y(z) is Re A Ry, where is R, the ROC of G(2) and R,, is the ROC of H(z). (a) yen} = gin—no}. Hence ve = Sytye = Yyxtu-ughe = Yatug-mro? 0S atmj2™ =2-0G 2), te In this case the ROC of Y(2) is the same as that of G(z) except for the possible addition or elimination of the point z = 0 or z = e (due to the factor 2"). Lelalea)* = Ge/a). (e) y[nJ=a%gin}. Hence, ¥(z) Dyer ‘The ROC of ¥(2) is jo, (© yla] = ngfn}. Hence Y(z) =) ng{n}z = Veale dw) __ n=l az) _ > 1 Now Ge= Yale Thus, = Zor = Se Dato : Thus Ye) = - 22S a (8) yin} = etn] @aln] =} elk1n{n-K]. Hence, [ Sees > atk) Spta-ig2" kee Keene | In this case also Y(z) will converge wherever both H(z) and G(z) converge. Thus ROC of Y(z) is RyRy (h) yln] = g{nJh{n]. Hence, ¥(@2) =} eln}h{n]2™. From Eq. (3.107), 521 ne y 1 n-igy |e" ato) ==Lfav""lav. Thus, Ye) Sino foo ws anid ~ “foo( 3 Senor ore fowmar 1 w Sst “Ini af > *injv" va jfoones Ivey lv. 3.69 (a) Expnading in a power series we get X\(2)= a Ye 2, z|>1. " ‘Thus, xy{ay = {CDS ifn =3kand n20, oe { 0, elsewhere Using partial fraction, we get i 1 X\@=—= ‘Therefore, OTe ay ala} 1)" uf} + 3.cos(ne/ 30] xatol = Lentutnye §($-5°R) plod 4 ie! 3yfn} = DP an] + Fei"! Syn] Thus xj{n] = ee if n=3kandn20, 0, elsewhere. () Expnading in a power series we get X3(2)= z|>1 1, ifn=2k andn20, os ctel= (5 omer Using partial fraction, we get X(2)=—ty=—2,+—2,. Therefore, 1-2? teat 1-2t ola] = ula] + 5(-DMala] _ fy ifn=2kandn20, ‘Thus, xafn) = a elsewhere. 3.70 (a) X,(z)=log(t-a2""), _fe]>|o. Expanding Jog(I-c2~') in a power series we get Thus, x,{n] (b) X,(2)=log(1-Bz), }<1/B}. Expanding tog(I-Bz) in a power series we getThus x9{n]=$—p[-n-1} @+h = 371 HO) = 5 e+ 005) Since the inverse z-transform h{n] of H(z) is a right-sided sequence, the ROC of H(z) is given by |z|> 0.5. By partial fraction expansion we get H@) ~ -8+ +———_ . Thus. bn] = -88{n) +4(0.5)" ulin] + (-0.25)" ula, 1-0.5z 22k>f20 2 4g gnlya! 42-9 a +01 #7 ag! +25, 22g alent 2 237 +25!25 +29 on wp +z; (P+alzg JO zzh aay! (et ay" eq! +27? +29! gm! (5! 2y eq! +25? +25) ise a+ (co Gl -2h@5! +2; +29!