Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
2.11 LE't us denote the signal in question by g(t) and its energy by E1/' For parts (a> and (b)
(c) Efl ;:" sin
2
t at =  dt   cos 2t dt = 71' + 0 = 11'
1
·" 1 1
4
" 1 1
4
"
k 2 k 2 k
(d)
1
2" [112.T 11
2
.. ]
ElI = 0 (2 sin t)2 cit = 4 2' 0 tit  2' 0 cos 2t dt = 4[11' + OJ = 471'
Sign change and time shift do not affect the signal energy. Doubling the signal quadruples its energy. In the
same way we can show that the t'nergy of is k
2
/::/I'
2.12 <al J:.T == J:(1)2"t = 2. Ell = J;(1)2dt + J1
2
(1)2dt =:2
2.13
1
"12 f,,./2 f2
W
= 0 (2)2dt + } .. /2 (O)2dt + }:>,,,,/'2 (_1)2
111
= 411'
Similally. we can show that E:c = 411' Therefore E;r±1I = E" + E". We are tempted to conclude that =
Er  1::. in general. Let us see.
in general EZ±!I :f; Ez + Ell
1 lTD c21TO
P
,I
= T, c
2
cos
2
(wot + 9) dt = 2'l II + cos (
2w
o' + 29)] dt
o 0 0 0
c
2
I
T
'' ] C
2
C
2
= 2To 0 til ... · 0 co:! (
2w
ot + 29) dt = 2To [To + 01 = T
2.14 This problem is identical 10 Example 2.2b. excE'pt that WI <f.;..I2. In thil!i case. the third intt'gral ill Pg (see p. 19
is not zero. This integral is given by
2.15
2.16
2.11
Therefore
1 /2
P
q
= 4' _.2(1) r/t = 64/7
2
(b) P211 = i 12 (
2t
:l):2dt = 4\64/7) = 256/7
112 :l:2 :2
(c) Peg =  (ct) dt = 64!: /7
4 2
Sign change of a signal does not affect its power. Multiplication of a signal by a constant (. increases the power
by a fartor ,.2.
<a>
(b)
1 J" 1 11</2
P
II
=  VI2(t) ,it =  dt = 0.5
2". _or 2". 1</2
(c) Pg ==  dt ==  dt = 1
1 jTOIZ 1 (Tu/2
To To/2 To JToI2
(d)
1 j:.! .,
PIJ::  (±lrdt = 1
4 2
(e)
.p _ 1.. [2fr tit =!
g  2"..10 2r. 3
ThE' integrals of the crossproduct terms (when k =1= ,.) are finite because the integrands are periodic signals
(mnde lip of sinusoids). These terms. when divided by T  00. yield zerO. The remaining terms (k == I') yield
2.18 (a) Power of a sinusoid of amplitude (' is ('2/'J. (Eq. (2.6a)j regardless of its frequency (:..I =1= 0) and phase.
Therpforc. in this r.ase P = (10)2/2 == 50.
(b) Power of a sum of sinusoids is equal to t.he sum of the powers of the sinusoids (Eq. (2.6b)!. Therefore. in
this case P = + == 178.
(c) (10 + 2 sin 3t) cos lOt = lOcos lOt ... sin 13t  sin 3/. Hence from Eq. ('l.6b) P = + i == 51.
(d) lOcos 5tcos lOt = 5(eos 5t + cos 15t. Hence from Eq. (2.6b) P = + ¥ = 25.
(c) JOsin 5tcos lOt = 5(sin 15t  sin 5t. Hence from Eq. (2.6b) P = ¥ + (_g)2 = 25.
(f) 1'.1<1' cos ..... ot =! [1'.1(0+"'0)1 + ri(a""O)I]. Using the result in Prob. 2.17. we obtain P = 0/4) + (1/4) = 112.
2.21 For a rcai 1'1
at 2 2eal
J
'" <><
Ell == .. ,..,,(r ) dt = J.., r. ,It = 00
TI'J TI2
P I
. I! (ol}2 I \. 1! 2eal d
g = lin  I' , t = 1m  r t = ex·
T",, T T/2 T,,,, T T12
For imaginal)" fl. let 1'1 = .i:r. Then
1 JT
I
2 . 1 fT!2
P
g
= IlimT_", (,.)%1)(,.1%1) dt == lim  tit == 1
T T/2 T",. T . T/2
2
2.31
•
A
t+
"
t ...
• j.
s
t
:!a
,.... .r::::
9·(\)
0
a
,
'1 ...
3&J
't ..
1
Fic. S2.32
Clearly. if 0 is real. (,,al is neithf!r energy not power signal. However. if a is imaginary, it is a power signal with
powcr 1.
92(t) == .q(t  1) + 9;{t  1). 9:1(t) = g(t  1) + 91(1 + 1). = .q(t  0.5) + ql(t.;. 0.5)
ThE' !>ignal q!)(t) can be obtained by (i) delaying g(t) by 1 seC'ond (replace t with t  I J. (ii) then time·expanding
by >I. faclor 2 (replan' , with t:'2). (iii) then multiply wit.h 1..'j. Thus 95(t) =  1).
2.32 All the signals are shown in Fig. S2.3·2.
2.33 All the signals are shown ill Fig. 52.3·3
2.34
FiC. S2.33
EIJ = = 1'.: ,l(t)lit = Eg • E'('I) = !:lg(_t)]2
dt
:: 1: i(:r.}tiX = Eg
Eg(tT) =  T)j
2
df = !:l(:r;)tlT == EI/' Eg(al):: 1:[q(ot)]2
tlt
=; 1: g2(:r)II:r: = r.g/O
2.41 Lsing the fact that q(.T)"'(.T) = g(O)/)(x). WP. have
(8) (l (b) i"(..:) (c) !"'(t) (d) ill(t  1) (e) 2:.1:\/)(": + 3) (f) k6(w) (use L' H6pita\'s rille)
2.42 In Thcse problems I'emembr.r that impulse "'(3:) is located at x = O. Thus. an impulse "(t  :) is located at 'T = f.
and so on.
(a) The impulse is locaTed at .. = t and g(T) at T = t is q(t). Therefore
3
1: g(1") ... (t  1") 1/1" = .q(t)
(b) The impulse "'(1") is at 1" = 0 and R(t  T) at 1" = 0 iii get). Therefore
1: "(T)R(t  T) 1/7' = R(t)
Vp;ing similar arguments. we obtain
(e) 1 (d) 0 (e) r!l (f) 5 (K) g(I) (h) _p'
2
2.43 Lelting nt == .1'. we obtain (for (J > 0)
lP(t)li(at) dt =  tIJ(!.)b(x)dx = .p(0)
1
_.,., a _."" a a
Similarly for (J < 0, we show that this integral is !tIJ(O). Therefore
Therefore
1 1 J ....
_."" lP(t)"(nl) tit = j;jtIJ(O) = j;;i _"" tIJ(t)6(t} dt
h(n.t) = 1..,"(1)
\a
2.51 Tri\'ial Take the derivative of le:
2
",ith respect to (; and equate it to zero.
2.52 (a) [n this case E% = fo1 dt == 1. and
1 '10
1
111
(. =  .q(I):r(I) rlt =  t dt = 0.5
E% • 0 ) 0
(b) Thus.qlt) 0.5.T(/). and the error ret) = t  0.5 o\'er (0 !S t !S 1). and zero outside this inler\'ui Also EI/
and E. (t he of I he elTor j are
The error (t  0.5) is orthogonal to .r(/) hecause
11 (/ 0.5)(I)dt = 0
t hnt EI/ = ,.
2
E", + E
r
. To t'xplain these results in terms of vector cOllcepts we observe from Fig. 2.15
thaI the error \'ector e is orthogonal to the component (?t. Because of this ort.hogonality. the lengthsquare of
g lenergy of gU)l is equal to the sum of the square of the lengths of ex and e [sum of the energies of and
r(t)J.
2.53 In this case EI/ = fol .q2(t) tit = .Fo
1
,2 tit = 1/:3. and
111 11
.r(t).q(t)dl:::3 f(it=1.5
EI/ 0 0
Thus. :r(I) 1.59(1). and the error r(1} = :ret)  1.59(1) = 1  1.51 over (0 f 1). and zero outside this
intelv:.J. Also (the energy of the error) is E" = Jo
1
(1  1.5t)2 dt = 1/4.
2.54 (a) In this case E", = f:sin227rldt = 0.5, and
1 11 1 11
(. = .q(t):r(f)dt = 0 tsin 27rtdt = 1/7r
E", 0 .5 0
(b) Thlls. 9(t) ::::: (1/lr):rU). and the error I'(t) = t + (l/lr) sin 27rt over (0 t 1). and zero outside this
interval. Also E!/ and (the energy of the error) are
4
Eg = t g2(f) (It = f' f2 Jt = 1/3 a.nd = f' [t  (1/7r) sin 27rt)2 df =  2!2
Jo.lo Jo
The errOl It + (1/7r) sin 21l'II is orthogonal to .r(t) because
11 sin 27rt!t + (l/1l') sin 27rf)dt = 0
Note that Eg = (.:! Ex + Ee. To explain these results in terms of vector concepts we observe from Fig. 2.15 that
the errOl vector e is orthogonal to the component ex. Because of this orthogonality, the length of f [I'nergy of
.q(t}l is equal to the sum of the square of the lengths of (!x and e [sum of the energies of ex(t) and dt)j.
2.55 (a> If :ref) and U(t) are ort.hogonal. then we can show the energy of x(t) ± U(t) is E", + EI/'
1:.IX(t) ± 1/(t)1
2
elt = 1:r(t)1
2
dt + 1: 1,,(t)1
2
dt ± 1: x(t),/(t) dt ± I: XO(t)1/(t) dt (1)
= f: IX(f)I' dt + f: 11I(t)12 dt (2)
The last rf'Sult follows from the fact. that because of orthogonality, the two integrals of the ('ross products
.r(f)lI°(t) and .r·(t)U(f) are zero [sP.e Eq. (2.40)J. Thus the energy of x(t) + u(t) is equal to t.hat of .T(t)  y(t) if
:r(f} and !I(t) arE' orthogonal.
(b) Csing similar argument. Il,e can show that the ellergy of CIT(t) + is equal to that of "I X(I)  ('211(t) if
1'(t) and 1I(t) are orthogonal. This energy is given by h1
2
E", + 1('21
2
E
II
,
(e) If zit) = 1·(t) ± lI(t). then it follows from Eq, (1) in t.he above derivation that
2.56 gl (2. 1). K2( 1. 2). g:l(O. 2). K4(1.2). Ks(2. I), and ge(3,0). Fl'om Fig. 52.56. we sec that pairs (g3. gel·
(g 1. g4) and (g2. al E' orr hogonal. We can verify this also analytically.
IPg. S:U6
g!l . 1;6 = (0 x 3) + (2 x 0) = 0
gl . 14 = (2 xl) + (1 x 2) = 0
I' . g5 = (1 x 2) + (2 xl) = 0
\".'1' can show that the cOl'I'esponding signal pairs are also orthogonal.
1': In(t)Ilf;(t)rit = 1:[:r2(f)J!3X1(t)Jdl = 0
= ['l:rl(f)  T2(t)J[r.l(t) + 2X2(t)]dt = 0
I: .Q2(1)IlS(t) (It == I:[Tl(t) + 2T2 (t)J!2.T I (f) + :r2(t)ltlt = 0
2.61
In deriving these l'es'IItS, we used the fact that J::' = L: = 1 and L::rl (1)1:2(t) dt = 0
(1)
(3)
We shall compute ,." using Eq. (2.48) for each of the 4 cases. Let us first comput.e the energies of all the signals.
1Isin
2
211'tdt=0.5
In the same way we find Ell! = E,n = EII3 = Ell. = 0.5.
t:sing Eq. (2.48), the correlation coefficients fol' four cases are found 8l:i
J I( r
Sin2
11'tsin411'tdt=0 (2) I tCsin21rIHSin211't)dt=1
(0.5) ./0 V(0.5)(0.&j 10
J 1 (I 0.707 sin 211'tlif = 0 (4) J I [10.:1 0.707 sin 21f'tdt _ t O.707sin 211'tdt] = 1.414/11'
(0.5)(0.5) 10 0 10.5
Signals 3'(t) and .Q2U) provide the maximum protection against noise.
2.81 Here To = 2. so that ..... 0 = 211'/2 = 11'. and
""
get) = a·o + L 11" cosn.1I" + b
n
sin n1l't
"=1
where
2/1 4( 1'"
an =? f
2
cosn1l'frif = .
 I 1r n
I," = t t
2
sinn1l'tdt::: 0
 11
Tht'refole
1 4 "". (1)"
qlt) =  +   (,051111't
. 3 r,2 n2
 1 t 1
"=1
Figult' 52.81 shows q(f) = f2 for all f and the conesponding Fourier series l'epreS('nting q(t) O\'er (1. 1).
1'( ... )
.:.c.. /), fl..L':=::.. L> ,"'"
The power of g(t) is
Pg =  t
4
dt = _
1/
1
1
2 1 5
Moreover. fl'om Parseva!'s theorem [Eq. (2.90)J
2 (1)2 1 (4(_1)")2 1 8 1 1 8 1
P
g
= Co + T = 3 + 2 L = 9 + ;4 n
4
= 9 + 90 = 5
I "=1 ,,=1
(b) If the JVtt'rm Fourier !';t'ries is denoted by :r:(I). then
!VI
I 4 L (1)"
3'(1) =  T' "2 cosn1l't
3 11' n
'l=)
Tht' powel' Pz is required to be 99%P
g
= 0.198. Therefol't'
Nl
1 8 L 1
Pr =  + "'7 4' = 0.198
!) II
n=1
6
 
._
"
For N = l. = 0.1111: fOI· N :: 2. Pz :: 0.19323. For N = 3. = 0.19837, which is greater than 0.198.
Thus . .IV = 3.
2.82 Here To = 2w. so that Wo :: 2w /2w = 1. and
.....,
get) = 0.0 + Lan cosnt + b
n
sinn.t
,,:1
where
1 j"
0.0 = 2lf tilt = 0,
..
2 j'"
On = 211' tcosfltdt = 0,
,.
2 jf<. 2(I)n+1
bn =  t Sin TI t dt = '
211' _" n
Therefore
q(t) = 2(1) .. +1 !.sinfl1.
Ln.
.... I
Figure 52.82 shows get) = t for all t and the corrp.sponding Fourier series to represent get) over (w, 11').
Fig. S2.82
The power of qft) is
1 /.. 2 ;r2
Pg :::  (t) dt = 
2r. 3
f<
flOm Parse\,al's Iheorem [Eq. (2.90)J
(b) If the StPlm FOllrifn series is denoted by .r(t). then
N
:r.(t) = 2( 1 ),,+1 L * sin fl71't
" ... \
The power P", is requirt'd to be O.9D x * = 0.311'2. Therefore
N
1",4 .J
P", = 2' L. ;;2 = 0.311'"
n_l
For.\' = 1. p", = 2; for S = 2. P", = 2.5. for N = 5, = 2.927. which is less than 0.371'2. For N = 6. P", =
2.9k25. which is greater than 0.311'2. Thus. N = 6.
2.S3 R('l'all that
1 f
Ta/2
110 = T. .q(l.) dt
o To.'2
2 f
TON
Il
n
= T. .q(t) cos n.Jot dt
o To/2
2 jr
on
I, .. =  g(/)sin fl...:otdt
To Toi2
..
,

(la)
(1b)
(lr.)
 ... ..  .. ..
Recall also that cos Tl..Jot is an e\'en function and sin nwot is an odd function of t. If get) is an even function of
t. then .q(l) cos Tl""'o' is also an even function and g(t) sin n .... ot is an odd function of t. Therefore (see hint)
2 (To/2
no = To Jo .q(t) df (2a)
4 lTO/2
n. .. =  g(f)cos'l7;.!otdt
To.o
(2b)
lIn = 0 (2r.)
Similarly. if .q(t) is an odd function of t, then g(t) cos nwof. is an odd function of t and get) sin TlWot is an even
function of f. Therefore
110 = n'n = 0
(3a)
41
TO/2
b .. = To 9(t) sin lI;.)ot dt
o 0
(3b)
Observe that. because of symmetry. the integration required to compute the coefficients need be performed over
only half t he period.
2.84 (a) To = 4 . ..Jo = ¥o = Because of even symmetry. all sine terms are zero.
"'"
g(t) = 110 + L 11" cos (n21r t)
"",1
IJo = 0 (by inspection)
nn = [1
1
cos ('I727r f) df 1
2
cos C; f) Ilf] = sin
Therefote. the FourieI series fOI l1(t) is
. 4 ( trt 1 31rt 1 51!'t 1 77rt )
q{t) =  cos  cos + cos  cos + ...
. r. 2 3 2 5 2 7 2
Here I," = O. and wt' al\ow C" to take negati\'e values. Figure S2.8·4a shows tht' plot of Cn .
(b) To = JO:r .... ·0 = ¥o = i· Bt'cause ofe\'en symmetry. all the sine terms are zero.
g(t) = no + t n" cos + lin sin
n=1
no = (by inspection)
a" =..!... {" cos (!2. t ) tit = ..!.. sin (!!:t) I" = sin
107r Jtr 5 51r '17 5 _if 7m 5
lin =..!... {" sin (!!:t) tit = 0 (integrand is an odd fUllction of t)
101r Jtr 5
Hele btl = O. ano we allow Cn to tllke negative values. Note that C
n
= /1 .. for n = 0, 1. 2. 3, . ". Figure
shows t he plot of en.
(c) To = 27r. ""'0::; 1.
and
", .
. q(f) = no + 2: aft cos nt + lin sin 'I7t with 110 = 0.5 (by inspection)
,,::1
11
2
" t
aft =  2 cos n.t df = O.
11' 0 11'
11
2
" t 1
II .. =  sinntdt =
1!' 0 21r 1rn
() 0
 1(. 1'2
1
'3
1
'4 )
g t = .:>  ; Sin t + 2 S1I1 t + 3 Sin t + 4 sm t + ...
= 0.5 + ;. [cos (t + + cos (21 + + cos (3t + i) + ' .. J
8

c,. r
" ...
t
"
..
,
, (b)
.
\
"'" ea.Ieo
\
,
..
of
.,
/AJ.,
fI
, 11. I':;' I. (
c.t ·"1,
I I I t
•
(,e)
I
., ..
5
(,
w..,..
w ..
t
.,
t!
G.
V
.
I
(d ')
•
2. ..
lao
ct
Sf
JtI
t\
()
..,.
(e)
,
yo
cJ ...
tJ
;J
qC'
• •
,
FiS. S2.84
ThE' reason for vanishing of the cosines terms is that when 0.5 (the dc component) is subtracted from n(t). the
remaining function has odd symmetry. Hence. the Fourier series would contain de and sine terms only. Figt.:re
S:2.8·4c shows the plot of en and 8n.
(d) To = 1r . .o.IO = 2 and .q(t) =
flO :: 0 (by inspeC'tion).
lin = 0 (71 > 0) because of odd symmetry.
41"14 4 . 2 (2 . 71'n 1r1l)
lIn =  tsm2ntdt =  sm   eos
1r 0 71' 1rn 71'n 2 2
( )
4'?t 1. 4 4. 6 1. 8
.q f ;; :; sm _ ....  Sin t   sm t   Sin t + ...
7r 7r 911'2 271'
= ...,.. cos 2t   +  ('os 4t   +  cos 61 +  +  cos 8t +  + ...
. 4 ( 1r) 1 ( 10) 4 ( 71') 1 ( 7r)
2 7r 2 911'2 2 1r 2
2.85
Figure S2.84d shows the plOl of C" and fin.
(e) To '"" 3. "'0 = 2'71"/3.
IlO =! t t tit = !
3.10 6
211 27' 71" 3 2'71"71 2Jrn 21r"
an =  t cos tlit == [cos  +  sin   IJ
3 0 3 21\"
2
71
2
3 3 3
211 . 2mr 3.. 27T7I 27Tn 2'71""
lin =  t 9111 tdf = Ism    COS J
3 0 3 2'71"2,,2 3 3 3
Therefore Co = i and
C
3 [ 2 411"2,,2 2 271"71 47T1,. 2ro71
"= 271"2,,2 \ + 9  cosT  TSlnT
(
1lm cos l!.!l  sin )
1 I 3 a
and fin = t.an 21!.!! fu.
cos :r + 3 sm :l  1 ,
(f) To = 6 . ...10 = 7T/3. "0 = 0.5 (by inspection). Even symmetry; b .. :0: O.
=!
3 0 3 11 3
6 [mr 27171"]
= 7I"2n2 cos T  cos
3
. _ 6 ( 71" 2 1 5r. 1 )
q I t I D.C! +  cos  t   cos rot +  cos  t T  cos  t + ...
, 7r
2
3 9 25 3 49 3
th?! even hllrmonics vanish. The reason is that if t.he tit: (0,5) is subtracted from .fI(f), the
function has halfwan' symmt'try. (See Prob, 2.86). Figurc S2,84f shows the plot of C",
An ('\"en fUIICI ion g.1 t) ann all odd (unct ion g,,(I) have the property that
= q.(I) and /lo{f) = go( f)
En'l sigul11 .q( I) ran be expressed as a sum of cven and odd components bE'cause
g(t) =! [a(t) ..... q(f)J + [g(t)  g(t))
., , .. '
.... ...
odd
(1)
From the definitions in Eq, (1). it can be seen that the first component on the righthand side is an even
function. while the second compont'nt is odd. This is readily seen from the fact that replacing t by t in the
first component yiE'lds thE' same function. The same manf'uver in the serond component yields the nE'gative of
that component.
To find the odd and the (!ven components of .q(t) = l1(t). \\'e have
.q(t) = g.(t) + .!lo(t)
",helo:' [from Eq. (I)J
and
1
qo(l) = ! If/(tl  fI(t)! = 2sgn(l)
The even and odd components of the signal '1 (t) are shl)wn in Fig. S2.85a.
Similarly. to find the odd and the even components of ,q{I) = eClt,,(t). we have
.q(t) = .q. (t) + 90(#)
10
G.s
c
I
Fig. S2.8li
floft) = [,,all/(t)  r. .. t,,(t)]
ThE' e,"PIl and odd l"omponents of the signal ,,al,,(t) are in Fig. 52.8ab.
For fI(/) = ,..". we have
"jt :;:: .9.(t) + .9o(t)
nnd
2.86 (a) For half wa\'e symmetry
(
. TO)
nU) = g , "2
and
9
c
(t)
elS
0
c·s
2 l
TO
2 l
TO
/2 lTO
and fI" = r: .9(t)COS71...lot dt = 7f g(t)C'os71...lotdt + fI(t) cos '1wiOt dt
o 0 0 0 To/2
Let :r = t  To/2 in the second integral. This gives
? [l
TO/2
l
TO/2
. 7i ) ('Ii)]
tl
n
= ;0 0 gU) cos 11...10t. dt + 0 .9 (:r. + 2
0
cos 11...10 X + 2
0
tl:r
2 [l
TO/2
l
TO/2
]
:;:: To 0 g(t) COSll...lot dt + 0 .9(xH  cos 11wo:r.J d:r.
4 [(01
2
]
= To Jo g(t) COSTli';Ot dt
In a similar way we can show thal
41
TO/2
/'r= g(f)sinn...Jot,ft
To 0
(b) (i) To:: 8. "':0 = 7' no = 0 (by inspection). Half wave symmetry. Hence
11
4 ( nr. nr. nr. )
= T12r." cos T + T sin T  1
(n odd)
::  slOl
4 (flr.. 1I1r )
TJ21r 2 2 2
(11 odd)
Th('refore
{
('2"  1) 11 = 1,5.9.13 ... ·
an =
(¥+1) n=3.7.11.15, .. ·
Similarly
112 t . 111:' 4 ( fl1r n1r 1171') 4 (fl1r)
b"::  smttlt::  sin  cos = sin 
2 0 2 4 T121r2 2 2 2 fl21r2 2
(fl odd)
and
.q(t) =
n1r . n1r
a" cos "'4 t + b" sm ""4 t
.
(ii) To = 270. ":0 = 1. no:: 0 (hy inspection). Half wave symmetry. Hence
"'"
9(f)= 2: a"COSTlt+iJ"sin1lt
"=1,.1.5 ..
21"
n" =  rtllOcosnttlt
To 0
(O.JCOSflf+T1sinTlt)]"
1I'T1+.1 0
(11 odd)
= [ (0.1)  :1 1001 (0.1)]
T.'T1+.1 n+.
2 "'/10 00465
= + 0.01) (p . 1) = T1 2'+ 0.01
and
'1"
II" =.:. ('t/iOSiIlTlttlt
11' 0
= [ (O.lsinnt _ ncosTlt)] " (T1 odd)
71' " + . 1 0
= 2" «(,"/10 _ I) = 1.461Tl
(fl2 + 0.01) fl2 + om
2.91 (a): 10:: 4 ..... 0:: r./2. Also Do = 0 (by inspection).
1 :1
D
1 1 ··/(",,/2)t It f J(",,/2)t it 2. nr.
"::  ('.. I  (' ( :: 5111
21r 1 1 11'".. 2
In\ 1
(b) To = 1071' ..... 0 = 21!,/10r. = 1/5
.""
.q(t) = L where
1 1" j ( .. nr.) I. (11")
D.,. =  (' J" rtt =   2J Sin  =  Sin :
IOn " 2?1'fl 5 1!'n ;)
.,,=...,.,.,
12
o T I
•• :L
I
(c)
# •• )t\
(J»
,
,
'1>" )
dO'r r, +
.,
Fig. S2.91
"'"
q(t) = Do + L Dn".i"t. where. by inspection Do == 0.5
1 1
2
... t "I j J { i
D" == 2 2 "J dt:: 2 . so that IDnl:: 2 • and LD" =
7r 0 71" 7rn 7rTl T
Tl > 0
n<O
(f)
(d) To == 71". "0 = 2 and D" = 0
...
.q(t) = L D""/2nt.
where
1 j " / ~ 4t _ ~ " t j (2 . 7rTl 11"Tl)
Dn =   ( ~ J dt ==   Sin   cos 
To ",/4 11" 11"n 11"n 2 2
n=.,...
13
(e)
2.92
(e) To = 3 . ...,0 = 1,f .
....
q(f) = 2: DnrJ"Tt. where
111 ,2Jut 3 [_.b.!t (i27rTl ) ]
Dn = 3 0 trJ,t dt = 411'2112 r Jr 3 + 1  1
r._,..".
Therefore
. 47r
2
n:2 27rTl 47rn 27rfl
2+ 2cos sin
9 3 3 3
(
 )
and LDn = tan
1
\"n 2'!n' :i"n:l
cos r + ,sm T  1
(f) To = 6 . ...,0 = 7r/3 Do = 0.5
..,.,
'" J.¥!
g(t} = O.S + L.., Dft ('
3
(.Pro. ,
2
.1
0 1
S 8 fl ...
(Q)
.n:
an
.lP".
•
:>
$"
, 5
..
!
".5
. ,
C
j.
lit> )
4
!1
.3 ..
Fig. 52.92
,q(t) = 3cost + sill (St  i)  2eos (St  j)
n>
For a ("ompact trigonometric form. all terms must have cosine form and amplitudes must be positive. For this
reason. WE:' rewrit.e g(f) as
,q(,) = 3 cos f + cos ( 5f  i  + 2 cos (St  i ,.. )
= 3 ("os f + cos ( St  2;) + 2 cos ( St _ 4;)
Figure 52.92a shows amplitude and phase spect.ra.
(b) By inspection of the trigonometric spect.ra in Fig. 82.92a. we plot the exponential spectra as shown in Fig.
S2.92u. By inspection of exponential spectra in Fig. 52.92a. we obtain
.q(t) = + r
Jt
) + [r1(St1f> + + [r3(8t¥1 + e
J
(8t¥>}
= + arJ¥) r1r>1 + (ri¥) ri8t + + G,..J1f) (,j:lt + (,.j¥) (,1
St
14
2.93 (n)
2.94
11'
get) = 2 + 2cos(2t  71") + cos(3t  2')
= 2  2 cos 2t + sin3t
(b) The exponential spectra are shown in Fig. S2.93.
(c) By inspection of exponential spectra
get) = 2 + [,.(2t.. ) + r.
j
(2t")] + ¥ + (,;(3tt)]
= 2 + 2 cos (2t  "If) + cos (3t  i)
(d) Observe that the two expressions (trigonometric and exponential Fourier series) are equivalent.
7(
Fig. 82.93
1 [f
T
O/2 j
T
O/2 ]
!(t)cosn...Jottltj !(t)sinTl;.Jotdt
o To!2 To/2
If t) is ('\'en. 'he se("ond term on the righthand side is zero because its integrand is an odd function of t.
Hen("e. D" is real. In contrast. if q(t) is odd. the first term on the righthand side is zero because its iT'llegranrl
is an odd function of f. Hence. D" is imaginary.
15
3.1·1
3.1·2
3.13
Chapter 3
C( .... ) = l:.Q(t)r
iW
'dt= l:g(t)coswtdtj 1: g(t)sin .... ·t dt
If 11(1) is an even function of I. lI(t)sin.olt is an odd function of t, and the Sf>cond integral vanishes. Moreover,
q(t) co!'!,.;t is an ('ven function of 1. and the first integral is twice the int.egral over the interval 0 to oc. Thus
when .q(l) is even
C(w) = 21"".Q(t) coswt dt
Similar argument shows that when q(t) is odd
G(w) = 2j 1 ..... q(t)sin.oltdt
If q(l) is also real (in addition to being e\'en), the integral (1) is real. Moreover from (1)
C(,.;):: 21'"" g(t)cos.oItdt = C(.,,:)
(1)
(2)
Hellce C;.·) is real and even fUllction of,.;. Similar argument.s can be used to prove the rest of the properties.
Since !C( •• :): is an even fUllction and 9,,(,.;) is an odd function of w. the integrand in the second integral is an
odd function of ... :. and thf'refore vanishes. Moreover the integrand in the first integral is an even function of.,.;.
and therefore
.q(t) =  IG(w)jcos[wt+9,,(w)]d ... •
11"'"
To 0
For .q(l) == ,,alll(t). G(.,.;) == ;:;t.;. Therefore IG("")I = 1/v'.,.;2 + 4
2
and 9
11
{",,) ==  Hence
" =  cos .01'  tan  d.w
at 1 1"'" 1 [ I (W)]
r. 0 Jw
2
+ n.:! a
Therefore
and
16
3.14 (a>
(b)
3.15 (a)
1
1 /2 4 2 ,... 2 j2",
G( .. ,;) = 4r J"" tit + 2,,1"" dt =  t' .  r.
o I JW
(b)
G(W) = r _!..r·
i
..,' tit + iT !..r.j..,t dt = + w1'sinw1'  I]
lT l' 0 'r 'rW
This result could also be derived by obsen'ing that g(t) is :l.n even function. Therefore from tbe result in Prob.
3.11
3.16 (a)
? 1" 2 G( ..... ) =.:. tcos..,Itdt = ,Icosw'r +w1'sinw1' I}
l' 0 TW


,
2
Flp:. S3.1·6
 2) sin ..,lot + 216101 COS..,Iof
'1ft'
(
+
, o •
(b) The oeri\'ation can be simplified by oLsen'ing tha.t G(;.;) ('all be expressed a sum nf \1\'0 gatl! functions
Cd...;) find fl.S "hown in Fig. S3.16. Therefore
g(#)= dw= t·) /lw+ ('J dw =...;....
1 f' '.." 1 {f2 · ... t fl .... ,} !'in 2t + sin t
2'1f 2 2'1f 2 1 1ft
3.11 <a)
(b)
Because G(w) is even fUllctioll. t he second integral on tbe rightband side vanishes. Also the integrand of the
fi rst term is an even fUllct iOIl. Therefore
17
1
I:'
,
3.18 (a)
()
1 1,",0 ...J .J.. 1 [CostiIJ + twsint;.v)OoIO
g t =   ('os t;.v <HI = 
. 7r 0 ;.vo 7rWO t
2
0
1 .
= 2 [COS ""ot + wot Sin ;.vot  1\
1T""ot
.'1(1) = i /0010 rJ ... t0r' ... t d;.; = ...!.. fWO ,.J..,(tto) dw
*7r wo 2r. "'0
1 j..,(I .. tO) ,..,0 sin wo(t  to) Wo. [ (t t)]
= r = = sine wo .  0
(2"').i(t  t.o) 'If(t  to) 7r
0010
(b)
g(t) = .ir·,..,t 11..1 + 1"'0 jci"'t d""]
_ 1 j",t 1
0
1 jW'I""O _ 1  cos;.;ot
 ('  (' 
27rt _ "'0 27rt 0 7rt
C I t:..
J
,
t'"
Co
(L)
(f)
w
SiT t
Fig. S3.21
3.21 Figure S3.21 shows the plots of various functions. The funct.ion in part (a) is a gate function centered at the
OIigin and of width 2. The fUllction in part (h) can be expressed as 6 (.). This is a triangle pulse centered
at the ol"igin and of width 100/3. The fUllction in part (e) is a gate function rect(i) delayed by 10. In other
words it is a gate pulse centPred at t = 10 and of width 8. The function in part (d) is a sinc pulse centered
at t.he origin and the first zero occurring at T = lI', that is at w = 5. The function in part (e) is a sine pulse
sinc( f) delayed by lOr.. For the sinc pulse sin('( 1}. the first zero occurs at f = 11'. that is at w = 571'. Therefore
the function is a sinc pulse centered at ;.; = 1071' and its zeros spaced at intervals of 57r as shown in the fig.
S3.2Ie. Thl! function in part. (0 is a product of a gate pulse (centered at the origin) of width 10". and a sine
pulse (also centered at the origin) with zeros spaced at intervals of 571'. This results in the sine pulse truncated
beyond the inTerval ±.l7r (If I 57r) as shown in Fig. f.
3.22 The function reet (t  5) is centered at t = 5. has n width of unity. and its value over this interval is uuity. HencE'
GI
.)   jwl dt _ 1  J.JI _ 1 [  j4.S", j&.s",\
l
!UI IS'S
...  ('  f'   r  r
. 4.5 jw 4.5 j..;
r
j
5..l· . r
jlSw
[ JJ]
= _._[,.)",/2 _ ,.,..,/21 == : 2jsin
Jw 2
. ("") ·",5..,
= SlnC  ('
2.
18
3.23
,..1101 .. • jlO1
:::: . [2) sm 7.t) = slne( r.t),.
J27r..J
3.24 Obser\,t' that 1 + sgn(t) = 2u(t). Adding pairs 7 and 12 in Table 3.1 and then dividing by 2 yields the desired
result.
3.25 Obser\,t' that
cos (..Jol + 9) = [p:1("'OI+f) + ,.j(woI+l)]
1 .i' j"'Ot 1 ill )"'1)1
= 2"' P. +2P. I'
Fourier transform of the abo\'e equation yields the desired result.
3.31 (a)
1
1/(1) 7r,o,(w) + :
........... j..J
g(l) y'
G(..,)
Application of duality yields
7rMt) + ..!.. 27r1t( ..J)
}I ..
• ... ' 2"9("')
G(I)
01
* [hit) + 1/(..J)
_ J':rt
Application of Eq. (3.28) yields
! [t(t) __ 1_] u(.')
2 J7r1
But "11 is an e\'en function. that is "(1) = h(t). and
1 .
Ih(t) + LJ tl(..J)
2' Trt
(b)
cos ..Jol Trlh(..; + ,",,0) + "(..J  ;""o}l
.......... ' ,
g{t) Gt ... )
Application of duality property yields
7r [h(t + ..Jo) + h(t  wo») ¢=> 211' COli (,"'ow) = 271' cos (wo .... )
, "'.."
¥ •
G(t) 2"11("')
Setl ing "':0 = T yields
I:(t + T) + h(t  T) 2 cos Tw
(e)
sin ..vo' jr.!h(w + ..vo)  h( ....  ;.10)]
 .. ,
1/(1) GC ... l
Application of duality property yields
+ .'':0)  h(t  '..II)l <=> 2r.sin(;.Io .... ) = 27r sin(:..ro .... )
.. .I..,
Y
G(I) 2 ... g(··,.,)
SCI ling "':0 = T yit'lds
lO
bet + T)  bet  T) 2jsin Tw
3.32 Fig. (b) .91(t) = .q(t) and
G1 (w) = G( ·w) = 2..2 /rJ"" + )0.1('.'""  I]
W
Fig. (c) q2(f) :"q(1  1) + .9dt  1). Therefort'
G1 (w) = IG(;.;) + G1(w)}ri'" = !G(w) + G( w»)eJ""
2('.)"'
= r(cosw + w sin..,  1)
w
Fig. (d) .Q1(t) = 9(t  1) + gtft + 1)
Fig. (e) .94(/.) = get  + ql!l + i). and
Fig. (f) can be obtained in three steps: (i) timeexpanding .9(1) by a factor 2 (ii) then delaying it by 2
seconds. (iii) and multiplying if by \.5 [we may interchange the sequence for steps (i) and (ii)!. The first step
(timeexpansion by a fact or 2) yields
SE'conu step of lime delay of 2 sees. yields
1(
12) 1 (,2.., .? ,2 ... ),2 .. 1
11
.ft
2 <==> 2 ... ,2 (.  .I_we  1 r' = 2,.:2'   ('
The third step of multiplying the resulting signal by 1.5 yields
= 1.5f ('; 2) (1  .i2w _ ,..)2",,)
o
Fig. S3.33
3.33 (a)
(
t + T/2) (t T12)
g(t) "" reet. r  reet r
reet (f) Tsine (Wi)
recl (f ± ::?'/2) = Tsinc C.;{) p±J",T/2
20
and
G(",,) = Tsinc (W{) [r.;IoIT/2 _ r.
jIolT
/2)
2
'T' (WT). wT
= J SlnC T SInT
= (W{)
(b) From Fig. S3.33b we verify that.
.iI(t} = sin t u(t} + sin(t  1I'}u(t  11')
!':ote that sin(t  11'}"('  :r) is sin ,,,(t) delayed by 11'. Now. sintll(t) <==* t[h(w  1)  6(..1 + 1)] + 1_1"" and
sin(t  1I')!J(t  11') {211'. [h(w  1)  h(w + 1») + _1_ .. }(']"'"
J lw·
Therefore
Recall that g(.I·)M:r  xo) = g(.roHI(:r  :ro). Therefore h(w ± 1)(1 + ('] ..... ) = O. and
CI .• :) = _1_(1 + ,.Jw'J)
1 _;.;2
(c) From Fig. 53.3& we verify that
q I I) = cos t [" (I)  " (f  i)] = cos I lJ (t)  cos t 1/ (I  i)
But sinlt  i) = cost. Therefore
.q(l)::= cost 1/(1) + sin (I  %) 1/ (I
(
• 11' [ ) ») j;.l { :':' I () I} ) ..... '2
G .• :)=;;t(..Il +,+ +2 "
• 1 ..,; J 1  ",,'
Also because q(.r)h(:r  :ro) ::= q(.To)M1'  1'0).
Therefore
(d)
g(t) = ,,
lIt
lll (t) !/(t  T)] = ""'!J(t)  ,,"',*  T)
= (III,,(t)_ (,oT r.oCtT)I/(f _ T)
1
,,7 1
G(..,;) ::: .   .!:..._p;",T = :11 _ f!(o+.'IoI)T)
./..1 + 0 .IW + a J ..... + a
3.34 From timeshifting propert.v
Thcl'efOle
.q(I + T) + q(t  T) e=:;. G(..I)r..1
oo1T
+ G(;.I),·J;.>T = 2C( ..... ) cos .... T
Wc can use thib result to derive transforms of signals in Fig. P3.34.
(8) Here .q(l) is 8 gate pulse as shown in Fig. S3.348.
21
.q(t) = reet <=* 2sine(w)
Also T = 3. The signal in Fig. P3.34a is .q(t + 3) + .q(t  3), and
.q(t + 3) +g(t  3) 4sinc(w)cos3w
(b) Here 9(1) is a triangular pulse shown in Fig. S3.34b. From the Table 3.1 (pair 19)
.q(I.) = D. (4) sinc
2
Also T = 3. The signal in Fig. P3.34b is .q(t + 3) + g(t  3). and
.9(t + 3) + g(t  3) 2sinc
2
cosaw
I
FiB. S3.34
3.35 Fl'equencYbhift ing propel I)' st ates that
Therefore
q(1) sin :: 21 I },.I"'ol + .q(t)1' 1,",0
1
] = 21. [G(w  ,.:0) + G(..:  ..... 0)]
.I )
Timeshifting property sta!E'S that
Th<>refOJe
Ilnd
'21 [q(t ... T)  .q(t  T)] <=* G(..J) sin T..J
.I
ThE' signal in Fig. P3.35 is .q(' + 3)  .q(1  3) where
.q(t) = reet <===* 2sinc(:.I)
Therefore
g(t + 3)  g(t  3) 2j[2sinc(..J)sin3w) = 4j sinc(...:) sill 3..J
:i.36 Fig. (8) The Signal R(t) in this case is a triangle pulse (Fig. S3.36) multiplied by cos lOt .
. q(t) = ) cos lOt
Also from Table 3.1 (pair 19) <===* 7I'sinc
2
(¥) From the modulation property (3.35), it follows that
(
t ) 11' {. ., [7I'(..J  10)] . 2 [7I'(W + 10)] }
R(t) = 2;r cos lOt <===* '2 sine· 2 + sine 2
The Fourier transform in this case is a real function and need only t.he amplitude spectrum in this case as
bhown in Fig. 53.36a.
Fig. (b) The signal.q(t) here is the same as signal ill Fig. (a) delayed by 271'. From time shifting property.
its Fourier transform is Ihe same as in part (a) multiplied by ,,).;(2 .. ). Therefore
22
..
if
""
.....
,
t ...
, "
<
}
C't:oLk en .
...
t ....
( "')
,&11,1)) ,
.....
... ...
... ..... ,D (11
.,iII'# ......
=  '.
o
o
•. \ =  SInC + smc  "
G(
' 1: {. 2 [11'(";  10)] . 2 [11'(;..' + 10)]} )2" ...
2 2 2
Thr FOllrif'1' transform in this case is the sam'e as that in part (a) multiplied by ,,J2",",. This multiplying fllctor
a linear phllSe spectlllm 2l1''';. Thus we have an amplitude spectrum [same as in part (a)i as well as
a linl'ar phase !.pectrum .::G( •• :) = 21r": 11.'> shown in Fig. S3.3·6b. the amplitude spectrum in this c(lSe as shown
in S3.36h.
1'\ott': In the above solution. we first multiplied the triangle pulse by cos lOt and then delayed the result
b.\· 211. This nlf'ans the signal in Fig. (h) is ('xprefst'd as ) cos 10(t  211).
"'e could have interchangecJ the operation in this particular case. that is. the triangle pulse is first delayed
lw '2r. and then tht' result is multiplied by cos lOt. In this alternate procedure. the signal in Fig. (b) is expressed
as .:3. ( I ) cos 1 Ot .
This inl t'rchange of operation is permissible here only because the sinusoid cos lOt executes integral number of
cwles in the interval 2r. Because of this both the expressions are equivalent since cos 10(t  211') = cos lOt.
Fig. (e) In this case t he signal is identical to that in Fig. b. exC'ept that the basic pulse is rect (21,..) instead of
a triangle pulse :'\ow
rect 211 sinc( lI'w)
t:sing the same argument as for part (b). we obtain
G(",,) = 1I'{sinc[lr(.u + 10)) + sincl7i(w  10)]}r
j
!!,.. ...
(
w4) ("'+4)
G(",,) = rect 2 + reet 2
;'Sinc(t) rect
Therefore
g(t) =
• 1f
(b)
Also
23
ThE'refolc
get) = !sinc
2
(t) eos4t
11"
3.38 Flom the frequency convolution property. we obtain
The width propel"ty of convolution states that if (;1 (x) .. C2(X) = y(x), then the width of !I(x) is equal to the
slim of the widths of ('J(:r.) and ('2(.'7'). Hence, the width of G(w) .. G(",,) is twice the width of G(",,) , Repeatp.d
application of this argument shows that the bandwidth of g"{t) is nB Hz (n times the bandwidth of g(t»,
3.39 <a>
3.310
(b)
find
(c)
G()
j
o JWCd iT J';JCd 2 [ T])4 . 2 (""T)
"" = " t  (! t =  ":' 1  cos ~ . =  sm .
T 0 JW ..J:2
(
t+T/2) (tT/2)
,q(f) ::: reet T  rect T
rert (f ) ~ Tsinc (..J[)
(
t ± T /?) (..JT) ± 'r '2
rect ;p= <==> Tsinc 2"" ~ J..J ,
df, :: h{f + T)  2h{t) + h{t  T)
"
The Fourier transform of this equation yields
Therefort!
)wG(w) = (!j",T  2 + (!jwT = :2[1  cos ""T] = 4sin
2
( ~ )
2 ,
)4 (WT)
G{ ..... ) = :; sin
2
T
A basic demodulator is shown in Fig. S3,310a. The product of the modulated signal q(f)cos;..;ot wit.h 2cos..Jot
yields
g(f) cos •• :ot x 2 cos ..Jof :.: 2gft) co.r;2 ..Jof = .q(t)[1 + cos 2..:ot] = g(t) + g(t) cos 2 .• .:ot
ThE' product contains the desired g(f) (whose speet rum is centered at..J = 0) and the unwanted signal ,q(f) cos 2""ot
with sptctmm ! IG(..J+2..Joj+G(",,2..Jo!. which is centered at ±2..Jo, The two spectra are nonoyerJapping because
24
3.41
3.51
12
,
tb
_ .!l • ..ttii)
,i.". ,{1rr to ....
~
Fig. 53.41
Ir < .'0 (See Fig, 53,310b). We can suppress the unwanted signal by passing the procluct through a lowpass
filter as shown in Fig 53.310a.
Gd ... ·) = sinc(doo) and G::(w) = 1
Figure S3.4j shows GI(":). G'l(":). Hd..:) and H2(":)' Now
Yd ... ·) = Gdw)HI(w)
Y2(W) = G2(w)H2(W)
The spectra YI(":) and Y:z(..:) are also shown in Fig. 83.41. Because 1/(t) = 1I1(t)1/2(t). the frequency convolution
property yields }'(..:) = }'! (..:) * Y2(W}. From the width property of convolution, it follows that the bandwidth of
1'(..:) is the sum of bandwidths of }'!(..:) and Y2(W). Because the bandwidths of Y1{w) and Y2(W) are 10 kHz. 5
kHz. respectively. the bandwidth of Y(..:) is 15 kHz.
Csing pair 22 (Table 3,1) and timeshifting property, we get
This is noncausal. Hence the filter is unrealizable. Also
1
"" IlnIH(w)11 I/.;.; = 1''''' ~ d w = 00
..:2 + 1 ;.)2 + I
 ~  ~
25
3.52
hlf)
Figure S3.:;"1
Henr.e the filt.er is non causal and therefore unrealizable. Since h(t) is a Gaussian function delayed by to. it looks
as shown in the adjacent figure. Choosing to = 3v'2k. "(0) = r.
4
.
11
= 0.011 or 1.1% of it.s peak value. Hence
to = 3m is a reasonable choice to make the filter approximately realizable.
H(w) = 2 X 10
5
,.j"'lo
w
2
+ 10
10
From pair 3. Table 3.1 and timeshifting property, we get
I/(f) =
The impulse response is noncal1sal. and the filter is unrealizable.
h Lt""
Io'igure S3.:;"2
The eXj.>onential delays to 1.8o/t at 4 times constants. Hence to :;: 4/a:;: 4 x == 40/<s is a reasonable choice
to makE' this filte. approximately realizable
3.53 FlCm the results in Example 3.16
1 1\
11 =  = 10
RC
Also HlO) = 1. Hence if"';l is the frequency where thp. amplitude response drops to 0.95. then
10
6
IH("';I)I = :;: 0.95 =* .... 1 = 3:28.684
v;.,)f + 10
12
Moreover. the time delay is given by (see Example 3.16)
td(W) = _ a ===> trl(O) = ! = 10
6
...;2 + a
2
a
If J.i2 is t he frequency where the time delay drops to 0.98% of its value at W = O. then
td(W2) =
2106
0
'''1 = 0.98 X 10
6
===> W2 = 142,857
w2 + 1 •
We select the smaller of"';l and ,.)2. that iR w = 142,857, where both the specifications are satisfied. This yields
n frequency of 22,736.4 Hz.
3.54 There is a typo in this example. The time dela.y tolel'ance should be 4% instead of 1%.
The band of A .... = :2000 cent.erp.d at W = 10
5
represents t.he frequency range from 0.99 x 10
5
to 1.01 X 10
5
. Let
us consider the gains and the time delays at the band edges. From Example 3.16
a
I H (...; ) i = .... :;:2 =+=I1:;r,
a = 10'
At the edges of the band
26
t.
,;
3.61
The gain variation over the band is only 1.99%. Similarly, we find the time delays at the band edges as
and
The time delay variation over the band is 4%. Hence, the t'ransmission may be considered uistortic.nless. The
signal is transmitted with a gain and time delay at the center of the hand, that is at. ;.I = 10
5
. We also find
IH(105):::; 0.01 and fd(l05)::::: 1_1. Hence, jf g(t) is the input. the corresponding output iii
t.+
Fig. 83.61
= G(...;) rect (..:::!.....) (' J( ... ·to+k.,n"'T)
4r.B
:::::G(...;)rect ( ;.IB)llj/rSin;.lTJr.1 ...
t
o
t
.471'
This follows from the farl that r'" :::; 1 +:r when.7 « 1. Moreover. G(...;)rert = G( ..... ) hecause G(.;)
bandlimileJ 10 B Hz. Hence
n .. :) = G{;./)r
PJto
 jkG( •• :)r.in.,.;Trj..,t
o
:\loreo\,el. we can show that (see Plob. 3.35)
Hence
d. !.q(f + T)  g{t  T)] G(.,.;) SillwlT
.1
1/(1) = g(t  to) +  to  T)  g(t  to + T)]
Figure 53.61 shows .q(1) and 1/(1).
3.62 Recall that the transfer function of all ideal time delay of T seconds is (,j",T. Hence. the transfer fuuction of
the equalizer in Fig. P3.62 is
Heq(w) = ao + all'j ... + a21·i2 .... + ... +
Ideally. we require the equalizer to have
[Jleq("';)]desircd = 1 +
= 1 _ Qr + Q2
r
.I
2w
AI _ + ... + ...
The C'qualizel in Fig. P3.62 approximates thili expression if we select ao = 1. al = Il. a:! :.: 0
2
. "', an =
(1)" C\ ".
27
3.11
EI/ = j'"" .q2(t)dt = (OU ('._t
2
/tr
2
dt
_"., 271"17 i_
Letting = and consequently tit =
E" = r",
2/2
dx = ..j7f; = _1_
2u
2
V2 _.,.. 2V2lfl1 211';;
Also from pair 22 (Table 3.1)
G(w) = ('._"2,,,2/
2
Eg = 1:, IG(w)1
2
dw ;: 211f 1: ('._"2,,,2 d .. ,
Letting rr ... • = and consequently dw = Ldx
Ell = ...!.._1_ f"" (,",2/'ld:r: = y'2; = _1_
271" 11v'2 ou 21f11v2 '2,,';;
3.12 Consider a signal
q(l) = sinc(kt) and G(w):=
:1.73 Rerall that
Tilerefore
Interchanging the roles of gl(t) and .q2(t) in the above development. we can show that
3.74 In the generalized Parsevars theorem in Prob. 3.73. if we identify gl(t) = sine (211"Bt  '"11') and .92(t) =
sine (211"81  717r), then "
Gdw) = _I_ reet
28 471"B •
1 ( w ) .I..!!.!!!.
and G2(w}=reet  r'nr
2B 411'B
Therefore
But rect = 1 for 1 .... ·1 $ 271" B. and is 0 otherwise. Hence
28
n=m
In evaluating t.he integral. we used the fact that r±j2trlc = 1 when k is an integer.
3.7·5 Application of dualit.y property [Eq. (3.24)J to pair 3 (Table 3.1) yields
The signal energy is given by
Th(' energy contained within the band (0 to W ) is
E 4
l
w  2o", J. 271" [1 
2aW
J
IV= 71" I' ",';=1'
o a
If EI\' = 0.99E". then
,,2a'" = 0.01
IV 2.3025 d/ 0.366 H
==> t"y = ra s=  z
(/ a
3.7·6 Ifq2(1) <==> 04(;.;). then the outpul Y(;.;) = A(;.;)H(w). where H(..;) is the lowpass filter transfer function (Fig.
S3.76). Becanse this filrer band 11/  O. we may express it as an impulse function of area 4r.1l.f. Thus,
Hl're we used the property 9(:1')/)(1") = .q(O)1.>(.1") [Eq. (1.23a)J. This yields
1/(t) =
because 9
2
(1) o¢::::;::::> A (..;). we have
Hence . .,(t) = 2E,,::"/.
rv

whel'e
1 jT/2
'Rry(T) = lim r 7(/)1/(t +"T)dt
T.,.., T/2
29
If we let 9\(1) = Cl COS(";lt + 81) and 92(t) = C2COS(W2t + 82). t.hen
1 jT
I
2
R
II1II2
(1')= lim T C
1
C
2
COS(Wlt+8
1
)C05(W2f +";21'+82)dt
T,.." T/2
According to t he argument used in Example 2.2b. the integral on the righthand side is zero. Hence. Rl/JlI2 (T) = O.
t: sing t he same argument. we have RII291 (1') = O. Therefore
C
2
C
2
Rg(1') = R 9, (1') + RII2(1') = t COSW11' + + COSW21'
This result can be extended to a sum of any number of sinusoids as long as the frequency of each sinusoid is
distinct. hence. if
.....
get) = L Cft cos(n..;ot + 8 .. ) then
,,=1
Moreover. for qo(t) = Co. RfIO(T) = and
1 jT
I
2
R gOgl (1') = lim T CoCI C05(W\t + WI l' + 9d dt = 0
T"" T/2
Thus. we can generalize the result as follows. If
"'
q(f) = Co + L C" cos(n..;ot + 9,,) then
"" C
2
Rg(1') = C5 + L T COSTlW01'
"",I .. =\
and
51/(";) = 27rC5b( __ ) + i L  11.";0) + b(..; + n,.;o)]
,""1
:\.82 riglll"P 53.82a shows the wa\'eforms T(t) and :r(f  1') (or l' < Tb/2. Let T = NT". On the average. there are
,"1//2 pulses in the waveform of duration T. The area under the product :r(t):r(t  1') is N/2 times (If  1') as
shown in Fig. S3.82b. Therefore
1 J
T
/
2
R;r(')= lim T 3'(t)T(t1')dt
T.,., T/2
= 4 (4 
For 1t :s 11'1 :s T". there is no overlap between pulses. and Rz(1') = O. For T" :s 11'1 :s pulses again overlap.
But on the average. only half pulses overlap. Hence. Rz(1') repeats every T" but (lnly with half the
magnitude. as shown ill Fig. S3.82c. We can express Rz(1') 85 a sum of two components. as shown ill Fig.
S3.82d. Thus. R.,(1') == Rl(T) + R2(1'). The PSD is the sum of the Fourier transforms o( 1\'.1(1') and R2(T).
Hence
T" . 2 (wT")
Sz("') = 16 SIDC 4 + 52(";)
,,'here 52(";) is the Fourier transform of the periodic triangle function, shown in Fig. S3.82d. We find the
exponential Fourier series (or this periodic signal to be
....,
R2(T) = 2: D .. r
i
""','"
"=r"'JIL,.
Csing Eq. (2.80). we find D" = fgsinc'("l"). Hence. according to Eq. (3.41)
52( .• 1) = i t sine 2 ("211') ,,(..;  ll'b)
"=,..,..,
30
..:'"
Therefore
0211
'1:
...,. ...
1[11
T. 16
:I
'I
"1
(c J
Fig. 93.82
""
1.

:L.
"

Tz,
1b 2 (WTh) 11' 2(n1l') ( )
S",(...,) = 16 sine 4 +"8 L sine T II w  nwb
"=I'"I1I.I
and
31
.2.r.
I
't' ....
(b) .T
2
(1)=_ dw=
""" 1 11 1
7r 0 7r
(c) :7
2
(t) =  h(..,;  l)d..,; =   11"'"' 1
7r 0 7r
and
and
~ 111 1 1
y2 (t) =  uWJ = 
• 7r 0 WJ2 + 1 4
~ ( ) 11"'" h{WJ 1).1 _ 1 1"" h(WJ  1) I _ 1
'1'= .. ..,; 1"';
. 7r 0 ..,;2 + 1 7r 0 2 211"
3.84 The ideal difi'el'entiator transfer function is j...;. Hence, the transfer function of the entire system is
H(",)= . (jw)=.
(
1) jWJ
J'" + 1 JW + 1
and
~ 1 1"" (:J)..,;2 1 11 WJ2 1 ( To)
11 (I) =  rect  .,  dw =   dw =  1   ;;; 0.06831
11 0 2..,; + 1 11 0 w
2
+ 1 7r 4
32
"
," ~
"
.,'
:.;
4.21 (i) For", (t) =: ("os 1('001
Chapter 4
'POSB.SC(t) = m(t)coslO,OOOt == cos lOOOtcos 10,OOOt
1
=: 2 + ;05 11.000,)
LSB V;B
(ii) FOI "d') = :l cos 1000, + cos 2000t
. (t) =: //I (f) cos 10. ooot =: [2 cos 1000t + cos 2000f\ cos 10. OOOt
1
= ("os 9000f + cos 11.0001 + 2 [cos 8000t + cos 12. OOOf]
III
= icos90001 + 2 co.s8000tJ + [cos 11.000t +  cos 12.000t)
, ., 2 ,
". ..
LSB
L'SB
(iii\ F"l ",'/) = cos 1000t cos3()OOt
1
"cnsB'c (t) = m (I) cos 10. OOOf = :i[cos 2000t + cos 40001) cos 10. OOOt
1 1
= '2 [c058000f + cos 12.000t] + 2!C056000, + cos 14,000tj
I, 1 I,
== 2 8000t + cos 6OMt) + 2 12. OOOf + cos 14. 0009
This information is summarized in n table below. Figure 54.21 shows various spectra.
ModuJ4ted 5 0ntlt
4.4)..,.
\ iD
t
t t t t t
«.11 tt
T t, t t
Fig. S4.21
33
(a)
I Mew)
, ,rlleo
T
100
too
, ,
(\I )
w
lllt)
... .
IC.CDO
I
Iaeo
,
' ...
Fig. 84.22
FI,. S4.23
(b) & ((!. )
(b) {ei )
D
case Basp.blLnd frequency DSB frequency I LSB frequency I USB frequency I
i 1000 9000 and 11,000 I
ii 1000 9000 and 11 ,000 I
2000 8000 and 12.000.1
iii 2000 8000 and 12,000 I
4000 6000 and 14,000 I
4.22 The relevant plots are sho"'n ill Fig. 54.2·2.
4.23 The relevant plots ale shown in Fig. 54.23.
4.24 (a) Thp. signal at voint h is
9000
9000
8000
8000
6000
ga(f) = m(t)cos:4"';ct
I
11.000
I
11,000
I
12,000
I
12,000
1
14,000
= 711 (t) [i COSWc
t
+ C06 3...Jct]
34
I
I
I
I
_I
The t.erm is the desired modulated signal, whose spectrum is centered at The remaining
term i",(t)coslwet is the unwanted term, which represent.s the modulated signal with canier frequency 3w
c
with spectrum centered at ±3.'
c
as shown in Fig. 84.24. The bandpass filter centered at ±we allows to pass
t.he desired term bllt sllpprl!llfies the unwanted term t",(t)cos3wct .. Hence. thill; system works as
desired with the output im(t) cos wet. .
(b) Figure 54.24 shows the sper.tra at points band c.
(c) The minimum usable value of ";c is 21r B in order to avoid spectral folding at dc.
(d)
2 ", (t)
1fI(t)COS wet = 2 [1 +cos2:.Jetl
1 1
= 2m(f.) + 2m(t)cos2wc
t
The signal at point b consists of the baseband signal !m(t) and a modulated signal tm(t) cos2wct., which has a
carrier frequency 2we . not t.he desired value iJJ
c
• Both the components will be suppressed by the filter. whose
center center frequency is We. Hence, this system will not do the desired job.
(E') The readel may verify that the idrntity for cosn .... ·et contains a term coswet when n is odd. This not true
when 71 is even. Hence. the lI;ystem works for a carrier cos
ft
""ct only when n. is odd.
U.
I
W
Qt@
DI
L,:>....
..
lAJ .... 3"'6
I
W
I
Q
et®
•
tOe...
cO...:.
Fig. S4.24
4.25 WE' use the ring modulator lI;hown in Fig. 4.6 with the carrier frequency ic = 100 kHz = 20071' X 10:l). and
thp. output bandpass filter centered at Ie = 300 kHz. The output 1Jj(t) is found in Eq. (4.ib) as
".(t) = i [",(t) cos wet  !m(t)cos 3w
e
t + cos 5wct + ... J
7f 3 ,)
The output bandpass filt.er suppresst's all the terms except the one centered at 300 kHz (corresponding to thE'
carrier Hence. the filter output is
4
.11(t) = 3 m(t) COli;
.1r
This is the desired output Jr", (t) l'Os";ct. with Jr = 4/31r.
4.26 Tht' resistance of each diode is ,. ohms while conducting. and 00 when off. When the carrier A cos";ct is pOliiti\'e.
the diodes condu(" (during the entire positivE' half cycle). and when the carrit'r is negativ<> the diodes are Op<>11
(during the entire negative half cycle). Thus. during the positive half cyde. t.he voltage appears a.:r05"
ench of tht' rel';istors R. During the negative half cycle. the out.put voltBge is zero. Therefore, t.he diodes act as
a gate in the circuit that is b!lSically a voltage divider with a gain 2R/(R + 7'). The output is therefore
2R
"o(t) :::: Ul(t)m(/)
R+/'
The period of Idt) is To = 2r./..;c. Hence. from Eq. (2.75)
35
1 2[ 1 1 ]
lI'(t) = '2 + ; cos Wet  3 cos 3wJct + 5' cos 5;';c
t
+ ...
The output I'ott) is
I'o(t) = n
2R
I/'(t)m(t) = [.!. + (cos wet  .!. cos lwet + cos s.wct + ... )J
+,. n+r' 2 11' 3
(8) If we the output 1'0(1) through a bandband filter (centered at We). the filter suppresses the signal "t(t) and
II/(t)cos n"':et fol' alln "" 1, leaving only the modulated term intact. Hen('e. the system actl;
as a modulator.
(b) The same circuit can be used as a demodulator if we use a basepass filter at the output" In This case, the input
is tP(t) = '11(1) COS wet and the output is
4.27 From the results in Prob. 4.26. the output = Irm(t)cos ..... "t. where Ir = In the present case.
",(t) = sin (..Jet + fl). Hence. the output is
1'0(1) = /rsin(:""et + 8)coswcf = + sin(2;';ct + R)]
The 10wpCUls filter suppresses The sinusoid of frequency 2wc and transmits only the dc term sinB.
CDs 2.0DODt
,
t I.. f f 1+'>.,.""_2._( 'f )
4,28 (A) Fig. 54.28 shows thE.' signals at points a. b. and c.
(b) From the spectrum at point c. it is clear thaI I.he channel bandwidth must be at least 30.000 rad/s (flom
.1)000 to rad/s.j.
(c) Fig. 54.28 shows the rel'eiver 1.0 rp.covar .",lIt) and m2lt) from the received modulated signal.
4.29 (a) S4.29 shQWS the output signal sVf'ctrum Y(o.:).
(b) Ob:;er\'e tlla! Y(...:) is the same as All.') with the frequency spect.rum inverted. that is. the high frequenciE'1;
are shifted to lower frequencies and \'ic'e Thus, I.he scrambler in Fig. P4.29 inverts the frequency spectrum.
36
, '
'0
o
fA. ;.0' '5
,
Fla. 54.32
I"
 ')It ..
To gel back the spectrllm M(,.:). WI' need to invert the spectrum Y(;.;) once again. This can he dOlle by
p!l.bsing I he scrambled signal ,,(1) Through the same scrambler,
4.210 \\'e use the ring modulator shown in Fig. 4.6. except. that the input is /1I(t)cos(2r.)10bt instead of m(t), TI.e
caniE'r frequency is 200 kHz [;""c ::: (400r.)10
1
tj. and the output bandpnss filter is centered at 400 kHz. Tht?
output !',(t) is found in Eq. (4.7b) as
I', (t) = [1//11) cos(2r.)106
'
]lI'o(t) = 2",(t) cos(2r.)106t [(:os (400?T)lO't  .! cos 3\400?T)lO"t + CO"" 5(4007:')10" .... ' ,1
'II' 3 OJ J
The product of the Terms (1/3)C05 3(4001l')lO't and (4/II')m(t)cos(211')106t yiclds the desired term
 (;/11(1) cos (SOOr.)IO't. whose' spectrum is centered at 400 kHz. It alone passes through the bandpass filter
(centered at 400 kHz). All the other terms arc sl1ppre58ed. The desired output is
!I(t) =  3: m(t)cos (8001T)lO't
4.31q,,(t) =!A. m(t)] cos ""<t, Hence.
qb(t) = IA + met)] cos
2
wet
= + m(t)! + klA + m(t)Jcos2:.1ct
The first term is a lowpass signal because iTS spectrum is centered at = O. The lowp.us filter allows this telln
to pass. hut suppresses the second term. whose spectrum is centered at ±2..Ic . Hence the output of the luwpass
filter io;
,,(t) = A + 111 (t)
When this signal is passed through a dc block, the de term A is suppressed yielding the output TIl (,), This
show!; that the system can demodulate AM signal regardless of the value of A. This is a synchronous or coherent
demodulation.
37
4.32
(a)
/II" 10
~ A=20 ,,=0.5=  =
.4 A
(b)
11Ip 10
~ A = 10 I' = 1.0 =  =
A A
(c)
,."" 10
:::>.4=5
I' = 2.0=  =
A A
(d)
rn 10
I' = OC = ::.2 ==  => A = a
A A
This means that I' = 00 represents the DSB·SC case. Figure S4.3·2 shows various waveforms.
4.33 <a) According to Eq. (4.lOa), the carrier amplitude is .4 = in,,/II :::; 10/0.8 :::; 12.8. The carrier powe)' is
Pc = .4.
2
/2:::; 78.125.
tv'(t)
10
.. '.
t ~
,
..
TO
",
......
'.

..
i
Fiy'. S4.33
(b) The sideband power is '" 2(1 )/2. Because of symmE'tl'y of amplitude values e\'ery quartp.r cyclE'. the power of
",(I) may be complJted by averaging the signal energy over a quarter cycle only. Over a quarter cycle mit) can
be representE'd as mit) = 40t/To (see Fig. S4.33). Ho!nce.
~ 1 r
To/4
[40']2
", (t) = To/4 io To dt:::; 33.34
The sideband power is
,.".,.,..,.
P. = m:{f) :::; 16.67
The efficiency is
p. 16.67 00 9 6 6 ~
'#:::; p. + p. = 78.125+ It:,67 )( 1 = 1. 70
4.34 From Fig. S4.3·4 it is deal' that the envelope of the signal m (t) cos ...Jet is Im(t)l. The signal !A + r" (t)l cos ..... c! is
identical to ",(1) COS..let with m(,) replaced by A + m(t). Hence. using the previous argument. it is clear that
its envelope is IA + rut'):. Now, if A + ",(1) > 0 for all t, then .4 + TrI(t) = IA + m(t)j. Therefore. the condition
fo\' demodulating A ~ I signal using envelope detector is .4 T m(t) > 0 for all t.
Fig. S4.34
38
mLt)
4.35 Whcn an input t.o A DSB·Sc generator is "'(t), the corresponding output is m(l) coswct. Clearly. if the input is
.4 + 1II(t), the corresponding output wi\1 be IA + m(t») COS;Jet. This is precisely the AM signal. Thus. by adding
A de of value .4 to the baseband signal me'). we can generate AM signal using a DSB·SC generator.
Thc converse is generally not true. However, we can generate DSBSC using AM generators if we use two
identical generators in a balanced scheme shown in Fig. S4.35 to cancel out the carrier component.
4.36 \\'hen an input to a DSBSC demodulator is met) cos ... ·ct. the corresponding 01.ltput is met). Clearly. if the
input is [A 4 ,,, (t)] COs"'e'. the corresponding output. will A + met). By blocking the dc component A from this
output. we call the Ar.l signal using II DSB·SC demodulator.
The converse. unfolt unately. is not true. This is becamle. when an input to an A:\1 demodulator is m (1) Cos "'et.
the corresponding output is Im(t)1 [the p.nvelope of TII(t)1. Ht'nce. unless m(tl ? 0 for all t. it is not possible to
dernodulatt' DSB·SC signal using an demodulator.
n
I
_..6
\I
v " \J
Fi,. S4.31
4.3·1 ObM'I"!' thaI ",2(1) = .4
z
for all t. Hence. the time average of ffl2(t) is also A2. Thu,>
The carrier amplitude is A = "'1'/1' == tn, = A. Hence Pc == A
2
/2. The total power is P, == Po: + P. = .4
2
• The
power efficiency is
P A2j')
1/ = p; )( 100 = A2  x 100 = 0.5
The A:\I Signal for I' = 1 if; shown in Fig. 84.37.
4.38 'fh'" signal Pot point a is [.4 + ",(t)Jcos wet. The signal at point b
2 2 + 2.4m(t) + m
2
(t}
.T(t) = \..1 + m(t)J cos "'e
t
= 2 (1 + cos 2 .... ·c
t
)
The lowpa.c;s filter suppresses t.hf' term containing cos 2..l
c
t. Hence. the Signal at point. c is
l'su:dly. ",(I)/A « 1 for most of the time. Only when m(l) is neal' its peak, this condition is violated. Hencp..
the output at point d i.o;
39
A2
1/(t) :::; T + Am(/.)
A blocking capacitor will suppress the dc term A
2
/2. yielding the output Am(t). From the signal wit). we see
that the distortion component is 7//
2
(1 )/2.
4.41 In Fig. 4.14. when the carrier is cos HA .... )t + b] or sin (A .... )t + Ii), we have
.Tdt) = 2Iml(t) cos..,;"I. + "'2(t) sin .... ctlcos [( .... e + Aw)t + h)
= 2m1 (t) cos .... et cos [( .... e + Aw)t + h] + 277t2(t) sin Wet cos [(w" + Aw)t + ,Ii]
= "'1 (t){ COS [(A .... )t + b] + cos [(2wc + Aw)t + b)} + Tn2{t.){sin {(2 .... c + A .... )t + bJ  sin [(A .... )t + hI}
Similally
.T2(tl = '" 1 (t){ sin [(2..;" + A..;)t 't' hI + sin [(Aw)t + hI} + rn2(t){ cos [(A..;)t + h]  cos {(2..;c + A ... ·)t + hI}
After 1'J tt) and :T2(t) are passed through lowpass filter. the outputs are
TIl (t) = m dt) cos [(A..;)t + 6J  1112(1) sin HAw)t + II]
"'2(1) = 1111(1) sin [(A .... )t + bj + Tn2(t)COS !(A",·)t + h]
4.51 To genernte a DSB·SC signal from met). we multiply met) with cos",,,,I. However. to generate the SSB Signals
of thE:' same relative magnitude. it is convenient to multiply m(t) with 2C05..;,t. This also avoids the nuisance
of the fractions 1/2. and yields the DSB·SC spectrum M (w  We) + At ("" + ""e). We suppre55 the \.:SB spectrum
(above .... ·e and below  .... e) 1.0 obtain the LSB spectrum. Similarly. to obtain the USB spectrum. we suppress
the LSB spectrum (bl'twecn  .... e and..;e) from the DSBSC spectrum. Figures 54 .. 51 <l. band c show the three
C'ilSes.
4.52
(a)From Fig. a. we can express .ft.S8(t) = cos900t and \Pl:S8(t) = cos 1100t.
(b )From Fig. h. we can express <?'LS8 (t) = 2 cos 700t + cos 900t and 'Pl"58 (t) == cos 1100t + 2 cos 1300'
(c)From Fig. c. we can express hss(t) = + cos600lJ and IPns(l) = 1400/ .. cos 1600(
and
(a) fII(t) = cos 100' and 1IIh(t) = sin lOOt. Hence,
ifLSB (t) = cos lOOt ('081000# + sin lOOt sin 1000t = cos(lOoo  l00)t :: cos900t
'Pl"SB (t) = cos lOOt cos 1000t  sin lOOt sin lO00t := cos(1000 + l00)t := cos 1100t
(b) UI (t) ::; cos lOOt + 2 cos lOOt and m" it) = sin lOOt + 2 sin 3OOt. Hence,
.; (t) = (cos lOOt + 2 cos 300t) cos 1000t + (sin lOOt + 2 sin 300t) sin 1000t = cos 900t + 2 cos 7001
.; (1) = (cos lOOt + 2 cos 3OOt) cos 10001  (sin lOOt + 2 sin 300t) sin 1000t = cos 1100t + 2 cos 1300/
(1:) wit) = cos 1001 cos 500t = 0.5 cos 4001 + 0.5C05600t and Tn,,(t) = 0.5 sin 400t + 0.5sin600t. Hence.
(t) = <0.5 cos 4001 + 0.5 cos 600t) cos 1000t + (0.5 sin 400,. + 0.5 sin 6OOt) sin 1oo0t = 0.5 cos 400t + 0.5 cos 6001
= (0.5 cos 4001 + 0.5 cos 600t) cos 10001  (0.5sin400t + 0.5sin600t) sin 10001:= 0.5 cos 1400t + 0.5eos 1600t
40
!
2.Jr
I.'O(l ~ ~
1
1 tit t ;
1.31< II K.
o l4) ... II '" 13t<..
( h)
Lt! )
Fig. S4.51
41
".,  ' ~ D  ~ "'0 0 'tW ,D, 11,.0 "CD
LV'"
j
.2.tCDif
Icoooji
,
n
1 ltc})
r
2.. D
o
o C'rl0l)T7
Fig. S4.53
<a)
(b)
Fig. 84.55
4.53 '.1) Figurp S4.53a shows lhe specTrum of ",(t) and Fig. S4.53b shows the corresponding DSBSC spectrum
'2",11\ cos ft) OlIcrrt
(hi Figure S4 .. ,)3c shows the conesponding LSB spectrum obtained by suppressing the eSB spectrum.
(c) Figure S4.53d shows the corresponding t,;SB spectrum obtained by suppressing the LSB sppctrum.
\\'p now find Thp in\'erse Fouri!'r nansforms of the LSB and USB spectra from Table 3.1 (pail' 18) and the
fl'I'(I'H!nry shifting as
.f"sa U) = lOOOsinc(10001l't)cos 900011"
'PlSB(t) = lOOOsinc (lOOO1l't) cos 1l.0oo1l't
4.54 Because MIt(w) = jU(..:) sgn (w). the transfer function of a Hilbert transformer is
H(;.J) = jsgn(,.)
If w(' apply", h (t) at the input of the Hilbert. t.ransformel'. Y (w). the spectrum of the output signal 1I(t) is
This !>hows that the Hilbert transform of "'h(t) is m(t). To show that the energies of lII(t) and 11lh(t) an! ,;
pqual. we have
42
The incoming SSB signal at the receiver is given by [Eq. (4.17b)]
'fLSD(t) = m(t)coswct + m,,(t)sinwet
Let the local carrier be co.'! [(We + + .'I]. The prodeuct of the incoming signal and the local carrier is I'd(t).
given by
('d(t) = 'PLSS(t) cos [(We + + .'I]
= 2[1II(t) cosw"t + 11'1" (t) sinwct] cos [(We + + II]
The lowpass filter suppresses the sum flequency component centered at the frequency (20.1< + and passes
only the difference frequency component centered at the frequency £:.w. Hence, the filter output r.o(t) is given
by
roU) = 111 (t) + h)  Tn ,,(t) sin(6w)t + h)
Obser\,1.' that if both .... ' and 1'1 are zero. the output is given by
1'0(1) = met)
as expected. If only" = O. then the output is given by
1'0(/) = m(t·)cos(6w)t  "".(t)sin(6..J)t
This is an l'SB signal corrE'sponding to a carrier frequency as shown in Fig. S4.55b. This spectrum is the
saml.' as the spectrum M ( .... ) with each frequency component shifted by a frequency This changes the sound
of an audio signal slightly. Fo!' voice Signals. the frequency shift within ±20 Hz is considered tolerable. Most rs
systems. howevel'.lestricl I he shift to ±2 Hz.
(b) When only = O. the lowpass filter output is
We now show that this is a phase distortion, where each frequency component of .U(..;) is shifted in by
amollnt ,.. The Fourier transform of Ihis equation yields
Bul from ECj. (4.J4b)
and
Eo(..;) = M (..J) COS," 
{
jM(..tJ)
IIh( •• :) = jsgn (;..I)M(w) =
M(;..I)
It follows that the amplitude spectrum of l'o(t) is M( .... ). the same as that for met). But the phase of each
component is shifted by 1'1. Phase dist.ortion generally is not a serious problem with voice signals, because the
human ear is somewhat insensitive to phase distortion. Such distortion mlly change the quality of speech. but
the \'oice is st ill intelligi ble. In video signals and data I ransmission, howe'er, phase distortion may be intolerable.
4.56 We showed in prob. 4.54 that the Hilbert transform of rTlh(t) is m(t)_ Hence, if m,,(t) [instead of l1/(t)] is
applied at the input in Fig. 4.20. the USB output is
y(t) = 1/1 h (I) cos .. ),.t  1fI U) sin ..Jet.
= TTl (t) cos (weI + I) + 711 h (t) sin (...:c
t
+ I)
43
 ~
,:).
, e
~
'
;
~ J ( ~
() # '7 ~
~ ~ i
Ho(ciJ
I
..
:=
~
.  , , ~
(1, )
: .'. ...
4 . .:2.
 I
4
:L.
4
f
k.
tl
b
Fi •. S4.61
Thus. if we apply m,,(t) at the input of the Fig. 4.20. the USB output is an LSB !;ignal corrf'sponding to ',,(I).
The calrier also acquires a phase shift 1r /2. Similarly. we can show that if we apply 711/,(1) at the input of the
Fig. 4.20. the LSB output would be an USB t;ignal corresponding to m(t) (with a carrier phase shiftE'd by 1r/2).
4.61 From Eq. (4.20)
Ho( .. :) =. 1
H.( ... : + we) T H.{w  We)
Figul'e S4.61a shows H.( .... :  .... c) and H,(...; +w,;e). Figure S4.61b shows the rE'cipl'ocal. which is Ho{ .... ).
4.81 A station can be heard at its allocated frequency 1500 kHz as well as at its image frequency. The two frequencies
are 2/1F Hz apart. In the present case. /IF = 455 kHz. Hence. the image frequency is 2 x 455 = 910 kHz apart.
Thel'"fore. thE' :;tation will also be heard if the receiver is tuned to frequency 1500910 = 590 kHz. The l'easCin for
this is. as follows. When the receiver is tuned to 590 kHz. the local oscillator frequency is fLo = 590+455 = 1045
(4"
). :
.'
kHz. :.low this frequency fLo is multiplied with t.he incoming signal of frequency Ie = 15(10 kHz. The output l
yit'lds the two modulated sigllal:; "'hose carrier frequencies are the sum and difference frc"<!uencies. which arr,
1500 ; 104., = 2545 kHz and 1500  1045 = 455 kHz. The sum carrier is suppressed. but the difference carrier
passes through. and the station is receh·ed. /'
4 .. 82 The local oscillator generates frPquencies in the I'ange 1+8=9 MHz to 30+8=38 MHz. When thl? receiver setting
is 10MHz. ho = 10 + 8 = 18 :\fHz. ","ow. if there is a station at 18 + 8 = 26 :\lHz. it will beat (mix) with
/Lo = 18 MHz to produce two signals centered at 26 + 18 = 44 MHz and at 26  18 = 8 MHz. The sum
component is suppressed by the IF filtel'. but the difference component. which is centered at 8 MHz. passes
through the IF filter.
44
. ,
,
y
Chapter 5
mtt)
C1
,
r
r
f1
"
f·
.
v
v
v \,
Fig. S5.11
5.11 In this case Ie = 10 /rip = 1 and = 8000.
For FM :
= A'fll/p/2":r = 21T )( = lOS Hz .. Also Ie = 10
7
. Hence, (f.)m/Ax = 10
7
+ = 10.1 MHz. and
(j,)mlTl:= 10
7
 10!> = 9.9 :'.lHz. The carrier frequency
increases linearly from 9.9 to 10.1 MHz over a quarter (rising) cycle of duration a seconds. For the next (/
seconds. wh('n 10(/) = 1. the carrier frequency remains a.t 10.1 MHz. Over the next quarter (the falling) cyclc of
duration n. the carrier frequency decreases linearly from 10.1 MHz to 9.9 MHz., and over the last quarter cycle,
whell 1/1 (t) = 1. the carrier frequency remains at 9.9 MHz. This cycles repeats periodically with the period 4n
seconds as shown in Fig. 55.1·1 a.
For PM:
= :: 50r. )( ROOO/2'1l' = 2 x lOs Hz. Also Ie = 10
7
. Henee. (j.)max = 10
7
+ 2 x lOS = 10.2 MHz.
and (/,)mm = 10
1
 2 X 1O!> = 9.8 :'.1Hz. Figure 85.11b shows m(t). We conclude that the frequency remains at
10.2 the (rising) quarter cycle, where r;l(t) = 8000. For the next a seconds, Th(t) = O. and the carriel'
frequency remains at 10 MHz. Over the next a seconds. where m(t) = 8000. the carrier frequency remains at
9.8 Over the last quarter cycle 1;/(t) = 0 again, and the carrier frequency remains at 10 MHz. This cycles
periodically with the period 41.1 seconds as shown in Fig. 85.11.
5.12 III this case Ie = 1 MHz. 1111' = 1 and = 2000.
For FM:
= I:f1ll
p
/2rr = 20,000:: /2rr = 10
4
Hz. Also Ie = 1 MHz. Hence. (Ji)mu. :: lOS + 10
4
= 1.01 MHz. and
(f.)min = lOS  10" = 0.99 MHz. The carrier frequency rises linearly from 0.99 MHz to 1.01 MHz over the cycle
(over the inrerval < f < 10;3). Then instantaneously, the carrier frequency falls to 0.99 MHz and starts
rising linearly to 10.0] MHz over the next cycle. The cycle repeats periodically with period 10:'1 as shown in
Fig. 55.12a.
For PM:
Here. because m(l) has jump discontinuities. we shall use a direct approach. For convenience. we select the
origin for flirt) as shown ill Fig. S5.12. Over the interval to If. we can express the message signal as
711(1) = 20001. Hence.
(I) = cos [2rr(1o)6t + (f)]
= cos (2rr(lCr)lI
t
+ i2oo0t]
:..: cos [2rr(10)6f + loo01rt] := cos [27f (10
6
+ 500) t]
At the diSC'ontinuity. the amount of jump is Trld = 2. Hence, the phase discontinuity is = r.. Therefore,
thE' carrier frequency is constant throughout at 10
6
+ 500 Hz. But at the points of discontinuities. there is a
45
.
'.
"
'VUYlf\ n ill n '
v v VV (TG ta
Fill' 9&.12
phILlIe discontinuity of 11' radians 8& IIhown in Fig. SS.I2b. In this case, we must maintain k" < '" because there
is a discontinuity of the amount 2. For k" > 11', the phase discontinuity will be higher than 2lr giving rise to
ambig\lity in demodulation.
(a) CPPM(t) = A cos [Wet + k"TT/(t)} == 10 cos{lO,ooOt + k"m(t)]
We are given that = 10 cos (13.000f) with k" = 1000. Clearly. mIt) = 3t over the intervalltl ::; 1.
(b) o;n.l(t) = A cos [ ..... et + k, /' m(Q)dQ] == 10 cos [10.0001 + kf /' TII(a)da]
Therefore k, II rr,(a)tlQ = 1000/
'
m(a)tJQ = 3000t
Hence 3t = /' m(o) do = 1II(t) = 3
5.21 In this case Ir/ = 1000'11' and kp = I. For
mIt) = 2 cos lOOt + 18 cos2ooor.t and 711(t) == 200 sin lOOt  36.000l!' sin 2000r.t
Therefore TIl" == :,w and == 36.000", + 200. Also t.he baseband signal bandwidth B '" 2000", /2'" = 1 kHz.
For FM: : ll.! = k/I11,,/2l!' = 10.000. and Bnl == 2(ll./ + B) = 2(20.000 + 1000) = 42 kHz.
For PM: : ll.! = == 18.000 + Hz. and Bpt.! = 2(A! + B) = 2(18.031.83 + 1000) ;: 38.06366
kH:!.
5.22 (I) = 10 COS(Wet + 0.1 sin 2oo0r.t). Here. the baseband signal bandwidth B = 2oool!'/2l!' = 1000 Hz. Also,
..,,(1) = We + 20011' cos 2()()(hrt
Therefore. A ..... = 200lr and A! = 100 Hz and BEM ::: 2(Af + B) == 2(100 + 1000) = 2.2 kHz.
5.2·3 o;£:.II(t) = 5 cos(Wet + 20 sin 1000lrt + 10 sin2oo01l't).
Here. the baseband signal bandwidth B = 200011'/211' = 1000 Hz. Also,
w,(t) = We + 20.000", cos 1000lrt + 20,00011' cos 2000lrl
Therefore. A"" = 20.000",+20. OOOlr = 40.000lr and Af = 20 kHz and BEM = 2(A! + B) = 2(20.00011000) =t2
kHz.
5.2·4 The baseband signal bandwidth B = 3 x 1000 = 3000 Hz.
For FM: ll.! = 1.;2:'· = It,x I = 15.951 kHz and BF'II.I = 2(ll.f + B) = 37.831 kHz.
k m'
For PM: = = 211;!OOO = 31.831 kHz and BpM = 2(ll.f + B) = 66.662 kHz.
5.25 The baseband signal handwidth B Ii x 1000 = 5000 Hz.
For FM: = = = 1 kHz and Bnl = 2(Af + B) = 2(2 + 5) = 14 kHz.
46
;'
. '
'.
For PM: To find BPM. we observe from Fig. S5.12 that IfJFM (t) is essentially a sequence of sinusoidal pulses
of width T = 10:1 seconds and of frequency Ie = 1 MHz. Such a pulse and its spectrum are depicted in Figs.
3.22c and d. respectively. The bandwidth of the pulse. as seen from Fig. 3.22d, is 4", IT radls or 21T Hz. Hence.
BpI\! = 2 kHz.
5.26 (a) For FM: AI = = 200.T,? .... 1 = 100 kHz and the baseband signal bandwidth B = = 1 kHz.
Therefore
Bf'M = 2(AI + B) = kHz
For PM: AI = = loxI?" = 10 kHz and BpM = 2(AI + B) = 2(10 + 1) = 22 kHz.
(b) ",(/} = 2 sin 2oo07ft. and B = 20oo7f/27f = 1 kHz. Also rn" = 2 and = 4oo01T.
For FM: = kf'e = 200.QpotrX2 = 200 kHz. and
" "
Bn.l = 2(AI + B) = 2(200 + 1) = 402 kHz
k m'
For PM: AI = T = IO"teoo" = 20 kHz and Bpt.! = 2(AI + B) = 2(20 + 1) = 42 kHz.
(c) m (t) = sin 4000",t, and B == 4000", /27f = 2 kHz. Also m" = 1 and = 400071'.
For FM: = Ie',,"'e = 200'WO"XI = 100 kHz and
.u ,
Bnl = + B) = 2(100 + 2) = 204 kHz
For PM:
I.. m' 000
AI = = IO"lr" = 20 kHz and Bp).1 = 2(AI + B) = 2(20 + 2) = 44 kHz.
(d) Doubling the amplitude of ", (I) roughly doubles the bandwidth of both FM and PM. Doubling the frequency
of /II(t) [expanding the spectrum .U(;.;) by 8 factor 2] has hardly any effect on the FM band ..... idth. However. it
roughly doubles tht bandwidth of P:\1. indicating that spE'ctrum is sensiti\'e to the shape of the baseband
sprrtrum. F"1 sptctrum is relatively insensitive to the nature of the spectrum Af(;.;).
5.21 From pair 22(Tnble 3.11. we obtain r
1a
y'i r",2/
1
. The spectrum M(w) == y'i 1'_",2/4 is a Gaussian puisf'.
which rapidly. Its 3 dB bandwidth is 1.178 l'ad/s=0.187 Hz. This is an extremely small bandwidth
compared 10 f:j,J.
l 2/4
Also ,;.(t) = _2/,,1 12. The !Spectrum of ,;'(t) is /1/'(..,) = j..,A/(..,) = jy'i",,'('wJ . This spectrum also deca:;s
rapidly away from the origin. and its bandwidth can also be assumed to be negligible compared to !:lJ.
For FM: = lr:" = == 3 kHz and Bnl:::: 2AI = 2 x 3 = 6 kHz.
For PM: To find we set the derivative of r;I(t) = _2tr
t2
/
2
t>quaJ to %ero. This yields
and == ,il( = 0.858. and
!:ll = k,:'p = == 3.432 kHz and BpM :::: 2(Af) = 2(3.432) = 6.864 kHz.
5.31 The h)ork diagram of the design is shown in Fig. S5.31.
{ :100 5 MH.!;
at oX lJt5
Af= 10Mb Af = ':2S0 H.
Fi,. S5.31
47
....
Fi l·8&.32
5.32 Tht' block diagram of the design is shown in Fig. 55.32.
5.41
(a) IPPM(t) = A cos ["'et + kl'm(t)]
When this 'l'PtI(f) is p&SSl'd through an ideal FM demodulator, the output is kpT;/(t) This signal. when passed
through an ideal intt'grator. yields kpm(t). Hence. FM demodulator rollowed by an ideal integrator acts as a PM
demodulator. However. if met) has a discontinuity. met) :::; 00 at the point(s) of discontinuity. and the system
will f:\il.
(b) <PYM(t) = A cos [wcf+klj\·71(o.)do.]
\\'h"n this signal .pnl (t) is passed through an ideal PM demodulator. t he output is k f r Tn (0) do. When this
signal is passed through an ideal differentiator. t.he output is klm(t}. Hence. PM demodulator. followed by an
ideal differentiator. acts as FM demodulatl,)T regardless of whether mit) has jump discominuities or not.
5.42 Figure S5.42 shows the waveforms at point.s b. c. d. and e. The filure is self explanatory.
5.4·3 From Eq. (5.30). the Lapla('e transform of the phase error 8.(t) is given by
Ilnd
= .• , e.(.,)
,. + AI\ H(s)
for fI,rt) = kt
2
. e,( .• ) and
•
The steadyst ate phase error [Eq. (5 33)J is
lim 8.(/.) = lim .. ) = ( 2\K) = :)0
1.... .1) " +
Hence. the incoming signal cannot be tracked. If
and
,,+a
,
"
then
lim 8,(') = lim 11ge ( .. ) = lim 2 ) =
1. .0 _08 +, II + a Ako
lienee. the incoming signal can be tracked within a constant phase 2k/Aka radians. Now. if
2
H()" +n!;+b
.• 
then
lim = lim .,ee(") = lim !I I) = 0
,._"". .0 .0.,· + + a., + I
In this ('ase. t he incoming signal can he tracked with zero phase error.
48
1'1 '/1'
J
r
Fig. S5.42
49
Chapter 6
G.l·l The bandwidths of Rl(t) and 112(t) are 100 kHz and 150 kHz, respectively. Therefore t.he Nyquist sampling rates
for 111 (t) is 200 kHz and for I1'Ct) is 300 kHz.
Also 1112 (t) <=> ';;Rl("';) •. qt{\OJ}, and from the width property of convolution the bandwidth of 1112(t) is twice
the bandwidth of .ql(f) and that of 112:\(t) is three times the bandwidth of 112(t) (se also Prob. 4.310). Similarly
the bandwidth of 111 (t)R2(t) is the !um of the bandwidth of Rl (t) and R2(t). Therefore the Nyquist rate fol' RI2(1)
is 400 kHz. fol' 112:\(t) is 900 kHz. for .ql(t)112(f) is 500 kHz.
0.12 (a)
sinc100",') <=> O.OIrect
The bandwidth of this signal is 100 1\' rad/s or SO Hz. The Nyquist rate is 100 Hz (samples/sec).
(h)
sinc
2
(1001l'f)
The band\\'idth of this signal is 20011' radls Clr 100 Hz. The Nyquist rate is 200 Hz (samples/sec).
(c)
sine (lOOr.f) + sinc (507l't) <=> O.OIrect 0.01 + Q.02rect
The bandwidth of the first term on the righthand side is 50 Hz and the second term is 25 Hz. Clearly the
bandwidth of the compositE' signal is the higher of the two, that is. 100 Hz. The !\yquist rate is 200 Hz
!sarnplesjsec).
(d)
Sinc(IOO:rt) + 3sinc
2
(601l't) 0.01 + 10 t:.(2:W")
The bandwidth ofrect(1fo;) is 50 Hz and that of is 60 Hz. The bandwidth of the sum is the higher of
the t\\'o. that is. 60 Hz. The :Oyquist sampling rate is 120 Ih.
(e)
sinc(50",t) <=> 0.02 rect( 163 .. )
sinc(lOO"'t) <=> 0.01 rect(m;)
The t\\'o signals have bandwidths 25 Hz and SO Hz respectively. The spectrum of the product of two signals is
1/21r times the convolution of their spectra. From width property of the convolution, the width of the convoluted
signal is the sum of the widths of the Signals convolved. Therefore. the bandwidth of sinc(S01l'f)sinc(lOO",t) is
25 + 50 = 75 Hz. The Nyquist ra.te is 150 Hz.
6.13 The pulse train is a periodic signal with fundamental frequency 2B Hz. Hence, \OJ. ;: 21r(:lB) = 411' B. The period
is To = 1/2B. It is an even function of t. Hence, the Fourier series for t.he pulse train can be expressed as
""
PT.(t) = Co+ LCnCOS)')..!.t
"1
Csing Eqs. (272). we obtain
1 /1/1
6
8 1
no = Co ""  lit = .
To 1/16S 4
? /1/1
68
2
an = en =..:.. cosnw.fdt =  sin
To 1/II;S mr 4
Hence.
1i( t) = g(t )PT. (t)
= .!.q(t) + 2 sin (!!) g(t) cos )');0).'
4 £ Il7r 4
n=l
::'0
0.2 G<<al)
(b)
1
<al
,

5 S "(Hzl
I
\
9;1)
(d)
;
\
(e)
I
\ t
t \
I
\
/
\
\
0.4 0.2 0 1 «>It lOK
20 5 S 20 "(Hz)
I
.......

\
,9(1)
<0
(e)
I"'" 401( 10K
10K
20 5
f
5
20 ,. (Hz)

!
9(')
Practical Filter
"4
(h)
(g)
FiK. S6.14
6.14 For g(,) == sinc
2
(S",') (Fig. S6.14a). the spectrum is 0(101) = (Fig. S6.14b). The bandwidth I){ this
signal is !; Hz (1071' rad/s). Consequently. the Nyquist rate is 10 Hz, that is. v.e must sample the signal at a rate
no less than 10 samples/so The Nyquist interval is T == 1/2B = 0.1 second.
Recall that the sampled signal spectrum consists of (1/T)G(",') = Y repeating periodically with a period
equal to the sampling frequency f. Hz. We present this information in the following Table for tfiree sampling
rates: f. = 1; Hz (undel'sampling). 10 Hz (Nyquist rate). and 20 Hz (oversampling).
sampling frcquenr}' J. sampling interval T
_tG(w) I
comments
I
5 Hz 0.2 I Undersampling I
10 Hz 0.1 I Nyquist Rute I
20 Hz 0.05 I Oversampling I
In the first rase (undersampling). the sampling rate il!l 5 Hz (5 samples/sec.), and the spectrum tC( ..... ) repeats
e\'ery S Hz (lOr. rad/sec.). The successive spectra overlap, as shown in Fig. S6.14d, and the spectrum 0( .. ;) is
not reco\'elable from G( ..... ). that is. g(t) cannot be reconstructed from its samples get) in Fig. 56.14c. If thl?
sampled signal is passed through an ideal lowpass filtt!1' of bandwidth I) Hz. the output bpectrllm is reel (.;0,,).
51
and the output signal is 10Ilinc(201rt). which is not the desired sinc
2
(51rt). In the second case. we use
the !'Iyquist sampling rate of 10 Hz (Fig. 56.14e). The spectrum G(w) consists of b.!.cktoback. nono\'erlapping
repetitions of tG(w) repe3ting every 10 Hz. Hence, G(w) can be recovered from G(w) using an ideal lowpass
filtpl' of bandwidth 5 Hz (Fig. 56.14fL The output is 10sinc
2
(5r.t). Finally, in the last case of oversampling
(sampling rate 20 Hz). the spectrum G(w) consists of nonoverlapping repetitions of tG( ... ·) (repeating every
20 Hz) with empty band between successh'e cycles (Fig. S6.14h). Hence, G(w) can be recovered from G(w)
using an ideal lowpass filter or even a practical lowpass filter (shown dotted in Fig. S6.I4h). The output is
20 sinc 2(51ft).
6.15 This scheme is analyzed fully in Problem 3.41. where we found the bandwidths of 1/1 (t). met). and,,(t) to be 10
kHz. 5 kHz. and 15 kHz. respectively. Hence, the Nyquist rates for the three signals are 20 kHz. 10 kHz. and 30
kHz. respectively.
6.16 (a) When the input to thill filter is h(t). the output of the summer is "(t)  bet  T). This acts as the input to
the integratol'. And, "(f). the output of the integrator is:
1/(t):: 1'(,,(1')  b(1'  T)I d1' = vet)  fI(t  T) :: rect (t ;.')
Thp impulse response "(f) is shown in Fig. 56.16a.
(b) The transfer function of this circuit is:
and
H(w):: TfillC( ",,;),.j..,T/2
IH(",,)!:: T lSinc( w;) I
The amplitudp response of the filter is shown in Fig. 56.I6b. Observe that the filter is a lowpass filter of
bandwidth 2rr/T radls or l;T Hz.
The impulse response of the circuit is a rectangulal pulse. When a sampled signal is applied at the input. each
sample f;enerates a rectangular pulse at the output. proportional to the corresponding sample value. Hence the
output IS a staircase approximation of the input as shown in Fig. 56.16c.
,
1,'1 .,
I ....
.
T
4T
e
6.1·7
Figure S6.1·6
(8) Figure S6.Iia shows the signal reconstruction from its samples using the firstorder hold circuit. Each
sample gpnprates a triangle of width 2T and centel'ed at the sampling instant. The height of the triangle is equal
to t hI:' sample value. The resulting signnl consists of straight line !legmt!nts joining the sample tops.
(b) The transfer fllnction of this circuit is:
1I(,.;) :: .r{h(tH = .r {A ) } :: T sinc
3
(w{)
Because is positive for all iJJ, it also represents the amplitude response. Fig. S6.17b sholl.·s the impulse
response h(t) = A( ,*). The corresponding amplitude response H(:..J) and the ideal amplitude response (lOWPll.£s)
required fol' signal reconstruction is shown in Fig. 56.l7c.
(c) A minimum of T sees delay is required to make h(tl causal (realizable). Such a df!lay would cause the
reconstructed Signal in Fig. S6.17a to be delayed by T sees.
(d) When the input to the first filter is "(t). then as shown in Prob. 6.14. its output is a rectangular pulse
l'U) = 'let) 1I(t  T) shown in Fig. S6.l4a. This pulse pet) is applied to the input of the second identical filter.
The output of the slimmer of the second filter i5p(t)  p(t 1') = fI(t)  211(t  T) + ,,(t  2T), which is applied
to the integrator. The output II(t) of the integrator is the area under p(t)  p(t  T). which. as
h(l) = 1'(11(1')  211(1'  T) + u(r  2T)Jtlr = tu(t)  2(t  T)u(t  T) + (t  2T)'l(t  2T) = A C
shown in Fig. S6.Jib.
6.18 As.c;ume n signalll(f) that is fiimuitaneously timelimited and bandlhnited. Let g(iJJ) = 0 for Iwl > 21f D. Therefore
,q( •• :;rect<'l:u') = 11( .. :) fOI B' > B. Therefore from the timeconvolution property (3.43)
l1(t):: get) * [2B'sinc(21rB't)j
= 28'I1(t). llIinc(27rB't)
H(oJ)
rT .
C .. ) t. ...
o
. .
(A)..,
(c:! )
Filure S6.1T
Because R(t) is timelimiteti. g(t) = 0 for It I > T. But .q(t) is equal t.o convolution of g(t) with sinc(271'B't) which
is not timelimitt'd. It is impossible to obtain a timelimited signal from the convolution of a timelimited signal
with a nontimelimited signaL
6.21 (a) Since 128 = 2". we need 7 bits/charact.er.
6.22
6.23
(b)For 100,000 characters/second. we need 700 kbit.s/second.
(a) 8 bit.s/charact.er and 800 kbits/second.
<a) The bandwidth is 15 kHz. The Nyquist rate is 30 kHz.
(b) 65:;36 = 216. so that 16 binary digits art' needed to encode each sample.
(e) 30000 x 16 = 480000 bits/so
(d) 44100 x 16 = 705600 bits/so
(a) ThE' ]';yquist rate is 2 x 4.5 X 10
6
== 9 MHz. The actual sampling rate == 1.2 x 9 == 10.8 MHz.
(b) 1024 = 210. so that 10 bits or binary puLces are n!'eded to encode each sample.
(e) 10.8 x 10
6
X 10 = lOS x 10
6
or 108 Mbits/r..
6.24 If TIIp is tht' peak sample amplitude. then
6.25
6.26
quantization error < (0.2)(fllp) =
 100 500
!ecnll5e the maximum quantization error is == == T, it follows that
!!!J!. ==> L = 500
L  500
BecausE' L should be a power of 2. we choose L "" 512 = 29. This requires a 9bit binary code per sample The
rate is 2 x 1000 == 2000 Hz. 20% above this rate is 2000 x 1.2 == 2400 Hz. Thus. each signal hc.s 2400
samples/second. and each sample is encoded by 9 bits. Thp.refore. each signal uses 9 x 2400 == 21.6 kbits/second.
Fi\'t' such signals are multiplf'xed. hence. we need a total of 5 x 21.6 = 108 kBits/second data bits. Framing
and synchronization requires additional 0.5% bits. that is. 108,000 x 0.005 == 540 bits, yielding a total of 108540
bit.s/second. The minimum transmission bandwidth is ¥ = 54.27 kHz.
Nyquist rate for each Signal is 200 Hz.
The sampling rate f. = 2 x Nyquist rate = 400 Hz
Total numbt'r of samples for 10 signals = 400 x 10 = 4000 samples/second.
Q
.. < O.2
00p
!:!.!
uantlzatlon error _ 1 = 400
Moreo"er. quantization error = = = T =:ii ==> L = 400
Because L is a power of 2. we select L = 512 == 2". that is, 9 bits/sample.
Therefore., the minimum bit rate = 9 x 4000 = 36 kbits/second.
ThE' minimum cable bandwidth is 36/2=18 kHz.
For n sinusoid. == 0.5. The S],;R = 47 dB =50119. From Eq. (6.16)
mp
So = = 3L
2
(0.5) = 50119 ==> L == 182.8
No 111;;
Because L is a ?ower of 2. we select L = 256 == 28. ThO! SNR for this "alue of L is

== 31.:1 == 3(256)2(0.5) = 98304 == 49.43 dB
''0 111;;
53
"
......... ,.
".
'.
Fi •. 56.2.7
.. ..
'.
6.27 For this periodic mIt). each quartp.r cycle takes on the same set of amplitude values. Hence, each quarter cycle
contributes identical energy. Consequently, we can compute t.he power for this signal by averaging its energy
over a quartel' cycle. The equation of the first quarter cycle u shown in Fig. 86.2·7 is met) = 4A/To. where A is
the peak amplitude and To is the period of m(t). The power or the mean squared value (energy averaged over
a quarter cycle) is
mJ(I) A2{, J.
[ence. = • = .\.
m
p
,.. .
",:Z(t) = __ _ dt =
1 lTot .. (4A)2 A2
To/4 0 To 3
The rest of the solut ion is identical to that of Prob. 6.26. From Eq. (6.16). SNR c.f 47 dB is a ratio of 50119. iF;
L is a power of 2. we select L = 256 = 2
s
. The SNR for this value of L is
So 2
;:; = 3L =r = 3(256) (1/3) = 65536 = 48.16 dB
TTlp
0.28 Here " :::: 100 and the S:'IiR = 45 dB= 31,622.77. From Eq. (6.18)
So 3L
2
No = (In 101)2 = 31,622.77
== 1. = 473.83
Becallsl' L is a power of 2. we select L = 512 = 211. The for this value of L is
§. _ 3(512)2 _ ? _
No  (In 101)2  369_2.84  45.67 dB
6.29 (a) :'Iiyquist rate = 2 x 10
6
Hz. The actual sampling rate is 1.5 x (2 x 10
6
) = 3 X lOb Hz. Moreover. L = 256
and I' = 255. From Eq. (6.18)
So 3L
2
3(256)2
No = (In(,1 + 1)]2 = (In 256)2 ... 6394 ::: 38.06 dB
(b) If we reduce the sampling rate and increue the value of L so that the same number of bits/second is
maintained. we can improve t.he SNR (because of increased L) with the same bandwidth. In part (a). the
sam.,ling rate is 3 x 10
6
Hz and each sample is encoded by 8 bits (L :::; 256). Benct'. the transmission rate is
8 x 3 X 10
6
:::; 24 Mbits/second.
If we reduce the sampling ratf! to 2.4 x 1011 (20% above t.he Nyquist rate). then for t.he same tran!imission rate
(:24 we can hswe (24 x 10
6
)/(2.4 x 10
6
) = 10 bits/sample. This results in L :::; 2
10
== 1024. Hence. the
new S!'R is
.§. _ 3L
2
_ 3(1024)2 _ ? _
No  Iln(l£ + 1)]2  (In 256)2  10_300  50.1 dB
Clearly. the SNR is increased by more than 10 dB.
6.210 Equation (6.23) shows that increasing n by one bit increases the 8NR by 6 dB. Hence. an increase in the SNR
by 12 dB (fro111 30 to 42) can be accomplished by increasing n from 10 to 12, that is increasing by 20%.
6.4·1 (a) From Eq. (6.33)
so that 1
 (".)(64.000) 008
 211' x 800 ==> 11  . { .1
54
(b)
N = .,.2l! = (0.0785)2(3500) = 1 12 )( 104
o 3/. (3){64000) .
(c) Here So = tf = 0.5. and
So 0.5 446 10:1
No = 1.12 )( 104 =. )(
(d) For uniform distribution
r ,,,2 1
So = 111 (t) = 1! = 
3 3
so that
So 0.333 = 2.94 x 10:1
No = 1.12 X 10
4
( ~ ) For onof!' signaling with a bit r a t ~ 64 kHz. we need a bandwidth of 128 kHz. For a bipolar case. we need a
bandwidth or 64 kHz.
Chapter 7
7.21 For full width rect pulse p(t) = rect(
P( aJ) = 7j, sinc ( aJ? )
For polar signaling [see Eq. (7.12»)
Ip(tUt (tU7i )
Sy(aJ) = = 7j, sinc:
2
Tb 2
For onoff case [see Eq. (7. lib)]
I
p(tUt [ 2 GO ( 2)]
Sy(aJ) = 1+" r 6 aJ=
47b 7b _0 7b
= It sinc
2
(tU
7b
) [1 + 2tr f 6 (aJ _
4 2 7j, 11=_ 7j, J
21171
for aJ =  for all n _ 0 ,and = 1 for n = O. Hence,
T6
But sinc
2
( ) = 0
Sy (aJ) = 7b sinc
2
(aJ
7b
) + .!. 6( aJ)
4 2 2
For bipolar case [Eq. (7.20b)]
Sy(aJ) = Ip(aJt sin2 (aJ7j,)
7b 2
:: 7b sinc
2
( ) sin2 )
The PSDs of the three cases are shown in Fig. S7.21. From these spectra, we fmd the bandwidths for all
three cases to be Rb Hz.
The bandwidths for the three cases, when halfwidth pulses are used, are as follows:
Polar and onoff: 2Rb Hz; bipolar: Hz.
Clearly, f"r polar and onoff cases the bandwidth is halved when fullwidth pulses are used. However, for
the bipolar case, the bandwidth remains unchanged. The pulse shape bas only a minor influence in the
bipolar case because the tenn sin
2
(tUl ) in Sy{tU) detennines its bandwidth.
l'
S!jl"')
o
,.l.r
___ eJ'/\  Off
"R&
Fia·57.21
S6
7.11
YH:)
R • · I I •
. nrnn r
llJ crOlOlT't....,.
I
(b)
FiC·
57
•
1
1
(
t+.!i] (t.!i]
P{tj ,%4 __ ,%4
and
P(",) = tSinC ( "'.:0 ) e14/1b
14
+ t
SinC
( "'1 ) ej4/1b/
4
= j16 sinc ( O1:b ) sin ( 01: )
S ( )
Ip(01t 1i . 2 (01Tb) . 2
Y 01 = = b SIDC  SID
16 4 4
From Fig. 57.22, it is clear that the bandwidth is 4" rad I s or Hz.
16
7.13 For differential code (Fig. 7.17)
= lim ..!...[.!(1)2 + N (1)2J= 1
N+«l N 2 2
To compute R\, we observe that there are four possible 2bit sequences II, 00, 01, and 10, which are
equally likely. The product 0kok+\ for the fU'St two combinations is 1 and is 1 for the last two
combinations. Hence,
R\ = lim 0
N+rs> N 2 2
Similarly, we can show that Rn = 0 II> 1 Hence,
Sy(.,j • (!f) .... ' ( "n
7.14 (a) Fig. 57.24 shows the duobinary pulse trainy(t) for the sequence 1110001101001010.
(b) To we observe that on the average, half the pulses have ak = o and the remaining half
have Ok = I or I. Hence,
Ro = lim ..!...[N (±1)2 + N (O)]=.!.
N+«l N 2 2 2
To determine Rl' we need to computeakak+\' There are four possible equally likely sequences of two bits:
11,10,01,00. Since bit 0 is encoded by no pulse (Ok = 0), the product of 0kok+l = o for the last three of
these sequences. This means on the average 'E:! combinations have akok+l = 0 and only N combinations
4 4
57
i
I
I
have nonzero akak+I' Because of the duobinary rule, the bit sequence 11 can only be encoded by two
consecutive pulses of the same polarity (both positive or both negative).
This means ak and ak+l arc 1 and 1 or 1 and  1 respectively. In either case akak+l = 1. Thus,
these: combinations have akak+1 .. 1. Therefore,
Rl = lim ..!.[N
N"'rI!l N 4 4 4
To compute R2 in a similar way, we need to observe the product akak+2' For this we need to observe all
possible combinations ofthrcc bits in sequence. There are eiaht equally likely combinations: 111, 101,
110, 100,011,010,001, and 000. The last six combinations have either the first andlor the last bit O.
Hence, akak+2 = 0 for all these six combinations. The fU'St two combinations arc the only ones which
yield nonzero akak ... 2' Using the duobinary rul:, the flfSt combination is encoded by three pulses of the
same polarity (aU positive or negative). Thus ak and ak+2 arc 1 and I or 1 and I, respectively, yielding
akak+2 = 1. Similarly. because of the duobinary rule, the fIrSt and the third pulses in the second bit
combination 101 are of opposite polarity yielding akak+l. I. on the average, akak+2 = 1 for
N N 3N
 termS,1 for terms, and 0 forterms. Hence,
884
R2 = lim N (1)+ 3N (0)]== 0
N ... c N 8 8 4
In a similar way we can show that Rn = 0 n> I , and from Eq. (7.1 Oc), we obtain
Ip(cut Ip(cut (CU1i )
S (cu) = (1 +coscu7b) = cos
2
_b
y 2T" T" 2
1 1 1 0 0 0 1 1 • 1 0 0 1 0 1 0
onn, , , au ' n , , n , a '
'1&57.14
For halfwidth pulse P(I) ... rect(211T,,}.
Sy(OJ)'" sinc
l
(OJ: )cos2 (OJ: )
From Fig. S7.24 we observe that the bandwidth is approximately Rb 12 Hz.
7.31 From Eq. (7.32)
(I +,)6000 1
4000= =>,=
2 3
58
7.32 Quantization error AV = mp S 0.0 Imp => L:2! 100
2 L
(a) Because L is a power of 2, we select L = 128 = 2
7
(b) This requires 7 bit code per sample. Nyquist rate = 2 x 2000 ... 4 kHz for each signal. The sampling
rate I, = 125 x 4000 = 5 kHZ.
Eight signals require 8 x 5000 ... 40,000 samples/sec.
Bit rate .. 40,000 x 7 = 280 kbitsls. From Eq. (7.32)
= l.2x280)(103 = 168kHz.
2 2
7.33 (a) BT = 2Rb => Rb = 1.5 kbitsls.
(b) BT = Rb => R" = 3 kbitsls.
1+, 1.25 .
(e) BT  2 R". Hence, 3000 = 2 R" => Rb = 4.8 kbltslS.
(d) BT = Rb => R" = 3 kbitsls.
(e) BT = Rb Rb == 3 kbitsls.
7.34 (a) Comparison of P{t'l» with that in Fig. 7.12 shows that this p{t'l» does satisfy the Nyquist criterion with
t'l> b = 21r )( 10
6
and, = 1. The excess bandwidth t'l> x = Ir X 10
6
.
(b) From Table 3.1, we fmd
p(t) = sinc
2
(lrX 106t)
From part (a), we havet'l>b = 2lrx 10
6
and Rb = 10
6
. Hence, 1j, = 106. Observe that
p(nTb)t n=O
Hence p{t) satisfies Eq. (7.36).
(e) the pulse transmission rate is..!...::: Rb = 10
6
bitsls.
Tb
7.35 In this case = I MHz. Hence, we can transmit data at a rate Rb = 2 MHz.
Also, BT = 12MHz. Hence, from Eq. (7.32)
1.2 x 10
6
.. I;, {2x 106)=>, = 02
7.36 h = 700 kHz. AlSO,.!} = 500 kHz and Ix ... 700  500  200 kHz.
Hence, r ... = 0.4 and II = Rb  Ix = 500  200 ::: 300 kHz.
2
7.37 To obtain the inverse transform of P{CI), we derive the dual ofEq. (3.35) as follows:
g(t  T) <=> G(t'l>)e}TAI and g(1 + T) <=> G(t'l»efl'AI
Hence,
g(t + T) + g(1 T) <=> 2G(aJ) cos Tt'l>
(1)
Now, p(t'l» in Eq. (7.34a) can be expressed as
p{ CI) = 1. +
2 41rRb 2 41tRb 2Rb
(2)
59
7.38
7.39
Using Pair 17 (Table 3.1) and Eq. (1) above, we obtain
pet)  Rb
sinc
(2trRbl)+tSinC 2!b)]
= (2K R"t)+iSinC + K)+ tSinC (2K Rbi K)]
[
Sin(2trR"l) 1 sin(2trR
b
l+tr)' ,I Sin(2trR
b
tK)]
=R" + .fL
2tr Rbi 2 2tr R"t + K 2 2K  K
= R,,[Sin(2trRb/) _! Sin(2KR"t) _ t
2tr Rbt 2 2tr Rbt + tr 2 2tr R"t K
R
. (2 R)[ 1 1/2 112]
• b sm tr b
t
   ::
2tr (2tr R"I + tr) (2K  tr)
= Rb ( 1 2)]
12trR,,114R,,2,
2RbCOStrRbtsintrRbt R"COStrRbt. ( R )
= = smc tr bl
2tr t4R,,2t2
Hence,
p(t)., sinc (tr Rb
t
) +sinc [tr R,,(t 
sin tr Rbi sin( tr R"t tr)
=
tr Rbt tr Rbi  tr
• sin tr R"t _ sin K Rbl:= sin K R"t
tr tr Rbi tr tr R,,1(1 Rbi)
The Nyquist interval is =.La 1b. The Nyquist samples are p (±"r,,) for" = 0, 1,2, ......
Rb
From Eq. (7.16), it follows that
p( 0) • P( r,,) = 1 and p( ±nTo) = 0 for all other n.
Hence, from Eq. (6.10) with T, = r", and B = = _1_.
2 21b
PIt) sInc "/101 +smc
= sin K Rbi _ sin K R"t = sin K R"I
tr Rb tr Rbt  tr tr Rbl(l Rbi)
60
"
The Fourier transfonn of Eq. (1) above yields
P(m) = _I + _I reet(",,!!}jOJ/ Rt
R" 2" Rb R" 2" R"
= _1 LjOJ/2Rt
Rt, 2"Rb r
= :b co{ ) ree{ 2:R
b
} 
7.310 (a) No error because the sample values of the same polarities are separated by even number of zeros and
the sample values of opposite polarities are separated by odd number of zeros.
(b) The fU'St sample value is 1 because there is no pulse before this digit Hcnce the first digit is I. The
detected sequence is
11000100110110100
7.310 The fU'St sample value is 1, indicating that the transmissions starts with a positive pulse, that is, fll'St digit I.
The duobinary rule is violated over the digits shown by underbraeket.
12000 200
Following are possible correct sample values in place of the 4 undcrbracket values: 220 2, or 2 0 2 2,
or 0 00 2, or 2 0 0 O. These sample values represent the following 4 digit sequence: llOO, or 1000, or
0100, or 1010. Hence the 4 possible correct digit sequences are
where is any of the four possible sequences 1100,1000.0100. or 1010.
7.41 S = 101010lOOOOOIll
From example 7.2
T = (I ED D3 ED D
S
ED D6 ED D9 ED D
IO
ED Dli ED Dl2 ED D13 ED DISED ... )S
R = (lED D3 ED DS)T
T = 101110001101001
R = 101010100000111 = S
7.41 S == 101010100000111
T. (lED D2 ED D4 ED D6 ED D8 ED DIOED Dl2 ED D
I4
ED ••• )S
>
R • (1 ED D2)T (see Fig. 57.42)
T
T .. 100010000000110
R = 101016100000111 = S
7.43 S 101010100000111
T==(IED DED D2 EDD4 EDD' EDD8 ED D9 ED Dli ED D14)S
R={IEDDEDD
3
)T (seeFig.S7.43)
T ... 110111101001011
R = 101010100000111 "" S
61
I ! I
>
FII57.41
ct
T
R
FII. S7."3
7.51 From Eq. (7.45), we obtain
[
C_l] [ I
CO" 0.1
cl
0.3
_0.07]1[0]
0.3 I = 1.07
1 0 0.1
A
7.61 Ca).J!.. = 5
(0 Forpolarcase p. Q(5)=2J7 x I0
7
(il) P, Q(512)=0.00621
(iii) For bipolar case p, = 1.5Q(5/2) = 0.009315
In the following discussion, we A p = A, the pulse amplitude.
(b) Energy of each pulse is E p = A2 Tb 12 and there are Rb pulses/second for polar case and
pulses/second for onotT and bipolar case. Hence, the received powers are
A21; A2 (0.0015)2
Ppolar = = .. = l.l25 x 106
222
A
2
16 A2 6
P onoft' = x " =   0.5625 x 10
2 2 4
A2Tb A2 6
l\ipolar = 2 )( 2 ="4 = 0.5625 x 10
Cc) For onotT case:
We require P(e) == 2J7 x 10
7
.. Q(Ap 120',,). Hence,
Ap 120',," 5 andA
p
" lOa" = 0.003
A2 (0.003)2 6
Ponotf = .. = 225)( 10
4 4
For bipolar case:
A
P( e) == 2J7)( 10
7
= 1.sQ( Ap 120',,) =>.J!.. == 5.075
a"
Hence
A "" A p = 5.075 x 20'" = 0.003045
and
7.61 For onotT case:
P, = 106 S,./ => Ap
20'"
62
all =10
3
=> Ap
for onoff case, half the pulses are zero, and for halfwidth rectangular pulses, the transmitted power is:
Sj '"'  :.:.P ,",:.:J!...... .. 22.56 )( 106 watts.
I(A 2) A 2 (95)(10
3
)
2 2 4 4
There is an attenuation of 30 dB, or equivalently, fa ratio of 1000 during transmission. Therefore
Sr loooS; = 22.56)( 10
3
watts
7.63 For polar case:
P
e
.. lOoo{, = ,.[ = Ap .. 4.75 => Ap = 4.75)( 10
3
all
For polar case with halfwidth rectanaular pulse:
5, ( 1I28xI0"'_
Sr .. (I000Xll.28)( 10oo{,W) = 1l.28)( 10
3
\yatts
For bipolar case:
p. .. lOoo{, = 1 c"/ Ap )= Ap =4.835 andAp =4.835)(2)(10
3
.. 9.67)( 10
3
""¥l2a" 2a"
For bipolar (or duobinary), half the pulses are zero and the receive power Sj for halfwidth rectangular
pulses is
A 2 1 2
Sj = : = 4(9.67)( 10
3
) '"' 23.38)( 1000{, watts
Sr .. (IOOO)S; = 23.38)( 10
3
watts
7.7'1 Sampling rate .. 2)( 4000)( 1.25", 10,000 Hz.
m
Quantization error.. { .. 0.00 Imp => L = 1000
Because L is a power of2, we select L = 1024 = 210. Hence, n 10 bits/sample.
(a) Each 4ary pulse conveys log
2
4 .. 2 bits of information. Hence, we need = 5 4ary pulses/sample,
2
and a total of5)( 10,000 = 50,0004ary pulses/second. Therefore, the minimum transmission bandwidth is
50,000 .. 25 kHz. .
2
B
R" (I + ,) 50,000( 1.25) 3.2 kHz.
(e) r""" 1 S
2 2
7.73 (a) Each Sary pulse carries log2 8 = 3 bits of information. Hence, the bandwidth is reduced by a factor of
3.
(b) The amplitudes of the 8 pulses used in this Sary scheme are±AI2, ±3AI2, ±SAI2, and ±7AI2.
Consider binary case using pulses ± A 12. Let the energy of each of these pulses (of amplitude ± A 12 )
be Eb' The power of this binary case is
l\iuy = EbR"
Because the pulse energy is proportional to the square of the amplitude, the energy of a pulse ± is
,,2 Eb' Hence, the average energy of the 8 pulses in the 8ary case above is
63
Hence,
Therefore,
,..1f)I = 7 l\illll')'
7.71 (a) M = 16. Each 16ary pulse conveys the infonnation oflog2 16 ... 4 bits. Hence, we need
12000
'4 = 3000 16ary pulses/second.
Minimum transmission bandwidth = 3000 ... 1500 Hz.
2
2
(b) From Eq. (7.32), we haveR" =Br. Hence,
1+,
2
3000 = 1.2 Br = Br = 1800 Hz.
7.74 (a) For polar signaling. Rb bits/second requires a bandwidth of Rb Hz. The halfwidth rectangular pulse of
amplitude has energy
2
Eb = 7b = A
2
7b
2 2 8
A27j A2
The power P is given by P = EbRb ... Rb =
8 8
(b) The energy of a pulse ± is k
2
E
b
• Hence the average energy of the Mary pulse is
EM. +2(±3)2 +2(±5)2+ ..... +2[±(M I)t Eb]
M2
= 2Eb t (2k + 1)2
M tO
M2_1
Each Mary pulse conveys the infonnation oflog
2
M bits. Hence we require only Rb Mary
log2 M
pulses/second. The power PM is given by
P = EM& • (M2_I)Rb E
b
=(M
2
1)A
2
== M2A2
M 1012 M 310g
2
M 241012 M 2410g
2
M
7.75 Each sample requires 8 bits (256 = 2
1
). Hence: 24,000)( 8 = 192,000 bits/sec.
Br 30kHz
2 2
R =  Br 0= I (30,000) .. 50,000 Mary pulses/sec.
1+, .2
We have available up to 50,000 M.ary pulses/second to cransmit 192,000 bits/sec. Hence, each pulse must
transmIt at east ... Its.
. 1 (192,000) 384b'
50,000
64
= choose 4 bitslpulse
= M  16 is the smallest acceptable value
7.81 (a) Baseband polar sip at a rate of 1 Mbitslsec
and using full width pulses hasBW = IMHz. PSK
doubles the BW to 2MHz.
(b) FSK can be viewed as a sum of2 ASK signals.
Each ASK signal BW = 2 MHz. The first ASK signal
occupies a band I cO ± 1 MHz, and the second ASK
signal occupies a band lei ± 1 MHz. Hence, the
bandwidth is 2 MHz + 100 kHz = 2.1 MHz.
Fie·57.8l
7.12 (a) A baseband polar signal at a rate 1 Mbitslsec using Nyquist criterion pulses at , = 02 has a
BW= (1+') R" .12 xl0
6
=6.0xl0'Hz.
2 2
PSK doubles BWto 1.2 MHz.
(b) Similar to Prob. 7.81.
BW
FSK
= 0.6 MHz + 0.6 MHz + 100kHz
BWFSK = 1.3 MHz
7.83 log2 M ... 2 for M = 4.
7.91
We need to transmit only OJ x 10
6
4ary pulses/sec
(a> BW is !educed by a factor of2.
BW
FSK
= 1 MHz
(b) In FSK, there are four center (carrier) frequencies
lei, te2' 1c3' and/
e
4, each separated by 100kHz.
Since ASK signal occupies band Ie ± OJ MHz, the total
bandwidth is
OJ MHz + 0.5 MHz + 100kHz + 100kHz + 100kHz = 1.3 MHz.
l t""'.....,
"'.z.Lt) t...,
,
."", It)
I
6S
Fie. 57.83
,
I:
"'.rt>."..,.£'
"
I  t
\ tc. /
'r
Fla. (b)
Fla. S7.91
Either figure (a) or (b) yields the same result.
ml(t) has 8400 samples/sec.
·
4 c.: 7..z00
mz(t), m3(t), m..(t) each has 2800 samples/sec.
Hence, there are a total of 16,800 samples/sec.
7.92 First, we combinemz(t), m3(t), and m4{t) with a commutator speed of 700 rotations/sec. This combined
signal is now multiplexed with ml (t) with a commutator speed of 2800 rotations/sec, yielding the output of
5600 samples/sec.
1, c: 21DO roTatioflS
') Sec..
Fia. S7.92
66
7.93
rt'I.lt')
.f:.; .:: 72CO
f :::. 2. J &00
'"
kb/.s
c.. I
> •
g.
of:; :2.¥Co
Cll r ......
" £
0
'((\if:: ')
G. \l49.;t'1:1.("
'>
•
Cohr
,:zq'b /tb/S
,
/
'.

,/
·fs2.ifC{I
l'iNet.1L1J1I...6
• !,
Q:leJe(,
a frt:l r
c
tt
)
(b)
Fla S7.93
67
Chapter 8 Exercises
tUl If a plesiochronous network operates from Cesium beam clock which is accurate to ± 3 parts
in 10
12
, how long will it take for a DS3 signal transmitted between two networks to become out'
of sync if a 1/4 bit length time error results in desynchronization?
Solution: A DS3 bit is transmitted in 1/(44.736.10
6
) = 2.235336.10
8
sec. In the worst case,
both network clocks will be out of synchronization by 6 parts in 10
12
.
2.235336.10
8
/(6.10.
12
) = 3922.27 seclbit or 980.57 sec/lA bit
8.12 For the bit stream 011100101001111011001 draw an AMI waveform.
Solution:
Note that typically, for illustrative purposes, the waveform is given as
8.13 For the following waveforms, determine if each is a valid AMI format
for a DS 1 signal. If not, explain why not.
a.
JU'lllllllllllllllnlll'U
Solution: No. 16 O's violation
Solution: No. bipolar violation
c.
JUI
I I I I I I n I I , , I I , I ,
Solution: No.1' s density violation
d.
""U
JUI'II'I,nl""IIIIU""
Solution: Yes
68
.'
, '
8.14 a) You have received the following sequence of ESF framing pattern seQ.uence bits
•.• 00110010110010110 ...
Is this a"legitimate framing bit sequence in order to maintain
synchronization between the Tl transmitter and receiver?
Yes __ No __
H yes, why? H no, why not?
Solution: No. The bit sequence 0011 cannot be in an ESF framing bit sequence.
b) The following TI AMI signal is received:
,....

r
 


Is this an acceptable Tl signal"?
Yes_No __ _
a. H yes, explain.
b. If no, explain why not (what, if any. OS 1 standards are violated) and
draw on the figure the AMI wavefonn which would be transmitted by the DSU?
~
Solution: No. 16 O's violation. The 16 O's will be replaced by a pattern of 1 's by the
DSU.
8.15 The signal 11010000000000000001 is received by the DSU in a Tl data stream which uses a
B8ZS format. Draw the output of the DSU for this signal? The first 1 is already drawn.
Show the bit stream which is substituted by the DSU.
Solution:
8.16 T·l synchronization at two distant locations is controlled by separate crystal controlled
oscillators which differ in frequency by 125 parts per million. H the terminal equipment doesn't
maintain sync in how many complete D4 superframes will the faster oscillator have generated (at
most) one more time slot (8bit) than the slower oscillator? Circle the correct answer.
69
a) 5
b) 10
c) 15
d) 20
e) None of the above  if "none", what is the number of D4 superframes before an extra time slot
is generated? __ _
Solution: e) The faster oscillator will generate 125.1061.544.10
6
= 193 bits per second more
than the slower oscillator. This is one frame/sec = 24.125 time slots. Hence, a time slot
difference will be generated in 1124.125 = 0.04164498 frames or 0.0034704 superframes. .r
8.17 Two plesiochronous digital networks, A and B, utilize Cesium beam clocks accurate to 3
parts in 1013. The networks are operated by independent long distance companies and are
synchronized to each other by means of a UTe signal.
a. If a company leases a T1 line with D4 framing which is tenninated at one end in
network A and at the other end in network B, how often must the networks be resync'd to
each other to avoid a framing bit error in the customers Tl signal in the worst case? {You
may assume a framing bit error occurs when the two networks are out of sync by ~ 112 of
a T1 "bittime".}
Solution: A TI bit time is 1/(1.544.10
6
) = 6.47668·10,7 seclbit. In the worst case, the
two clocks would be off by 2·3 ~ 6 parts in 10
13
or 6·10,13 errored bits per bit transmitted.
Hence, 6.47668·10,7 seclbit 16.10.
13
errored bits per bit = 1.07945.10
6
seconds per errored
bit or 5.39723.10
6
seconds per errored halfbit.
b. UTe operates via GPS satellites which are approximately 23,000 miles above the Earth.
How long. in terms of Tl bits, will a correction signal take to be transmitted to
the network switches?
Solution: The speed of light is approximately 186000 miles/sec.
23000x2 = 46000miles up and down. 46000/186000 = 0.247 sec
0.247x1544000:: 381850 bits
10
. '
, .
10.11 (a)
52 2
1+ 1 I
(b) P(black queen) = 52 = 26
(e) P{picture card) =.!.!. = 2
52 13
4 1
(d) P(7) S2 =13
20 5
(e) P(nsS)"Sij'3
Chapter 10
10.12 (a) S = 4 occurs as (1,1,2),(1,2,1),(2,1,1). There are atotal of6x 6x 6= 216 outcomes.
Hence, P(S '" 4) = .!...._1
216 72
(b) S = 9 occurs as (1,2,6), (1,3,5), (1,4,4), (1,5,3), (1,6,2), (2,1,6), (2,2,5), (2,3,4), (2,4,3), (2,5,2), (2,6,1),
(3,1,5), (3,2,4), (3,3,3), (3,4,2), (3,5,1), (4,1,4), (4,2,3), (4,3,2), (4,4,1), (5,1,3), (5,2,2),
(5,3,1), (6,1,2), (6,2,1)
25
P(S=9)= 216
(e) S = IS occurs as (3,6,6), (4,5,6), (4,6,5), (5,4,6), (5,5,5), (5,6,4), (6,3,6), (6,4,5), (6,5,4), (6,6,3)
10
peS = 10) = 216
10. J 3 Note: There is a typo in this problem. The probability that the number i appears should be ki not k
j
•
6 I
1= rki = k +2k+3k+4k +5k+6k ",21k = k =
;=\ 21
P(i) = ;1 (i = 1. 2, 3, 4, 5. 6)
10.14 We can draw 2 items out of 5 in 20 ways as follows: 0,0
2
.0
1
0
3
, OIP" 01P
2
• 0
2
0" 0
2
0
3
, 02P" 02P2, 0
3
0.,
O)P .. 03P2, PIO .. P102, PIO
,
• PIP2• P
2
0" P
2
o" P
2
P,. All these outcomes arc equally likely with
probability 1120.
(i) This eventE\ = 011\ U021\ U0
3
P\ UI\
O
I UP\02U1\03UP:z01 UP202 UP
2
0
3
Hence. P(EI) =.!.!. =
20 5
(Ii) This event E2 • 1\ P:z U 1\
2 1
Hence, P(
E
2) = 20 = 10
(iii) This event E3 = 0102 UO\03 U
0
2
0
1 U
0
2
0
3 U030l U0302
6 3
Hence, P(E3) = :;;
. 20 10
(iv) This event E4 :;; E2 U E3 and both E2 &. E3 are disjoint.
Hence, P{E4 ) = P(E2)+ P(
E
3) = "" t
71
10.1·5 Let xo\ be the event that the flJ'St chip is oscillator and x 1\ be the event that the fU'St chip is PLL. Also,
let x0
2
and x Pj represent events that the second chip drawn is an oscillator and a PLL, respectively. Then
P{I osc and 1 PLL) 0: 1'(xo\ ,xPj) +
= P(XO\ )p(x Pj Ixo\)+ p(x,,\ IXI\)
5 4 5 4 5
10.16 Using the notation in the solution ofProb. 10.15, we find:
(a> p(xO:z Ix 1\) =
(b) p(xO:z Ixo
l
) =
10.1·7 (a) We can ways of getting two I's and eight O's in 10 digits
(
10)= 10! 45
2 2!8!
P (two 1 's and eight 0'5) = .. 45(Ojt _ ..
2 1024
(b) peat least four O's) = l[P(exactlyone 0)]+[ P(exactly two O's)]+[ P(exactly three O's)]
p(one 0) = (IP"O.5)IO = _5_
j\ 1024 512.
P(twoO's) = (If''o.s)IO
j\ 1024
P (three 0'5) '"' (1f"'O.s)IO = 120
" 1024
(
5 45 120 ) 849
P(atleastfourO's)l ++ 
512 1024 1024 1024
10.18 <a> Total ways of drawing 6 balls out of 49 are
= 49! = 13,983,816
6!43!
Hence, Prob(matching all 6 numbers) 1
13983816
(b) To match exactly 5 number means we pick S of the chosen 6 numbers and the last number can be
picked from the remainina43 numbers. We Cart choose S numbers of our 6 in (f) = 6 ways and can choose
one number out of 43 in(4?) '"' 43 ways. Hence, we have 43 x 6cornbinations in which exactly 5 numbers
match. Thus,
P(matching exactly 5 numbers) = 43 x
6
= IJ.4Sx 10
S
13983816
(c) To match exactly 4 numbers means we pick 4 out of the chosen 6 number in(:) = 15ways and choose
2 out of the remaining 43 numbers in ( 4j) "" 903 ways. Thus there are 1 S x 903 ways of picking exactly 4
numbers out of 6 and
• 15)( 903 4
P (matchmg exactly 4 numbers) '"' 13983816 = 9.686 x 10
72
(d) Similarly, we can pick three numbers exactly = 20x 12341 = 246820 ways. Hence,
. 246820
P (matchmg exactly 3 numbers) = • 0.0 176S
13983816
10.19 (a) Let f represent the system failure. Then
P(i)= (1_0.01)10 = 0.90438
P(f) = 1 P(i) = 0.0956
(b) p(i):: 0.99 and P(/) "" 0.01
If the probability of failure of a subsystem $j is p, then
P(i). (1 p)ID or 0.99. (1 pt = p. 0.0010045
10.110 Iff represents the system failure andfu andfL represent the failure of the upper and the lower paths,
respectively, in the system, then:
(a) P(/) = p(/ufd = p(/u)P(fd = [P(fu)t
P(fu) = 1 p(ill) = 1(1_0.01}10 = 0.0956
and
P(f) = (0.0956)2 = 0.009143
Reliability is P{i) = 1 P(/) = 0.9908
(b) P(i) = 0.999
P(f) = 10.999 = 0.00 1
P(f
u
) = .JO.OOI = 0.0316
PUll) = (1 p)10 = 10.0316 P = 0.003206
10.111 Let P be the probability of failure ofa subsystem ($1 or $2) .
For the system in Fig. a:
The system fails if the upper and lower branches ran simultaneously. 'I1!e probability of any branch not
failing is
(1 P)(I p) = (1 p)2. Hence, the probability of any branch failing is 1(1 p)2.
Clearly, PI' the probability of the system failure is p/ • [1(1 p)2 11(1 p)2];; 4p2 P« 1
For the system in Fig. b:
We may consider this system as a cascade of two subsystems Xl and .12 , where.l
l
is the parallel combination
of $1 and $1 and .12 is the parallel combination of $2 and $2' Let p/ (Xi) be the probability of failure of.%1 .
Then
P/(XI):: P/(X2} = p2
The system functions ifneither Xl nor .12 fails. Hence, the probability of system not failing
is (1 p2)( 1 p
2
). Therefore, the probability of system failing is
p/ = 1(1 p2X1 p2) = 2p2 _ p4 ;;:2p
2
P« I
Hence the system in Fig. a has twice the probability offailure of tile system in Fig. b.
73
10.112 There are (Sf) ... 2598960 ways of getting 5 cards out of 52 cards. of ways of drawing 5 cards of
the same suit (of 13 cards) is(1l) ... 1287. There are 4 suits. Hence there are 4 x 1287 ways of getting a
tlush. Therefore,
P{tlush} = 4)( 1287 = 1.980ax 10
3
2598960
10.113 Sum of 4 can be obtained as (1,3), (2,2) and (3,1). The two dice outcomes are independent. Letxt be the
outcome of the regular die and Xl be the outcome of irregular die.
1 1 1
PXJ%2 {1,3} = pot\ (1}P%2 (3) = '6)( '3 = Is
1
P
ott
%2 {2,2} ... P.q (2)P%2 (2) = '6 x 0 = 0
1 1 1
P
XJ
%2 {3,1) ... p
ot
\ (3)P%2 (I) ... '6)( '6 = 36
1 1 1
Therefore = is + 36 ... 12
Similarly,
P'(S) p., •• (1,4) + p.". (2,3) + p.", (3,2) + p.'" (4,1)
I III III
xO+x+xO+x= 
6 6 3 6 6 6 12
P{B) ... P{A)P{B\A) + p( AC)P(
;6
P{AIB) a P(AB) _
P{B) J.. 51
26
10.115 CI) Two 1'5 and three 0'5 in a sequence of5 digits can occur in(n = 10ways. The probability one such
sequence is
p .. (0.8)2(0.2)3 ... 0.00512
Since the event can occur in 10 ways, its probability is
lOx 0.00512 = 0.0512
(b) Three 1'5 occur with probabilityUX0.8)3(0.2)2 = 0.2048
Four I's occur with probability ( 1 X 0.8)4 (0.2)
1
... 0.4096
Five 1'5 occur with probability ( = 03277
Hence, the probability of at least three 1'5 occuring is
P = 0.2048 + 0.4096 + 03277 = 0.9421
10.116 Prob(no more than 3 error) = p(no error) + P{1 error) + P(2 error) P(3 error)
... (1 p')I00 +(lOO)P,(I p')99 +OOO}p, 2(1_ p,)98 +OOO)p'3{1 P, )97
.. (1IOOP
e
)+ lOOP
e
(199P
e
)+4950P
e
2(1 .... 98P
e
)+ 161 JOOP/(197 Pel
74
10.1·17 Error can occur in 10 ways. Consider case of error over the first link
p
c
( correct detection over every link) = (11\ )( 1 ... (1 Plo)
P£ =1P
c
=1(II\XIF,z) ... (Il\o)
= 1[1 (1\ + + ... + 1\0) + higher order tenDs]
== PI + .. ·+I']o Pt «I
10.1·18 p(e)  ±e )Pe'(l p.)5
j
=10P.
3
(1 p,)2 + SP, 4(1_ p.)+ P,
S
j=3
10.1·19 (a) P(success in 1 trial) = = 0.1
(b) P(success in S trials):= 1 P(failure in all S trials)
= 1 Pit Ph Pf3 Pi. PIs
PI) = Prob(failure in I" trial)., 9/10
Ph = Prob(failure in 2l1li trial) = 8/9
Similarly, Ph =718, PI. =6/7, andP" =5/6
Hence, P(success in S trials) = 1 =11.. = 0.5
10 9 8 7 6 10
10.1·10 Let x be the event of drawing the short straw and the Pt (x) denote the event that ith person in the sequence
draws the short straw.
Now, 1'1 (x) = 0.1
... Prob(111 person does not draw the short straw»)( Prob(2
nd
person draws the short straw)
= [1I\(X)]t .. t) = 0.1
Similarly,
P
3
(x) ... Prob(neither I" nor 2l1li person draws the short straw»)( Prob(3
n1
penon draws the short straw)
 [1 ( 0.1
Similarly. P4(X)  ps(x} = ... I']o(x) = 0.1
10.1·21 All digits are generated independently
(a) p(alllO digits are 0) ... (OJ)IO
(b) There ofamnging eight 1'5 and two 0'5. Hence,
p(cight 1'5 and two 0'5)'"
(c) P(at least five O's).p(five O's)+p(six O's)+ ... +p(ten 0'5)
= (lfXo.7)s(OJ)S +cgXO•7)4 (0.3)6 + CY)(0.7)3(03)
7
+( 18Xo.7)2(03)1 + (J8XO.7)(03)9 + (03)10
10.1·1 Py(O) = Pxy(l,O) + Pxy(O,O) = Px(I)Pyjx(O!I)+ Px(O)Pyjx(ojO)
... 0.6)( 0.1 +0.4[1 Py/x(IIO)]  0.06+032 = 038
Py(l) = 1 Py(O) = 0.62
7S
Py/x(l,l)Px(l) (1 Pe)Q
10.22 (a) pxjy(lll) Py(l) = (lQ)P,+(IP,)Q
10.23
10.24
(b) Pxly(O\I)'" 1 Pxjy{I/I)
(note that Py(l) and Py(O) are derived in Example 10.10)
(a) p(x 1) = r.!xe%dx =!
1 2 e
Pro ( )
f
1 %..1.. 1:
1
% 1 3
(b) b 1 < x:so 2 = xe '" + xe dx = 1
1 2 2 e 2e
2
[
2 1 3
(e) Prob(xS2) xe%dx
2
....., 2 2e
o
Fil. SI0.24
Since this is a halfwave rectifier, YaSsumes only positive values. So .p(y < 0) = O.
Hence, Fy(Y)'" 0 (fory < 0) and P(y < 0+)  i· Hence, Fy(O+) ... i
10.25 x is a gaussian r.v. with mean 4 and U
x
 3
Hence,
(a)
(b) P(x 0) = 10.09176= 0.9083
(e) P(x 2) = 2
3
4) .. 1 Q(2) 10.0227S = 0.9773
Fie SI0.25
10.2' (a) From the sketch it is obvious that x is not gaussian. However, it is a unilateral (rec:tified) version of
Gaussian PDF. Hence, we can use the expression of Gaussian r.v. with a multiplier of 2.
For a gaussian r.v.
p (y)= I e%2132 withu .. 4
y 4& y
(b) Hence, (I) p(x 1) = 2P(y 1) = 2q = OJ026
(il) P(I < x S 2) = 2P(1 <y S 2) { 0.1856
o
(e) If we take a Gaussian random variable y
Py(y) ... 4$e
i132
Fil· SI0.26
and rectify y (all negative ofy multipJed by I), the resulting variable is the desired random variable x.
76
' .
. "1
" .
10.17 The volume V under pxy(x,y) must be unity.
10.28
10.2·9
V = !(J x I)A = ~ = I, A = 2
2 2
Px(x) = J P"Y(x,y}dy
y
But y = x + 1 and the limits on y are 0 to 1 x. Therefore,
Px(x) ... J ~  . t 2dy = 2 •
1
1
% {2(1%) OS.t:S4
o 0 oIhawisc
{
2(Iy) OSy:S1
Similarly, Py(y) =
o odIcrwisc
P (x,y) 2 { 1I1y OSy:S1
PxlY(xly) = ;y(y) = 2(1 y)'" 0 ochcJvisc
{
IIIx OSx:S1
Similarly, PYlx(,ylx) =
o odIerwisc
Clearly x and y are not independent.
<a>
Similarly,
and
(b) From results in <I), it is obvious that x and y are independent.
Therefore
77
Fi .. SI0.27
10.210 Kf:f:
e
= = 1
But f: = Jie;rl/4 and, KJiJ:e
3
;r2/
4ea
= KJ'ii( fl.) = 1
Because ofsymmetry of pxy(x,y) with respect to x andy. Py(y) = /3e
3i
/
4
v 4;'
(
I .. ) _ pxy(x,y) _1 4"
Pxly xj.Y  () = ,.... e
Py y vtr
and
(xl) p..,(x,y) I • .,..; 1
Pyl
x
y = Px(X) .. ;;e
Since P"y(X'Y)  Px(X)Py(y}, x and y are not independent
10.211 P, = P(£jl)P;r(I}+ P(£jO)P:r(O}
If the optimum threshold isa, then
p( ti0) = a : : p )
P, = q a::
p
)P;r(O)
dP, = L[e (A
p
a)l/2ts,? P;r(I)e P;r(O)] = 0
da 2trer"
H
(Apa)l/2ts"lp, (1) (Ap+a)2/2ts/lp, (0)
ence, e ;r  e z
And a = er"
2
In[ P;r(O)]
2Ap P;r(l}
_ ( ) 1 (;r2)2/18
10.31 x=2. er
x
=3, Px x =:E='e
3'1/2tr
78
Fit. SI0.2·11
10.31 px(x)"'!lxle1zI
2
10.33
10.34
10.36
Because of even symmetry of Px (x). x = 0 and
x
2
= 2 J; x
2
Px(x}dr = 2 J: x
2
i
xe

Xdt
= fax
3
e
X
dt = 3! = 6
x
2
=u
x
2
+x
2
= u ~ =3!=6
The area of the triangle must be 1. Hence K =:! and Px (x) ... .!. (x + 1)
2 8
( )
4
_ 3 1 1 y3 y2 1 64 16 5
x=J xp (x)dt=rty(yI)dy=  =() ... 
\ x 8 Jo 8 3 2 8 3 2 3
o
"2 1 3 2 1 x4 x
3
x 1 x (x+I}dr ...  +
( )
3
8 \ 8 4 3_\
... .!.[!!+ 27 _! +.!.] ... !!
8 4 3 4 3 3
u 2 :x
2
_(x)2 .!!_ 25 =!
x 399
12 1 2 3 4 5 6
x'" rx;px(x;) = (2)+(3)+(4)+(5)+(6)+(7)+
;2 36 36 36 36 36 36
5 4 3 2 1 256
(8)+(9)+(10)+(11)+(12) =  = 7
36 36 36 36 36 36
.'
 12 1 2 3 4 5 6
x
2
= rx? Px(x,) = (4)+(9)+(16)+(25)+(36)+(49)+
;=2 36 36 36 36 36 36
5 4 3 2 1
(64) +(81) + (100)+ (121)+ (144) = 54.83
36 36 36 36 36
79
Fig. SIO.31
ISxs3
Fig. SI0.35
10.37 x" = Jr,;IGOx"e.r2/2a:dr. Fornodd,the integrand is an odd tUnction ofx. Therefore x" =0.
ax 21r GO
For n even, we fmd from tables
_ ." even
XII =
o "
10.38 Let Xi be the outcomes of the ith die. Then,
_ 1+2+3+4+5+6 7
x· = =
, 6 2
i = 1,2, ... , 10
Fil·510.41
10.41 Px{x) = 0.46(x)+0.66(x3)
1 2/1
Pn{n) = 2S
e

1I
and
()
1 _y2/1 3 {y3)2/1
>tI2" lW2"
Fia·510.42
10.43 px(.r} = Qt5{x 1)+(1 Q)t5(x + I), Pn{n) = P6(nl)+(IQ)6(n+ 1)
Py(y) = [Qt5(yI)+{I Q)t5(y+ 1)]·[ Pt5(yl)+{1  p)t5(y+ I)l
= (P+ Q2PQ)t5(y) + PQ6(y2} +(1 P)(l Q)t5(y+2}
80
•
10.44 pz(Z) = Px(x)* Py(y)
Taking Fourier transfonn of both sides, we have
Taking inverse Fourier transfonn we get
( )
I
Pz Z = e
+ tTn
It is clear that z = x + y and C7! • C7! +
::2 .
10.51 For any real a, [a(x x)(yy)] 0, or +tT; 2atTxy O. Hence, the discriminant of this
quadratic in a must be nonpositive, that is:
 SO, that is, I tT xy lSI or IPI S 1
tTxtTy
10.52 When y = K)x + K2 Hence, y:: K)x + K2
10.53
tT;  KftT; and tT
xy
= (xx)(y y) = = Hence,
Pxy = tT xy :: / = 1 if K) is positive. If K) is negative, tT.I)' :: K)tT; is negative.
tTxtTy
But tT x and tTy are both positive. Hence, Pxy = 1
x = r21r cosOp(O}iO= _1 r21r cosO dO = 0 Similarly, y =0
Jo 2"JO
tT.I)' = Xy ... cos Osin 0 = .!. sin 20 = .!. L021r sin 20 p( O}iO = _1 sin 20 dO = 0
2 2 4"
Hence, tT xy = X Y ... 0 and x, y are uncorrelated. But x
2
+ y2 = 1.
Hence, x and y are not independent.
JO.6J In this case
Rl\ = R21 = R33 = ml P'"
R)2 = R
2
) ... R23 = R32 ... Ro) = OJ2SP",
R)3  R31 = Ro2 = 0562P",
Ro3 = 0.308P",
Substituting these values in Eq. (10.86) yields: a) ... 1.1025, a2 = 02883, a3'" 0.0779
From Eq. (10.87), we obtain
6
2
= [1(O.82Sa) +0.562a2 +0.308a3)]P", =027S3P",
Hence, the SNR improvement is
1010..{ P'" ) = 5.63 dB.
,027S3P",
81
Chapter 11
11.11 This is clearly a nonstationary process. For example,
amplitudes of all sample ftmctions are zero at same
instants (one is shown with 1 dotted lind). Hence, the
statistics clearly depend on t.
11.12 Ensemble statistics varies with I. This can be seen by
fmding
 ,. (\00
x(t) = Acos(at+O) = A Jo
= ro1oo cos( at + 01Jtl). This is 1 function of t.
lOOt Jj
Hence, the process is nonstationary.
11.13 This is clearly a nonstationary process since its
statistics depend on t. For example, att = 0, the
amplitudes of all sample functions is b. This is not
the case at other values of t.
11.14 x(t) = + 0)
x(t) == + 0) + 0) = cos(ox + 0) tAaP1(a)da
= + 0)/2AJ ada = 0
R
x
(t\.t2)·.2 COs(U\ +0)COs(U2 +0). COs(U\ +0)12
A a
2
= cos(ox\ + 0)Cos(tJJt2 + 0) LA tta
A2
+8)COs(tJJt2 +8)
83
I
 e:=.
Fia. SI1.11
Fla. 511.12

f'la. SU.I3
A A
Fia·511.14
11.15
11.16 x(t)". aI +b = It +b. But a'" 0 Hence, %(t)". b
 2' 2 2 1 a
3
12 4
Also, a=O. a = L
2
a p{a)da 
4 3 2 3
Rx(llh) = (all +b)(at2 +b) = a
2
tlt2 +a(t
l
b+t
2
b)+b
2
= a
2
tlt2 +t2)+b
2
= i tlt2 +b
2
3
I PC"')
11.17 (b) x(t) = K = ° t
(c) j. • .s.
R
x
(tltt2) = KK = K2 = JI K2 lJI K
2
dK". l
I 2 c 3
(d) The process is W.S.S. Since x(t) = ° and R
x
(tlo
t
2) = 1.
3
(e) The process is not ergodic since the time mean of each
sample function is different from that of the other and it is
not equal to the ensemble mean (x ".0) •
 1
(f) x
2
=RxO=
3
11.18 x(t) = acos(Ct)ct + 8)
2'1
a ".
3
(b) x(t)';' acos(mc' +8)". a cos(a1ct+8) = 0
(c) R
x
(tl,t2)  a
2
cos(mctl +8)COs(Ct)ct2 +8)
+/2)+28]}
= iCOSa1c(/lt2)+ t2)+28}i8
".lcOSCt)c(tl/2)
3
(d) The process is W.S.S.
84
_______________ ___________
.. : t'+
Fla. 511.1'
1\ A
.,/' "'==" t;,.
Fig. 511.18
..
(e) The process is not ergodic. Time means of e&l:h sample function is different and is not equal to the
ensemble mean.
 t
(f) x
2
.. R,,(O)="3
11.21 (a), (d), and (e) are valid PSDs. Others are not valid PSDs. PSD is always a real, nonnegative and even
function ofQ}. Prcx:esses in (b), (c), (0, and (g) violate conditions.
11.22 (a) Let x(t) .. Xl and x(t + r) = x2 Then,
11.23
222 22
(XI±X2) =Xl +x2 +2Xlx2:ii!O, Xl +x2 :ii!± 2xlx2
But, x,x2 = R,,(r) and x\2 ZX22 =R,,(O) Hence,
(b) R,,(r) = x(t)x(t + r), lim R,,(r) = lim x(t)x{t+r)
f+ao f+OO
Asr + co, x{t) andx(t+r) become independent, so lim Rx(r) = x(t)x(t+r) = (XXx) = x
2
f+ao
R,,( r) = 0 for r = and its Fourier transfonnSz(Q}) is bandlimited to B Hz. Hence, R
z
( 1') is a
2B
wavefonn bandlimited to B Hz and according to Eq. 6.1 Ob
Rx(r) = :E (2/rBrn). Since ° for aUn exceptn .. 0.
,,:ao 2B 2B
R,,( 1') = R,,(O) sinc (2/rBr) andS" (ClJ) = R,,(O) rect(.!!...). Hence, x(t) is a white process bandlimited
. 2B 4/rB
toB Hz.
11.24 R" (t) = P
XiX2
(I, 1) + P"I"2 (1,1)  P"I"2 (I, I)  P
X1X2
(1,1)
But because of symmetry of 1 and 0,
•
P"I"2(I, 1) P"I"2(I,I) and P"I
X
2(I, I) P
x1x2
(I,I)
and Rx (r) = 2[ P"I
X
2 (I, I)  P"I
X
2 (1,1)]
= 2P"1 (1)[ P"21
x
l (111) P
x2
1
x
I (III)]
= 2P
x1
(l)[ P"21xl (111) ( 1 P"21
x
l (III))J  2P"21
x
l (111)1
Consider the casen16 «n+ 1)16. In this case, there are at least"nodcs and a possibility of (n+ I)
nodes Prob«n+ I)nodes] = rIIT
b
= ..!..II
lb lb
Prob( n nodes)  1 Prob« II + 1 )nodes] = (II + 1)  ..!..
7b
The event ( x 2 = llx I = 1) can cx:cur if there are N nodes and no state change at any node or state change at
only 2 nodes or state change at only 4 nodes, etc.
Hence, P"llxl (111) ... Prob«n+ l)nodes] Prob{state chanle at even nmnber of nodes) +
PrOb(1I nodes) Prob(State changes at eveen nmnber of nodes)
The number of ways in which changes at K nodes out of N nodes cx:cur is Hence,
P"21x1 (111)  [( S+'XO.6)O(O.4)"+' + ..... ] II )+
8S
•
[(3)(0.6)°(0.4)" + .... .J (n+ 1
andR
x
(r)=2P
x1
I
x
l(*)1 Thisyields
R
x
(r)=1
12
M
1b
(n=O)
= 0.44+
024
11 1b < 2lb
1b
(n·.I)
=0.J360.0481:i 2lb
16
(n=2)
and soon .
1·2
d't:
.....
••
I
I
_ •. a¥ 0,,, f'
 ,,2.
 • . aI'
.;a..'"
'ta. SII.l4
The PSD can be found by differentiating Rx (r) twice. The second derivative d
2
R;x / dr
2
is a sequence of
impulses as shown in Fig. SI1.24. From the timedifferentiation property.
86
11.15
:1'22 Rx(T) ++ (jtU)2 Sx(tU) = tU
2
S
x
(tU). Hence, recalling that 6(1' T) (:) e
jAlT
, we have
_tU
2
Sx (CtI):;;: ;b [2.4 + 1.44( efAl7b + e  jAl7b )  eflAl7b + e  j2A17b )+ .... ]
= ..!..[2.4 +2.88costU16 0.576cos2Ct116 + 0.1152 cos3Ct1Tb+ .... ]
Tb
and
Sx(tU) = costU16 1.COS2tU16 +..!..cos3tUTb __ I_COS4tUTb+ .... )]
16tU '\ S 25 125
Because Sm(cu) is a white process bandlimited to B, Rm(T) = Rm(O} sinc (2Bt) and
n=±l, ±2, ±3 ...
This shows thatx(t>{t + = 0
Thus, all Nyquist sample are uncorrelated. Now, from Eq. 11.29,
Sy(tU) '"' Ip(cut [Ro + fRm cosnCtlo16j
Tb ".1
Rn = a.ak+,. = 0 n 1 and whereak is the kthNyquist sample.
 
Ro ... al = x
2
= Rm(O). Hence,
Sy(w) = Rm(O):; 2BR
m
(O)jP(w)2j since Tb =
11.16 For duobinary
Pat (1) = Pat (1) = 025 and Pat (0) = 0.5
ak :;;:(I)1.+(I).!.+J.!.) .. o
4 4 '\2
Ro = al = (1)2 .!. + (_1)2.!. + 02(.!.) = .!.
4 4 2 2
RI :; akak+1 = L La.ahIPI.I'+1 (ak
a
.. +l)
ak ahl
Because ai and ak+l each can take 3 values (0, I, I), the double sum on the righthand side of the above
equation has 9 terms out of which only 4 are nonzero. Thus,
R) ... (t)(I)P
I
•
I
.+
1
(1,1) + (I)(I)P
a
•
ahl
(1)(1) + (1)(I)P
a
•
a
.+
1
(IXl)(l)(l)PI.ahl (1)(1)
Because of duobinary rule, the neiihboring pulses must have the same polarities. Hence,
Pa,lhl (1,1) == Pat (I)P
I
••
1
1a. '"' iG) ... i
Similarly, P
I
•
a
.+
1
(1,1) = i Hence, RI = i
Also R2 ... akak+2
In this case, we have the same four terms as before. buu. andak+2 are the pulse strengths separated by one
time slot. Hence, by duobinary rule,
Pa•
ah2
(1,1) ... Pa. (I)Pah211. (III) ... i( i)=
Similarly, P
I
•
a
.+
2
(1,1) = 116
87
In a similar way, we can show that p •••• +
2
(1,1) = P'.'
k
+
2
(1,1) ...
Hence,R2"' 0
Using a similar procedure, we can show that R" ... 0 for n 2. Thus, from Eq. (11.29) and noting that R" is
an even function or.., we _in Sy( .. ) + <os .. r,,) ] ... ,( "':. )
For halfwidth rectangular pulse p(tv) ... t Sinc (tv,: )andSy(tv) ... SinC
2
( W1 )cos
2
( w: )
11.27 ak =(I)Q+(I)(1Q)=2Q1
Ro'" ai _(1)2
Q
+(_1)2{I_Q)= I
Because all digits are independent,
Rn = a.ahl .. i
k
i
k
+
1
== (2Q_l)2 Hence,
Sy(w) = Ip(w)1
2
[1+ 2(2Ql)2(fcosnw7b)il
7b nel J
11.28 Approximate impulses by rectangular pulses each of height h and width & such thath& = 1 and & .0
(Fig. S 11.28a)
Rx(r) = rr%I.l"2Pxt
X
l (Xl%2)
XI X2
Since x I and x2 can take only two values hand 0, there will only be 4 tenns in the summation, out of which
only one is nonzero (corresponding lOxl = h, x2'" h). Hence,
Rx( r) = h
2
P
XtX1
(h,h) 5 h
2
P
X
) (h)PX1I
xl
(hjh)
Since there are a pulses/second, pulses occuPYa& fraction of time. Hence,
P
Xt
(h) = a& and Rx( r) ... h
2
a&P
X2
1xI (hjh),. ahP
x2
lx) (hjh).
Now, consider the < &. P
x2
1xI (hjh) is the
Prob(X2) = h, given that xI = h. This means XI lies on one
of the impulses. Mark off an interval ofr from the edge of
this impulse (see fig. SI1.28b). If XI lies in the hatched
interval, x2 falls on the same pulse.
Hence,
PX21x (h/h) = Prob(XI lie in the hatched reaion) ... !:.!. ... 1!.
I G &
and Rx(r)=an(I;)
Since Rx( r) is an even function of r, Rx( r) ... a{ I_I;)
In the limit as & + 0, Rx ( r) becomes an impulse of strength a.
Rx(r)=a4(r)
When r > G, Xl and x2 become independent. Hence,
P
x2
1xI (hjh),. P
X2
(h) = a&
Rx( r) = a1h& '"' a
2
Hence, Rx(r) ,",@(r)+a
2
88
Fig. Sl1.28
1;,.
11.29 In this case the autoc:OlTelation function at r = 0 remain same as in Prob 11.28. But for r > 0 whenever
x( t), x( t + r) are both nonzero, the product x( t )x( t + r) is equally likely to be h
2
and  h
2
• Hence,
Rx(r)O, r9l0andR,;(r)=a6(r)
11.210 The process in this problem represents the model for the thermal noise in conductors. A typical sample
fUnction of this process is shown in Fig. SI1.210. The signalx(t) changes abruptly in amplitude at random
instants. The average number of changes or shifts in amplitudes are fJper second, and the number of
changes are Poissondistributed. The amplitude after a shift is independent of the amplitude prior to the
shift. The f1l'Storder probability density of the process is p( %; t). It can be shown that this process is
stationary of order 2. Hence, p(%; t) can be expressed as p( %). We have
Rx( r) .. I: f: %1%2PxtZl (%It%2)taldr
2
... J:J: %1%2Px\ (%I)PX2 (%2/%1 .. %1)dcldr
2
To calculate PX2 (%2IXI = %1), we observe that in r seconds (interval between XI and x2)' there are two
mutually exclusive possibilities; either there may be no amplitude shift(X2 = XI), or there may be an
ampJitude shift(X2 9l XI)' We can therefore express PX2 (%2/XI = %1) as
PX2 (%2/
X
I = %1) = PX2 (x2lxl = x\t no amplitude shift)p(no amplitude shift) +
PX2 (x2lxl = Xl, amplitude shift)P(amplitude shift)
t
Fla. 511.210
The number of amplitude shifts are given to have Poisson distribution. The probability of" shifts
in r seconds is given by
Pi(r)  (P:t e
ftr
(1)
where there are on the average fJ shifts per second. The probability of no shifts is obviously Po ( r) , where
po(r)_e
ftr
The probability of amplitude shiftl po(r) = l_e
ftr
. Hence
PX2(X21xI =xI)=e
ftrpX
2(X2!XI =xI, no amplitude shift) + (te
ftr
)
PX
2 (X2/XI =xI, amplitude shift)
(2)
when there is no shift, x2 = XI and the probability density ofx2 is concentrated at the single value XI'
This is obviously an impulse located at x2 = X I' Thus,
PX2(X2IxI =xI, no amplitude shift) =4(X2 XI) (3)
whenever there are one or more shifts involved, in general, x2 ¢ x I' Moreover, we are given that the
amplitudes before and after a shift are independent, Hence,
PX2 (x2lxl I: X\' amplitude shift) = PX2 (X2) = p(x) (4)
89
where PX2 (X2) is the fU'Storder probability density of the process. This is obviously p( x). Substituting
Eqs. (3) and (4) in Eq. (2), we get
PX2 (x2lxl = XI)· e
pr
0(X2 XI)+(1e
Pr
)
PX
2 (X2)
aepr{0(X2 xl)+(e
Pr
1)
Px
2(X2)]
Substituting this equation in Eq. (1), we act
Rx{r) = e
Pr
I:I:xlx2PX\ (XI)[8(Xl xl)+(e
pr
t)PX2(Xl)}ir1dr2
.. epr[J:I:xlxlPX\ (XI)6(X2 XI)ctrldr2 + I:I:xlx2(eP
r
t)PX\ (XI)PX2 (Xl)dr\dr2]
== epr[J: x? Px\ (XI)dr1 +(eP
r
l)C xlPx\ (XI)drIJ: x2PX2 (X2)
dr
2 ]
= e
pr
[ x
2
+ (e
pr
t)x
2
]
where x and xl are the mean and the meansquare value of the process. For a thermal noise x = 0 and
Eq. (5) becomes
Rx(r) === x
2
e
pr
r > 0
Since autocorrelation is an even function of r, we have
and
11.31 For any real numbero, (Q y)2 0
11.32
0
2
xl + y2  20xy 0
Therefore the discriminant of the quadratic in 0 must be nonpositive .. Hence,
2   2
(2XY) <4xl .y2 or (iY) < x
2
y2
Now, identify x with x(t\) and y with y(t2) , and the result follows.
1\.( r} "" u(t)u(t + r). [x(t) + y(t}Ix(t + r}+y(t + r)]
.. Rx(r)+ Ry(r) + Rxy(r) + Ryx(r) .. Rx(r) + Ry(r)
since x( t) and y( t) are independent.
Rv( r) = [2x(t) + 3y(t)I2x(t + r)+3y(t + r)]
.. 4R
x
(r)+9R
y
(r) since Rxy(r) = Ryx(r) =0
Ruv( r) ... [x(t) + y(t)I2x(t + r)+ 3y(t + r)] = 2Rx( r)+ 3Ry( r)
Ruv( r) === Ruv( r} = 2Rx( r) + 3Ry( r)
11.33 Rxy( r) = ABcos(mot + r) +n;]
= +nMo(t + r)+(n+ + r)Mol +(nl);])
90
(5)
11.34
+n010(1 + r)+(n+ 1);] = _I (211' c...J'01ot+ n010(1 + r)+(n+ I);)d; = 0
. .
+ r) aJOI +(nl);] = 0 and Rxy( r) ... 0
CD
X(I) = Co + l:CllcOSn01o(lb)+Oll
11=1
CD
... Co + LCII(naJolnaJob+Oll)
,..1
Since b is a r.v. unifonnly distributed in the range (0, Tb), O1ob = 2nb is a r.v. uniformly distributed in the
16
range (0, 2").
Using the argument in problem 11.33, we observe that all harmonics are incoherent. Hence the
autocorrelation function of Rx( 1') is the sum of autocorrelation function of each term. Hence follows the
result.
11.41 (a) SI(01) = 2KTRI and S2(aJ) = 2KTR2
Since the two sources are incoherent, the principle of superposition applies to the PSD.
If SOJ (aJ) and SOl (01) are the PSD's at the outputterminals due toSl(01) and S2(aJ) respectively, then
SOJ (aJ) = IH
1
(01t SI(aJ) andS
o
2(aJ) = IH
2
(aJ)1
2
S2(aJ)
where
c
(a)
(b)
Similarly,
H
2
(aJ)'" Rl ... R
R2(jt»RIC+I)+RI jt»R.IR
2
C+R
I
+R2
v
o
S (01)= 2 KTRIRI and S (a» 2KTR2Rf
OJ a>2RlRJC2+(Rl+R2)l Ol O1
l
RlRJC
l
+(RJ+R2)2
2KTR
1
R
2
{R
J
+ Rl)
Svo(m)=SOJ(aJ)+SOl(01) 1 1 1 2 2
aJ R
J
R2 C + (R
J
+ R
2
)
(b) H(m)= lIjaC = R
J
+R2
_1_+ R,R
2
joCR,R
2
+(R, +Rl)
jt:£ R, + R2
Svo = IH(01)112KTR
I
R
2
RI+R2
c
(c)
Fi,. Sl1.41
(R, +R2)2 2KTR,R2 2KTR,R2(RJ +R2)
= a>
2
c
2
RlRI+(RJ+R2)2' RJ+R2 =
which is the same as that found in part (a).
91
11.41 y(t) ... I: h(a)x(t  a)da

Rxy(:) = x(t)y(t +:) ... x(t}I: h{a}x(t + :a)da
= I: h{a)x(t)x(t + t a)ia ... I: h{a)Rx(: a)da ... h{ r). Rx( r) and Sxy (CI» ... H{a»Sx{CI»
I
In Fig. 11.13, H(CI» .... I
R+ jDJRC+I
joC
andS
lIVo
(td) = 2KTR l(jtdRC + I) and Rnvo (:) ... 2KTR e"rlRCue r)
11.43 (a) We have found Rx(:) of impulse noise in Prob. 11.28
Rx{r) ... ac5(:) + a
2
,and Sx(m),.. a +2mz
2
c5(CI»
Hence,
Sy(a» IH(mt[a+2Ira
2
c5(CI») ... 2lra
2
lH(ot 3(CI»+a/H(tdt
and Ry(:) = + all{ :).h(:)
(b) h{t) = !l..eI1rll(t}. H(CI»'" 1.1.
r r. 1
jCl>+
:
/H(a>t = q2 ,and Rx(t) = a2q2 + (fll[ q2 ]_ a
2
q2 + aq2 .eWr
I+CI>2
t
2 I+CI>2:2 2r
t ttt t t t tt m t
xU)
1
11.5t n(t) == "c(t)COSCI>ct + ns(t)sina>ct
H(w)
h(l)
Fl •. Stt.43
y(t)
c
The PSD of"c (t) and,,( t) are identical. They are shown in Fig. S 11.5·1. Also, n
l
is the area under ','
Sn(td) ,and is given by n
2
_ 2[..AI x 10
4
+ 10
4
= 1.2S x 10
4
..A1
2· 2 2 2
()  [ ..All ]
or is the area under Snc (tV), and is given by = 11K 2 SOQO..,\J + 2". '2 x 5000 = 1.25 x 10
4
..AI
.5P...
Fig. SIl.5t
92
<,
11.52 We follow a procedure similar to that of the solution of Prob.
11.5t except that the center frequencies are different. For the
3 center ftequencics Sflc (It) or S ... (It)) ] are shown in Fig.
S t t .52. In all the three cases. the area under SIIc (0.1) is the
same, viz., 1.25)( 1
Thus in all 3 cases
= = 1.25)( 10
4
,OK
5k 'OK
..
f ... /0 E k.
15 K ", Jtc.
fHa*
q5 It.
Fll·511.52
6
()
Sm(w) ?:;r _ 6
11.54 <a) Hop W = () () = 6  2 = 2
Sm 0) +Sn 0) +6 641 +60 0.1 +10
9+ 0.1
2
(b) h (I) =
Of' 2,,10
Fil. Sl1.53
(e) The time constant is Hence, a reasonable value of timedelay required to make this filter
,,10
realizable is :: 0.949 sec.
,,10
(d) Noise power at the output of the filter is
N 1 fao Sm(O))Sn(O.I) '" _ 1 roo 6 d 6 tanI 0) lao _ 3
o = 21r ao Sm ( 0) + Sn (0.1) tv  21r Jao It)2 + 10 tv:: 21rJiO To aD  110
The signal power at the output and the input are identical
1 lao 6
SI :: So s  2 dO) "" 1
21r aD 9 + It)
SNR:: .& = J10 "" 1.054
No 3
93
4
11 5 5 () H ( ) Sm{t») ;r:;4
•  • op t» = Sm(t»)+Sn(t») = _4_+ 32
'oJ2 + 4 t»l. + 64
t»2 +64 1 [I 53.33]
... 9t»2 +96 ... 9' + t»2 + 10.67
(b) "'(/)'" !6(t)+ 8.163e 3.l66jtj
(e) The time constant of the filter is 0.306 sec.
A reasonable value of timedelay required to make this fiher realizable is 3 x 0306 ... 0.918 sec.
(d) Noise power at the output of the filter is
No = _I r. Sm(t»)Sn(t») dt» = _I 1
00
32 dt» ... 0.544
211' Sm(t»)+Sn(t») 211'
The signal power is
1 1
00
4
S; =So"'
211' 4+t»
Sn 1
¥... = = 1.838
No 0.544
94
Fig. SI1.55
Chapter 12
5 5·
11.11 .::; = r =
o
5·
r = 1000 = .' 51 = 0.08
2 x 10 )( 4000
Also, Hc(e») = 10
3
• Hence, 5r = 5; 2 = 8 x 10"
IHc(e»)!
5r = _1 ,8 [2 )( 800011''1 = 8 x 1 0
4
.8 ... 10
211'
11.11
11.11
H+w.
Fla. SI1.11
5
no
(IV) = 5
n
(IV)IHd (IVt = 10
10
( a = 80001r
No""!" r 1O
10
(IV
2
+a
2
tIV = 10
10
(e»3 +a2IVl
a
.. 32 x 10
7
Ir 0 a
2
r a
2
1r 3 0 3
50 =>5
0
= 3.37 x 10
3
No 32 x 10
7
3
10
3
But So(/)" m(t}. Hence,
a
10
6
 
50 = 2 m
2
(t} = 3.37 x 10
3
=> m
2
(t},.. 21S.7 x 10
9
a
Also, m
2
.. _1 fao {Jdil1'" pa .. 8000.8 .. 21S.7 x 10
9
21r GO Ir
Hence, .8'" 26.96 x 10
6
and5
m
(iI1) = 26.96 x
1 t
a
\l \1
2
1 r
a
6( 10
6
}
5
1
,, Sm(e»IIHc(e»/1 dOJ"
J10
26.96 x 10 2 2 iI1
Ir Ir OJ +a
26.96 1 e» la 26.96 2.68 104
tan  = x
air a 0 4a
Sr = .!.. r
a
Sm (iI1 )dOJ ... ..!.. r
a
26.96 x 10
6
dOJ'" 26.96 x 10
6
a = 68.65 x 10
9
Ir Jo Ie Jo Ie
S 51 5· 4
(a) 30 dB .. 1000 .. ..:::2.. ... r == ' =>5; ... 4)( 10
No JJB 10
10
)(4000
(b) From Eq. (12.7), No .. .. 10
1
°(4000) ... 4)( 10
7
(c) 5; .. IHc(IVt Sr and 1O
8
5
r
= 4 x 104 => 5r = 4 x 10"
95
11.21
(a) So = 1000 = Sj = Sj S = 4 x 104
No 10
10
x 4000 I
(b) No =JJB = 10
10
x 8000 = 4 x 10
7
(e) Sj = IHc(mt ST = 108
ST
= 4 x 104 ST = 4 x 10
4
11.13 Let the signals ml(t) and ml(l) be transmitted over the same band by carriers of the same
trequency (t» c), but in phase quadrature. The two transmitted signals are J2[ ml (t) cost» cl + ml (I) sin 11I c
l
]
Milot: ) .... V\,,{ot:)
Fic. 512.13
The bandpass noise over the channel is nc(l)cOSl1lcl + ns(l) sin l1I
c
l. Hence, the received signal is
[Jim) (I) + nc(I)]cosmcl + [Jiml (I) + n
s
(I)]sinl1l
c
l
Eliminating the high frequency tenns, we get the output of the upper lowpass filter as ml(t)+ *nc(l)
Similarly, the output of the lower demodulator is ml (t) +"* ns (I)
These are similar to the outputs obtained for DSBSC on page 535. Hence, we have = r for both QAM
No
channels.
11.14 (a) p = [m(t)]min ... !!!.e..
A A
m
1
S I·
(e) For tone moduiatioDK
1
= // =2 andforp=l • .:::L:r=L
"'p 2 No 2+1 3
96
12.15
S ,i
(a> From Prob. 12.24, .::2.. = 2 2 ,.
No K +p
and when p >= I, .§.sz. = .J., = L
No 9+1 10
S (05)2
(b) When p = 05, .:.JL = , • L
No 9+(05)2 36
11.16 For tone modulation, letm(t) ... pAcost»",t. For BSBSC,
; DSB(t) ... /ipAcosOJ",t ·cosOJet
= *[COs(OJ
e
+ 4I",)t + cos(me  41",)1 J
S. = ,,.,2.,42 + p2.,42 = p2.,42 and m = = t::
2
,,A
, 4 4 2 p Hence, the peak power
Sp WhereSj=.!.Sp
No ..AlB 4..AIB 4
ForSSBSC
;SSB(t) = m{t}cost»el + m,,(/) sin t»el
::;.UA cosm",t COSOJci + pA sin m",1 sin OJet = pA cOs(OJ
e
 OJ",)I
p2.,42 S S 2,42 S
Sj =and mp =pA. Hence, Sp = p2.,42 and =E=L
2 No ..AlB 2,"",B 2JJB
For AM
, .AM(t) = .,4(1 + p cos m",t ) cos OJet
.,42 m
2
.,42 p2.,42
5=+=+
, 2 2 2 2
mp = ,4(1+ p)and Sp = ,42(I+p)2.
Hence,
Sp(2+p2) So p2A2/2 Sj (p2 i Sp(2+p2) 1
Sj= 4(I+p)2 and No'"" .1.2+;:2'= ,42+{p2,42/2)·JJB
Under best condition, ie., for p ::; 1, !R.. =
No I(UJB
Hence, for a given peak power (liven S p) DSBSC has 6dB superiority, and SSBSC has 9dB superiority
over AM. These results are derived for tone modulation and for p = 1 (the case most favorable for AM).
11.17 For 40" loading, mp .. 40"", and the camer amplitude A'"' mp = 40"m (for p = I). For Gaussian m(/),
m
2
;; (assuminl iii .. 0 )
Prob(E A) = Ii '"' e.A
2
/ltr
Il
2
::; 0.01
.A O"n
.04
2
0"2 ,42 +m2 160"2 17
Hence, 4.60Sand Sj'" = m m
20"n O"n 2 2 2
Therefore, rThresh =_1 =(4.60S)=9.79dB
S 17 0"2 17(40"2) 17
..AlB 8..AIB 8
97
12.31
5
.!L = 28dB = 631. Hence,
No
631 x9
Therefore, y =  = 473.25
12
(a) Also,y =yvfB = 473.25x2 x 10
10
x 15000 = 1.4197 x 10
3
dJB
kIm k
/
{30'm)
(b) p ..  = p ::> 2 = ::>"10' m = 20,OOOn
21dJ 2trB 30,ooon
 2
50 =(104) (20,ooon)2 ... 4n2
5
(c) No = R = 0.0199
631
12.32 mp = B,
1l.33 m(t) = cos
3
CtJol and mp = 1
m(t) = 3CtJ 0 C05
2
CtJ otsin CtJ ot and m(t) = 3CtJ
o
[ CtJ
o
C05
2
CtJ 0' cosCtJot  2CtJ 0 cosCtJ ot sin
2
CtJot]
For a maximum
m(t) = O. This yieldsc05
2
wot = 2sin2 CtJot
1
·2 2.2 . 1 (2
or SID CtJo'= SID CtJ
o
t::>SIDCtJ
o
t='J3' COSCtJ
o
t=V3
and
12.34 m(t) = al coswlt + a2 c05w2', mp = al +a2
m(t) = (atCtJ \ sinCtJlt +a2CtJ 2 sinw2t), mj, = alCtJl +a2CtJ2
(5
0
/N
o
)PM (21d1}2m; +a2)2
= 2" 2
(5o/No)FM 3mj, 3(alWt+a2CtJ 2)
,.
12.35 Error in this problem. There should be 41r2 in the denominator (see below).
12.36
12.37
S,;.(w) = W
2
S
m
(M). Hence,
C[m(t)t tit ... 1: Sria(2Jf)cU'  1: 41r2 1
2
Sm(2 tt/)cU'
From Eq. (12.42a)
ro 12 cU' 31ao .x
2
 J....., 1 + (III. )2k 10 ri dx
8
2 0 «I1+x
m= =
ro 1 dl J.l
ao
_1_dx
J....., I + (II 10)2k 0 «11 + x
2k
1.
2
[ Ir J
c • 2hm(*) /.'
tr 0 • (31r)
2ksin(2:) sm 2k
The definite integrals are found from integral tables.
k
2 2 sin(Ir/2k) 2 (1r/2k) 1 2
As + <Xl B = I. + I. =1.
'm 0 sin(31r/2k) 0 (3tr/2k) 3 0
S (M)  m
2
 2 reo dM 2
m  2 , Jo 2 ...
0' 0'
Hence, the normalized PSD's is..f1.eGl2/2a2
20'2
  3 1 ,
If W = 21C8, then W
2
= (21C8)2 "" 2 ('II 12rr d(j) = 20'2. If P(W) is the power within the
Jo 20'
bandW to W.
p(W) = 2 rW ..!!!.::e oh2t1
1
dw = 2[1 e ],
Jo 20'2
P(W) W
2
/2 2
p(co)2,and p(co) =le tI B=0.4820'
= 0.9S => W = 2.4SO', B = 039SO'
= 0.9 = W = 2.1 So', B = 03420'
x = 0.99
w
2

3 = 3.060'2 > W
2
=> PM superior
99
}
x = 0.95
x=0.9
w
2

 = 2.000'2 = W
2
=> PM and FM equal
3
W
2

 ... 1.540'2 < W
2
=> FM superior
3
12.38 m(t) = al COSCi)I' +a2 COS Ci)2t, and
a
2
a
2
Sm(l) .. +[6(1  11)+6(1 + 11))+1[6(1  12)+6(1 + 12}]
m
2
= J: Sm(l)4f == (a? + ai) 12
= (fIX) f 2Sm(l)dl Q)= _2_[arll + alIi] = a?f?+a1fl,
00 /111 a? +a1 2 2 at +af
.
Smce B = h,PM is superior to FM ifff > 2 2 '
al +a2
th t
· ·f(ada2}2(1t112)2+1 1 ·fl 2 1 l+x2
a IS, 1 2 <  or 1 + X Y <
(al/a2) + 1 3 3
11.39
So 2 m
2
1 2  1 1 2
<a> =3/J y=p y. Sincem
p
=30', m
2z
0'2 and23.4dB218.8, 218.8=3
P2Y
=3(2) r
N m
2
3
o p
218.8 x 3 .
r = 4 = 164.1. Also, Y'Tlucsh '"' 20(P+ 1)
So PThrcsh .. 164.1 1 = 721
20
==1. p2y = 1.(721)2(164.1} = 2844 = 34.53 dB (40 dB == 10,000)
No 3 3
So 12 1 (Y17l20)2 212)(10
7
(b)  = P y =  Y17l = 10,000 or(Y17l20) = => y;; 242.5
No 3 3 20 Y17I
Required increase in Y = 242.5 = 1.479 = 1.7 dB
164
12.310 From Eq. (12.40) p2 .. .!.[ . (.2.; )]
3 1+ m
2
m
2
p
(1) Tone modulation IP =.!. (1.....) => P = 0.47
3 1+0.5
(2) Gaussian with 30'  loading p2 = '!'(L) => fJ .. 0.547
3 1+ 1/9
(3) Gaussian with 40'  loading p2 .. .!. (1) => P .. 0.56
3 1+ 1/16
where For tone modulation,
m
2
=0.5
m
2
p.
2 2
For Gaussian modulation with 30'  loading, m 2 = = 1. .
mp (30") 9
100
F G
. od I' ·th 4 I d' m
2
a
2
1
or aUSSlan m u atlon WI a  oa mg, 2 = 2 ==6
mp (4a) I
12.311 Let us fU'St analyze the L+R channel. In this case, the demodulator output signal, when passed through the
, " 2
015 kHz (Iowpass) filter, is given by(L + R)' + no (I), whereS
n
(OJ) = "OJ
2
[see Eq. (12.33)].
o A
When this signal is passed through the deemphasis filter Hd(QJ) = OJ) ,the signal is restored to (L+R)
jQJ+tV)
and the output noise power is given by
1 rWI 12 ..AI rW QJttV2 ..AIQJt [ 1 w]
=:  Jo Hd(tV) Stlo (m)dQJ == 2 Jo 2 2dtV = 2 W OJ) tan 
If 1rIf QJ + tV I 1rIf QJ 1
Let us now consider the (LR) channel.
LetQJ, = 2". x 38,000 andQJI = 2". x2100.
The received signal is FM demodulated (Fig. 5.19c). The PSD of the noise at the output of the FM
demodulator isS
no
(tV) == ..AItV
2
J A2 [see Eq. (12.33)] The output of the FM demodulator is separated
into ( L + R)' over 015 kHz and (L  R)' cos 01 ,t over the band 38± 1 S or 23 kHz 10 S3 kHz. Let US consider
the signal over this passband, where the noise can be expressed as nc(/)cOSOJ,t + ns(/)sinQJcl. The signal
is (L  R)' cosm,'. Hence, the received signal is[(L  R)' + nc(/)]cOSOJ
c
+ ns(/)sinQJcl. This signal is
multiplied by2cosmcl and then lowpassfiltered to yield the output(L  R)' + nc(/). But
Sfic (01) == Sn(m+OJc)+Sn(mQJ,) == [(01+01,)2 +(mOJ
c
)2]
When this signal is passed through deemphasis filter Hd(m) = . 4)1 ,the signal is restored to (LR) and
jOJ+ml
the output noise power N:; is given by
" 1 rWI 12 (W[ 2 2J mf
NO==Jo Hd(m) Sn (OJ)dm=2Jo (QJ+QJc) +(0101,) 2 2dQJ
". ; 1rIf 01 +011
 + tan 
2..A1mr [w  tVr 1 w]
mof2 01) 01)
Hence, the (LR) channel is noisier than (L+R) channel by faaor!jf, liven by
o
2(W + OJ: +QJt 2(B+ f,2 + 1.
2
tanI..!)
N"
01
1 lOl Ii II
= W t( W ) ""' B ' t( B)
01) tan ;;  Jl tan II
Substituting B '"' 15,000, I, == 38.000, II == 2100 in this equation yields:
Nil
== 166.16 = 222 dB.
101
W = 2". x 15,000
12.41 L = M
n
=> n = IOiM L
!..£. = 3L2 m
2
(t)
N m
2
o p
3M"(:': )
11.41 = 55 dB = 316200
No
For uniform distribution
 1
m
2
=I"'Pm
2
dm
2m III
p P
I
=m
3 p
(0) 316200 = 3{2 )'. [:,: )
=3(2)2n(i)=2
2n
2n = 18.27
Since n must be an integer, choose n = 10 and L = 1024
(b) = 3(2)20 1. = 1.048576)( 1 0
6
60.17 dB.
No 3
B PCM = 2nB = 90 MHz (assuming bipolar signaling)
o
Fig. SJ1.41
(c) To increase the SNR by 6 dB, increase n byl, that isn = II. Then the new bandwidth of transmission is
22 x 4.5 = 99 MHz.
12.43 Sj = 2 BnEp , Ep = 2)( 10
5
, B = 4000, n = 8
Sj = 2 )( 4000)( 8 )( 2 x 10
5
= 128
S 128 A2
r = _I = = 2.56)( hi
..)JB 2)(625)(10
7
)(4000
t;) Q.m = 7.569 )( 10
9
B", = nB = 8)( 8000 = 64 kHz (assuming bipolar line code)
S 3(2)2n ( m
2
}
(0) "ff,;= ;;;
where rr 2x2xl0
5
=J2S6 =.ffi
V; ..)J 2 x62Sx 10
7
8
S 3(2)16 (1)
So. it; = 1+4(216 1)Q(J32) 9" ;:21845= 43.4 dB.
(b) Ifpower is reduced by 10 dB, theny = 25.6, Q(J3.2) = Q(1.79) = 0.0367 and
3(2)16 (1.);:227.3.56dB.
No 1+4(216_ 1)Q(m) 9
102
The table below gives SNR for various values ofn under the reduced power.
Hence, n = 3 yields the optimum SNR. The bandwidth in this case is Bm = 3)( 8000 = 24 kHz.
12.44 1 PE = P (correct detection over all K links) + smaller order terms
== (1 p,)KI(I_ == [I(K 1)p,Il == 1 (K 1)P,
So PE = +(K 1)Pe
(b) r" 2S dB= 316.2, r = 23 dB= 199.5
P, = = Q(6.287) = 1.6)( 10
10
.. = Q(4.994) ... 3)( 10
7
P
E
.. 99)( 1.6)( 10
10
+3)( 10
7
.. 3.16)( 10
7
=
12.45
As noted on Pg. (570), the optimum tilters for OSBSC and SSB·SC can be obtained from Eqs. (12.83a)
and (l2.83b), provided we SUbstitutei[ Sm(t» + Q),,) + Sm(Q)  Q),,)] for S.(Q1} in these equations. Let
Sm(Q)) ... irSm(Q) + Q)t:)+Sm(Q) Q)t:)]
+a' + (.,.,:;, +a']
a
2
(Q)2 +a
2
) alOGO",
.. 2
(Q)2 +a
2
) "c .. 2",xIOS
We shall also require the power ofSm(t»).
I:; _I roo Sm(Q)):/Q)
2n Joo
We can simplify the evaluation of this integral by recognizing that the power of the modulated signal
m(l)cosQ)"t is half the power ofm(t). Hence,
1= Sm(tV):/Q) =  dQ) = tan  =
1 1 Joo 1 tao a
2
a _I tV lao a
2 2n 00 2n Q)2 + a
2
211' a 0 4
We shall use the POE system shown in Fig. 12.19
103
(1)
(2)
(a) For this system
IH p(t» t = __ S=m ... ( t» .... ) =
I
(\Ir«' Sm(t»)Sn(t») df
He t»I1J_ IHe(t»)I.
Because H c( t») and Sn (t») are constants, we have
I
(
\/2 STJ"""I/S"m(m) _
Hp mil '"'  1 f't.I"::\
J: "/Sm(t»)df ;if: "Sm(m)dt»
wbereSm(w) is found in Eq. (1). AI50from Eq. (12.83b)
\/1 G
l
J: JSm(m)df 10
4
$; JSm(OJ)dOJ
IHd(m" =  =
ST ./i/ Sm{m) 10
3
tr/ JSm(t»)
(b) The output signal is Gm(t) . Hence, So" a
l
m
2
(t)
We have already found the powerofm{/) to be2(a/4) .. 2. Hence
ala (10
2
)2 (3ooo/r) 3/r
So =2= 2 = 20
To fmd the output noise power No, we observe that the noise signal with PSD Sn( OJ) = 2 x 109 passes
throu&h the deemphasis filter Hd(OJ) in Eq. (4) above. Hence, Sn{t») the noise PSD at the output of
Hd(OJ) is
A 2 2xlO
16
J: JSm(m)dm
Sn(t») = = ;1.
/r ,jSm(m)
(3)
(4)
2   1
Also, the output noise power is n(;(/)/ Ii and No = 1 [see Eq. where '" n
2
= ; f: Sn(OJ)dOJ
and £. = 3tr/20 • 3tr
2
N. ;. r S.(e»)de» lOr S.(e»)de»
12.S.% Similar to Prob. 12.51
12.S.3
Hence, the improvement ratio is
104
13.11
A
Chapter 13
outf't t Clf'
eutpllt of''*'''f.
"Lt:.)
f' 1'; It" r ft\ 4IItt.htd
0
ofWt ATb
tc PLt:)
Tb
t:. ....
1b
t
t..,.
(p.)
C,b)
tc! )
Fig. S13.11
For the integrate and dwnp filter (1&0), the output is the integral of p(t). Hence, att = T", Po(T,,) = AT
b
.
If we apply 6{t} at the input of this filter, the output h(t) = u(t)  u{t 7j,) .
Hence,
and
and
p2 = p;(Tb) = A2Tl = 2. E
n;(t) vVlb/2 V¥ p
This is exactly the value of p2 for the matched filter.
13.12 The output Po{t) of this RC filter is p'lt:)
Po (I) = A(l_e
t
/
RC
)
..
 .. .,  ....
..
I>T".
t
The maximwn value ofPo{/) is Ap. which occurs atTb:
Ap  Po(T,,) =
2 l...Nrc dO) ...N
0" n = 2". . '2 Jco 1 + 0)2 R2 C2 = 4iiC
and
lOS
13.21
We now maximize p2 with respect to RC. Letting x = 1bIRC, we
and
This gives
and
x == 1.26
Hence,
Observe that for the matched filter,
2 4.A216 (1_e;r)2
p=_.
x
or
or
2
:: (0.816) 2.A 1j,
. 2 2Ep 2A21j,
PmlX ""::
=
The energy of p( t ) is 1j, times the of p( t) .
Hence,
Similarly,
Hence,
and
1 1.26
=
RC 1j,
13.22 Let C I be the cost of error when I is transmitted, and Co be the cost of error when 0 is transmitted. Let the
optimum threshold bea
o
in Fig. SI3.22. Then:
q ::qoP(E1ml)qoQ(
Co:: COl P (Elm = 0) = COl Q ( .Ap(T:a
o
)
106
The average cost of an error is
(1)
For optimum threshold dC/do
o
= O. Hence, to computedC/do
o
, we
observe that
P ~
po "r 4c Ap Y"
and
Fig. SI3.22
Hence,
Hence,
and
But
Hence,
13.23 We follow the procedure in the solution ofProb. 13.22. The only difference is Pm{l} and Pm (0) are not
0.5. Hence,
C  P
m
(1) q + Pm (0) Co = Pm(l) qo Q ( A p u ~ Q O )+ Pm(O) COl Q ( Apu:
ao
)
and
Hence,
In [P",(O) C
OI
]= laOAp =>ao = u ~ In [P",(O) COl]
P",(l} qo u ~ 2Ap P",(I) C
IO
107
But
Hence,
Qo = c.Alln [P",(O) COl ]
4 P ... (I) C
IO
13.51
s
r
Fia. SI3.51
2 c.AlE p
tf,.=
2
The thresholds are ± E p /2 and
= )
13.52 Here, p( I) and q( t) arc identified with 3 p( I) and p( I), Hence,
H( Q») .. [3 p( Q) )  p( Q») )e jeur" 2 p( Q) )e:' jc»r"
and
11(/) .. 2P(7b I)
Qo = i[E3P Ep] =i[9E
p
 Ep]" 4Ep
But multiplication of II( t) by a constant does not affect the performance. Hence we shall choose II( t ) to be
p( 1b  t) rather than 2 P( 7b I). This will also halve the threshold to Q
o
= 2E p. This is shown in Fig.
SI3.52. Also,
108
and
9E +E
The energylbit is Eb .. P 2 p = 5Ep Hence,
13.53 For M = 2, c.AI = 2 x 10
8
For 256,000 bps the baseband transmission requires a minimum bandwidth 128 kHz. But amplitude
modulation doubles the bandwidth.
Hence
BT = 256kHz
10
7
= Eb = 2.7 x 10
7
Sj = = 2.7 x 10
7
x 256,000 = 0.069W
For M = 16
This yields Eb .. 5.43 x 106
S; = EbRb .. 5.43 x 106 x 256,000 "" 1.39W
For M=32
B = 256,000 "" 512 kHz
T 1012 32
P,J,I =
10
12 32"" S x 10 =  L
7 )
32 1023c.A1
This yields Eb = 1.719 x 10
5
Sj "" = 1.719 x 10
5
x 256,000 ... 4.4W
13.54 For M '" 2 andc.Al 2 x 10
8
This case is identical to MASK for M = 2
10
7
= Eb" 2.7)( 10
7
S; = EbRb = 2.7 x 10
7
)( 256,000 = 0.069W
BT = 256,000 )( 2 = 256 kHz
2
109
I
For M = 16
S; = EbRb = 1.67 x 106 x 256,000 = 0.4275W
In MPSK, the minimum bandwidth is equal to the number ofMary pulses/second.
Hence,
For M = 32
B = 256,000 s 64 kHz
T 1012 16
Sj :: EbRb :: 5.24 x 106 x 256,000 = 134W
B = 2'6,000 ='12 kHz
T 1012 32
110
Chapter 14
14.11 The following signals represent 2 sets of S mutually orthogonal signals.
I
I
~ 
Fig. SI4.11
14.11
itl
~
~
, ~
:2.T
t,....:..

'"
\J'f:
tJ
~
!l
'1 "'r
t.fTe
._ ..L
2.;T
,. e
I
~
ve
~
~
_.L
\ffc,
_.1.
Z"1:
FiC·SI4.Il
111
14.13 I)
"To
21r
"'0 =
To
:l/vro
1/v=e
Fig. SI4.13
112
14.14
~ ) f, (ot: )
~
a.
~
.. .L..
""0
\lTD
(1;2
/
', ') .. )
If
I) it)
"f
ft
.. "
..i.
"'t
(3,::2.1 '3,4f) /)
b) The energy of each signal is:
£3 = 9 +
4
+
9
+ 16 + 1 10 = 39
To
. f ~ ic)
It
'fII!
la, I )  ~ J ",2)
"0 ...... ~ Si +*
wi
r., ft.)
1\
~
J'bt t 'b
't'"
 a...
tfFo
(::2) "I) lJ 2
/
.5")
FiR. SI4.14
c) F
3
· F4 = (68 +6+8 +0) = O. Hence, 13(1) andf4(t) are orthogonal.
14.11 Let X(I) = xI. X(l + 1) = x2 x(t + 2) = xl
We wish to detennine
Since the process x(t) is Gaussian, x .. xl, xl arejointl)' Qaussian with identical variance
(
(12 == (12 = (12 .. R,,(O) == 1). The covariance matrix is:
"I "2 "3
[U
2
(1"1"2
"I
K = (1"2"1
(12
"2
(1"3"1 (1 "3"2
so
U,.,. 1
(1"1"3
Also
(12
"3
(1 "1"2 == 0'''2''1 = i;X;" s x(t) x(t + 1) .. Rx (1) = .!.
e
 1
(1 "2"3 = 0'''3''2 .. x2
X
3 .. x(t + 1) x(t + 2) .. R" (1) .. 
e
(1"1"3 = 0'''3''1 .. xlx3 = x(t) x(t +2) = R,,(2) = ~
e
113
14.31
1
K= 
e
J
e
e
2
e
Hence
and
Hence
1
e
2
1
e
•
S,
and
t •
~
J
i
at
•
..;a.
. }_.. ......
5""
FiR. SJ4.31
P(CjmJ) = PrO{ nl < i) and p(CjmM) = PrO{ nJ > ~ a )
114
Hence the average pulse energy E is
Also
E (M2 1)a
2
Eb = Iog
2
M = 1210g
2
M
Hence
Which agrees with the result in Eq. (13.52c)
14.32 p( qmd = p( C;m4) ... p( C;ms) = p( Clms)
p( Cl'"2) = p( C\m)) = p( C/m6) = p( C\m, )
and
p( qmd = n 1 < i, n2 > ;)
P(C\
m
2) = < i. 82 >
= [1 [1 Jh )]
P( C) = H p( qm.) + p( C\
m
2) J = t[ 1 Jb )] [
2
3q Jh )]
P,M
 &,/2
....
J
c. ...
r; S,,1. + t.... \..... .$
r .. s&2.
•
Fie. S14.Jl
115
The average pulse energy E is
E=i({(i)' + (i)} {(i)'
and
P.,. iQ( rn [S.3Q(JJ)]
= 25Q ( assummg Q «1
This perfonnance is considerably better than MASK in Prob. 14.31, which yields
P,M = 1.7SQ ) for M = 8
14.33 In this case, constants a" 's are same for k = 1, 2, ..... M. Hence, the optimum receiver is the same as that
in Fig. 14.8 with tenns a" 's omitted.
We now compare ,·s), "s2, ..... '·sM .
Since r· s" = .fir cosO" is the angle between' and Sic , it is clear that we are to pick that signal Sic with
which , has the smallest angle. In short, the detector is a phase comparator. It chooses that signal which is
at the smallest angle with ,.
14.34 Because of symmetry,
p(qmd = P(q
m
2)=·····= p(qmM)
where M =2N
and
Let
Then
d d
5·· =: or
!) 2 2
Nd
2
E) = E2 = ..... = EM =E
4
(
d d d)
5)= 2' 2' .. · .. '2
116
j
M :,
•
•
...
•
1
•
N:,.
•
• •
Fit. SI4.34
I.
14.35
14.36
and
p(C} = P(C\ml)
P,u = IP(C) = Jr
Here, M = 2 N • Hence, each symbol carries the infonnation 1082 M EO N bits .
Hence
and
Also
Hence
..AI P(mo) d I
P = In+ ,.,
2d p(md 2
= Fli. SI4.35
2d 2
p(cjm,) p(cjm,) = !'rob.[n, > (d  p)l = J
P(Cimo)= Prob an,l < p)= J
P{ C) = p( qmo) + ± p( + ± p( C\mI)
 )  )
2E_1n2 2E+1n2
P.M=1P(C}=2
1
+
P,M = 1 P(C)
p(C) = p(q'"2)]
P(C\ml) __ 1 lJ
e
[<qld
Y
+tll''''' tiqltiq2
JI..,\j RI
3d
=...!.. TJq11ln(K/B) e [<qld)l+qU""]tiq dq
Jl..\I J ql tIn(K/B) 2 I
o
117
and
't,'"
•
Fig. 514.36
(Q
~
~ 4
Sa
•
•
,
,
,
.
14.37
.1.. ~ tI>, ( /;; )
1
rr;
lb
t ; ~
Jtt
c1,:a.lt)
}I
I.,
I
t.
Fla. 514.37.
Note that
d d
51 = "2;\ "2;2'
d d
5) = "2;\ "2;2'
118
0
.9.
.JTj,
(.(F)
$;,./i ')
5 ... ('1:)
.$Sl!J

I
=i"
..
i r"
1b
0
i "1b
0
!Iz
"Ii. t.. c
t:. 0
::a.
_d

Fia. S14.37b
(c)
assuming J :Ii ) « 1
20'"
and
p{£!ms)=4
J
; )
l2 20'11
We also observe that E. the average energy is E = ) = O.4d
2
E = O.4d
2
= 02d
2
and and d =JSE
JJ v\I 20'" 4JJ 2JiO'II &.AI
Therefo.. Plq .. ,) J J
The decision region R2 for "'2 is shown in Fig .• and again in Fig. CI. R2 can be expressed as the ftrst
quadrant (horizontally hatched area in Fig. CI)  AI' Thus
1'( Cf"'2) = noise originating fi'om s2 lie in R2
= p( noise lie in 1st quadrmt)  p( noise lie in A I )
 [1 J r 1'( noise oriainatina fiom" lie hi AI)
But P( noise lie in AI) = ±[ p( noise lie within outer square)  p( noise lie within inner square) J (See
Fig. C2)
=i[ {Inll. in21<i){lnll.
119
t ..
T,b
= 4  +4
1[ d) d )]
 4 20'n 2J2O'n
and
and
Fig. S14.37c
Moreover, by symmetry
P(ti
m
3) =
Hence
14.38
."'4
.>''b\P
..
", ;A
4 •. ( .:
• • "'3
""t ,.
'e "' ..... ;
.iL  •
, .'
.•.. ..,
'R,
•
•
Fig. S14.38
1 4 1
p(c) =  I: 4 P(C/m; ):: [ P(q
m
1) + P{Q'"2) + P(q
m
3) + P(C!m4)]
16 ' .. 1 4
The decision region R1 for ml (see Figure) can be expressed as
R1 = outer square of side d.fi  (outer square  inner square of side d)
:; .!. outer square of side dJ2 +.! inner square of side d
4 4
120
Now p( qml) = Prob( noise originating from ml lies in R
1
)
= .!. p( n lie in outer square) + ! P( n lie in inner square)
4 4
Similarly R2, the decision region for "'2 (see figure above) can be expressed as
R2 = outer square of side d.fi .!. (outer square  inner square of side d)
2
= 1.. outer square of side d.fi .!. inner square of side d
2 2
and p( <=1"'2) = noise originating from "'2 lie in R2
Zl
·· , ..... ,' ...
\...... t " '.
r .••. <{ ...... :'t.:" ',.
,..J ..... .... 1>· d
,. ._ :;e:,: ... .' .
.. I •• '
' .... _ ..  , .....
The decision region R3 for m) can be expressed as
and
R3 = R,4 + Rs  Rc
p( C1m3) = Prob(noise originating from m) lie in R3)
". P(noise in R,4)+ P(noise in Rs)  P(noise in Rc)
= p(nl 11n11.
121
p.,. .....
·  ".. · 1 . ... "'.'
2A
.r f
•••• 1
The decision region 14 for m. can be expressed as
I4=R
A
RB
and
P(q
m
4) = P(n) > d, n2 > d)  In21 < d) In21 < 12)}
+{d8)]' ±[J2{d8)]'
For any practical scheme Q(.)« 1, and we can express
[lkQof s;12kQ(')
Using this approximation, we have
Hence
Now
Therefore
And
so that
Therefore
Moreover
Hence
And
P(q
m
2) == h )2q
p( q
m
4) == 1 { d 8)a( h)
E) = d
2
, £2 = 2d
2
, £3 = 4d
2
, and £4 = &1
2
.
E .. i(d
2
+2d
2
+4d
2
+&1
2
) _ d
2
E E
Eb"=
108216 4
!L .. E
..AI 16.Jj
122  
P.M = 1 P(C) = 8 Eb)
e 2 IS JJ
Comparison of this result with that in Example 14.3[Eq.(14.S7)] shows that this configuration requires
approximately 1.5 times the power of the system in Example 14.3 to achieve the same performance.
14.39 If 51 is transmitted, we have
=a=1i n2
b_
l
=E+a.[i nl
b2 aIi n2
and
Note that
hi> b_
1
implies E + a+1i nl > E +a,[i nl or nl >Ii
b
l
implies E +a+1i nl > a+IE n2 or "2 < Ii +nl
b
l
>b_
2
implies E +a+JE 01 >aIi n2 or "2 > (JE +n1)
Hence
Similarly
Hence
P(Cjmt}=Prob.{b
t
>hlo b2. b2. b), b_3 .. ·· bA;. b_k)
= prob.[nl > JE. In21 < (nl + JE). I
n
31 < (nl + IE) . .... Inkl < (nl + JE)]
Since nl. n2 .... ok are all independent gaussian random variables each with variance JJ/2.
P(q
m
l) = [p(nl > Ii) POn21 < nl +JE)P(l
n
31 < n1 +1£) ... POnkl < nl + Ii)]
= 1 j enlIJJ[r+JE enlIJJdntldn
,J1VJ JE (IIJ+JE) j I
= I j enlIJJ[I2) yJE)J
N

1
dn
l
J1VJ JE \. JJ/2
L
n1 +/i
ct Y 

{
1U)2}
I GO N 1
Je' [12Q(Y)] 
,,2tr .
Also
E E
b = log2
2N
14.41 The onoff signal set and its minimum energy equivalent set are shown in Figs. (a) and (b). respectively.
The minimum energy eqUivalent set of orthogonal signal set in Fig. (e) is also given by the set in Fig. (b).
Hence. onoff (Fig. a) and orthogonal (Fig. c) have identical error probability. The set in Fig. (b) is polar
with half the energy of on off or orthogonal signals.
123
""'"0 .... .... ···
CIt
(A.')
14.42 Here
Therefore
SI{I) = 50 ;2(1)
S2(1) =.rs ;I{I)
S3(1) = 15 ;1(1)
......
d 0
i:
c;,)
d

FigSI4.41
II = l" 4>2
'2 = 15 4>1
13 = 15 4>
Fig. SI4.42
14.43 To fmd the minimum energy set, we have II = 1.(11 +12 +$3 +14) = tJi ;,.
4 .
Hence the new minimum energy set is
II ;,.."Ji fSJ 12 =tJi =13 fSJ ;,.. i4 Ji;,.
124
14.44
Note that all the four signals fonn vertices ofa square because (i
l
i
2
). (i2 i
3
). (i) i
4
). and (i4 i
1
) are
orthogonal. The distance between these signal pairs is always 2J2. This set is shown in Fig. SI4.43a.
............ s
Observing symmetry we obtain
P(C) = P(qml) = P(CI"'2):: P(qm3):: P(qm4)
= 1'( nl > .J2 and n2 > ./2)
Fia. 514.43
=f!(X116)j'
P'M = 1 pee) ::1(tQ(3.16)J
2
:: I'sb 10
3
¢:a. (t.....;) __ ....
IOV';'D r
I J I
$, S,
t6.3 t... ;J ".,lr' e.. s'r.o
va) The... ortJ,0jcnal Set CD) TFie. "t9nQ1 MiniMuM C.M'f.!y«t
I ., .... ___ ...,
s.3
f
)
c o
,If t
Fig. 514.44
]25
and
P(Clm3) == P(qml)= J nl I_
J
y#.)= 1Q(7.07)
l 20.yS ,. vl05
p( Cl
m
2) == 11nl < 20rs) = 12Q(7 .07)
P( C)  j[2 P( Ciml) + P( C\"'2)]  j [2  2Q(7.07) + ( 2Q(7.07))
4
=: 13'Q(7.07)
P,M = 1 P(C) == == 1.03 x 10
12
Also E1 = £3:0: (s.ro r == 4 x 10
3
E2 = LoJwf +Lol.or =2xl0
3
E=j(EI+E2+E3>:o:ix102
Mean energy of the minimum energy set:
Emin == j(2 x 10
3
+0+2 x 10
3
) = x 10
3
14.45 The use ofEq. (14.76) and signal rotaiion shows that the minimum energy set in this case is identical to
that in Prob. 14.44. Hence the minimum energy set is as shown in Fig. SI4.44c. this situation is identical
to that in Prob. 14.35 with d == From the results in the solution of Prob. 14.3S, we have
10.y5
2
E
2
Also, we are given SII("') = .JJ == 10
5
• Hence, ..AI == 2 x 10
5
•
2
(a) From the solution ofProb. 14.35
P'M == tQ(7.02) + Q(7.l2) == 1.09 x 10
12
(b) and (c) identical to those in Prob. 14.44
14.46 (a) The center of gravity of the signal set is (II +ll)/2
Hence, the minimum energy signal set is
(11+12) 1112 (11+
1
2) h51
xI = 51 .. cl Xl =12  =.....l..
2 2 2 2
The minimum energy signals are
2 "'0 = 20001r
%1(1) = 0'sO.707sin OJot}
%2(1) = 0.707 sin OJ;'  O,s
OoOOI( OJ ,)2
(b) EXI = ! O'sO.707sin 2
0
dl=0.4984xlO
5
E%2 == EXI = 0.4984 x 10
5
. We arc given ..AI = Sx 106
Ij, = ) = Q( 4.46S) • OA Ix 10
5
126
(c:) We use GramSc:hmidt orthogonaJization procedure in appendix C to obtain
S,it)
$:/ll.1:. )
~
.i y , ~ )
...... .
o·cC! <:: O'ODI t,.
t:. ....
"
~ s ~
.•.••••.•.•..••....•...•.•..•.. tit
.
. St
Fia. S'4.46
127
Chapter 15
151.1 PI:: 0.4, = 0.3, PJ = 0.2 and P
4
:: 0.1
H(m) = (PI log PI + log + PJ log I'] + P4 log P
4
)
=1.846 bits (source entropy)
There are 10' symbols/so Hence. the rate of infonnation generation is 1.846 x 10' bits/so
15.12 Infonnation/element == log2 10 = 3.32 bits.
Infonnation/picture frame" 3.32 x 300,000:: 9.96 x lOS bits.
15.13 Infonnationlword == log2 10000 '" 13.3 bits.
Infonnation content of 1000 words ... 13.3 x 1000 = 13,300 bits.
The infonnation per picture frame was found in Problem 15.1·2 to be 9.96 x lOS bits. Obviously, it is not
possible to describe a picture completely by 1000 words, in general. Hence, a picture is worth 1000 words
is very much an underrating or the reality.
15.14 (a> Both options are equally likely. Hence,
/ = log{ts) = I bit
(b) P{2 lanterns) = 0.1
/(2 lanterns) = log210 = 3.322 bits
15.15 Ca) All 27 symbols equiprobable and P(Xj) =
HI(x) :: 27(f, log2 27):: 4.755 bitst symbol
(b) Using the probability table, we compute
27
Hw(x) =  rp(x;)log P(X/) = 4.127 bits/symbol
1.1
(e) Using Zipfs law, we compute entropy/word Hw(x).
8727
Hw(x) =  r Per) log Per)
,=1
8717
=  jlJ.log(jIJ.):: 9.1353 bits/word.
'" , ,
,=1
HIIetter =11/8215.52.14 bits/symbol.
Entropy obtained by Zipfs law is much closer to the real value than HI (x) or H2 (x).
128
7 63
15.11 H(m) = L P; Jog P; =  bits
1..,1 32
Message Probability Code 5,
~
5,
m, 112 0 112 0 112 0 112 0 112
m2 1/4 10 114 10 114 10 1/4 10 114
m3 118 110 118 110 118 110 118
110]'+114
m.
1/16 1110 1116 1110 1116
1110rll8
111
m, 1/32 11110 1/32
11110r1l16
1111
""
1164
111110r 1132
11111
m1 1164 111111
L = LP;L; = 1(1)+1(2)+1(3)+1..(4)+_1 (5)+1.(6)+_1 (6)
I 2 4 8 16 32 64 64
63 b' d"
• 32 mary 19lts
Efficiency TJ = H(m) x 100 = 100%
L
Redundancy r = (100 TJ) = ()o1o
7
15.21 H(m) ..  L P; log P; .. 2289 bits
Message
m,
m2
m3
m.
m,
m6
m7
;=1
.. 2289 = 1.4442 3  ary units
log2 3
Probability
1/3
t/3
1/9
1/9
1127
1127
1127
Code
o 1/3
I 113
20 1/9
21 1/9
220r }/9
221
222
o 113
1 113
20j1l3
21
22
o
I
2
7 1 1 1 1 1
L .. LP;L; =(1)+(1)+(2)+(2)+3(3)
; .. 1 3 3 9 9 27
13 3 d"
.. '9 ary 19lts
... 1.4442 3 ary digits
. H(m) 1.4442
EffiCiency TJ = = x 100 = 100%
L 1.4442
Redundancy r :; (1 TJ)loo .. 0%
129
s. 5,
0 112 0
lor 112
1
11
4
15.23 H(m) =  r P; log P; = 1.69 bits
Message
m.
m2
m]
rn.
Message
m.
m2
m3
m.
ms
15.24
Message
m.
m2
m3
rn.
m,
m.,
;=1
Probability
O.S
0.3
0.1
0.1
Code
o 0.5
1 0 0.3
110,..... 0.2
1 1 1J
o 0.5
101. 0.5
I1J
o
1
L = r P; L; = 0.5(1) + 03(2) + 0.1(3) + 0.1(3) = 1.7 binary digits
Efficiency" = H(m) x 100 = 1.69 x 100 = 992%
L 1.7
Redundancy r = (1 ,,)100 = 0.8%
For ternary coding, we need one dummy message of probability O. Thus,
Probability
0.5
0.3
0.1
0.1
0
Code s.
o 0.5
1 0.3
20[j0.2
21
22
o
1
2
L = 0.5(1) + 03(1) + 0.1(2) + 0.1(2) = 12 3 ary digits
H(m) = 1.69 bits = 1.69 = 1.0663 3ary units
log2 3
. H(m) 1.0663
EffiCIency ,,=  x 100= x 100= 88.86%
L 12
Redundancy r = (1 ,,)100 = 11.14%
Probability
112
114
118
1/16
1/32
1/64
1/64
Code
o 112
1 1/4
20 118
21 1/16
22U
1/16
221
222
o 112
1 114
20
J
1I4
21
22
o
1
2
3arydigits
16
From Problem 15.21, H(m) = bits .. 1242 3ary units
. H(m) 1242
EffiCIency '1"  )( 100 =  x 100 '"' 94.63%
L 13125
Redundancy r = (1 ,,)100 = 537%
130
15.15
Message Probability
m. 113
Code 51 12
1 113 I 1/3 113
m2 113 00 113 00 113 00 113
m3 119
m.
119
ms 1127
011 119 011 1/9
OIlJf2i9
0100 119 0100 1/9 0100 119
01010 1127
OIOlOr 119
0101
0101 lOr 2127 114
1127 01011
m7 1127 010111
L l: P;L; • : ~ .. 2.4074 binary digits
H(m) = 2289 bits (See Problem 15.2  2).
. H(m) 2.289
Efficiency 1];:: x 100=  x 100 = 9'08%
L 2.4074
Redundancy r = (11])100 = 4.92%
15.16 (a) H(m) = 3(1 log3)1.585 bits
(b) Ternary Code
Message Probability Code
m. 113 0
ml 113 1
m3 1/3 2
. 1.585
H(m) = 1.585 bits = = 1 3aryunlt
10g2 3
Efficiency 1] = H(m) x 100 = 100%
L
Redundancy r = (11])100 = OOA.
(c) Binary Code
Message Probability Code 5.
I ~ 2 1 3 0
00 113 1
01
ml 113
m2 1/3
m3 1/3
L = .!.(1) + (2).!.(2) = ~ = 1.667 binary digits
3 3 3
. H(m) 1.585
EffiCiency 1] = x 100 = x 100 = 95.08%
L 1.667
Redundancy r .. (11])100 = 4.92%
131
53
1 113
00 113
010 1/3
011
11213 0
00 113 1
01
(d) Second extension  binary code
H{m) = 1.585 bits
. H(m) 1.585
EfficIency" = x 100 '" x 100 = 98.391'A.
L 1.611
MasalC Frob a.
'1
52
.,
Ie
m,m, V9 001 219 01 2i9 01 2i9 01 2J9
01/'.
m,mz 119 0000 119
10 219 10 219 10 219
m,m, 119 0001 119 0000 119
II 219 Jl 219
m2m, 119 110 119 0001 119 0000 119
001]f219
000 219
m2m2 119 III 119 110 119 0001 119 0000 119 001
mlm, 119 100 119 III 119 110 119 0001
m,m, 119 101 119 100 119 III
m,m: 119 010 119 101
m,m) 119 011
15.41 (a) The channel matrix can be represented as shown in Fig. SI5.41
(b)
1 1
H(x) '" P(xl)log+ P(x2)log
P(xI) P(x2}
= 1.Iog2 3 + !.log
2
l = 0.918 bits
332
To compute H(xly), we find
132
5, So 51
00,'"
0
01 1/3 00 419 I
10 219 01
II
S
10
Fig. SIS.41
and
Thus.
Also,
H(xly) = P(Yl)H(xlYl)+ P(Y2)H(%IY2)
... .!!(0.779)+E.(0.624):: 0.6687
45 45
I(xly) = H(x)  H(xly) = 0.918 0.6687 = 0.24893 bits/ binit
H(y) = P(Yi) 10g_I_::.!! log 45 + 32 log 45 = 0.8673 bits/ symbol
I P(y;) 45 13 4S . 32
H(ylx) = H(y)/(xlY) = 0.86730.2493:: 0.618 bits/ symbol
j".
(p)
(eR)
X • ,. eX
, I
[
I 0%1 0]
Yj 0 P Ip
o Ip P
P
(a()
P(YjIXj)P(%/)
Now we use P(xjIYj) = to obtain
P(x; )P(yjlx;)
Fia·515.42
i
%i
P(x;ly;) matrix as Yj ; 1_0p]
o 1 p P
H(x) = = PlogP2QlogQ with (2Q= 1 P)
P(x;)
= [ Plog P+(l P)J = O(P) +(1 P)
1
H(xly) = I:I:P(Yj)P(%iIYj)log
I j P(%;IYj)
 Plogl +rI plog..!.+(I P)IOg_I_J+ J(1p)log_
1
_+ PIOg..!.]
, P Ip' Ip P
... 0 + 2QO(p) = (1 P)O(p)
I(xly) = H(x)  H(xly) = O(P) + (1 P)  (1 P)O(p)
= O(P)+(i pX1O(p)]
Letting P = 20(P) or O(p) = 10gP
I(xly) = O{P)+(1 PXIIog,8)
or .!..[O(P)+(IPXIJog,8)]=O. This means
dP dP
P){1 P)log(l P)(1log,8)] =0
dP
log P log(1 P) + [1logp] = 0
133
P
Therefore log  = 1 + log fJ
IP
Note: 1 + 10&2 fJ = log
2
2 + log2 fJ = 10&2 fJ
2
P fJ
=
1 P 2
:::> p=L. and I_P=_2_
fJ+2 fJ+2
so
fJ fJ+2 2 fJ+2 2 fJ+2
C = MAX l(xly) = log+Iog+ (IlogfJ) = log
/J+2 /J fJ+2 2 1+2' • ' fJ
loa.!
/J
15.43 Consider the cascade of2 BSCS shown in Fig. SI5.43. In this case
Pyfx(lll) = (1 P
1
)(11!z)+ l\1!z = 1 l\ 1!z
Pyfx (011) = (1 + 1\ (11!z) = l\ + 1!z  2l\ P2
1 .....  ...
Hence, the channel matrix of the cascade is
l\+1!z21\1!z J=[I1\
11\i!z21\P 1\
This result will prove everything in this problem.
(a)
(a) With = i!z = Pe, from the above result it foUows that the channel matrix is indeed M2.
(b) We have already shown that the channer matrix of two cascaded channels is MI M2 .
(e) Consider a cascade of k identical channels broken up as k 1 channel cascaded with the ktll channel.
If M
t
_
1
is the channel matrix of the first k 1 channels in cascade, then from the results derived in part (b),
the channel matrix of the k cascaded channels is M k = M kI M. This is valid for any It. We have
already proved it for k = 2, that M2 = Ml. Using the process of induction it is clear that Mk = Mk.
We can verify these results from the development in Example 10.7. From the results in Example 10.7, we
have, for a cascade of 3 channels
and
Now
1 P£ = (1 p,)l +3P/(1 P,)
= 13Pe +3Pe
2
 p,3 +3P,2 3P,3
= 13P, +6P,2 _4P,3
2 3
P£ = 3p, 6p, +4P,
134
I
f'
3 [IP
e
M ::
p.
Clearly
P£ ... 3p,  6P.
2
+4p'3
which continns the results in Example 10.7 for k = 3.
(d) From Equation 15.25
C, = I[P£ 1081+(1 P£
P£ IP£
where PE is the error probability of cascade of k identical channel.
We have shown in Example 10.7 that
and
P£ = 1[(1 Pe)k + I: . k! . p/(l Pe)k
i
]
j_2.4.6J!(k J)!
If kPe «I,
c, =
kP. lkP,
15.44 The channel matrix is
Also,
Y. Yz y, 
,x. q 0 p q = Ip
o q p
P(
I
)
P(Ytlxl)P(xt) ql2 1
xI YI ...   ...  ::
P(YI) ql2
P(x21Yt) = P(Ytl
x
2)P(x2) = 0
P(Yt)
P(XtIY2)= P(YZIXI)P(XI) ... O
P(yz)  
P(x I ) .. P(yzlxz)P(xz) = ql2 = 1
2 Yl P(yz) ql2
P(xIIY3) = P(Y3I
X
t)P(Xt) = p/2 s!
P(Y3) p 2
P(x2IY3) = PCnlxl)P(x2) = pl2 =!
P(Y3) p 2
135
Let
xI = 0 YI = 0
15.45
Therefore.
q
P(XloYl) = P(Xl)P(Yllxl) = '2
P(Xl,Y2) = P(xl)P(Yllxl) = 0
p
P(Xl,Y3) = P(Xl)P(Y3I
x
l) = '2
P(X2,Yl) = P(X2)P(Yllx2) = 0
P(x2,Y2) = P(x2)P(Yllx2) = f
P(xl,Y3) = P(X2)P(Y3Ix2) = 1!..
2
H(x) =  P(xl) log P(xl)  P{xl) log P(xl)
1 1
=+= 1
2 2
H(xly) = I: I: P(Xj,Yj) log ° 1
; j P(x;lYJ)
= 1(0)+0+.!. p+O+.!.q )(O+.!. p = p
2 2 2 2
l(xly) = H(x)  H(xly)
= 1 P bits I symbols
Note that for cascaded channel. the output z depends only on y. Therefore.
P{zkIYj,x;) = P(zkIYj)
By Bayes' rule
and
FiC. 515.45
It can be shown that the summation over x of the tenn inside the bracket is nonnegative. Hence, it follows
that
H(xlz)  H(xly) ~ 0
From the relationship for l(xly) and l(xlz), it immediately follows that
l{xly) ~ l(xlz)
136
if
•
M 1 JM
15.51 We have H(x) = LMPIOgr= _Mplogpdx and J ~ p d r = 1
Thus,
of
F(x,p)=plogpand Op =(l+logp)
;I(X,P) = p and 0;1 = 1
op
Substituting these quantities in Equation 15.37, we have
(1+logp)+al =0= p=e
al

I
and
Hence,
ealI =_1_ andp(x) =_1_
2M 2M
Also,
J
M 1 JM 1
H(x)= MP(x)logdx= Mlog2Mdx=log2M
 P(X)  2M
15.51 We have H(x) = J; plogpdx,
. of
F(x,p) = plogp and  = (1 + logp)
op
;.(x,p) = px and 0;1 = x
op
;2(x,p)=pand 0;2 =1
op
Substituting these quantities in Equation 15.37, we have
or
and
Hence,
1 d a .. I "
al =  an e" = al = .IS
A
so
{
I v
e 7..4
p(x)= :
x ~ O
x<O
To obtain H(x)
137
H(x) = J; p(x) log p(x)dx = J; {log A  lOge] dx
= 10gAl; p(x)dx+ f; xp(x)dx
• log A + loge = 10g(eA)
15.53 Information per picture frame = 9.96 x lOs bits. (See Problem 105.1). For 30 picture frames per second, we
need a channel with capacity C given by
C=30x9.96xI0
5
=2.988xIO' bits/sec.
But for a white Gaussian noise
C = B Jog (1 + ! )
We are given.!:: SO db = 100,000 (Note: 100,000 = SO db)
N
Hence,
B= 1.8 MHz
15.54 Consider a narrowband AI where A/ + 0 so that we may consider both signal noise power density to be
constant (bandlimited white) over the interval !if . The signal and noise power Sand N respectively are
given by
S = 2S
s
(OJ)/if and N = 2S,,(OJ)A/
The maxi..fllum channel capacity over this band 4f is given by
C4I =A/log [Ss(OJ)+Sn(tlI)]
N Sn(tlI)
The capacity of the channel over the entire band (/1, h) is given by
C = Jh log [Ss(tlI) + S,,(tlI)]df
II Sn(tlI)
We now wish to maximize C where the constraint is that the signal power is constant.
Ss(Q) df = S (a constant)
Using Equation IS.37, we obtain
(} I [Ss +Sn] OS, 0
og +a =
OSs Sn OSs
or
1
S + Sn =   (a constant)
S a
Thus,
This shows that to attain the maximum channel capacity, the signal power density + noise power density
must be a constant (white). Under this condition,
C=Jhlog [Ss(tlI)+Sn(tlI)]df=JhIOg [ 1 Jd
f
II Sn(tlI) II aSn(tlI)
=(12  I,) log ( [S,,(OJ)]dl
138
I..
15.55 In this problem, we use the results of ProbJem IS.54. Under the best possible conditions,
c= !1log J!t21og [Sn(4I)]eIf
constant
We shall now show that the integral J!t210g [Sn(CII)] elf is maximum when Sn(4I) = constant if the noise
is constrained to have a given mean square value (power). Thus, we wish to maximize
under the constraint
2 iR log (Sn (CII)] elf = N (a constant)
Using Equation 15.37, we have
or
and
a as
a (logSn)+a
n
=0
Sn aSn
)
+a=O
Sn
1
Sn(CII) =  (a constant)
a
Thus, we have shown that for a noise with a given power, the integral
i!t210g [Sn(CII)]d/
is maximized when the noise is white. This shows that white Guassian noise is the worst possible kind of
noise.
139
,:
Chapter 16
"
l'
3
16.11 211 L =
j=o
2048 1 + 23 + 23 x 11 + 23 x 77 = 2048
16.11 (a) There are (j) ways in which j positions can be chosen from n. But for a ternery code, a digit can
be mistaken for two other digits. Hence the number of possible errors in j places is
(jX31)" or 3" 3
k
t (j)21 + 3,.k t (j)2i
1=0 jO
(b) (11,6) code for t = 2
This is satisfied exactly.
16. J 3 For (18,7) code to correct up to 3 errors
3
211 I: (}B) or 211
i·O
Hence
3
211>I:e
B
)
j=O
18' 18' 18'
::: 1+' +'+' = 1+18+153+816=988
17! 2! 16! 3! 15!
Clearly, there exists a possibility of 3 error correcting (18,7) code. Since the Hamming bound is
oversatisfied, this code could correct some 4 error patterns in addition to all patterns with up to 3 errors.
16.11 GH
T
= [Ik p]
=pep
0
16.11 C::: dGwhere d is a single digit (0 or I).
Ford =0
c=O[1 1 1][000]
Ford::: 1
c = 1 [1 1 I] = [11 IJ
140
16.13 e = dG where d is a single digit (0 or 1).
Ford =0
e=O [11111]=[00000]
Ford = 1
e = 1[1 1 1 I 1] = [1 1 1 1 1]
Hence in this code a digit repeats S times. Such a code can correct up to two errors using majority rule for
detection.
16.14 0 is transmitted by [000] and 1 is transmitted by [11 1]
(8) This is clearly a systematic code with
G[ll1]
16.15 (8)
(b)
100 .. ·01
010·· ·01
p=[]
000·· ·11
.. f.......
'. P
Data word
000
o 0 1
010
o 1 1
100
1 0 1
110
111
Codeword
o 000
o 0 1 1
o 1 0 1
o 1 1 0
1 0 0 1
1 0 I 0
1 1 0 0
1 1 1 1
Note that m .. 1
(c) This is a parity check code. If a single error occurs anywhere in the code word, the parity is violated.
Therefore this code can detect a single error.
(d) Equation (16.9a) in the text shows that cN
T
= O.
Now
and
rNT ={ee,)N
T
=cN
T
etH
T
.,R
T
If there is no error , == 0 and
rN
T
=,n
T
=0
Also
NT .. [ ~ , J . But since m = 1, 1m = [1]
and
141
16.16
16.17
16.28
If there is a single error in the received word r, e has a single 1 element with all other elements being O.
Hence
,H
T
= eH
T
= 1 (for single error)
Data word Codeword
0 0 0 0 0 0 0 0 0
0 0 1 1 1 0 0 0 1
0 1 0 1 1 1 0 I 0
0 1 1 0 0 1 0 1 1
1 0 0 0 1 J 1 0 1
1 0 1 J 0 1 1 0 0
1 1 0 1 0 0 I 1 1
1 1 1 0 1 0 1 1 0
From this code we see that the distance between any two code words is at least 3. Hence dmin = 3.
Data word Codeword
0 0 0 0 0 0 0 0 0
0 0 1 0 0 J I 0
0 1 0 0 I 0 1 0 I
Observe that d
min
= 3
0
)
1 0 I 1 0 I I
1 0 0 1 0 0 0 I 1
1 0 1 1 0 1 I 0 1
I 1 0 1 1 0 I I 0
1 1 I 1 1 1 0 0 0
H T is a IS x 4 matrix with all distinct rows. One possible H Tis:
1 1 1 I
1 1 1 0
I I 0 1
1 100
1 0 1 1
) 0 ) 0
100 I
= [ ~ J
HT =
001 1
o 1 I 1
o I 10
0101
1000
0100
0010
0001
142
i
16.19
100000000001111
010000000001110
001000000001101
000100000001100
000010000001011
G=[l
k
p]= 000001000001010
000000100001001
000000010000011
000000001000 III
000000000100110
000000000010101
For d = 1 0 1 1 1 0 1 0 1 0 1
c = dG = [101 t lOt 010 t] G 10111010101 1110
Ca)
111
100111
1 I 0
010110 N
T
=
101
G= &.
Q..OJ UU
100
010
III
p
001
(b)
Data word Code word
0 0 0 0 0 0 0 0 0
0 0 1 0 0 1 1 0 1
0 1 0 0 1 0 1 1 0
0 1 1 0 1 1 0 1 1
1 0 0 1 0 0 1 1 1
1 0 I 1 0 1 0 1 0
1 1 0 1 1 0 0 0 1
1 1 1 1 1 1 1 0 0
Ce) The minimum distance between any two code words is 3. Hence, this is a single error correcting code.
Since there are 6 single errors and 7 syndromes, we can correct all single errors and one double error .
Cd)
• :dl
T
e s
1 0 0 0 0 0 1 1 1
0 1 0 0 0 0 1 1 0
0 0 1 0 0 0 1 0 1
0 0 0 1 0 0 1 0 0
0 0 0 0 1 0 0 1 0
0 0 0 0 0 1 0 0 1
1 0 0 1 0 0 0 1 1
143
(e) _____________ ;_
S • C d r
101100
000110
101010
110 010000 111100 111
110 010000 010110 010
000 00000o 101010 101
16.110 <a> done in Prob. 16.27
o 1 1
I 0 J
(b) HT =
1 10
100
010
001
t s
100000 011
010000 101
six single errors 001000 110
000100 100
000010 010
000001 001
1 double error 100100 111
16.111
[1000101]
0100111
G=[/k
P
]= 0010011
0001110
cdG
144
,
t'
4
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
11 01
1 J 1 0
1111
0000000
0001110
0010011
0011101
0100111
0101001
0110100
0111010
1000101
1001011
1010110
1011000
1100010
1101100
1110001
1I111JJ
! !
0000001 001
0000010 010
0000100 100
0001000 110
0010000 011
0100000 111
1000000 101
s = ,H T where, = received code
c,ee
c = corrected code
101
111
o 1 1
HT = 110
100
010
001
16.211 We observe that the syndrome for all the three 2error patterns 100010, 010100, or 001001 have the same
syndrome namely 111. Since the decoding table specifies, = 111 for, = 100010 whenever e = 100010
occurs, it will be comcted. The other two patterns will not be corrected. If for example, = 010100
occurs, s  111 and we shall read ftom the decoding table, = 100010 and the error is not corrected.
Ifwe wish to correct the 2error pattern 010100 (along with six single error patterns), the new decoding
table is identical to that in Table 16.3 except for the last entry which s}lould be
, !
010100 111
16.213 From Eq. on P.737, for a simple error correcting code
or
This is satisfied for n 12. Choose n = 12. This gives a (12, 8) code. HT is chosen to have 12 distinct
rows of four elements with the last 4 rows fonning an identity matrix. Hence,
145
16.214
001 1
0101
01 10
o 1 1 1
1001
H
T
"" 1010
1 0 1 1
1 1 0 0
1000
0100
0010
0001
p
G=
100000000011
010000000101
001000000110
000100000111
000010001001
000001001010
000000101011
000000011100
The number of nonzero syndromes = 16  1 = 1 S. There are 12 single error patterns. Hence we may be
able to correct 3 doubleerror patterns.
! !
0000 000000000000
0011 100000000000
0101 010000000000
0110 001000000000
0111 000100000000
1001 000010000000
1010 000001000000
101 1 000000100000
1100 000000010000
1000 000000001000
0100 000000000100
0010 000000000010
0001 000000000001
1 1 1 1 100000010000
1 1 10 001000001000
1101 000000010001
Data word Codeword
00 000000
01 011011
10 101110
1 1 110101
The minimum distance between any two code words is d
min
= 4. Therefore. it can correct all Ierror
patterns. Since the code oversatisfies Hamming bound it can also correc:t some 2error and possibly some
3error patterns.
146
16.214
001 1
0101
01 10
o 1 1 1
1001
H
T
"" 1010
1 0 1 1
1 1 0 0
1000
0100
0010
0001
p
G=
100000000011
010000000101
001000000110
000100000111
000010001001
000001001010
000000101011
000000011100
The number of nonzero syndromes = 16  1 = 1 S. There are 12 single error patterns. Hence we may be
able to correct 3 doubleerror patterns.
! !
0000 000000000000
0011 100000000000
0101 010000000000
0110 001000000000
0111 000100000000
1001 000010000000
1010 000001000000
101 1 000000100000
1100 000000010000
1000 000000001000
0100 000000000100
0010 000000000010
0001 000000000001
1 1 1 1 100000010000
1 1 10 001000001000
1101 000000010001
Data word Codeword
00 000000
01 011011
10 101110
1 1 110101
The minimum distance between any two code words is d
min
= 4. Therefore. it can correct all Ierror
patterns. Since the code oversatisfies Hamming bound it can also correc:t some 2error and possibly some
3error patterns.
146
(b)
1 1 10
1 0 1 1
T 1000 T
H = 0 1 0 0 and s = eH
0010
0001
6 siDgle error __ {
7 doubleerror patterns
2 tripleerror patterns {
16.31 Systematic (7, 4) cyclic code
g{x} = x
3
+x+ 1
For data 1111
100000
010000
001000
000100
000010
000001
IJOOOO
101000
100100
100010
100001
011000
010010
000111
001101
X
3
(X
3
+x2 +x+ 1) = x
6
+X
S
+x4 +x3
x
3
+x2 + 1
x
3
+x+ l)x6 +x
S
+x4 +x3
x
6
+ x4 + x
3
x
S
x
S
+x3 +x2
x
3
+x
2
x
3
+x+ 1
x
2
+x+ 1
s
1110
1011
1000
0100
0010
0001
0101
0110
1010
1100
I 1 1 1
0011
1001
0111
1101
c(x} '" (x
3
+x+ lXx
3
+x+ 1) = x
6
+x
S
+x4 +x3 +x2 +x+ 1
The code word is 11111111
147
x
3
+X+ I)X6 +x
S
+X4
X6 +X4 +X3
x
S
+x3
x
S
+X3 +X2
X
2
The code word is 1110100
A similar procedure is used to fmd the remaining codes (see Table 1).
(b) From Table I it can be seen that the minimum distance between any two codes is 3. Hence this is a
single error correcting code.
d
1 111
1110
11 01
1100
101 1
1010
1001
1000
0111
0110
0101
0100
0011
0010
0001
0000
c
1111111
1110100
1101001
1100010
1011000
1010011
1001110
1000101
0111010
0110001
0101100
0100111
0011101
0010110
0001011
0000000
Table 1
(e) There are seven possible nonzero syndromes.
for , = 1 0 0 0 0 0 0
s=101
x
3
+x+ l)x6
x6+
x
4+x3
x4 +x3
x4 +x
2
+x
x
3
+x
2
+x
x
3
+x+ I
x
2
+1
148
The remaining syndromes are shown in Table 2.
1000000
0100000
0010000
0001000
0000100
0000010
0000001
s
101
1 J 1
110
011
100
010
001
Table 2
(d) The received data is 1101100
r{x) = x
6
+x5 +.1'3 +.1'2
.1'3+.1'+1).1'6+.1'5 +.1'3+.1'2
.1'6 +.1''' +.1'3
x! +.1''' +.1'2
x
5
+.1'3 +.1'2
x" +.1'3
x4 +.1'2+.1'
x
3
+ x
2
+ x
s{x)=x
2
+1 x
3
x+l
.=101
From Table 2
,,=1000000
c=r$,,=110IIOO$1000000 0101100
Hence d = 0 1 0 1
16.32 g(x) = xii + x
9
+ x
7
+ x
6
+ x
5
+ x + 1
c{x) = d{x)g(x)
1.
and
and
d
l
=000011110000, d
l
{x).,X
7
+x6+x5+x4
d
2
= 1 01 0 1 0 1 0 J 0 I 0, d
2
(x) ... xli + x
9
+ x
7
+ x
5
+ .1'3 + x
cJ{x) = dJ{x)g(x) = xiS +.1'17 + .1'13 +x12 +xll +.1'9 +x
S
+.1' 7 +.1'4
CJ=00001100011101110010000
C2(X) = d2(X)g(x) = .1'22 + xiS +x17 +%15 +.1'13 +x
S
+%5 +%4 +x3 +.1'2 +x
C2 .. 1 0 0 0 1 1 0 1 0 1 0 0 0 0 1 0 0 1 1 1 1 1 0
149
16.33 x+ I)X3 +X2 +x+ 1
X
3
+x
2
x+1
x+1
o
Hence x
3
+x2 +x+l (x+ IXx
2
+ I) =(x+ 1)(x+ IXx+I)= (x+ 1)3
16.34 X+l)x5+ x4+x2+1
x
S
+x4
16.3·4
x
2
+1
x
2
+x
x+l
x+1
o
Now try dividing x4 + x + 1 by x + I, we get a remainder I. Hence (x + 1) is not a factor of (x4 +x + I).
The 2
nd
order prime factors not divisible by x + 1 are x
2
and x
2
+ x + 1. Since (x4 + X + I) is not divisible
by x
2
, we try dividing by (x
2
+ x + I). This also yields a remainder I. Hence x
4
+ x + 1 does not have
either a fll'st or a second order factor. This means it cannot have a third order factor either. Hence
x
S
+x4 +x2 + I = (x+ IXx4 +x+ 1)
Try dividing x
7
+ I by x + I
x
6
+x5 +x" +x3 +x2 +x + 1
X+l)x7 +1
x
7
+x6
x
6
+1
x
6
+x
S
x
5
+ I
x
5
+x
4
x4 +1
x4 +xl
x
3
+ 1
x
3
+x
2
x
2
+ I
x
2
+x
x+ 1
x+l
o
ISO
Now try dividing (X
6
+ x
S
+ X4 + X
3
+ X
2
+ x + 1) by (x + 1). It does ~ o t dividc. So try dividing by
(x
2
+ 1). It docs not divide. Try dividing by (x
2
+ x + 1). It does not dividc. Ncxt try dividing by
(x
3
+ I). It docs not dividc cithcr. Now try dividing by (x
3
+x+ 1). It divides. Wc fmd
(x
6
+x
S
+x4 +x3 +x2 +x+ I) =(x
3
+x+ IXx
3
+x2 + 1)
Since (x
3
+x2 + 1) is not divisible by x or x + 1 (thc only two firstordcr primc factors), it must bc a
thirdordcr prime factor. Hencc
x
7
+ 1 (x + IXx
2
+x+ lXx
2
+x2 + I)
16.36 For a single error correcting (7, 4) cyclic code with a generator polynomial
g(x)  x
3
+x2 +1
k:!:4 n=7
X
k

1
g(x)
~
x
k

2
g(x)
x
3
g(x)
x
6
+x
S
+x3
x
2
g(x)
x
S
+x4 +x2
==
==
x g(x)
x4+x3+
x
g(x) x
3
+x2 + 1
g(x)
Hence
[
1 101000]
0110100
G'=
0011010
0001101
Each codc word is found by matrix multiplication c  dG'
c[OOOO] [:i::r;:]= 0000000
0001101
c=[OOOl] [E:;!;:]== 0,001101
0001101 .
151
The remaining codes are found in a similar manner. See table below.
d
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1110
111 1
c
0000000
0001101
0011010
0010111
0110100
0111001
0101110
0100001
1101000
1100101
1110010
1111111
1011100
1010001
1000110
1001011
The desired fonn is
1000· .. Oh
l1
~ I ~ I ... ~
o I 0 0 . . . 0 hl2 hn h32 . . . ~
o 0 1 0 . . . 0 hl3 ~ 3 h33 . . . ~
G'= ...... .
pO 0 0 . . . 1 hl/e ~ k h3k . . . h",.
. ~ '     ~     ~
I. ;.
(b k) (k,X m)
The code is found by using c == dG
Proceeding with matrix multiplication, and noting that
o + 0 ... 0, 0 + 1 = I + 0 == 1, I + I ... 0 and 0)( 0 = 0, 0 )( 1 ... I )( 0 ... 0, I )( 1 == 1
we get
[
1000110]
0100011
cIS == [1 I IIJ 0010 t 11 = [11 l1111J
0001101
cl4 = [11 1 0] G = [111 00 I 0]
and so on.
152
d
1111
11 10
1101
1100
1011
1010
1001
1000
0111
0110
0101
0100
0011
0010
0001
0000
1111111
1110010
1101000
1100101
1011100
1010001
1001011
1000110
0111001
0110100
0101110
0100011
0011010
0010111
0001101
0000000
These results agree with those of Table 16.5
16.38 (a)
[
101 1000]
0101100
G'=
0010110
0001011
(b) The code is found by matrix multiplication. c=dG'
In general g{x} = glx,,k + g2x,,k1 + "'g,,hl
For this case gl = I, g2 = I, 83 = 0, g4 = 1
Since hlk = g2, ~ k = g3, h]k = g4' the fowth row is immediately found. Thus, so far we have
Next, to get row 3, use row 4 with one left shift.
[
0011010]
0001101
The I is eliminated by adding row 4 to row 3.
[
0010111]
0001101
Next for row 2, use row 3 with 1 left shift.
153
[
0101110]
0010111
0001101
The 1 is eliminated by adding row 4 to row 2.
[
0100011]
0010111
0001101
Next for row 1, use row 2 with 1 left shift.
[
1000110]
0100011
0010111
0001101
'Ibis is the desired fonn.
c
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1 J 10
1 1 1 1
d
0000000
0001011
0010110
0011101
0101100
0100111
0111010
0110001
1011000
1010011
1001110
1000101
1110100
1111111
1100010
1101001
(c) All code words are at a minimum diSllnce of3 units. Hence this is a single error correcting code.
16.39 g{x) = x
3
+ x+ I. Hence row 4 is 000 1 0 1 1.
G'[:!; ~ :;:]
0001011
Row 4 is ok.
Row 3 is left shift of row 4.
For row 2, left shift row 3.
And add row 1 to obtain row 2.
For row 1, left shift row 2.
And add row 1 to obtain row I.
0001011
0010110
0101100
0100111
1001110
1000101
154
+ row4
+ row 3
+ row 2
+ row I
[
1000101]
0100111
G=
0010110
0001011
16.41 The burst (of length 5) detection ability is obvious. The single emr correcting ability can be dem onstrated
as follows. If in any segment of b digits a single erTOr occurs, it will violate the parity in that segment
Hence we locate the segment where the error exists. This erTOr will also cause parity violation in the
augmented segment. By checking which bit in the augmented segment violates the parity, we can locate
the wrong bit position exactly.
16.51 The code can correct any 3 bursts of length 10 or less. It can also correct any 3 random errors in each code
word.
16.7·1 PEII = = 12Q(J2 x 9.12) = 9.825 x 106
P£c = (l' H ) r = (l'XQ( J9516S )]' ",12 x 10"
To achieve a value 9.872 x 10
9
for PEN' we need new value say Eb/,,IJ. Then
9.872 x 10
9
= )= )
Hence
) = 0.8227 x 10
9
and
J2£;, = 6.03 => Eb = 18.18
..... IJ ..AI
This means Eb must be increased from 9.12 to 18.18 (nearly doubled).
155
Therefore
2.15
Pq
1 = 4'
2
/2
_.2(1) r/t = 64/7
~:2
(b)
P211
=
i 12(
<a>
(b)
2t :l):2dt
= 4\64/7) = 256/7
(c) Peg = 
4
112
2
:l:2 (ct) dt = 64!::2 /7
Sign change of a signal does not affect its power. Multiplication of a signal by a constant (. increases the power by a fartor ,.2.
2.16
PII =  1
2".
J"
VI2(t)
,it
= 1
_or
2".
11</2 dt = 0.5
1</2
(c)
Pg
== (d)
1 jTOIZ
To
1/'~(t) dt == dt To/2 To JToI2
1
1
(Tu/2
=1
PIJ:: 
4
j:.! (±lrdt = .,
2
1
(e)
.p _
g 
1..
2"..10
[2fr (.!....)~
2r.
tit
=!
3
2.11
ThE' integrals of the crossproduct terms (when k =1= ,.) are finite because the integrands are periodic signals (mnde lip of sinusoids). These terms. when divided by T  00. yield zerO. The remaining terms (k == I') yield
2.18 (a) Power of a sinusoid of amplitude (' is ('2/'J. (Eq. (2.6a)j regardless of its frequency (:..I =1= 0) and phase. Therpforc. in this r.ase P = (10)2/2 == 50. (b) Power of a sum of sinusoids is equal to t.he sum of the powers of the sinusoids (Eq. (2.6b)!. Therefore. in this case P = ~ + (l;)~ == 178.
(c) (10 + 2 sin 3t) cos lOt
= lOcos lOt ... sin 13t 
sin 3/. Hence from Eq. ('l.6b) P
(d) lOcos 5tcos lOt = 5(eos 5t + cos 15t. Hence from Eq. (2.6b) P = ~ + 25. (c) JOsin 5tcos lOt = 5(sin 15t  sin 5t. Hence from Eq. (2.6b) P = + (_g)2 = 25. (f) 1'.1<1' cos .....ot [1'.1(0+"'0)1 + ri(a""O)I]. Using the result in Prob. 2.17. we obtain P = 0/4) + (1/4)
=!
¥
¥ =
= (l~)2 + i
~ ~ == 51.
2.21 For a rcai
1'1
Pg
'" J . = I I!
Ell ==
lin T",,
= 112.
.. ,..,,(r
at 2
) dt
= J.., r.
<><
2eal
,It
= 00
r 2eal d t
TI'J
T
(ol}2 , I t I'
T/2
\. = T,,,, T 1m
1!
TI2
T12
= ex·
For imaginal)"
fl.
let
1'1
= .i:r. Then
1 Pg = IlimT_",
T
JT I 2(,.)%1)(,.1%1) dt == lim 1 fT!2 tit == 1 . T/2 T",. T . T/2
2
•
• j.
A
t+
s
t
0
~(.t.)
a
'1 ...
1
Clearly. if 0 is real. powcr 1.
2.31
(,,al
:!a
9:1(t)
,. .
V~"'30
.r::::
3&J
,
" 9·(\)
t ...
't . .
Fic. S2.32
is neithf!r energy not power signal. However. if a is imaginary, it is a power signal with
92(t) == .q(t  1) + 9;{t  1).
= g(t 
1)
+ 91(1 + 1).
.q~(t)
= .q(t 
0.5) + ql(t.;. 0.5)
ThE' !>ignal q!)(t) can be obtained by (i) delaying g(t) by 1 seC'ond (replace t with t  I J. (ii) then time·expanding by >I. faclor 2 (replan' , with t:'2). (iii) then multiply wit.h 1..'j. Thus 95(t) 1.5g(~  1).
=
2.32 All the signals are shown in Fig. S2.3·2. 2.33 All the signals are shown ill Fig. 52.3·3
FiC. S2.33
2.34
EIJ
= .f.~Iq(t)1211t =
1'.:
,l(t)lit
= Eg •
EI/'
E'('I)
= !:lg(_t)]2 dt ::
1:[q(ot)]2 tlt
1:
i(:r.}tiX
Eg(tT)
= l~Lq(f  T)j 2df = !:l(:r;)tlT ==
Eg(al)::
=;
1:
= Eg
= r.g/O
g2(:r)II:r:
2.41 Lsing the fact that q(.T)"'(.T)
(8)
(l
= g(O)/)(x).
WP.
have
(e) 2:.1:\/)(": + 3)
(f) k6(w) (use L' H6pita\'s rille)
'T
(b) i"(..:)
(c) !"'(t)
(d) ill(t  1)
2.42 In Thcse problems I'emembr.r that impulse "'(3:) is located at x = O. Thus. an impulse "(t  :) is located at and so on. (a) The impulse is locaTed at .. = t and g(T) at T = t is q(t). Therefore
= f.
3
(t . and equate it to zero.q(I):r(I) rlt =  111 ) 0 t dt = 0. we obtain (e) 1 (d) 0 (e) r!l (f) 5 (K) g(I) 2. the lengthsquare of g lenergy of gU)l is equal to the sum of the square of the lengths of ex and e [sum of the energies of "~r(t) and r(t)J.jtIJ(O) = j.Fo1 .2 E". and (. and 0 (. = fol .5.q(t) Therefore = 0 iii get).43 Lelting nt == . Also E~ (the energy of the error) is E" (1 . and the error ret) = t .. = f:sin227rldt = 0. (t he ellerg~' of I he elTor j are The error (t  0.51 Tri\'ial Take the derivative of le: 2 ".54 (a) In this case E". Also E!/ and E~ (the energy of the error) are + (l/lr) sin over (0 ~ t ~ 1). we show that this integral is !tIJ(O). 2.=~ EI/ 111 0 .59(1).53 In this case EI/ 0. 11 0 .. and the error r(1} :ret) .T(/).ith respect to (. Because of this ort.qlt) ~ 0. (h) _p' 2 we obtain (for (J > 0) ~ lP(t)li(at) dt J_.5) is orthogonal to .0. :r(I) ~ 1. and the error I'(t) = t interval.q(t)dl:::3 11 0 f(it=1.r(t). + E r . = ~ E".59(1) 1 .T) = R(t) Vp."" tIJ(!.r(/) hecause 11 (/ 2.1") 1/1" = .5 (b) Thus.5t)2 dt 1/4. = 1 •'10 E% 1 .5 o\'er (0 !S t !S 1).51 over (0 ~ 1 intelv:.t) = 1..5)(I)dt =0 ~ote t hnt EI/ = .q2(t) tit = ."" lP(t)"(nl) ~ tit 1 1 = j. 1/7' "(T)R(t .1'. 2. Similarly for (J 1!~ = . and zero outside this 4 .T) at 1" 1: 1: g(1") ...ing similar arguments.1.J. and 1 (. _"" tIJ(t)6(t} dt h(n.1._.5.p(0) a a 1 < 0. Therefore ! Therefore _.. and zero outside this inler\'ui Also EI/ and E.5 11 0 tsin 27rtdt 27rt = 1/7r (b) Thlls...2 tit = 1/:3.hogonality.)b(x)dx = a . To t'xplain these results in terms of vector cOllcepts we observe from Fig. and zero outside this 2. = = Jo = = 1).5 f ~ Thus.i J. 9(t) ::::: (1/lr):rU)..1.15 thaI the error \'ector e is orthogonal to the component (?t.(b) The impulse "'(1") is at 1" = 0 and R(t .q(t):r(f)dt 1 = 0 ."(1) \a 2..52 (a) [n this case E% = fo1 dt == 1.
40)J.r·(t)U(f) are zero [sP. + 1('21 2 E II . the two integrals of the ('ross products + u(t) is equal to t. I).e Eq.~ ['l:rl(f) .Eg = Jo.l(t) 2T2 (t)J!2. Il. The last rf'Sult follows from the fact.y(t) if (b) Csing similar argument. 2). g5 = (1 x 2) + (2 xl) = 0 g!l . 52. 2. that because of orthogonality.e can show that the ellergy of CIT(t) + (~2!1(t) is equal to that of "I X (I)  ('211(t) if 1'(t) and 1I(t) are orthogonal.hogonal.15 that the errOl vector e is orthogonal to the component ex.r(f)lI°(t) and . then we can show the energy of x(t) ± U(t) is E".q~({I'/f = !.T I I: !:qdt).6 = (0 x 3) \".0).r(t) because 11 1:.q(t}l is equal to the sum of the square of the lengths of (!x and e [sum of the energies of ex(t) and dt)j. gel· (g 1.nd The errOl It + (1/7r) sin E~ = f' [t Jo (1/7r) sin 27rt)2 df =~  2!2 21l'II is orthogonal to . 1./(t) dt ± I: + EI/' XO(t)1/(t) dt (1) 11I(t)12 dt (2) :r(f} . 14 = (2 xl) + (1 x 2) = 0 I' .2). K2( 1. K4(1. (1) in t. Because of this orthogonality.56 gl (2.:! Ex + Ee. we sec that pairs (g3. We can verify this also analytically. 1': In(t)Ilf. This energy is given by h1 2 E".56.hat of .(t)1 dt ± 2 IX(f)I' dt + f: 1: x(t). then it follows from Eq. the length of f [I'nergy of . IPg.(t)rit = 1:[:r 2(f)J!3X 1(t)Jdl = 0 T2(t)J[r. (e) If zit) 1·(t) ± lI(t).Q2(1)IlS(t) (It + 2X2(t)]dt (f) =0 == I:[Tl(t) + + :r2(t)ltlt = 0 . 1).IX(t) ± 1/(t)1 elt 2 sin 27rt!t + (l/1l') sin 27rf)dt =0 Note that Eg = (.he above derivation that = 2. = 1~ 1:r(t)1 2 dt + = f: 1: 1. S:U6 + (2 x 0) = 0 gl .'1' can show that the cOl'I'esponding signal pairs are also orthogonal.55 (a> If :ref) and U(t) are ort. (2. g~) al E' orr hogonal. Thus the energy of x(t) and !I(t) arE' orthogonal. 2.lo t g2(f) (It = f' f2 Jt = 1/3 a. g4) and (g2. Fl'om Fig. and ge(3. Ks(2. g:l(O..T(t) . 2). To explain these results in terms of vector concepts we observe from Fig.
051111't 3 r.~ "=1 1 ~ (4(_1)")2 = 9 + .198.L':=::.707 sin 211'tlif = 0 1 ( (4) J (0.:.48) for each of the 4 cases.. /). ~ Pg 1'(. 1).81 Here To = 2. and get) where "" = a·o + L "=1 11" cosn.5)(0.0 = 211'/2 = 11'. E~= 1Isin 2 211'tdt=0..81 shows q(f) 1 "".5 In the same way we find Ell! = E.+ 4 ~ (1)" (.. Therefol't' Nl Pr 1 8 = ./0 I( r Sin2 11'tsin411'tdt=0 (2) V(0.. ~ 1r I.. the correlation coefficients fol' four cases are found (1) 8l:i J (0.1~ t ~ = f2 for all f and the conesponding Fourier series l'epreS('nting q(t) O\'er (1."'" .e the energies of all the signals.5 O.=1 9 + 90 = 5 1 8 1 (b) If the JVtt'rm Fourier !'.In deriving these l'es'IItS.414/11' Signals 3'(t) and ..5) o..~) I [10. fl.707 sin 21f'tdt _ t 0 10..+ "'7 L !) "'~ 1 4' II = 0. fl'om Parseva!'s theorem [Eq. then I 4 = ." =~ . Let us first comput.c..11 1 t t 2 sinn1l'tdt::: 0 Tht'refole .. .198 n=1 6 . so that .90)J P g = Co + ~ T = 3 I 2 ~ C~ (1)2 + 2 L.2 ~ n2 "=1 ..48).1I" + bn sin n1l't an 2/1 =? I f 2 cosn1l'frif = 4( ~n1'" .:1 0. (2. t:sing Eq.~) .5. (2.707sin 211'tdt] = 1.T' "2 L 3 11' !VI 3'(1) (1)" ~ cosn1l't n 'l=) Tht' powel' Pz is required to be 99%Pg = 0.. ." using Eq.61 J::' :r~dt = L: x~(t)dt = 1 and L::rl (1)1:2(t) dt = 0 We shall compute ._ . 2.Q2U) provide the maximum protection against noise...5) 10I 0. ) " L> The power of g(t) is = 2 1/ 1 t 4 dt =_ 5 1 1 n4 = Moreover.. (2.. = .t'ries is denoted by :r:(I)..5)(0.5)(0.qlt) Figult' 52.&j 10 I tCsin21rIHSin211't)dt=1 (3) J (0.n = EII3 = Ell.. = 0.. we used the fact that 2.4 ~ 1 1 8 ~ .
 .) . for N = 5. n sinn. P~ = 0. . = 2. which is less than 0. ... P". 11').t where 0.5.. = = = = 2.q(t) cos n..82 The power of qft) is Pg :::  1 /..82 shows get) =t for all t and the corrp.19323..J P". is requirt'd to be O.0 1 = 2lf j".. then :r.9D x * " ..82 Here To fOI· N :: 2. for S = 2. Figure 52. bn = 2 211' jf<.. For N = 3.. so that Wo :: 2w /2w = 1. which is greater than 0. 1 2 o Ta/2 o To.For N = l. f< .371'2..  ~~.IV = 3.. P". P~ = 0.. which is greater than 0.r(t).(t) = 2( 1 ).1111: Thus . Therefore . \ L* sin N N fl71't = 0. I Ln. Thus..\' = 1. P~ 2. S2. and get) = 0.. 1". For N = 6. +1 .....'2 TON ..:otdt To Toi2 (lr. . f = T.9k25.... N 6. (2..Jot dt I.q(l.311'2.198.  .+1 The power P". Fig.19837.~. 2.0 + Lan cosnt + b .311'2.r2 = 3 ~loreoveL flOm Parse\.sinfl1.) dt (la) .:1 . 2 (t) dt 2r.90)J (b) If the StPlm FOllrifn series is denoted by .sponding Fourier series to represent get) over (w. = 2 jron To/2 (1b) g(/)sin fl. = 2w..2 = 0.  tilt = 0. tcosfltdt = 0.. On 2 = 211' j'" .S3 R('l'all that 110 Il n f = T.. 2.927. ~ !..4 = 2' L.. P z :: 0. _" t Sin TI t dt 2(I)n+1 = 'n Therefore q(t) = 2(1) . p".311'" n_l For.al's Iheorem [Eq. 2.. .
::1 2 " t 2 cos n.)ot dt (3b) Observe that. Sin t [cos 1'2 '3 '4 + 2 S1I1 t + 3 Sin t + 4 sm t + . 3.5 + .1 "'" IJo = 0 (by inspection) nn =~ fOI [1 ( 1 cos ('I727r f) df 1 C.) + lin sin (~f) = . then .q(t) is an odd function of t.. 4 =r..) n.:>  1(..84 (a) To = 4. .Jo = ¥o = ~.!.. Therefore 110 = n'n = 0 = To (3a) b. ano we allow Cn to tllke negative values. (~) sin (!!:t) 51r 5 '17 n=1 no =~ (by inspection) a" lin =. = en. l1(t) is . Figure S2.q(f) = no + 2: aft cos nt + lin sin 'I7t ".t df = O.cos2 3 2 1 51!'t + cos5 2 1 77rt . {" sin (!!:t) 101r tit I" = _if ~ sin (~) 7m 5 Jtr 5 tit =0 (integrand is an odd fUllction of t) Hele btl O... Figure S2.. ""'0::. Therefore (see hint) no = To Jo = 4 =0 To.. .. (t + ~) + ~ cos (21 + ~) + ~ cos (3t + i) + ' .... is an odd function of t and get) sin TlWot is an even function of f.o 2 (To/2 . trt 1 31rt cos..cos7 2 Here I.. the integration required to compute the coefficients need be performed over only half t he period. Note that Cn shows t he plot of (c) To = 27r. = /1 ...5 2 " 0 (by inspection) aft =11' 11 0 . . then g(t) cos nwof. lIn lTO/2 g(f)cos'l7... 2 cos f) Ilf] = n~ sin )~7r + .. 2.. ) 1 = 0...!otdt Similarly.~·4h ..Jot is an e\'en function and sin nwot is an odd function of t.. Jtr cos (!2. = ¥o i· g(t) = no + t n" cos (~. ". 41 o 0 TO/2 9(t) sin lI. 1.q(t) df (2a) (2b) (2r. J 8 ..8·4a shows tht' plot of C n . ) Therefote. all sine terms are zero.!. for n = 0. the FourieI series q{t) . all the sine terms are zero.. (b) To = JO:r . . and wt' al\ow C" to take negati\'e values.. 2.ot is an odd function of t..·0 = Bt'cause ofe\'en symmetry. 1. If get) is an even function of t. with 110 = 0..!. because of symmetry.Recall also that cos Tl.q(l) cos Tl""'o' is also an even function and g(t) sin n . =1!' 1 11 t sinntdt 21r 1 =1rn and g () t = 0. t ) {" 107r 5 =. if ." = O. g(t) = 110 + L 11" cos (n21r t) "".. 11' II. . Because of even symmetry.
J~rIIIIIl w .o.I .~ ct yo ...('os ( 4t . 'G"...Ieo c.84 • ..8·4c shows the plot of en and 8n..= . cos ( 2t .IO = 2 and . t! I ''~ (d ') • 2.. 2 (2 .sm t . . I':. /AJ.eos1r1l) sm 71'n 1rn 71'n 2 2 4'?t .+ .+ . 6 1. .~ t ~ .... • w. 4 4.' (..Sin t .11.+ . the Fourier series would contain de and sine terms only. fI "'" ~ .. sm _ . G.q ( f ) . ""'~ o...+ . 8 . . . . I lao t 5 (. qC'tJ • FiS.q(t) = ~t. flO :: 0 (by inspeC'tion)..5 (the dc component) is subtracted from n(t).. .. lin 0 (71 > 0) because of odd symmetry.. Il~ . = = lIn =1r 7r 41"14 tsm2ntdt = 0 4 71' . :. t V .Sin t + .... . Figt. (e) cJ .. Hence.cos ( 8t + .... the remaining function has odd symmetry. 4 1r) 1 10) 71') 1 7r) 7r~ 2 2 911'2 2 1r 2 7r 1.. (d) To 1r .J Sf JtI t\ () .... ThE' reason for vanishing of the cosines terms is that when 0....~ '1"~. . (b) " of .t ·"1...4 cos ( 61 + . I ~ I .e) I. ( ea.. 7r911'2 271' .. S2... \ r ". \ c.:re S:2. ..
D.. S2. 27T7I 27Tn 2'71"2... This is readily seen from the fact that replacing t by t in the first component yiE'lds thE' same function..!lo(t) and qo(l) = ! If/(tl  fI(t)! 1 = 2sgn(l) The even and odd components of the signal '1 (t) are shl)wn in Fig.g(t)) .q(t) = l1(t). . (I)J = g.3 3 21\" 2 71 2 3 211 =. 3 sm :l = 7T/3.85 An ('\"en fUIICI ion g.0 t . we have = .(t). =! [11cos~lit+ r(2t)t:OS~ttlf] 3 0 3 11 3 = ..t .TSlnT and fin = t.. (t) + 90(#) 10 .. 3 2 9111 lin tdf 2mr 3 = 3 ..sin 3.2 [ \ 2 + (f) To = 6 . it can be seen that the first component on the righthand side is an even function.. ) rt'~. =! .w~n odd From the definitions in Eq.(t) + . 3 3 9 25 T  qIt I ~ . to find the odd and the even components of .85a. .(t) = q.fI(f).C! + . The reason is that if t. Similarly.1 t) ann all odd (unct ion g. O... \\'e have . b. Even symmetry.5) is subtracted from . To find the odd and the (!ven components of .Figure S2.. Figurc S2..1 .+ .t 7r 6 ( 71" 1 49 cos  7~ 3 t + . (See Prob.q(t) = . 271"71 47T1..86).an 1 :0: 1lm cos l!.ltlfillg Oh~t'rve th?! even hllrmonics vanish.2 3 C "= 271"2. _ 7I"2n2 6 [mr .(I) have the property that ~.84d shows the plOl of C" and fin.q(t) ".cos rot + . "0 = 0..q.!! ( cos : r + fu. 2.q{I) eClt.cos27171"] cos T 3 1 2 5r.(I) and /lo{f) = go( f) (1) En'l ~ sigul11 .84f shows the plot of C". the function has halfwan' symmt'try.2 cos .10 9.0 t cos tlit == [cos 3..q( I) ran be expressed as a sum of cven and odd components bE'cause g(t) [a(t) .he tit: (0.!l .Ism 3. (1)... (e) To '"" 3.helo:' [from Eq.2 3 COS 2'71"" J 3 Therefore Co =i and 411"2.. while the second compont'nt is odd.5 (by inspection)...cos . "'0 = 2'71"/3.. ...q(f)J + ~ [g(t) . ~ .. . IlO =! 3. ' ..10 t t tit =! 6 IJ an 211 27' 71" 3 2'71"71 2Jrn 21r" = . The same manf'uver in the serond component yields the nE'gative of that component.sin ~ ) I 3 a 21!. 2ro71 cosT . 2..
9 TO/2 0 (:r. dt TO/2 + 0 g(t) COSll... Hence 11 .9o(t) nnd 2. ~ "2 0 and Let :r fI" 2 = r: o l 0 TO .:: To 2 [l [l TO/2 gU) cos 11...lot dt + l l TO/2 . .G. "':0 = 7' no = 0 (by inspection).Jot..lot dt 2 = 7f 0 l TO /2 g(t)C'os71.. + 7i0 ) cos 11.J d:r. ] 4 = To [(01 Jo 2 g(t) COSTli'.) elS t~ I 9 c (t) 0 t~ (~) c·s Fig.9(t)COS71. "jt :.:: .ft (b) (i) To:: 8. 52. S2..10t.. This gives tl n = ? .9.all/(t)  r..8li floft) = ~ [. TO) = g (.(t)] ThE' e.s c ~elt...8ab..(t) + .10 ('Ii)] 2 + 2 tl:r 0 X 0 .0 :.... t.9(xH ...al.(t) are ~ho\\'n in Fig. For fI(/) we have = ..".. Half wave symmetry.lotdt + lTO To/2 fI(t) cos '1wiOt dt 0 =t  To/2 in the second integral.86 (a) For half wa\'e symmetry nU) and ."PIl and odd l"omponents of the signal .Ot dt ] In a similar way we can show thal /'r=To 41 0 TO/2 g(f)sinn.cos 11wo:r...
9(f)= Half wave symmetry.1 (' ~t rtt J" IOn " 1" 12 j = .. a"COSTlt+iJ"sin1lt n" = To 21" 0 rtllOcosnttlt =~[ .· a" cos "'4 t + b" sm .5."" where .t~~ (O.1..s l 1I1r 2 odd) Th('refore an = ~ ('2" .lsinnt _ ncosTlt)] " 1 0 (T1 odd) = (fl2 + 0.cos2 4 T121r2 2 2 2 4 = sin (fl1r) fl21r2 2 (fl odd) and .t~: 71' + . 21r 1 1 ··/(". sin nr. 11'". = .0 = 21!.. (11") Sin Sin :2?1'fl 5 1!'n ... Also Do D ":: ."/10 _ I) = fl2 1. = DnrJ~t.q(t) = "=1...q(t) = L . nr..7... 1 ('.11." TJ2 1r 2 ( nr.01) (p and 2 "'/10 ./2)t I It  f :1 (' J(".. 1) = T1 2'+ 0. cos T +T 2 nr. n1r .9..:\.. (O.. D.:..1)] = 101r(T1~ + 0.. Hence "'" 2: "=1.1) T..~..91 (a): 10:: 4 .2J.) ..(flr.4 ( sin.0:: r.13 ../2...=... nr.JCOSflf+T1sinTlt)]" 1I'T1+. 11' '1" 0 " ('t/iOSiIlTlttlt = ~ [ .1 0 = ~ [ ~"~I~ (0.4 = T12r.'T1+.01) 2" «(. 2 In\ ~ 1 (b) To = 1071' . · { "2~2 (¥+1) n=3. O ..01 00465 II" =. · Similarly b"::  2 112 0 t ./10r. T (11 ) 1 (n odd) :: 4.15. no:: 0 (hy inspection).461Tl + om 2..1 n+. ":0 = 1.5 . :1 (11 odd) 1001 (0. ""4 t n1r (ii) To = 270.1) 11 = 1.1 = 0 (by inspection)..( .) = I ... 1/5 ...l ) ./2)t ( it :: 1 21 1 1 5 . . 111:' fl1r n1r 1171') smttlt:: ..
./4 11" 11"n 2  cos  11"Tl) 2 13 . + '1>" ) I (e) . t j 2 " . J 71" so that J IDnl:: 2 • 7rTl and LD" 7rn ={ iT Sin 7rTl Tl >0 n<O (d) To == 71". "0 = 2 and D" = 0 ... . where.i"t.91 (c) q(t) = Do + "'" L Dn". where Dn 1 = ..q(t) = L n=.# •• )t\ .... by inspection Do == 0. S2." I dt:: 2. (J» o •• T :L I dO'rr. .5 1 D" == 2 7r 1 2 0 .j"/~ To (~ J 4t _ ~"t dt ==  j (2 . 11"n ".. (f) Fig. D""/2nt...
.. = O. 52. 82.sm T ..0 = 7r/3 Do = 0.j) For a ("ompact trigonometric form.. . ..b.1 (f) To = 6 .e g(f) as .92a.92a shows amplitude and phase spect.1 St .n: ~ an $" j. WE:' rewrit. r...2eos (St . S2.) (b) By inspection of the trigonometric spect. 2+.j:lt = ~(Jt + arJ¥) r1r>1 + (ri¥) ri8t + ~('Jt + G. we obtain . By inspection of exponential spectra in Fig. where Dn =3 111 0 .lP". + 2 cos ( St _ 4. .) = 3 ("os f + cos ( St Figure 52.j¥) (.92 ._.~) + 2 cos (St .Jt ) + ~ [r1(St1f> + r'j(51~IJ + [r3(8t¥1 + e.5 g(t} . we plot the exponential spectra as shown in Fig...Pro..".q(t) = ~(rJt + r....S + '" L. ...... n> (Q) ~ . For this reason.(e) To = 3 ...sin~ ) 2'!n' :i"n:l + ....J(8t¥>} (. all terms must have cosine form and amplitudes must be positive.. q(f) = 2: DnrJ"Tt. ~ .t dt = 411'2112 [_.ra.q(t) = 3cost + sill (St  i) . ! ".s i n  47r 2 n:2 9 27rTl 3 47rn 3 27rfl 3 and LDn = tan 1 ( cos \"n r ~cosh!! ... D ft (' J.3 lit> ) Fig.2Jut 3 trJ. ~ . .!t (i27rTl + 1) r Jr 3 . .5 ~ .. 52.92 4 !1 .¥! 3 2 a~ (.) = 3 cos f + cos ( 5f  i . ) 2.J1f) 14 + (..q(.f . .92a.1 0 1 S 8 fl .2cos. :> • 5 C r"I~ ~ 2.i .ra in Fig.1 ] Therefore .92u.0 = 1.
Jotdt ] o To!2 To/2 If q~ t) is ('\'en..(3tt)] = 2 + 2 cos (2t "If) + cos (3t . In contrast. 'he se("ond term on the righthand side is zero because its integrand is an odd function of t. Hence. (c) By inspection of exponential spectra get) = 2 + [. 7( Fig.93 (n) get) = 2 + 2cos(2t =2 2 cos 2t 71") + cos(3t ..94 Dn=~ 1 [f T O/2 !(t)cosn. if q(t) is odd.j (2t")] + ¥ [~(.2') 11' + sin3t (b) The exponential spectra are shown in Fig..(2t. Hen("e..2. D" is real. 82..'ti) + (.93. D" is imaginary. S2. 15 ..i) (d) Observe that the two expressions (trigonometric and exponential Fourier series) are equivalent.93 2. the first term on the righthand side is zero because its iT'llegranrl is an odd function of f.Jottltj j T O/2 !(t)sinTl.. ) + r.
G(.. the integral (1) is real.13 1"'" Jw + 0 2 1 n..olt is an odd function of t.• For . Similar argument.(.q(l) == . Q(t)r iW 'dt= l:g(t)coswtdtj 1: g(t)sin .1·2 Since !C( ••:): is an even fUllction and 9.·t dt If 11(1) is an even function of I.1·1 C( . 3.. 3..t.tan .:! . lI(t)sin. and the Sf>cond integral vanishes..egral over the interval 0 to oc.. Moreover the integrand in the first integral is an even function of...alll(t).:... Therefore IG("")I = 1/v'.s can be used to prove the rest of the properties.q(l) is even C(w) = 21"".I (W)] d.. q(t) co!'!.oltdt (2) If q(l) is also real (in addition to being e\'en)..:... q(t)sin.(w)]d ..w a Therefore and 16 . Moreover from (1) C(... ) = l:.. Hence 1 "at = r..):: 21'"" g(t)cos...Chapter 3 3..) is an odd function of w..01' tanI(~)..t is an ('ven function of 1.. and the first integral is twice the int..) == .Q(t) coswt dt (1) Similar argument shows that when q(t) is odd G(w) = 2j 1.... and therefore .2 + 4 2 and 911 {"...:) Hellce C. the integrand in the second integral is an odd function of . Moreover.. and thf'refore vanishes....·) is real and even fUllction of.q(t) =To 11"'" 0 IG(w)jcos[wt+9..oItdt = C(. Thus when ..) == cos [...
. S3.S "hown in Fig.Itdt 2 = ... 2 Flp:.) Cd..t dt 'r = ~[cos.11 ? G( ....I) fl.)]. Therefore 1 g(#)=2'1f IGI(... 1" 0 tcos.. j2".Icosw'r +w1'sinw1' I} TW 3.' tit + iT 0 !...2) sin ..n even function...)+G~(.1"" I dt = 4 ..1·6 ('all ( + be expressed ~ .14 (a> (b) 3... .. Therefore 17 .16 (a) (""'~f2 ..15 (a) G( ..) = 14r J"" (b) 1 o l' tit + /2 2. JW G(W) = lT r _!.r. ) =.+ ('J dw !'in 2t sin t 1 1ft 3. \1\'0 o • (b) The oeri\'ation can be simplified by oLsen'ing tha. Also the integrand of the fi rst term is an even fUllct iOIl..Iof f' 2 .r· l' i . " 1 dw=2'1f {f2 ·..t G(.. t he second integral on tbe rightband side vanishes.) find C~!...} =.3.j. t' 2 .. .... S3.lot '1ft' + 216101 COS.  r...16.J a sum nf gatl! functions '.. 3.:....... t·) t /lw+ 2 fl .''''r + w1'sinw1' 'rW I] This result could also be derived by obsen'ing that g(t) is :l... Therefore from tbe result in Prob...11 <a) (b) Because G(w) is even fUllctioll..2 ..
5 j jw 5..J../0010 *7r wo rJ .~).\ ' rr= _.21 shows the plots of various functions. and a sine pulse (also centered at the origin) with zeros spaced at intervals of 571'.. This results in the sine pulse truncated beyond the inTerval ±. J C I t:.jwl dt _ f' 1 . that is at w 5....J... Therefore the function is a sinc pulse centered at ...1\ 1T""ot 3.. The fUllction in part (h) can be expressed as 6 ( .27rt j". ) .21 Figure S3./21 == :J~ JJ] 2jsin2 Jw .j4.cos.i(t  t.5.J .5 jlSw  _  4....v 7r 0 . = . S3.r j&._[. Thl! function in part. 18 ..S". The fUllction in part (e) is a gate function rect(i) delayed by 10. In other words it is a gate pulse centPred at t = 10 and of width 8..ot 0 7rt (' ~11"£. .ir·.2 [COS ""ot + wot Sin .J. "'0 .t (' 0 1 _ "'0 1 _ .. This is a triangle pulse centered at the ol"igin and of width 100/3. the first zero occurs at f 11'.... that is at w = 571'.1.0 0010 = sin wo(t to)to) = Wo. (L) (f) wFig..t(i) .l7r (If I ~ 57r) as shown in Fig. f.... j. = SlnC ("") (' ·". and its value over this interval is uuity. The funct..... tO) ..)"..5) is centered at t = 5. [ 1 [r .'1(1) = i. has n width of unity. For the sinc pulse sin('( 1}..27rt jW'I""O .ion in part (a) is a gate function centered at the OIigin and of width 2. The function in part (e) is a sine pulse sinc( f) delayed by lOr. 2..) .he origin and the first zero occurring at = lI'. S3... = 1071' and its zeros spaced at intervals of 57r as shown in the fig. (0 is a product of a gate pulse (centered at the origin) of width 10".(tto) dw t)] 0 1 = (2"')....t0r'. The function in part (d) is a sinc pulse centered at t...s".22 The function reet (t .JI IS'S 4. HencE' T = = GI .."..vot . t'" ~l'" Co (~) ~et (.. fWO 2r..0 .. sine 'If(t 7r (b) g(t) = 2~ [1~~ . 3..1./2 _ .l· .18 (a) ..21 SiT ~iT t 3.2Ie. l !UI (' . g () t 1 = .....('os t.t d. = .v)OoIO = 7rWO t2 0 1 . <HI 1 [CostiIJ + twsint.. [wo (t.(I .t 11.!.o) r j.1 + 1"'0 jci"'t d""] _ 1 .vo .
..0) g{t) Application of duality property yields .~ 'Trlh(.!h(w .j(woI+l)] 1 ..II)l <=> 2r..10)] .') Application of Eq. that is "(1) = h(t).t' that 1 + sgn(t) result..i' 1 = 2"'..2"9("') ." • = 271' cos (wo ... + .o...7r [h(t + .!.. . = 27r sin(:. ~ h(t  '..24 3.. g(··.. ) G(t) 2"11("') Setl ing "':0 = T yields I:(t + T) + h(t  T) ~ 2 cos Tw (e) 1/(1) sin ..Jol + 9) = ~ [p:1("'OI+f) + ...t) J27r.J G(. • = slne( r.. Adding pairs 7 and 12 in Table 3.i7r~"(t + . I . jlO1 3.. GC. G(I) * _ [hit) + J':rt ~J ~ 1/(. j"'Ot +2P.. .. ill )"'1)1 I' Fourier transform of the abo\'e equation yields the desired result.Io. g(l) 1/(1) ~ 7r....J) u(.""o}l.....31 (a) 1 + : ..ro.. ) Y G(I) 2 . and 1 Ih(t) 2' + LJ Trt ....t' that = 2u(t).. ~ }I ' 27r1t( .3..."....Jol .. ..."'ow) "'......P.J) ..sin(.) SCI ling "':0 =T yit'lds lO .....(w) j. ~ tl(. Obser\... + Gt+)"(..J ...) y' Application of duality proper'~' yields 7rMt) + • 01 .... (3...[2) sm 7. .J .23 :::: ..1101 .J) (b) cos .... l Application of duality property yields ..'':0) .. 3.....Jo) + h(t ¥ wo») ¢=> 211' COli (.vo)  h( ..vo' ~ jr.25 Obser\..t).28) yields ! 2 [t(t) __ J7r1 1_] ~ But "11 ~ is an e\'en function.. ) .1 and then dividing by 2 yields the desired cos (..
)2"..bet .. "2~) The third step of multiplying the resulting signal by 1.. w Fig.reet (tr (f) ~ Tsine (Wi) recl (f ±::?'/2) =.1) + .T/2 reet 20 . S3. .91(t) = .. Therefort' G 1 (w) = IG(.33 (a) g(t) "" reet. yields 12) 1( 2 <==> 2.9dt ..9(1) by a factor 2 (ii) then delaying it by 2 seconds.5 yields 1. 1 r' 1 = 2.2 (.. .I~' .5 [we may interchange the sequence for steps (i) and (ii)!. (iii) and multiplying if by \.) = get  ~) + ql!l + i).. (f) q. 2) ~ 4~2 (1 . (b) ..) + G1(w)}ri'" = !G(w) + G( w»)eJ"" = r(cosw + w sin.1).33 3.q(t) and bet + T) .) o Fig. and Fig. (c) q2(f) :"q(1 ..'"" 2.2 I] Fig..3. (e) .? .2.1 q~(I) = .(' ft .2.... (d) .i2w _ ...1('. (.I_we  )..T) ~ 2jsin Tw G 1 (w) = G( ·w) = W /rJ"" + )0.5f ( ' .94(/.{) p±J".Tsinc C. The first step (timeexpansion by a fact or 2) yields f(~) ~2G(2w)= 2~2(r)2"'j2wrj2"'1) SE'conu step of lime delay of 2 sees.~(t) can be obtained in three steps: (i) timeexpanding .. ( t + T/2) T12) r.32 Fig..)"' 1) = 9(t  1) + gtft + 1) Fig.2.:2'11  .Q1(t) 2('...
S3...I)r. S3.* .348..(t)_ (.:ro) ::= q(.jIolT /2) = 2J'T' (WT).34. sintll(t) <==* t[h(w .i)] = cos I lJ (t) ..34 From timeshifting propert.T)] = ""'!J(t) .1'0).. cost 1/(1) • Therefore ."'.q(l) is 8 gate pulse as shown in Fig. .11') 1_1"" !':ote that sin(t ..xo) = g(. + { :..q(I + T) + q(t ..q(l)::= + sin (I ..To)M1' ..2 = cos t [" (I)  " (f .". " '2 Also because q(..... Therefore (d) (t) !/(t .>T = 2C( ....1I')!J(t ... [h(w ..:h(.Il ) +h(~'+1 ») j.. G(..) = Tsinc (W{) [r....~ sin~ (W{) (b) From Fig..1 oo1T + G(.h(w J and + 1») + _1_ }(']"'" ..l ..I·)M:r . Now.T) = (III...l :':' I () .3& we verify that q I I) = _1_(1 + ..r)h(:r .11') Therefore {211'.iI(t} = sin t u(t} + sin(t ~ 1I'}u(t . J I} ).7 1 1 G(...' 1 .·J.cos t 1/ (I  i) But sinlt  i) = cost.. lw· Recall that g(...) ::: ..%) (I ~) 1/ 11' [ • G ( .IoIT/2 _ r.·+l) +1 .IW +a J .1) . and CI .1 +0 ..oT r.33b we verify that. 21 .1 + 1)] + sin(t .:r) is sin . . P3.... wT SlnC T SInT = . 53... +a 3. Therefore h(w ± 1)(1 + ('] .2 + ..6(.11'}"(' .'IoI)T) g(t) lIt ll = .roHI(:r  :ro).v Thcl'efOle ...•:)=._p./.... T Wc can use thib result to derive transforms of signals in Fig.... ) = O.I).1) ..T = :11 _ f!(o+...(t) delayed by 11'..!:.Jw'J) 1 _..and G(".I)~".oCtT)I/(f _ T) .•:) (c) From Fig. ) cos .".t(.T) e=:. (8) Here .
.. '2 {sine· [7I'(.J 2 10)] + sine .he amplitude spectrum in this case as Fig. ).J . and . (b) The signal.3).3) ~ 2j[2sinc(..) (Fig. q(t) = ~ (2~ ) cos lOt Also from Table 3..J) sin T. (8) The Signal R(t) in this case is a triangle pulse ~(f.r ) cos lOt <===* 11' ..J)sin3w) = 4j sinc(.. T) .. 53.36a.. The signal in Fig.I ~q( I }.q(t) here is the same as th~ signal ill Fig.34b..q(t . S3..:) sill 3.".. 2 [7I'(W 2 + 10)] } The Fourier transform in this case is a real function and bhown in Fig.:0) + G(.q(I. (4) ~ sinc 2 (~) ~ (~) cosaw Also T = 3. P3..3) ~ 4sinc(w)cos3w (b) Here 9(1) is a triangular pulse shown in Fig.q(t) = reet (~) <===* 2sinc(:.34b is . and . .: . 3. it follows that R(t) t = ~ ( 2.9(t + 3) + g(t  3) 2sinc 2 I FiB.)..3).34 Fl'equencYbhift ing propel I)' st ates that Therefore q(1) sin ~'ol :: 21 ...q(t + 3) +g(t .I) Therefore g(t + 3)  g(t ..q(t) = reet (~) <=* 2sine(w) Also T = 3. ..(2 .1 (pair 19) ~(f.35 is .q(' + 3) .. From time shifting property.T)] <=* G(.1 (pair 19) . P3.) = D.. [G(w ) .q(t .. ~'e need only t...36 Fig.35). The signal in Fig.3) where .0)] Timeshifting property sta!E'S that Th<>refOJe Ilnd '21 [q(t . P3.36) multiplied by cos lOt .q(t)1' 1. Therefore 22 .35 S3.q(1 ..q(t + 3) + ...I"'ol + . S3.34a is . (a) delayed by 271'.J :i.q(t + 3) + g(t .I ThE' signal in Fig.0 1] = 21. From the Table 3.) <===* 7I'sinc 2 (¥) From the modulation property (3. its Fourier transform is Ihe same as in part (a) multiplied by ..
.. and then tht' result is multiplied by cos lOt. o ~) ... o . (b) is expressed as . 2 SInC 2 [11'(". 1'\ott': In the above solution.pectrum . could have interchangecJ the operation in this particular case..j !!.. we obtain G(".+ reet ("'+4) ~ rect .1)) . .t. .  2 Thr FOllrif'1' transform in this case is the sam'e as that in part (a) multiplied by .) instead of a triangle pulse ~(2'~)..arT'~· '.. .. .&11. ~. (h) is ('xprefst'd as ~(tz~" ) cos 10(t ..'Sinc(t) Therefore • (~) g(t) = ~sin('(t)cos4t 1f (b) Also 23 . the amplitude spectrum in this c(lSe as shown in Fi~ S3. '\ =  {.. + smc 2 [11'(..211).) by cos lOt and then delayed the result b. the signal in Fig.... This multiplying fllctor "f>p"f'~ents a linear phllSe spectlllm 2l1'''.iII'# .3·6b. ( "') ~ ..... .36h.."" . In this alternate procedure...::G( ••:) = 21r": 11. This inl t'rchange of operation is permissible here only because the sinusoid cos lOt executes integral number of cwles in the interval 2r. b.. .u + 10)) + sincl7i(w G(". .". . ( I i~" ) cos 1Ot .J2"..10)] 2 ..:3. the triangle pulse ~(2~) is first delayed lw '2r.) = rect w4) 2(2..D (11 ~(~ = ... S3.. Fig.. if C't:oLk en .. that is..... This nlf'ans the signal in Fig. Because of this both the expressions are equivalent since cos 10(t ..'> shown in Fig....' + 10)]} " )2".. " < .... (e) In this case t he signal is identical to that in Fig..\· 211.. Thus we have an amplitude spectrum [same as in part (a)i as well as a linl'ar phase !.211') = cos lOt.) = 1I'{sinc[lr(... exC'ept that the basic pulse is rect (21. 1: G( •. t:sing the same argument as for part (b). we first multiplied the triangle pulse ~(f... :'\ow "'e rect (2~) ~ 211 sinc( lI'w) 10)]}r. } t .
ThE'refolc get) = !sinc2 (t) eos4t 11" 3...sin 2 (WT) T 3.ot wit. Therefort! G{ .) is twice the width of G(". G(".J[) t±T rect ( .J 0) and the unwanted signal ..310a.38 Flom the frequency convolution property. The two spectra are nonoyerJapping because ! = 24 .Jo.) .:ot] = g(t) + g(t) cos 2.J:2 (""T) t+T/2) .T  2 + (!jwT = :2[1  cos ""T] = 4sin 2 (~) 2 . Repeatp..i T (! J'.. we obtain The width propel"ty of convolution states that if (...d application of this argument shows that the bandwidth of g"{t) is nB Hz (n times the bandwidth of g(t». Hence.J 'r . the width of G(w) . C2(X) = y(x)..T])4 sm 2 =. ..Jof :.Joj+G(".Jot yields g(f) cos ••:ot x 2 cos .h 2cos..q(f) ::: reet ( T.JT) ~ ± J.J+2.: 2gft) co.:ot ThE' product contains the desired g(f) (whose speet rum is centered at..39 <a> G() (b) "" = j T o "JWCd t ....q(f) cos 2""ot with sptctmm IG(..q(t)[1 + cos 2..310 A basic demodulator is shown in Fig. which is centered at ±2.) and ('2(..JCd t 0 2 = . The product of the modulated signal q(f)cos.Jo!.. :: h{f " )wG(w) + T) .Jof = .'2 2"" (c) df.'7')..•. ) )4 = :.. 3..2h{t) + h{t  T) The Fourier transform of this equation yields = (!j".":' [1 JW cos ~.2 ..p= find /?) <==> Tsinc (..r..rect (tT/2) T rert (f ) ~ Tsinc (.....1 (x) . S3.2.. then the width of !I(x) is equal to the slim of the widths of ('J(:r.
". respectively. 5 kHz. 53. G'l(":).. = = 3.:) and H2(":)' Now Yd .'0 (See Fig... Because the bandwidths of Y1{w) and Y2(W) are 10 kHz. 53.41 Ir < . . Hence the filter is unrealizable.. the frequency convolution property yields }'(. . it follows that the bandwidth of 1'(. = 1'' ' ~dw = + ~ .:) is the sum of bandwidths of }'!(. We can suppress the unwanted signal by passing the procluct through a lowpass filter as shown in Fig 53. we get This is noncausal.12 .i.41.:) }'! (. Also 1"" ~ IlnIH(w)11 .... the bandwidth of Y(...:2 + 1 I/..!l•.:) and Y2(W).4j shows GI(":). tb _ ~ .)2 I 00 25 .{1rr Fig. From the width property of convolution..51 Csing pair 22 (Table 3.. 3...310a.:) are also shown in Fig. Because 1/(t) 1I1(t)1/2(t). Hd...ttii) to .·) = sinc(doo) and G::(w) =1 Figure S3.41 Gd..:) * Y2(W}... 83.1) and timeshifting property..·) Y2(W) = Gdw)HI(w) = G2(w)H2(W) The spectra YI(":) and Y:z(.310b).:) is 15 kHz.
:2000 cent.. Hence if"'. Io'igure S3..98 X • 10.16 11 = .736.: 0.er is non causal and therefore unrealizable..===> W2 6 = 142.>onential delays to 1.53 FlCm the results in Example 3.erp.d at W 10 5 represents t. 11 = 0.1 = 10 RC 1\ Also HlO) = 1.54 There is a typo in this example.: 4/a:.i2 is t he frequency where the time delay drops to 0.. "(0) = r. we get X = 2 5 w2 I/(f) = ('lO~IIlol The impulse response is noncal1sal. Since h(t) is a Gaussian function delayed by to. At the edges of the band a a = 10' 26 .)2.95.1% of it. then 2106 w2 + 10 ' '1 = 0.684 = ! = 10. Hence to :. .:.. where both the specifications are satisfied..95 =* .he frequency range from 0."1 Henr. that iR w = 142.4 ..I)I = 106 v..l and n frequency of 22. and the filter is unrealizable. approximately realizable 3. then IH("'. From Example 3.. Let IH (.:2=+=I1:.16 = = X 105 .l is the frequency where thp. ) i = ~v'i=:... The time dela..01 us consider the gains and the time delays at the band edges.hlf) Figure S3...52 H(w) 10 ... Choosing to = 3v'2k.8o/t at 4 times constants. 3.98% of its value at td(W2) = = O.2 ===> trl(O) If J.: 4 x 1O~ == 40/<s is a reasonable choice to makE' this filte. it looks as shown in the adjacent figure.857.s peak value.857 This yields We select the smaller of"'. 3m 3.r. .e the filt.)f + 10 12 :..6 a W Moreover.011 or 1. h Lt"" ."2 The eXj.99 x 105 to 1. Hence to = is a reasonable choice to make the filter approximately realizable. Table 3.1 and timeshifting property. the time delay is given by (see Example 3. The band of A .:.16) td(W) = _ + a2 a .y tolel'ance should be 4% instead of 1%.. 1 = 3:28...j"'lo + 10 10 From pair 3. amplitude response drops to 0.4 Hz..
Hence.+ Fig.7 bandlimileJ 10 B Hz..T) ..in.g{t .99%. the transfer fuuction of = ao + all'j .~t + a21·i2.""~t = 1 _ Qr.. that is at. al = Il.... we can show that (see Plob...) SillwlT .. 83...·to+k.. 1 +:r when.35) PJto jkG( ••:)r.n"'T) :::::G(.IB)llj/rSin. .)]desircd = 1 + r.....1.. jf g(t) is the input.T.I = 10 5 .. we require the equalizer to have [Jleq("'..61 3.... t o t . P3.nless. we find the time delays at the band edges as and The time delay variation over the band is 4%.... Figure 53. 3. ~t + . Similarly.61 n~') = G(. G(. + (1"rj....:) = G{..) (' 4r.:::!. ) hecause G(..1 Hence 1/(1) =g(t  to) + ~Ig(t .1 .The gain variation over the band is only 1.Trj. Hence.. Moreover.62 Recall that the transfer function of all ideal time delay of T seconds is the equalizer in Fig...62 approximates thili expression if we select ao = 1../)r:\loreo\. The C'qualizel in Fig. We also find IH(105):::.: 0 2 .~t Ideally.B J( ....el.. a:! :. an = 27 . Hence «  1.)rert (~) = G( .)rect ( .q(f + T) .. The signal is transmitted with a gain and time delay at the center of the hand.·~t + Q2 r (1)" C\ ". + ..471' This follows from the farl that r'" :::.) rect (..to .j".61 shows . " ' .('··.I 2w AI _ (1t:l('j:l"'~1 + . .01 and fd(l05)::::: 1_1...g(t  to + T)] (... the t'ransmission may be considered uistortic.. !. the corresponding output iii t..to d. 3..lTJr..T)] ~ G(...) j~ n .... 0. P3.62 is Heq(w) Hence.q(1) and 1/(1).
_"2.92(t) = " = _I_ reet 28 (~) r~ • 471"B and 1 . we can show that 3..I..1) = ('.f"" r".!!.• =~ ~~ and consequently dw Ell = Ldx a.".. 3... 2 2u V2 _..!!!..717r). 3..'.: 211f 1: ('.2/ 2 2 Eg = 2~ 1:.. then Gdw) = sine (211"Bt .73. = _1_ 211'.!!.._1_ 271" 11v'2 f"" ou (... G(w) dx = 2V2lfl1 ..q2(t) in the above development.~ = .. 21f11v2 = _1_ '2. and is 0 otherwise.q2(t)dt = ~ _". if we identify gl(t) sine (211"81 .74 In the generalized Parsevars theorem in Prob...·1 $ 271" B._"2. G2(w}=reet ( .73 Rerall that Tilerefore Interchanging the roles of gl(t) and ...w ) r'nr 2B 411'B Therefore But rect (~) = 1 for 1. IG(w)1 dw .. Hence 28 .3.2 d.~) :1. 271"17 ~d:r 2/2 i_ (OU 2 ('...12 Consider a signal q(l) = sinc(kt) and G(w):= ~reet (.._t /tr dt 2 = ~ and consequently tit = E" = ~.j7f..'"11') and .11 EI/ Letting ~ = j'"" ..!. Also from pair 22 (Table 3..2/'ld:r: = y'2. Letting rr .
2o". ". then the outpul Y(.)H(w).1l. This yields 1/(t) = 2A(O)~/ ~ex'... (3.7·6 Ifq2(1) <==> 04(.(t) = 2E.. S3.1") [Eq.76). 271" [1 o a IV t"y = 0.) A(..f..q(O)1.01 ==> 2.O.::"/.. 3... Becanse this filrer band 11/ ....366 = ..Hz (/ 3..99E". because 9 2 (1) o¢::::.=1' .23a)J.::::> A (. we have Hence .).. then .>(.he integral.'. we used the fact that r±j2trlc = 1 when k is an integer.).r ad/s= a. = Hl're we used the property 9(:1')/)(1") = .2a'" = 0.7·5 Application of dualit. r jT/2 7(/)1/(t +"T)dt lim T/2 29 .  rv whel'e 'Rry(T) 1 = T.2aW J J.y property [Eq.) is the lowpass filter transfer function (Fig.n=m In evaluating t. Thus..1) yields The signal energy is given by Th(' energy contained within the band (0 to W ) is E IV= 471" If EI\' l w I' . ..24)J to pair 3 (Table 3.3025 0. where H(.. we may express it as an impulse function of area 4r. (1..
We can express Rz(1') 85 a sum of two components.I "' C" cos(n.. the integral on the righthand side is zero.1(1') and R2(T).. ... (1') + RII2(1') C = t COSW11' + 2 + C2 COSW21' This result can be extended to a sum of any number of sinusoids as long as the frequency of each sinusoid is distinct. J T 2 / 3'(t)T(t1')dt T/2 = 4(4 . Thus. (2.hen R II1II2 (1')= 1 lim T T.." jT 2C C I 1 T/2 2 COS(Wlt+8 1)C05(W2 f +".82d. ft cos(n. S3.82d.0) + b(. pulses again overlap.o)] :\.. as shown ill Fig. but (lnly with half the magnitude. as shown ill Fig.(1') == Rl(T) + R2(1').) .'" Csing Eq. Rg(1') =R 9.. + n.82c. . according to Eq.ot + 8."1//2 pulses in the waveform of duration T. hence. S3. D . we can generalize the result as follows.) is the Fourier transform of the periodic triangle function.82a shows the wa\'eforms T(t) and :r(f . Let T = NT".=1 ..•1) = i "=...ll'b) 30 . there are .... (3.If we let 9\(1) = Cl COS(". S3. ..80). if get) =L C . =\ .:'" and 51/(".. t.ot + 9. 1 R. But on the average. we have RII291 (1') O... R.82 riglll"P 53. and R gOgl (1') = 1 lim T T"" jT 2CoC I T/2 I C05(W\t + WI l' + 9d dt =0 C T COSTlW01' 2 Thus. The area under the product :r(t):r(t . only half pulses overlap. Therefore = = O. Hence 1t :s :s :s :s Sz("') T" .. we find D" = fgsinc'("l"). t sine 2 ("211') . If q(f) = Co + L "". S3..41) 52( .21'+82)dt According to t he argument used in Example 2."...) then Rg(1') = "" C5 + L .""1  11.. The PSD is the sum of the Fourier transforms o( 1\'..~~) = For 11'1 T".(.2b... 4 = 16 SIDC 2 (wT") + 52("... there is no overlap between pulses....1') (or l' < Tb/2....1') is N/2 times 1') as shown in Fig. We find the exponential Fourier series (or this periodic signal to be R2(T) = 2: "=r"'JIL.82b.. and Rz(1') O.lt + 8 1) and 92(t) = C2COS(W2t + 82). shown in Fig.'here 52("..) = 27rC5b( __ ) + i L C~rb(. ) then Moreover. Hence. On the average.. For T" 11'1 ~.... RfIO(T) = C~. r i ""'. Hence. Therefore (If.. Rl/JlI2 (T) t: sing t he same argument.r(')= lim T T. for qo(t) = Co. Rz(1') repeats every T" se~onds.. Hence.
93..(. Therefore S"... :L.82 Tz... 16 'I "1 ~ 1.I 2(n1l') II ( T w nwb ) and 31 ... '1: ..2.~/~ (c J 't'.  ~r.. .r.) = 1b sine 2 (WTh) +"8 16 411' "" ~ sine L"=I'"I1I... 1[11 0211 ~ ~'1i :I T. 'r'~ ~l 1j... I " Fig.
2 +1 _ 1 ... +1 dw 1 =11 11 0 WJ2 w2 +1 dw 1 = .+ 1 0 WJ2 1 uWJ =4 1 h(WJ .7r 1"" 0 2 211" 3..1) 1 "_ .06831 7r 4 :..T 2(1)=_ 7r 11 0 7r 1 dw=7r and ~ • y2 (t) =7r 111 ...)= ( 1) + ..JW + 1 jWJ and " .....1 ... ~ =1 11 (I) 11 1"" 0 rect (:J)...J'" 1 . .84 The ideal difi'el'entiator transfer function is j.2.. the transfer function of the entire system is H("." (jw)=. 32 . = 1 and ~() '1'=..To) .( 1 ..2 . 0. 7r 11"'" 0 h{WJ 1).... Hence..... .l)d.....I ' 1 (c) :7 2 (t) =  7r 11"'"' 0 h(.(b) """ 1 ..
SC(t) = m(t)coslO.11 tt Fig.) LSB V. 1 2~os 8000t + cos 6OMt) + 2I.B 1 (ii) FOI "d') = :l cos 1000. OOOt 1 = ("os 9000f + cos 11. t t 33 . LSB L'SB .D~MC (t) cos 10. OOOt 1 1 = '2 [c058000f + cos 12.. OOOf] I I = icos90001 +.s8000tJ +.21 shows various spectra.<m \ iD t t t tt t «. ooot =: [2 cos 1000t + cos 2000f\ cos 10.. + cos 14.4).OOOt =: 2~OOO$ + .21 (i) For".21 T t. (iii\ F"l ". 2 . 0009 ~08 ~B ~8 This information is summarized in n table below..000. Figure 54. Icos 12. OOOf 1 = :i[cos 2000t + cos 40001) cos 10.05 11. OOOf + cos 14.000t] + 2!C056000. .'/) = cos 1000t cos3()OOt "cnsB'c (t) = m (I) cos 10. (t) =: ("os 1('001 'POSB. ".2 co.000t + . 12.000t) [cos 11.0001 + 2[cos 8000t + cos 12..OOOt == cos lOOOtcos 10. 5 0ntlt $~e~trt. S4. + cos 2000t =: //I (f) ":.Chapter 4 4. ModuJ4ted 4.000tj == I. .
rlleo .100 I T (a) (b) & ((!. 54.000 12. iCl~ ' D . too IC.. 54.000 I I 9000 9000 8000 8000 6000 I I I 1 I 11.Jct] .000. 4.blLnd frequency 1000 1000 2000 2000 4000 DSB frequency 9000 9000 8000 8000 6000 and and and and and I LSB frequency I USB frequency I ii iii 11..000 11 .. (\I ) (b) ~ {ei ) wlllt) ~\ (c~) Iaeo . S4.22 The relevant plots are sho"'n ill Fig. .000 I 14.23 The relevant plots ale shown in Fig.22 TTf'~ FI.24 (a) Thp..CDO I ..000 14..000 11.1 12.2·2.000 I I I _I I 4.. 84. signal at voint h is ga(f) = m(t)cos:4"'.000 12. ~ .23..23 case i Basp.c t = 711 (t) [i COSWc t + ~ 34 C06 3.000 I 12. Fig. ) Mew) . 4.
The remaining term i".. Hence. When the carrier A cos".. desired with the output im(t) cos wet.) + 2m(t)cos2wc t The signal at point b consists of the baseband signal !m(t) and a modulated signal tm(t) cos2wct. (c) The minimum usable value of ". not t. t..J X 10:l). Hence. (d) 2 ". Hence. the filter output is . whose center center frequency is We..ystem works for a carrier cosft ""ct only when n..s the modulated signal with canier frequency 3w c with spectrum centered at ±3. lAJ .) 5w ct et® cO. (b) Figure 54... the lI.24.hown in Fig. The bandpass filter centered at ±we allows to pass t.(t) = 100 kHz (~c = 20071' The output 1Jj(t) is found in Eq.he diodes act as a gate in the circuit that is b!lSically a voltage divider with a gain 2R/(R + 7'). which represent. (t) l'Os"..istors R.!m(t)cos 3w et 3 The output bandpass filt.. and =i 7f [". output bandpass filter centered at "....:>. .The t. 2 (t) 1fI(t)COS wet = [1 +cos2:.ct is pOliiti\'e. system works as . The output is therefore "o(t) :::: Ul(t)m(/) R+/' 2R The period of Idt) is To = 2r.Jetl 1 1 =2m(f.1r COli. Qt@ L... S4... (E') The readel may verify that the idrntity for cosn ../.75) 35 . t..3~ U.(t) cos wet . Thus. Hence. Fig.24 4.6 with the carrier frequency ic thp.11(t) 4 = 3 m(t) . Therefore.24 shows the sper.'c as shown in Fig."f. thill. and when the carrit'r is negativ<> the diodes are Op<>11 (during the entire negative half cycle). is odd.. This i~ not true when 71 is even. I Q + ~m(t) cos .ib) as + . the diodes condu(" (during the entire positivE' half cycle).:.. ~ I DI I W ~. ohms while conducting. from Eq.26 Tht' resistance of each diode is .c.he desired value iJJ c • Both the components will be suppressed by the filter. which has a carrier frequency 2we . bllt sllpprl!llfies the unwanted term t". (4. Ie = 300 kHz. and 00 when off.:r05" ench of tht' rel'.(t)coslwet is the unwanted term. 84.. (2. whose spectrum is centered at ±w~. Hence. during the positive half cyde. this system will not do the desired job. 3"'6 • WtOe. ~ .ct.er suppresst's all the terms except the one centered at 300 kHz (corresponding to thE' carrier ~·cl)..c is 21r B in order to avoid spectral folding at dc.tra at points band c.he voltage ~¢(t) appears a. the out. 4.he desired term ~m(t)cos. During the negative half cycle. 3~ct This is the desired output Jr".. 4..put voltBge is zero.erm ~1T/(t)COS":et is the desired modulated signal. with Jr = 4/31r.(t)cos3wct.25 WE' use the ring modulator lI.·et contains a term coswet when n is odd.
_( 'f ) 4. CDs 2.rum inverted.26.er\'e tlla! Y(.29 shQWS the output signal sVf'ctrum Y(o. I. cos Wet .'. 4. P4.28 shows thE...3 cos 3wJct + 5' cos 5.j. it is clear thaI I. 36 . 54.!. Hence. (b) Ob:.28 shows the rel'eiver 1. n+r' 2 11' (cos wet  . the high frequenciE'1..27 From the results in Prob. (b) The same circuit can be used as a demodulator if we use a basepass filter at the output" In This case. the output ~o(t) ".. 54.'. leaving only the modulated term ~fn(t)COswet intact.~ )J ..1)000 to 3.lI'(t) The output I'ott) is I'o(t) 1 1 = '1 + . as a modulator.. (c) Fig. 4.29 inverts the frequency spectrum.. Hen('e. where Ir = ~.Jet + fl).lIt) and m2lt) from the received modulated signal.000 rad/s (flom .c t + .') with the frequency spect.~..(t) = sin (. (b) From the spectrum at point c. the system actl. = /rsin(:""et + 8)coswcf = ~[sinR + sin(2.".:).28 (A) Fig. f f 1+'>. Thus. In the present case.0 rp. 4.""_2. that is.ct + R)] The 10wpCUls filter suppresses The sinusoid of frequency 2wc and transmits only the dc term ~ sinB. + ~ +..he channel bandwidth must be at least 30.he scrambler in Fig.{~) t I.] 2 2[ = n2R I/'(t)m(t) = ~m(t) [..29 (a) S4..."t.:) is the same as All.covar . and c. the input is tP(t) = '11(1) COS wet and the output is ~TTI(t). the output is 1'0(1) = Irm(t)cos . cos lwet + ~ cos s.. b. m.!.' signals at points a.OOO rad/s. 3 ..wct + . are shifted to lower frequencies and \'ic'e v~r5a.0DODt .. (8) If we pas~ the output 1'0(1) through a bandband filter (centered at We). the filter suppresses the signal "t(t) and II/(t)cos n"':et fol' alln "" 1.
and the output bandpnss filter is centered at 400 kHz. Tht? output !'..31q. whose spectrum is centered at ±2.. (t) ..32 To gel back the ori~inal spectrllm M(.3: m(t)cos (8001T)lO't 4.).)IO't. 4...fA.) once again. (4.210 \\'e use the ring modulator shown in Fig.us filter allows this telln to pass.(t) = = A + 111 (t) When this signal is passed through a dc block. hut suppresses the second term. 37 . ' .. The desired output is !I(t) = [1//11) cos(2r.)106 ]lI'o(t) = 2".! cos 3\400?T)lO"t + ~ CO"" 5(4007:')10" 3 OJ = .""c ::: (400r./11(1) cos (SOOr. m(t)] cos ""<t.(. This is a synchronous or coherent demodulation. This show!. qb(t) = IA + met)] cos2 wet = ~[. that the system can demodulate AM signal regardless of the value of A.0' '5 .7b) as I'. All the other terms arc sl1ppre58ed.)10 1 tj.')It . 54..1 'II' J ' The product of the Terms (1/3)C05 3(4001l')lO't and (4/II')m(t)cos(211')106t yiclds the desired term .(1) Through the same scrambler. except.)106 t [(:os (400?T)lO't .(t) cos(2r.1ct The first term is a lowpass signal because iTS spectrum is centered at ~ O. WI' need to invert the spectrum Y(.bsing I he scrambled signal . whose' spectrum is centered at 400 kHz. This can he dOlle by p!l.e caniE'r frequency is 200 kHz [. .)10b t instead of m(t).. o '0 I" o~~~rH~~~~~~~ ~'''' . the de term A is suppressed yielding the output TIl (.(t) is found in Eq..:). Fla. Hence the output of the luwpass filter io.6. The lowp. 4. Hence. .(t) =!A. TI.Ic .. It alone passes through the bandpass filter (centered at 400 kHz). that the input is /1I(t)cos(2r.4 + m(t)! + klA + m(t)Jcos2:....
S4.8. This means that I' = 00 represents the DSB·SC case.33 i r To/4 TO ".125.33).. t~ '.. (b) The sideband power is '" 2(1 )/2./II The carrier powe)' is tv'(t) . 10/0. using the previous argument.. S4.66~ 70 4..3·2 shows various waveforms."" :::>.. the power of may be complJted by averaging the signal energy over a quarter cycle only.".0 /II" =10 .67 p.4.(1) COS. (t) The sideband power is [40']2 dt:::. Over a quarter cycle mit) can be representE'd as mit) 40t/To (see Fig..5= I' = 1. Hence..2 A == A => A =a :::.=0.(I) = ~ = To/4 io 1 ".. Fiy'..4 T. S4. .. p. S4.. ".let with m(.Jet is Im(t)l.r cyclE'. '... if A + ".m(t) > 0 for all t. Because of symmE'tl'y of amplitude values e\'ery quartp. the condition !A rut'):.. the carrier amplitude is . 78.34 From Fig. = fo\' demodulating A~I signal using envelope detector is ..33 <a) According to Eq.4 + TrI(t) IA + m(t)j. 4..34 38 .(1) > 0 for all t. Figure S4..32 (a) (b) (c) (d) . P. then .lOa). 2 /2:::. 33.8 :::. + p. Therefore. The efficiency is '#:::.0= I' . it is clear that its envelope is IA + Now... The signal + r" (t)l cos . 12.. . = m:{f) :::. = = in. 10 . Ho!nce.3·4 it is deal' that the envelope of the signal m (t) cos .c! is identical to ".) replaced by A + m(t)..4. 16..4 Pc . (4...67 )( 100 =1.4=5 = OC rn = ::.4 A A A =A =A 10 10 10 ~ ~ A=20 A =  11Ip = 10 I' = 2. 16.. Fig.125+ It:.67 9 = 78.34 To .
P )( 100 = A2j') x 100 = 0.lc t. we call del1lodulat~ the Ar. unless m(tl ? 0 for all t. Clearly. we can generate DSBSC using AM generators if we use two identical A~f generators in a balanced scheme shown in Fig. S4.31 4. 4. ".35 Whcn an input t. Hencp. the corresponding output.1 + m(t)J 2 cos2 "'e t = .4 + 1II(t).ltput is met).38 'fh'" signal Pot point a is [. the time average of ffl2(t) is also A2.P.36 \\'hen an input to a DSBSC demodulator is met) cos . 84. the corresponding output is Im(t)1 [the p. Only when m(l) is neal' its peak. if the input is .. 39 .> The carrier amplitude is A power efficiency is = "'1'/1' == tn.. 4.35 to cancel out the carrier component.. when an input to an A:\1 demodulator is m (1) Cos "'et.l signal using II DSB·SC demodulator. shown in Fig. + "..c.6 ~ I v " \J Fi.. Hence. = A. me'). == Po: +.nvelope of TII(t)1.3·1 ObM'I"!' thaI ".T(t) = \..4 z for all t..5 A2 The signal at point b i~ 2 The A:\I Signal for I' = 1 if. 1/ Hence Pc == A 2 /2. the corresponding output wi\1 be IA + m(t») COS.4 to the baseband signal we can generate AM signal using a DSB·SC generator. Thus.(I)/A « 1 for most of the time. is not true. the corresponding output is m(l) coswct.(t)Jcos wet.·c t ) The lowpa..Jet.o A DSB·Sc generator is "'(t). The total power is P. unfolt unately.4 . by adding A de of value . Thc converse is generally not true. the corresponding 01. However. n \I _.·ct.37. will A + met). the Signal at point...2(1) = . it is not possible to dernodulatt' DSB·SC signal using an A~1 demodulator. S4. if the input is [A 4. By blocking the dc component A from this output. Clearly..4m(t) + m 2 (t} (1 + cos 2.s filter suppresses t.4 2 • The = p. c is l'su:dly. The converse. (t)] COs"'e'. Hence. Ht'nce..hf' term containing cos 2. = . This is precisely the AM signal. Thu.o.. the output at point d i. This is becamle.mLt) 4. this condition is violated.4~ + 2.
.."I..14.. J...5eos 1600t :..){sin + Aw)t + .S8(t) cos900t and \Pl:S8(t) cos 1100t. )t + b] or sin (A .t..... it is convenient to multiply m(t) with 2C05.) A2 A blocking capacitor will suppress the dc term A 2 /2... (t) (t) .e) 1.. and yields the DSB·SC spectrum M (w .·)t + h] TIl ~ (t) 4... )t + bj + Tn2(t)COS !(A". Hence... c + A . L~'S (1) = (cos lOOt + 2 cos 3OOt) cos 10001  + 2 sin 300t) sin 1000t = cos 1100t + 2 cos 1300/ Hence... we can express <?'LS8 (t) (c)From Fig. band c show the three C'ilSes..·e and below . )t + hI} . et cos [(... to obtain the USB spectrum. )t + Ii).51 To genernte a DSB·SC signal from met)...... the outputs are = m dt) cos [(A. = = (b )From Fig." + A... From the signal wit)..... we see that the distortion component is 7// 2 (1 )/2... e + Aw)t + h) =2m1 (t) Similally cos . + "'2(t) sin . e and. However.5 cos 400t + 0. 4. = <0. (a)From Fig..)t + h] 't' 1'J cos {(2.T2(tl = '" 1 (t){ sin [(2. ifLSB (t) = cos lOOt ('081000# + sin lOOt sin 1000t = cos(lOoo 'Pl"SB l00)t :: cos900t (b) UI = cos lOOt cos 1000t ..51 <l.. We suppre55 the \. a.sin lOOt sin lO00t := cos(1000 + l00)t := cos 1100t (t) ::..e) from the DSBSC spectrum.... + 0. we can express hss(t) = 2 cos 700t + cos 900t and (t) == cos 1100t + 2 cos 1300' = ~!cos4oot + cos600lJ and IPns(l) = ~!cos 1400/ .5 cos 1400t +.ctlcos [(.5 cos 600t) cos 10001  40 .. (1:) wit) h~s (t) = cos 1001 cos 500t = 0.. 4..... cos lOOt + 2 cos lOOt and m" it) = sin lOOt + 2 sin 3OOt..5 cos 6001 (0.5sin400t + 0..Tdt) = 2Iml(t) cos.I. to generate the SSB Signals of thE:' same relative magnitude.5C05600t and Tn..c + A . cos 1600( 'Pl"58 4.5 cos 600t) cos 1000t + (0.5 sin 400.0 obtain the LSB spectrum. Figures 54 .l~B(t) = (0.~D = (cos lOOt + 2 cos 300t) cos 1000t + (sin lOOt + 2 sin 300t) sin 1000t = cos 900t + 2 cos 7001 (sin lOOt ..Ii] {(2 ..ft..5 cos 4001 + 0.0. we multiply met) with cos"..:SB spectrum (above .... we suppress the LSB spectrum (bl'twecn .41 In Fig.....5 sin 6OOt) sin 1oo0t = 0.. )t + b] + cos [(2wc + Aw)t + b)} + Tn2{t. c..·)t + hI} After tt) and :T2(t) are passed through lowpass filter. Hence... when the carrier is cos HA.5 cos 4001 + 0..(t) = 0. Similarly.5 sin 400t + 0... This also avoids the nuisance of the fractions 1/2.5sin600t) sin 10001:= 0.1/(t) :::... T + Am(/.)t hI + sin [(Aw)t + hI} + rn2(t){ cos [(A.5 cos 4001 + 0. h.1112(1) sin HAw)t + II] "'2(1) = 1111(1) sin [(A . e + Aw)t + h] + 277t2(t) sin Wet cos [(w" = "'1 (t){ COS [(A.)t + 6J .5sin600t. )t + bJ  sin [(A.We) + At ("" + ""e).. we can express . yielding the output Am(t). we have .52 and (a) fII(t) = cos 100' and 1IIh(t) = sin lOOt.
'O(l ~~ ! 2. S4.0 "CD LV'" Lt! ) Fig.51 41 .... '~D ~ "'0 0 'tW . ( h) ". 11.D. II '" 13t<. t.31< II K.I.Jr 1 1tit 1. o l4) ...
.11\ cos ft) OlIcrrt (hi Figure S4..53d shows the corresponding t..(t) and Fig.j .f"sa U ) = lOOOsinc(10001l't)cos 900011" 'PlSB(t) = lOOOsinc (lOOO1l't) cos 1l. To show that the energies of lII(t) and 11lh(t) an! pqual. t.1) Figurp S4.54 Because MIt(w) = jU(.53b shows the corresponding DSBSC spectrum '2".) If w(' apply". 2.J) = jsgn(.0oo1l't 4. ~ Dl~eooTr C'rl0l)T7 ltc}) (b) Fig.2.53 '.)3c shows the conesponding LSB spectrum obtained by suppressing the eSB spectrum.:) sgn (w). S4.. \\'p now find Thp in\'erse Fouri!'r nansforms of the LSB and USB spectra from Table 3. 42 .. h (t) at the input of the Hilbert.SB spectrum obtained by suppressing the LSB sppctrum.1 (pail' 18) and the fl'I'(I'H!nry shifting propcl't~· as .53 o r Icoooji 1 .. we have . Y (w).53a shows lhe specTrum of ". S4. the transfer function of a Hilbert transformer is H(. (c) Figure S4. the spectrum of the output signal 1I(t) is This !>hows that the Hilbert transform of "'h(t) is m(t)..ransformel'..tCDif o <a) n Fig. 84.55 4.
20.J = O. Hence.U) sin .' as the spectrum M (. but the \'oice is st ill intelligi ble. if m.. (b) When only ~.J)t This is an l'SB signal corrE'sponding to a carrier frequency ~.56 We showed in prob. the USB output is y(t) = 1/1 (I) cos .' = 111 (t) cos(~w)t + h) 1'0(1) Tn . (4. This changes the sound of an audio signal slightly. the filter output r. the output is given by = met) as expected.(t) [instead of l1/(t)] is applied at the input in Fig.'I]... because the human ear is somewhat insensitive to phase distortion.(t)sin(6. and passes only the difference frequency component centered at the frequency £:.J) COS. ). 4.. Such distortion mlly change the quality of speech. ('d(t) = 'PLSS(t) cos [(We + ~w)t + . (4.I) and It follows that the amplitude spectrum of l'o(t) is M( . Fo!' voice Signals... then the output is given by 1'0(/) = m(t·)cos(6w)t . If only" = O."".ortion generally is not a serious problem with voice signals..:c t + I) h 1fI TTl 711 h 43 .. = (t) cos (weI + I) + (t) sin (.J)sinh Bul from ECj..w.'! [(We given by + ~w)t + .. the same as that for met).~~ The incoming SSB signal at the receiver is given by [Eq. 4.Jet.. ) with each frequency component shifted by a frequency ~"".) is shifted in amollnt .lestricl I he shift to ±2 Hz. the frequency shift within ±20 Hz is considered tolerable... )....t .'I] = 2[1II(t) cosw"t + 11'1" (t) sinwct] cos [(We + ~w)t + II] The lowpass filter suppresses the sum flequency component centered at the frequency (20."  ... howevel'..U(. howe'er.17b)] 'fLSD(t) = m(t)coswct + m.J4b) IIh( ••:) = jsgn (.4.. phase distortion may be intolerable. Most rs systems.~h(. S4. 4. where each frequency component of ..55b. the lowpass filter output is We now show that this is a phase distortion. The Fourier transform of Ihis equation yields Eo(. In video signals and data I ransmission.54 that the Hilbert transform of rTlh(t) is m(t)_ Hence... But the phase of each component is shifted by 1'1....1< + ~w).' that if both ~..o(t) is given by roU) Obser\....(t) sin(6w)t + h) and 1'1 are zero.(t)sinwet The prodeuct of the incoming signal and the local carrier is I'd(t).I)M(w) = jM(.. Let the local carrier be co.J as shown in Fig.) phll~e by = M (...tJ) { M(. Phase dist.1. This spectrum is the saml.
When thl? receiver setting is 10MHz. 4.c) and H. but the difference component. When the receiver is tuned to 590 kHz. (4. tl b (1. = 2545 kHz and 1500 .( .61 From Eq. 4.20.61a shows H... ~ 4 :. the image frequency is 2 x 455 = 910 kHz apart.. as follows..I ~ ~i I 4 . 1500 . "'hose carrier frequencies are the sum and difference frc"<!uencies.low this frequency fLo is multiplied with t.61b shows the rE'cipl'ocal. and the station is receh·ed. it will beat (mix) with /Lo = 18 MHz to produce two signals centered at 26 + 18 44 MHz and at 26 . . ho 10 + 8 = 18 :\fHz. which is centered at 8 MHz. which arr. Thel'"fore. +w. which is Ho{ . The sum carrier is suppressed.:2. :. the USB output is an LSB !. 4. ". :L. = l /' 4 .. (4" ~J(~ . = = = . 82 The local oscillator generates frPquencies in the I'ange 1+8=9 MHz to 30+8=38 MHz.:)..he incoming signal of frequency Ie = 15(10 kHz.tation will also be heard if the receiver is tuned to frequency 1500910 = 590 kHz. Hence.. Figure S4... In the present case. we can show that if we apply 711/.(t) at the input of the Fig. . S4.' Fi•..(1) at the input of the Fig. passes through the IF filter.: + we) T H.~ . := .18 8 MHz..ignal corrf'sponding to '. e ~ 'Ho(ciJ ."ow. . The l'easCin for this is. The two frequencies are 2/1F Hz apart. 1 H. y 44 . 4.e). thE' :.ignal corresponding to m(t) (with a carrier phase shiftE'd by 1r/2)... the local oscillator frequency is fLo 590+455 = 1045 kHz.81 A station can be heard at its allocated frequency 1500 kHz as well as at its image frequency.:) = . The sum component is suppressed by the IF filtel'..~ .20...61 Thus.: .(I)..20) Ho( .(.104.{w  We) Figul'e S4.. ) ). f k. if there is a station at 18 + 8 = 26 :\lHz.. )..( . /IF = 455 kHz. The calrier also acquires a phase shift 1r /2... . () # '7 ~ 4 .....1045 = 455 kHz.'. if we apply m. but the difference carrier passes through. the LSB output would be an USB t. The output yit'lds the two modulated sigllal:.. Similarly.
99 MHz. '.9 ~fHz to 10. Hence.3).11b shows m(t). where r. If. Hence. (f. For the next a seconds.)mm = 10 .8 ~IHz.2 ~1Hz o~'er the (rising) quarter cycle.)m/Ax = 10 + IG~ = 10.11 ~ r v f1 ~ v J~ 5. f· .lHz. = lOS Hz . S5. whell 1/1 (t) = 1. Therefore.. wh('n 10(/) 1. because m(l) has jump discontinuities.000::. This cycles r(>peat~ periodically with the period 41. 1 and (/.)min = lOS . the carrier frequency remains at 9. and the carrier frequency remains at 10 MHz. For the next (/ seconds.11.1 seconds as shown in Fig. the carrier frequency decreases linearly from 10. 55.lHz. and the carriel' frequency remains at 10 MHz.10" = 0./2rr = 104 Hz. :: lOS + 104 1. )( ROOO/2'1l' 2 x lOs Hz.11 In this case For FM : Ie = 10 ~. the carrier frequency falls to 0. :: 50r. For convenience.99 MHz to 1. the phase discontinuity is ~pmd r. Over the next a seconds.2 MHz. Also Ie 107 . But at the points of discontinuities.1·1 a. The cycle repeats periodically with period 10:'1 as shown in Fig.2 X 1O!> 9. Over the last quarter cycle 1. Also Ie = 107 . 7 ~f = A'fll/p/2":r = 21T )( 10~/2r. /rip = 1 and m~ = 8000..9 :'. = = = = = 7 = = = 5. 85. and over the last quarter cycle.)max 10 + 2 x lOS 10./(t) = 0 again.Chapter 5 mtt) C1 " v \. Henee.: cos [2rr(10)6 f + loo01rt] := cos [27f (106 + 500) t] At the diSC'ontinuity. This cycles repeats periodically with the period 4n seconds as shown in Fig. we shall use a direct approach. 1111' = 1 and T1I~ = 2000. Then instantaneously. and (f. we select the origin for flirt) as shown ill Fig. the amount of jump is Trld = 2.1 MHz to 9. (j. The carrier frequency increases linearly from 9.12.01 MHz. The carrier frequency rises linearly from 0.10!> 9. there is a = 45 . Over the interval ~ to we can express the message signal as 711(1) = 20001.t 10.1 MHz.)mlTl:= 10 7 . Hence. the carrier frequency remains at 9.1 MHz. thE' carrier frequency is constant throughout at 106 + 500 Hz.p~ (I) = cos [2rr(1o)6 t + ~T1I (f)] = cos (2rr(lCr)lI t + i2oo0t] :. Hence. Over the next quarter (the falling) cyclc of duration n. r .0] MHz over the next cycle.1 MHz over a quarter (rising) cycle of duration a seconds.8 :'. Th(t) O..01 MHz over the cycle (over the inrerval ~ < f < 10. For FM: /2rr 20. For PM: ~f p = I:f1ll = = Here. Fig. 55.9 MHz.9 MHz. S5. and (j..12a.l(t) 8000. the carrier frequency remains a. where m(t) = 8000. (Ji)mu.12 III this case Ie = 1 MHz. For PM: = = ~f kpm~/2r. We conclude that the frequency remains at 10.99 MHz and starts rising linearly to 10.1Hz. Also Ie = 1 MHz. Figure 85.
' o.t 5. and Bpt.000 + 1000) = 42 kHz.I = 1.000lr and Af = 20 kHz and BEM = 2(A! + B) = 2(20.f + B) = 37. the baseband signal bandwidth B 200011'/211' = = Also.w and ".: 38. = 2(ll.£~1 (I) = 10 COS(Wet + 0. II 3t rr. = 200lr and A! = 100 Hz and BEM ::: 2(Af + B) == 2(100 + 1000) = 2. In this case.831 kHz and BpM = 2(ll. the baseband signal bandwidth B = 2oool!'/2l!' = 1000 Hz.2·3 o. Also t....831 kHz.1 sin 2oo0r.2:'· It.06366 kH:!. 5. OOOlr = 40...t).83 + 1000) .031. / ' m(Q)dQ] == 10 cos [10. . = = = = = 5.! = I = 15.22 :.ooOt + k"m(t)] 10 cos (13.n. For FM: ~f = Ic~:p 2OCJ~.'VUYlf\ nill n ' v v VV (TG ta Fill' 9&. the phase discontinuity will be higher than 2lr giving rise to ambig\lity in demodulation. SS.II(t) 5 cos(Wet + 20 sin 1000lrt + 10 sin2oo01l't). A .662 kHz. " ' phILlIe discontinuity of 11' radians 8& IIhown in Fig.(t) kHz. For mIt) = 2 cos lOOt + 18 cos2ooor. cos 1000lrt + 20.et ..(a)tlQ = 1000/ ' m(a)tJQ = 3000t = /' m(o) do =. and Bnl == 2(ll.000".36.. Also. Here.t and 711(t) Therefore TIl" == :.12 .00011' cos 2000lrl Therefore.he baseband signal bandwidth B '" 2000".13 (a) We are given that (b) <fp~dt) CPPM(t) = = A cos [Wet + k"TT/(t)} == 10 cos{lO.! = k/I11.21 In this case Ir/ = 1000'11' and kp = I.l(t) = A cos + k.25 The baseband signal handwidth B ~ Ii x 1000 5000 Hz. Here. A"" = 20.000. mIt) = 3t over the intervalltl ::./ + B) = 2(20.~ == 36.. .000l!' sin 2000r.(1) = We + 20011' cos 2()()(hrt = 1000 Hz.! 2(A! + B) 2(18. = 2(2 + 5) = 14 kHz.f + B) = 66..000 + ~ Hz. For k" > 11'.000".0001 + kf / ' TII(a)da] Therefore Hence k. For FM: ll.I2b. For FM: : ll.+20. ' 5. = We + 20.!OOO = 31.. Clearly. [. Therefore. './2l!' 10.1II(t) = 3 == 200 sin lOOt ..x = = For PM: ~! k m' = ~ = 211. we must maintain k" < '" because there is a discontinuity of the amount 2.000f) with k" = 1000..2 kHz.951 kHz and BF'II.. + 200.000".00011000) =t2 w.! k"rtI~/2'" == 18.£:. For PM: : ll.2·4 The baseband signal bandwidth B 3 x 1000 3000 Hz. 1. ~. /2'" 1 kHz. 5.)t1 = 1 kHz and Bnl 2(Af + B) = = = 46 .
31 47 .432) = 6.432 kHz and BpM :::: 2(Af) !:ll = k.22c and d.858 = 2(3.178 l'ad/s=0.12 that IfJFM (t) is essentially a sequence of sinusoidal pulses of width T = 10:1 seconds and of frequency Ie = 1 MHz.?.26 (a) For FM: Therefore AI = k~:p = 200.).~O. respectively. (I) roughly doubles the bandwidth of both FM and PM."'e = 200'WO"XI = 100 kHz . The !Spectrum of .. indicating that P~1 spE'ctrum is sensiti\'e to the shape of the baseband sprrtrum. S5. Hence. 3.1 = 2(AI + B) = 2(20 + 2) = 44 kHz. l 2/4 Also . ~~ I~' 5 MH. Bf'M = 2(AI + B) = 20~ kHz = = For PM: AI = Ie.. and = loxI?" " " Bn.bandwidth of P:\1. we set the derivative of r. which deca~'s rapidly. F"1 sptctrum is relatively insensitive to the nature of the spectrum Af(. For FM: ~I = kf'e = 200.! 2(AI + B) = 2(20 + 1) (c) m (t) = sin 4000".U(..858. and B = 20oo7f/27f = 1 kHz..""".. (b) ".. as seen from Fig...) by 8 factor 2] has hardly any effect on the FM band .22d. The bandwidth of the pulse.For PM: To find BPM.. m' 000 = ~ = IO"lr" = 20 kHz and Bp). 3.'. Also m" = 1 and m~ = 400071'. Such a pulse and its spectrum are depicted in Figs. Fi.QpotrX2 = 200 kHz.u ~ IO"teoo" = Bnl = 2(~1 + B) = 2(100 + 2) = 204 kHz For PM: AI I.A/(..:'p = SOOO. we observe from Fig. and == 3..~. 5. and . S5. IT radls or 21T Hz.J. and B == 4000". is 4". This yields and m~ == .2/4 is a Gaussian puisf'. Also rn" 2 and m~ = 4oo01T.il( ~) = 0.l k = Tm' = 2(AI + B) = 2(200 + 1) = 402 kHz = = = 42 kHz..t.. This spectrum also deca:. For FM: ~I = Ie'.21 From pair 22(Tnble 3.~ = 10 kHz and BpM = 2(AI + B) = 2(10 + 1) 22 kHz. Doubling the frequency of /II(t) [expanding the spectrum .1 12...(/} = 2 sin 2oo07ft.. { :100 ~~.(t) = _2/. For PM: AI = 20 kHz and Bpt.. This is an extremely small bandwidth compared 10 f:j.s rapidly away from the origin.31 The h)ork diagram of the design is shown in Fig.. (d) Doubling the amplitude of ".11. it roughly doubles tht.T.'(t) is /1/'(. = _2tr / 2 t>quaJ to %ero.. 1 = 100 kHz and the baseband signal bandwidth B = 2~" = 1 kHz. However. For FM: For PM: ~I = To find ".187 Hz. BpI\! = 2 kHz.) = j.. /27f 2 kHz.31. 5...!....at oX lJt5 Af= 10Mb Af =':2S0 H. 1a we obtain r~ y'i r".. idth.864 kHz. 5. Its 3 dB bandwidth is 1.I(t) lr:" = ~:'1 == 3 kHz and Bnl:::: 2AI =t22 x 3 = 6 kHz...2/ 1.) = jy'i". and its bandwidth can also be assumed to be negligible compared to !:lJ.'('wJ . S5.. The spectrum M(w) == y'i 1'_".
lim fI~(t) = . t he incoming signal can he tracked with zero phase error. If H(. Hence.. d.·!I + AK~~'"+ a. when passed through an ideal intt'grator. + AI\.(.pnl (t) is passed through an ideal PM demodulator...0 11ge ( .0 . H(s) e.(. if H()" .4·3 From Eq.. 00 at the point(s) of discontinuity. Hence.~e.)do.~) r = ..• 2 +n!.(.41 (a) IPPM(t) = A cos ["'et + kl'm(t)] When this 'l'PtI(f) is p&SSl'd through an ideal FM demodulator. 55.30). t he output is k f Tn (0) do.) = .. (5 33)J is lim lim 2\K) = 1. followed by an ideal differentiator.(') = lim ..• for fI.(/. 8. Now.+b . acts as FM demodulatl.. 5.32. 5. yields kpm(t). e... the Lapla('e transform of the phase error 8. ) = " (. FM demodulator rollowed by an ideal integrator acts as a PM demodulator..32 5.~ I In this ('ase.. . (b) <PYM(t) = A cos [wcf+klj\·71(o.(t) is given by e..4~( lim II + a) = ~ Ako lienee. However.~ + Hence. and • The steadyst ate phase error [Eq.. met) :::. and e.. c. 5. the incoming signal cannot be tracked.0.( . 48 . and the system will f:\il.ee(") = . The filure is self explanatory.~}= :)0 . lim 8.+a then " and 1. F i l·8&. (5. the output is kpT..) . When this signal is passed through an ideal differentiator..rt) = kt 2 .42 Figure S5._"". the incoming signal can be tracked within a constant phase 2k/Aka radians.he output is klm(t}.42 shows the waveforms at point. if met) has a discontinuity. t.• ) =~./(t) This signal.~2 then Ilnd .)T regardless of whether mit) has jump discominuities or not. + I) = 0 lim lim . PM demodulator.] \\'h"n this signal .32 Tht' block diagram of the design is shown in Fig. ) = _08 2 +.s b.1) .
1'1 '/1' J r Fig. S5.42 49 .
\OJ. (e) sinc(50". = Co "" To 1 8 /1/1 6 1/16S lit 1 = .') <=> O. It is an even function of t.qt{\OJ}.01t:.(~) is 60 Hz. the width of the convoluted signal is the sum of the widths of the Signals convolved. The period is To = 1/2B. The Nyquist rate is 200 Hz (samples/sec)..2 ~ Il7r 4 n=l sin (!!) g(t) cos )').f) + sinc (507l't) <=> O.he Nyquist sampling rates for 111 (t) is 200 kHz and for I1'Ct) is 300 kHz.: 21r(:lB) 411' B.310). The Nyquist ra. /1/1 68 1/II. .fdt = 2 mr sin (~). sinc 2 (1001l'f) ~ 0.q(t) + £.t "1 "" Csing Eqs. (272).OIrect (~) The bandwidth of this signal is 100 (h) 1\' rad/s or SO Hz..12 sinc100".01 (~) + Q.(2:W") The bandwidth ofrect(1fo. we obtain no Hence. 4.. 4 an ? = en =.te is 150 Hz. Hence. From width property of the convolution.(t) = = Co+ LCnCOS)'). and from the width property of convolution the bandwidth of 1112(t) is twice the bandwidth of . the bandwidth of sinc(S01l'f)sinc(lOO".he pulse train can be expressed as PT. ) sinc(lOO"'t) <=> 0.ql(f) and that of 112:\(t) is three times the bandwidth of 112(t) (se also Prob. The spectrum of the product of two signals is 1/21r times the convolution of their spectra. Similarly the bandwidth of 111 (t)R2(t) is the !um of the bandwidth of Rl (t) and R2(t). 6. Clearly the bandwidth of the compositE' signal is the higher of the two.) is 50 Hz and that of C. The :Oyquist sampling rate is 120 Ih.(~) The band\\'idth of this signal is 20011' radls Clr 100 Hz.02 rect( 163.l·l The bandwidths of Rl(t) and 112(t) are 100 kHz and 150 kHz. fol' 112:\(t) is 900 kHz.) •. (c) sine (lOOr. Also 1112 (t) <=> '.' 4 ::'0 .Rl("'. (d) Sinc(IOO:rt) + 3sinc 2 (601l't) ~ 0. that is. respectively.02rect (~) The bandwidth of the first term on the righthand side is 50 Hz and the second term is 25 Hz.13 The pulse train is a periodic signal with fundamental frequency 2B Hz.t) <=> 0.S To cosnw. that is. The !\yquist rate is 200 Hz !sarnplesjsec).OIrect 0.t) is 25 + 50 = 75 Hz. Therefore the Nyquist rate fol' RI2(1) is 400 kHz. 60 Hz.!. (t) = . (a) 0... the Fourier series for t. Therefore.!. 100 Hz. The bandwidth of the sum is the higher of the t\\'o.ql(t)112(f) is 500 kHz.01 rect(~) + 10 t:.Chapter 6 G. for . 4 1i( t) = g(t )PT. Hence. Therefore t.) The t\\'o signals have bandwidths 25 Hz and SO Hz respectively.:.0).01 rect(m. The Nyquist rate is 100 Hz (samples/sec).
1) (e)  5 S "(Hzl(d) / 0.0..). and 20 Hz (oversampling).9(1) <0 (e) I"'" 401( 10K 10K 20 5 Practical Filter 5 20 . S6.14a).. g(t) cannot be reconstructed from its samples get) in Fig.4 0...05 _tG(w) ~(E. t t I I \\ 0 9.14d.e must sample the signal at a rate no less than 10 samples/so The Nyquist interval is T == 1/2B 0.. and the spectrum tC( .. The successive spectra overlap.14 6.14c. Hz. Hz (undel'sampling).. 1.) == sinc 2 (S".) is not reco\'elable from G( ..). sampling interval T 0. and the spectrum 0( . .') (Fig.14b).. that is. S6. Consequently.2 G<<al) (b) 1 <al ...2~(~) (Fig. ) repeats e\'ery S Hz (lOr. as shown in Fig. 51 . v. = Y = sampling frcquenr}' J.1 0.. the spectrum is 0(101) = O. 56.. the output bpectrllm is reel (.0. 10 Hz (Nyquist rate).') = ~(~) repeating periodically with a period equal to the sampling frequency f... the sampling rate il!l 5 Hz (5 samples/sec.) 2~ 4~ 5 Hz 10 Hz 20 Hz I Undersampling I I I Nyquist Rute Oversampling I comments I ~) ~) I I In the first rase (undersampling)... (Hz) 9(') (g)  "4 (h) FiK. that is.1 second. We present this information in the following Table for tfiree sampling rates: f. S6. the Nyquist rate is 10 Hz.. Recall that the sampled signal spectrum consists of (1/T)G(". ).14 For g(. rad/sec.2 0.2 I \ \ \ \ \ 1«>It 20 lOK . The bandwidth I){ this signal is !..). I ! f \ 5 S 20 "(Hz) . If thl? sampled signal is passed through an ideal lowpass filtt!1' of bandwidth I) Hz. Hz (1071' rad/s). S6.
43) l1(t):: get) * [2B'sinc(21rB't)j = = 28'I1(t). S6. Hence.l4a. The height of the triangle is equal to t hI:' sample value..16a. This pulse pet) is applied to the input of the second identical filter... 56.211(t . The output of the slimmer of the second filter i5p(t) .!.2T)..17a to be delayed by T sees.T)u(t ..~ is positive for all iJJ. where we found the bandwidths of 1/1 (t).. 56..T Hz.. 6. 6. 6. I . The impulse response of the circuit is a rectangulal pulse.') Thp impulse response "(f) is shown in Fig. The output II(t) of the integrator is the area under p(t) .. "(f)...£s) required fol' signal reconstruction is shown in Fig. S6.p(t 1') fI(t) .14. Such a df!lay would cause the reconstructed Signal in Fig. in the last case of oversampling (sampling rate 20 Hz). 1. 5 kHz...*). In the second case. . then as shown in Prob.enerates a rectangular pulse at the output.T) + . 56. the Nyquist rates for the three signals are 20 kHz.16 (a) When the input to thill filter is h(t).'1 ..q( ••:. G(w) can be recovered from G(w) using an ideal lowpass filtpl' of bandwidth 5 Hz (Fig.r{h(tH = .T) shown in Fig.cktoback. 56. T Figure S6.T) + (t . The corresponding amplitude response H(:. which is applied to the integrator. respectively. . 56. The resulting signnl consists of straight line !legmt!nts joining the sample tops. and 15 kHz.(t) to be 10 1'(.18 As.:) fOI 0 for Iwl > 21f D.211(1' .(t . Each sample gpnprates a triangle of width 2T and centel'ed at the sampling instant. The spectrum G(w) consists of b. llIinc(27rB't) . G(w) can be recovered from G(w) using an ideal lowpass filter or even a practical lowpass filter (shown dotted in Fig. (b) The transfer function of this circuit is: H(w):: TfillC( ". met). the output of the summer is "(t) . (d) When the input to the first filter is "(t)..Iia shows the signal reconstruction from its samples using the firstorder hold circuit.·I) shown in Fig. as (8) Figure S6. S6.2T)'l(t . Observe that the filter is a lowpass filter of bandwidth 2rr/T radls or l. The output is 20 sinc 2(51ft).I4h).I6b. And. S6. Hence. which is not the desired ~nal sinc 2 (51rt). Fig.and the output signal is 10Ilinc(201rt).j.l7c. kHz...). Therefore B' > B.ume n signalll(f) that is fiimuitaneously timelimited and bandlhnited.~. Finally.T). S6. it also represents the amplitude response.16c. This acts as the input to the integratol'.r Because H(". and.. (c) A minimum of T sees delay is required to make h(tl causal (realizable)..15 This scheme is analyzed fully in Problem 3..·s the impulse response h(t) = A( .) :: .41. When a sampled signal is applied at the input.1·7 (2~ T sinc 1I(.rect<'l:u') = 11( . 56.bet .c. Let g(iJJ) .2T)Jtlr = tu(t)  2(t .) I The amplitudp response of the filter is shown in Fig. respectively.2T) = A C~.14fL The output is 10sinc 2 (5r.Jib. we use the !'Iyquist sampling rate of 10 Hz (Fig..1·6 4T e 6. its output is a rectangular pulse l'U) = 'let) 1I(t .)!:: T lSinc( w.·) (repeating every 20 Hz) with empty band between successh'e cycles (Fig.T). proportional to the corresponding sample value.T)I d1' = vet)  fI(t  T) :: rect (t . each sample f. (b) The transfer fllnction of this circuit is: {A )}:: 3 (w{) = h(l) = 1'(11(1') .. Hence the output IS a staircase approximation of the input as shown in Fig.17b sholl.p(t .14e).t). the spectrum G(w) consists of nonoverlapping repetitions of tG( . which.14h).T) + u(r .J) and the ideal amplitude response (lOWPll. nono\'erlapping repetitions of tG(w) repe3ting every 10 Hz. and 30 kHz. the output of the integrator is: 1/(t):: 6. . Hence.. Therefore from the timeconvolution property (3.T/2 and IH(".(1')  b(1' . 10 kHz. S6.
500 ==> = 500 BecausE' L should be a power of 2.26 For n sinusoid.6 108 kBits/second data bits. . so that 10 bits or binary puLces are n!'eded to encode each sample. But . and each sample is encoded by 9 bits.27 kHz. g(t) 0 for It I > T. the minimum bit rate = 9 x 4000 = 36 kbits/second. yielding a total of 108540 bit.s/charact.s/second. we need 7 bits/charact. The sampling rate f. The Nyquist rate is 30 kHz. hence.5 X 106 == 9 MHz.21 = (a) Since 128 = 2". Fi\'t' such signals are multiplf'xed. :\'~'quiM = ¥ 6. 108. 111.6 kbits/second.23 (a) ThE' ]'.000 characters/second.8 MHz. .o == 31.5.H(oJ) rT . (b)For 100. we need 700 kbit. Thus. quantization error ~ ~= ==> L 400 Because L is a power of 2..o convolution of g(t) with sinc(271'B't) which is not timelimitt'd.. C ) . Therefore. ThE' minimum cable bandwidth is 36/2=18 kHz.  = 256 == 28.R mp ~ = 47 dB =50119. The minimum transmission bandwidth is = 54. This requires a 9bit binary code per sample The rate is 2 x 1000 == 2000 Hz. _ 1 = !:!. (b) 1024 = 210.. 20% above this rate is 2000 x 1.36 = 216. <a) The bandwidth is 15 kHz.:1 ~ == 3(256)2(0. each signal hc.2 x 9 == 10.q(t) is equal t. then quantization error < (0.s 2400 samples/second.16) No So = 3L:I~ ~ = 3L 2 (0.24 If TIIp is tht' peak sample amplitude. The actual sampling rate == 1.. ~ == 0. ThO! SNR for this "alue of L is = 98304 == 49. Filure S6.! 400 Quantlzatlon error < O.005 == 540 bits. It is impossible to obtain a timelimited signal from the convolution of a timelimited signal with a nontimelimited signaL 6. L .er. that is. Framing and synchronization requires additional 0. (A). ... we select L ''0 ~.2 == 2400 Hz.8 Because L is a ?ower of 2. (a) 8 bit.s/second.2)(fllp) 100 !ecnll5e the maximum quantization error is ~ == ~ == =~ 500 T.1T o .25 Nyquist rate for each Signal is 200 Hz. t. we choose L "" 512 = 29. = 00p = = = = = T =:ii = 6.43 dB 53 . 6. The S].. that is.yquist rate is 2 x 4. 9 bits/sample. it follows that L ~ !!!J!.er and 800 kbits/second. (e) 30000 x 16 = 480000 bits/so (d) 44100 x 16 = 705600 bits/so 6.000 x 0. so that 16 binary digits art' needed to encode each sample. Thp.5) = 50119 ==> L == 182. 2 x Nyquist rate 400 Hz Total numbt'r of samples for 10 signals 400 x 10 4000 samples/second..5% bits. each signal uses 9 x 2400 == 21.. we need a total of 5 x 21. (e) 10.refore.. we select L = 512 == 2"..22 (b) 65:. From Eq.8 x 106 X 10 = lOS x 10 6 or 108 Mbits/r. 6.2 Moreo"er. (c:! ) Because R(t) is timelimiteti.5) 111. (6.
.45.ling rate is 3 x 106 Hz and each sample is encoded by 8 bits (L :::. Hence. mJ(I) ~ m = A2{. Consequently.67 dB (a) :'Iiyquist rate = 2 x 106 Hz. BE'c~uSE' L is a power of 2. From Eq.84 .Iln(l£ + 1)]2 ..~o So ~~ = 3(256) 2 (1/3) = 65536 = 48.. The power or the mean squared value (energy averaged over a quarter cycle) is ~ = __ 1 ". = • J..33) so that 1 . 6. The equation of the first quarter cycle u shown in Fig. 6. an increase in the SNR by 12 dB (fro111 30 to 42) can be accomplished by increasing n from 10 to 12..26. '.. each quarter cycle contributes identical energy. 2 10 == 1024.77 ==.06 dB (b) If we reduce the sampling rate and increue the value of L so that the same number of bits/second is maintained.. (a) From Eq.he same tran!imission rate (:24 ~1bits/s). each quartp.2. the SNR is increased by more than 10 dB.622..= 3L =r . SNR c. L = 256 No = (In(.)(64.83 Becallsl' L is a power of 2. (6.10_300 . the transmission rate is 8 x 3 X 106 :::.(In 256)2 . '. 24 Mbits/second.28 Here " :::: 100 and the S:'IiR = 45 dB= 31.:Z(t) To/4 ~l ~ lTot.. (6. (6.. Benct'.§.4·1 Equation (6.. The S~R for this value of L is §. The rest of the solut ion is identical to that of Prob. Hence..1 + 1)]2 = (In 256)2 So 3L 2 3(256)2 .29 3(512)2 _ ? _ No . . _ 6.18) So 3L 2 No = (In 101)2 = 31. we can hswe (24 x 10 6 )/(2.. 6394 ::: 38.he SNR (because of increased L) with the same bandwidth. we can improve t.1 54 .23) shows that increasing n by one bit increases the 8NR by 6 dB. that is increasing by 20%.369_2.\.16). " .he Nyquist rate). iF.2·7 is met) = 4A/To.50. 256).210 6. .4 x 1011 (20% above t..Fi•. . [ence.1.1 dB Clearly. This results in L :::. If we reduce the sampling ratf! to 2. Hence. the sam. { . then for t. (4A)2 0 _ To dt =  A2 3 ..f 47 dB is a ratio of 50119. (6. we select L = 512 = 211. In part (a). the new S!'R is .(In 101)2 .77.000) 211' x 800 ==> 00811 . we select L = 256 = 2s ...27 For this periodic mIt).7 6.4 x 106 ) = 10 bits/sample.r cycle takes on the same set of amplitude values. where A is the peak amplitude and To is the period of m(t). _ 3L 2 _ 3(1024)2 _ ? _ No . 56.".5 x (2 x 106 ) and I' = 255. From Eq.622. The actual sampling rate is 1.:. Moreover. p .18) =3 X lOb Hz. From Eq.16 dB TTlp 0. we can compute t.he power for this signal by averaging its energy over a quartel' cycle. The SNR for this value of L is .. = 473.(". 86.
we need a bandwidth of 128 kHz. and So No = 1.. (0.12 X 0. we need a . rat~ 64 kHz.4 so that 0.2l! = 3/.(b) N o = .333 10.= 1! = 1 . 446 )( 10:1 (d) For uniform distribution So (~) = r..5 =.5. (c) Here So = tf = 0. For a bipolar case.94 x 10:1 bandwidth For onof!' signaling with a bit or 64 kHz..4 = 2.0785)2(3500) (3){64000) = 1 12 )( 10.2 (t) 111 3 3 So No = 1.4 ..12 )( 10.
Chapter 7
For full width rect pulse p(t) =
7.21
rect( ~)
P( aJ) = 7j, sinc ( aJ? )
For polar signaling [see Eq. (7.12»)
Sy(aJ) =Ip(tUt = 7j, sinc:2 (tU7i ) ~
Tb
2
For onoff case [see Eq. (7. lib)]
Sy(aJ) = p(tUt [1+"
I
47b
2 _0 6 aJ= 2)] r 7b 7b
GO (
=It sinc2 (tU7b ) [1 + 2tr
4 2
7j,
11=_
f
6 (aJ _
~)l 7j, J
Hence,
But sinc 2 (
tU~b ) =0
for
aJ = 21171
T6
for all n _ 0 ,and
=1 for n = O.
2
Sy (aJ) = 7b sinc2 (aJ7b ) + .!. 6( aJ)
4 2
For bipolar case [Eq. (7.20b)]
Sy(aJ) = Ip(aJt sin2
7b
(aJ7j,)
2
:: 7b sinc 2 ( tU~ ) sin2 (tU~b )
The PSDs of the three cases are shown in Fig. S7.21. From these spectra, we fmd the bandwidths for all three cases to be Rb Hz. The bandwidths for the three cases, when halfwidth pulses are used, are as follows: Polar and onoff: 2Rb Hz; bipolar: ~ Hz. Clearly, f"r polar and onoff cases the bandwidth is halved when fullwidth pulses are used. However, for the bipolar case, the bandwidth remains unchanged. The pulse shape bas only a minor influence in the bipolar case because the tenn sin 2 (tUl ) in Sy{tU) detennines its bandwidth.
l'
S!jl"')
,.l.r
b:~ob(,
___ eJ'/\ 
Off
o
S6
"R&
Fia·57.21
7.11
YH:)
.
R•nrnn r · llJ crOlOlT't....,.
I I •
I
(b)
FiC· 57•11
P{tjand
t+.!i] (t.!i] ( ,%4 __ ,%4
"'.:0 ) e14/1b 14 + t
SinC (
P(",) =tSinC (
"'1 )
4
ej4/1b/
4
=j16 sinc ( O1:b ) sin ( 01: )
SY ( ) =
01 Ip(01t = 1ib 16
SIDC
. 2
(01Tb) SID 2 (~) . 4
From Fig. 57.22, it is clear that the bandwidth is 4" rad I s or 2~ Hz.
16
7.13
For differential code (Fig. 7.17)
~=
N+«l N
lim
..!...[.!(1)2 + N (1)2J= 1
2 2
To compute R\, we observe that there are four possible 2bit sequences II, 00, 01, and 10, which are equally likely. The product 0kok+\ for the fU'St two combinations is 1 and is 1 for the last two combinations. Hence,
R\
=N+rs> ..!...[~(I)+.!(l)J= 0 lim N 2 2
II>
Similarly, we can show that Rn
=0
1
Hence,
Sy(.,j
7.14
.I~t •(!f) ....' (
"n
=oand the remaining half
2
(a) Fig. 57.24 shows the duobinary pulse trainy(t) for the sequence 1110001101001010. (b) To compute~, we observe that on the average, half the pulses have ak have Ok =I or I. Hence,
Ro =
N+«l
lim
..!...[N (±1)2 + N(O)]=.!.
N 2
2
To determine Rl' we need to computeakak+\' There are four possible equally likely sequences of two bits: 11,10,01,00. Since bit 0 is encoded by no pulse (Ok =0), the product of 0kok+l =ofor the last three of these sequences. This means on the average 'E:! combinations have akok+l
4
=0 and only
N combinations
4
57
have nonzero akak+I' Because of the duobinary rule, the bit sequence 11 can only be encoded by two consecutive pulses of the same polarity (both positive or both negative). This means ak and ak+l arc 1 and 1 or 1 and  1 respectively. In either case akak+l =1. Thus, these: combinations have akak+1 .. 1. Therefore,
Rl
=N"'rI!l lim
..!.[N (I)+~(O)]"'! N 4 4 4
To compute R2 in a similar way, we need to observe the product akak+2' For this we need to observe all possible combinations ofthrcc bits in sequence. There are eiaht equally likely combinations: 111, 101, 110, 100,011,010,001, and 000. The last six combinations have either the first andlor the last bit O. Hence, akak+2 =0 for all these six combinations. The fU'St two combinations arc the only ones which yield nonzero akak ...2' Using the duobinary rul:, the flfSt combination is encoded by three pulses of the same polarity (aU positive or negative). Thus ak and ak+2 arc 1 and I or 1 and I, respectively, yielding akak+2 = 1. Similarly. because of the duobinary rule, the fIrSt and the third pulses in the second bit combination 101 are of opposite polarity yielding akak+l. ~ I. Thu~ on the average, akak+2 =1 for N N 3N  termS,1 for terms, and 0 forterms. Hence, 8 8 4
R2
= lim ..!..[~(I)+ N (1)+ 3N (0)]== 0
N...c
In a similar way we can show that Rn
=0
N 8 8 4 n> I , and from Eq. (7.1 Oc), we obtain
S (cu) =Ip(cut (1 +coscu7b) = Ip(cut cos2(CU1i ) _b
y
2T"
T"
2
onn, , , au ' n , , n , a '
1 1 1 0 0 0 1 1 •
1 0 0 1 0 1 0
'1&57.14
For halfwidth pulse P(I) ... rect(211T,,}.
Sy(OJ)'" ~ sincl (OJ: )cos2 (OJ: )
From Fig. S7.24 we observe that the bandwidth is approximately Rb 12 Hz.
7.31
From Eq. (7.32)
4000=
(I +,)6000
2
1 =>,=3
58
35 In this case Also.2x280)(103 =168kHz. n=O =106.35) as follows: g(t . we select L =128 =2 7 (b) This requires 7 bit code per sample. ~ =0.4 and II ~/2 = Rb . The excess bandwidth t'l> x (b) From Table 3. Hence. 7.. 40. g(t + T) + g(1. =125 x 4000 = 5 kHZ. p(t'l» in Eq.. Hence. (7.1. The sampling rate I..!} =500 kHz and 2 Ix .T) <=> G(t'l>)e}TAI and g(1 + T) <=> G(t'l»efl'AI Hence. 3000 =2 R" => Rb =4. Observe that Hence p{t) satisfies Eq.25 ..34a) can be expressed as (1) p{ CI) =1. 1.::: Rb Tb =106 bitsls. 1j. =02 7.500  200 kHz.!.0 Imp => L:2! 100 L (a) Because L is a power of 2..5 kbitsls. BT ~ =I MHz.34 1+. we havet'l>b =2lrx 106 and Rb =106. 7.36). Hence.000 samples/sec. from Eq..7.. =Rb ~ Rb == 3 kbitsls. BT =Rb => R" =3 kbitsls.000 x 7 =280 kbitsls.36 h =700 kHz. (7.33 (a) BT (b) BT = l. 7. (7.. p{t'l» does satisfy the Nyquist criterion with (a) Comparison of P{t'l» with that in Fig. = Rb => R" =3 kbitsls. Nyquist rate =2 x 2000 . {2x 106)=>.8 kbltslS. r .... we fmd =Ir X10 6. 40. AlSO. (7.. (e) BT  (d) (e) BT 7. we can transmit data at a rate Rb =12MHz. 2 =2Rb => Rb =1.32) BT_(I+')~ 2 7. From Eq. we derive the dual ofEq. (3.T) <=> 2G(aJ) cos Tt'l> Now. =1.12 shows that this t'l> b = 21r )( 106 and. Bit rate .. Hence.Ix = 500 . 4 kHz for each signal..200 ::: 300 kHz.32 Quantization error AV 2 =mp S 0.37 To obtain the inverse transform of P{CI). (e) the pulse transmission rate is. 700 . rec{~) + .2 x 106 .32) 1. =2 MHz. p(t) =sinc 2 (lrX 106 t) From part (a). 2 R"..m:t(~)co{~) 2 41rRb 2 41tRb 2Rb 59 (2) .!. Eight signals require 8 x 5000 . I. p(nTb)t Hence.
K 2tr Rbt • =R.114R.Using Pair 17 (Table 3. it follows that =.16)... p(t).Rb sinc (2trRbl)+tSinC [2K~(t+ 2~J]+l}SinC [2tr~(I.1(1.::b . bt ~t (2K ~I .2....I Sin(2trRbtK)] + .. sinc (tr Rb t ) +sinc [tr R.~(t..38 Hence.) =1 and p( ±nTo) =0 for all other n. The Nyquist samples are p (±"r.(2tr R"I + tr) .10) with T.2)] 1 ( 1 " 2RbCOStrRbtsintrRbt R"COStrRbt. Hence. 2 21b PIt) sInc "/101 +smc [..tr • sin tr R"t _ sin KRbl:= sin KR"t tr ~I tr Rbi tr tr R.Rbi) =sin tr Rbi tr Rbt +~~~ 7. (6.39 The Nyquist interval is ~ Rb From Eq.1) and Eq. ( R ) = smc tr bl 2tr R"t(I4~2/2) t4R..) for" =0.~)] = sin tr Rb KRbi _ sin KR"t tr Rbt .fL 2tr Rbi 2 2tr R"t + K 2 2K ~t ...2t2 7. (7. =r". (1) above.. . we obtain pet) .La 1b.. 1.tr = sin KR"I tr Rbl(l.2!b)] =~[SinC (2K R"t)+iSinC (2K~t + K)+ tSinC (2K Rbi K)] =R" [ Sin(2trR"l) 1 sin(2trRbl+tr)' ...2. p( 0) • P( r.[Sin(2trRb/) _! Sin(2KR"t) _ t Sin(2tr~t)] 2 2tr Rbt + tr 2 2tr R"t K 1/2 112] Rsm(2 tr R)[2tr1 . and B =~ =_1_. from Eq.J 2trR.tr) = Rb = sin(2tr~.(t  ~J] sin( tr R"t tr) tr Rbi .Rbi) 60 .
43) T .41 . 12000 200 20. 110111101001011 R =101010100000111 "" S 61 . (lED D2 ED D4 ED D6 ED D8 ED DIOED Dl2 ED D I4 ED•••)S R • (1 ED D2)T (see Fig.. 100010000000110 R = 101016100000111 = S 7.41 S == 101010100000111 T. (1) above yields P(m) =_I rect(~) + _I reet(". 2"Rb r } =:b co{ 2~b )ree{ 2:Rb jOJ/2~ 7. fll'St digit I.42) T I I> > ! ct FII57. or 1010. that is.The Fourier transfonn of Eq.. (b) The fU'St sample value is 1 because there is no pulse before this digit Hcnce the first digit is I.. or 1000.!!}jOJ/ Rt R" 2" Rb R" 2" R" =_1 rect(~) [ejOJ/2~ +eJOJ/2~ LjOJ/2Rt Rt. Hence the 4 possible correct digit sequences are where 11111)~ is any of the four possible sequences 1100. These sample values represent the following 4 digit sequence: llOO. or 1010. or 0 00 2..S7. or 2 0 0 O. indicating that the transmissions starts with a positive pulse.. The detected sequence is 11000100110110100 7.2 T = (I ED D3 ED DS ED D6 ED D9 ED DIO ED Dli ED Dl2 ED D13 ED DISED . 7. or 0100.ZO~ ~02202 Following are possible correct sample values in place of the 4 undcrbracket values: 220 2.41 S =101010lOOOOOIll From example 7.1000.0100. The duobinary rule is violated over the digits shown by underbraeket. 57.43 S 101010100000111 T==(IED DED D2 EDD4 EDD' EDD8 ED D9 ED Dli ED D14)S R={IEDDEDD3)T (seeFig.310 The fU'St sample value is 1..310 (a) No error because the sample values of the same polarities are separated by even number of zeros and the sample values of opposite polarities are separated by odd number of zeros.~ T .)S R =(lED D3 ED DS)T T =101110001101001 R =101010100000111 =S 7.. or 2 0 2 2.
3 I = 1 0 [~328] 1.00621 p. Hence..T FII. the pulse amplitude.l25 x 106 222 2 P onoft' l\ipolar R~ = A 216 x ".1 ~. =106 S./~) => ~2a" 62 Ap 20'" ~4. (7.61 I 0." 5 andA p " lOa" =0. =l. == 5. =A2 Tb 12 and there are Rb pulses/second for polar case and ~ pulses/second for onotT and bipolar case.=A2 .002 _0..009315 In the following discussion.61 For onotT case: P.003 Ponotf =.)..003045 and 7.."3 R 7.J!. Q(5)=2J7 x I07 P. =5 A (0 Forpolarcase (il) Foron~otTcase (iii) For bipolar case p.) =>.075 x 20'" =0.1 0.0.003)2 6 = 225)( 10 4 A For bipolar case: P( e) == 2J7)( 107 =1. Q(512)=0. Hence.07]1[0] 0. A2 (0.. Ap 120'.113 Ca).7S .51 From Eq.075 Hence A "" Ap a" =5. the received powers are A21.3 0.5625 x 106 2 4 = A2Tb ~ A2 6 2 )( 2 ="4 =0..5625 x 10 Cc) For onotT case: We require P(e) == 2J7 x 107 . we assum~ Ap (b) Energy of each pulse is Ep = A.07 ~. Q(Ap 120'. =1.J!. S7.sQ( Ap 120'.0015)2 Ppolar =~R" =.45)... we obtain C_l] [ CO" [cl 7.5Q(5/2) =0. 4 A2 (0..
9.000( 1...5XI03)2 1I28xI0"'_ Sr . Quantization error. the average energy of the 8 pulses in the 8ary case above is 63 . ( Y)'~(4. and for halfwidth rectangular pulses. = 22.[~) e ~lall = Ap .. Consider binary case using pulses ± A 12. 22. ±SAI2.67)( 103) '"' 23. (IOOO)S.25". 2 2 7.56)( 103 watts 7.73 (a) Each Sary pulse carries log2 8 = 3 bits of information. Let the energy of each of these pulses (of amplitude ± A 12 ) be Eb' The power of this binary case is l\iuy =EbR" Because the pulse energy is proportional to the square of the amplitude.75)( 103 all For polar case with halfwidth rectanaular pulse: 5.2 Eb' Hence.000 . the transmitted power is: Sj '"'  2 ) I(A 2) . = 1c"/ Ap ""¥l2a" )= Ap =4.000 =50.. the minimum transmission bandwidth is 50..63 For polar case: P ..000 Hz. Hence.. { .:. and a total of5)( 10. (b) The amplitudes of the 8 pulses used in this Sary scheme are±AI2.. = . we need ~ 2 m =5 4ary pulses/sample.835 andAp =4. watts Sr . lOoo{.. .".:.. 25 kHz..P2 4 4 There is an attenuation of 30 dB.28)( 103 \yatts For bipolar case: 2a" For bipolar (or duobinary). Therefore Sr loooS. 10.:J!. 2 bits of information. ±3AI2. we select L = 1024 =210.. (95)(103 .. half the pulses are zero. Br " " " 1 S (e) (a) Each 4ary pulse conveys log2 4 .25) 3. Therefore. half the pulses are zero and the receive power Sj for halfwidth rectangular pulses is A 2 1 2 Sj = : = 4(9. n 10 bits/sample. (I000Xll. Hence. 0.all =103 => Ap ~(4. the energy of a pulse ± ~ is .W) = 1l. 2)( 4000)( 1. fa ratio of 1000 during transmission. 2 R" (I + .28)( 10oo{.5)(IO3 for onoff case..38)( 103watts 7. 4. = 23. the bandwidth is reduced by a factor of 3.0004ary pulses/second. or equivalently.00 Imp => L = 1000 Because L is a power of2.".56 )( 106 watts.. A2 :. ..38)( 1000{.835)(2)(103 . lOoo{.) 50.7'1 p. and ±7AI2.7SX2)(103).67)( 103 Sampling rate .9..2 kHz. Hence..75 => Ap = 4..
we need 12000 '4.000 bits/sec. Hence we require only pulses/second.. each pulse must . The power PM is given by Rb Mary log2 M P M = EM& 2 2 • (M2_I)Rb E =(M 1)A == M2A2 b 241012 M 2410g M 1012 M 310g2 M 2 7.+2[±(M I)t Eb] M2 M tO =2Eb t (2k + 1)2 ..000 M.. .000)( 8 =192.000) . Hence. ~ Rb =  A27j 8 8 (b) The energy of a pulse ± ~ is k 2 Eb • Hence the average energy of the Mary pulse is EM.000) ..74 2 =Br =1800 Hz.75 Each sample requires 8 bits (256 =21 ). Therefore. .71 (a) M =16. we haveR" =Br. Rb bits/second requires a bandwidth of Rb Hz. The halfwidth rectangular pulse of amplitude ~ has energy 2 Eb =(~)2 7b = A 7b 2 2 A2 8 The power P is given by P =EbRb .. Hence. ~[2Eb +2(±3)2 +2(±5)2+.Hence.000 64 .000 bits/sec. transmIt at 1 (192. 4 bits. east 50.... Br 30kHz 2 2 R =.ary pulses/second to cransmit 192. 50.32).1f)I =7 l\illll')' 7.=3000 16ary pulses/second..~Eb M2_1 Each Mary pulse conveys the infonnation oflog2 M bits. 2 (a) For polar signaling.. 2 2 (b) From Eq. Hence. Minimum transmission bandwidth =3000 . 1+. 384b'Its. 3000 = 1.Br 0= I (30.. (7.. Each 16ary pulse conveys the infonnation oflog2 16 ...000 Mary pulses/sec. 1+. 1500 Hz. Hence: 24.2 We have available up to 50.2 Br 7..
.3 MHz 7.81. Fie.= = 7. and the second ASK signal occupies a band lei ± 1MHz... (b) FSK can be viewed as a sum of2 ASK signals.12 (a) A baseband polar signal at a rate 1 Mbitslsec using Nyquist criterion pulses at . Fie·57..81 choose 4 bitslpulse M .83 log2 M . 7. 2 2 PSK doubles BWto 1.z. the bandwidth is 2 MHz + 100 kHz =2. =02 has a BW= (1+') R" .3 MHz. I WI~lt) .12 xl06 =6. there are four center (carrier) frequencies lei.l~ l t""'..6 MHz + 100kHz BWFSK =1. It) 6S . each separated by 100kHz. BWFSK =0. Each ASK signal BW =2 MHz. 2 for M =4...8l 7. Since ASK signal occupies band Ie ± OJ MHz. "'.1 MHz. We need to transmit only OJ x 106 4ary pulses/sec (a> BW is !educed by a factor of2.16 is the smallest acceptable value (a) Baseband polar sip at a rate of 1Mbitslsec and using full width pulses hasBW =IMHz. .91 "J."". BWFSK = 1 MHz (b) In FSK. The first ASK signal occupies a band I cO ±1 MHz.. (b) Similar to Prob. 57.0xl0'Hz.Lt) t.2 MHz.5 MHz + 100kHz + 100kHz + 100kHz =1. PSK doubles the BW to 2MHz..83 7...6 MHz + 0. the total bandwidth is OJ MHz + 0.. Hence.. te2' 1c3' and/e4.
.£' " . Fia. S7.800 samples/sec. m.{~L Fla.DWlM~ s~ 1.  'r / t I:£~J 4c. 7. (b) mz(t). This combined signal is now multiplexed with ml (t) with a commutator speed of 2800 rotations/sec. .92 First. ml(t) has 8400 samples/sec.92 66 .~. I \ tc. yielding the output of 5600 samples/sec.. m3(t)..: · 7. and m4{t) with a commutator speed of 700 rotations/sec. Hence. I: t.z00 ro1idilrn\. m3(t).91 Either figure (a) or (b) yields the same result. we combinemz(t). S7.rt>.. there are a total of 16.."'.(t) each has 2800 samples/sec. c: 21DO ') roTatioflS Sec.".. Fla.
7. \l49. !.ifC{I G.:zq'b /tb/S ""4t~) ~ • l'iNet.1L1J1I.s .~r g.f:. 2....s fc.93 . .1rtb £ " Dt1tfJ~t '((\if::') ~::1...6 Q:leJe(.'HJl'J '> • ·fs2.  / . J &00 r~(":> I Cll r ~1·{. 21·'~.93 67 .: 72CO rt'I. ~. :::.....1. .¥Co 0 QUt1.(" Cohr ~ ~"~ .t'1:1.i kb/. '. :2..lt') > '~2fof\ ~ '" • of:./ AHerV1~ a frt:l rI~f71~" r ctt) (b) Fla S7..
235336. how long will it take for a DS3 signal transmitted between two networks to become out' of sync if a 1/4 bit length time error results in desynchronization? Solution: A DS3 bit is transmitted in 1/(44. both network clocks will be out of synchronization by 6 parts in 1012.27 seclbit or 980.8 sec. a. JUI I I I I I I n I I . In the worst case.10. bipolar violation c. 16 O's violation Solution: No. determine if each is a valid AMI format for a DS 1 signal. Solution: Note that typically.235336.' .736.106) 2. I I .108/(6. JU'lllllllllllllllnlll'U Solution: No. .1' s density violation d. JUI'II'I.Chapter 8 Exercises tUl If a plesiochronous network operates from Cesium beam clock which is accurate to ± 3 parts in 1012. the waveform is given as 8. explain why not.10. 12 ) =3922. for illustrative purposes. .nl""IIIIU"" Solution: Yes 68 .12 For the bit stream 011100101001111011001 draw an AMI waveform. I .57 sec/lA bit 8. If not. " " U Solution: No.13 For the following waveforms. ' = 2.
16 T·l synchronization at two distant locations is controlled by separate crystal controlled oscillators which differ in frequency by 125 parts per million.uence bits •. Solution: 8... Show the bit stream which is substituted by the DSU. 16 O's violation. if any. .•00110010110010110. H the terminal equipment doesn't maintain sync in how many complete D4 superframes will the faster oscillator have generated (at most) one more time slot (8bit) than the slower oscillator? Circle the correct answer.14 a) You have received the following sequence of ESF framing pattern seQ.. explain. Is this a"legitimate framing bit sequence in order to maintain synchronization between the Tl transmitter and receiver? Yes_ _No_ _ H yes. b) The following TI AMI signal is received: . why? H no. why not? Solution: No. H yes..8.  r ~ . b.. The bit sequence 0011 cannot be in an ESF framing bit sequence. explain why not (what.. 69 . The 16 O's will be replaced by a pattern of 1's by the DSU. OS 1 standards are violated) and draw on the figure the AMI wavefonn which would be transmitted by the DSU? Solution: No. Is this an acceptable Tl signal"? Yes_No_ __ a. 8.15 The signal 11010000000000000001 is received by the DSU in a Tl data stream which uses a B8ZS format. Draw the output of the DSU for this signal? The first 1 is already drawn. If no.
0034704 superframes. a. 23000x2 = 46000miles up and down. Hence. .247x1544000:: 381850 bits .125 time slots. 46000/186000 0. 6. A and B. will a correction signal take to be transmitted to the network switches? Solution: The speed of light is approximately 186000 miles/sec.1061. utilize Cesium beam clocks accurate to 3 parts in 1013.106 seconds per errored bit or 5. How long.17 Two plesiochronous digital networks.106 seconds per errored halfbit.} Solution: A TI bit time is 1/(1. The networks are operated by independent long distance companies and are synchronized to each other by means of a UTe signal.544. b.07945. In the worst case.000 miles above the Earth. in terms of Tl bits.7 seclbit.if "none".r 8. = = .247 sec 0. what is the number of D4 superframes before an extra time slot is generated?_ __ Solution: e) The faster oscillator will generate 125.5 b) 10 c) 15 d) 20 e) None of the above .47668·10. UTe operates via GPS satellites which are approximately 23. 13 errored bits per bit 1. a) = .544. = 10 . the two clocks would be off by 2·3 ~ 6 parts in 1013 or 6·10. If a company leases a T1 line with D4 framing which is tenninated at one end in network A and at the other end in network B.7 seclbit 16.10. a time slot difference will be generated in 1124.13 errored bits per bit transmitted. Hence. This is one frame/sec =24.47668·10.125 =0. how often must the networks be resync'd to each other to avoid a framing bit error in the customers Tl signal in the worst case? {You may assume a framing bit error occurs when the two networks are out of sync by ~ 112 of a T1 "bittime".106 193 bits per second more than the slower oscillator.106 ) 6.39723.04164498 frames or 0.
S2 20 5 (e) P(nsS)"Sij'3 10.2).. • P I P2• P20" P2~' P2o" P2P. (6.5. =252 13 1 =13 1+ 1 4 (d) P(7).!.3).. P(E3) = :.6). (2. PIO. (2.4.4.1).5).5).4). (1. (5.1 (i = 1.2.!.2).1). 2.1. (6. (2.=\ =k +2k+3k+4k +5k+6k ". (3...14 We can draw 2 items out of 5 in 20 ways as follows: 0.2.1. (6. (6. (3.3.6..3.3.3).5.2. (5.3.2. (2.2). OIP" 01P2• 020" 0203. (4. 5. 030.2).1). O)~. =~ 20 5 (Ii) This event E2 • 1\ P:z U ~ 1\ 2 1 Hence.3).!.2).0 1 3.2. P 102. There are atotal of6x 6x 6= 216 outcomes.21k P(i) = . P(EI) =.4.6.. J3 Note: There is a typo in this problem.(2. Hence.!. 6) = I k =21 10.5.6).!. (4.5.1). (6.2. (5. (4.Chapter 10 p(redcard)=~=. (5. (2.1. (1. The probability that the number i appears should be ki not kj • 1= rki .02. (2.11 (a) (b) P(black queen) = (e) I 52 =26 P{picture card) =.6).6.6).5.6. 20 10 (iv) This event E4 :. (5.3. (i) This eventE\ =011\ U01~ U021\ U02~ U03 P\ U03~ UI\ O UP\02U1\03UP:z01 UP202 UP2 03 I Hence.1). PIO ..1.. (5.3). (1. (3.!.1.4).5). 0 O)P . E3 are disjoint. (6.5).3.5).. E2 UE3 and both E2 &.3) 10 peS =10) =216 6 10. P(S '" 4) =. All these outcomes arc equally likely with probability 1120. 03P2.12 (a) S =4 occurs as (1. (4.. (4. 216 72 (b) S =9 occurs as (1.4). (4.4).5).2).4. P( E2) = 20 =10 (iii) This event E3 =0102 UO\03 U 0201U0203 U030l U0302 6 3 Hence. 02P" 02P2.2.4).4. 52 2 10. 3.5. (3.4.(1.6.6.1).1.6).2). 4. (1.1._1 (e) S =IS occurs as (3.4).3).6). (3.1) 25 P(S=9)= 216 Hence. P{E4 ) =P(E2)+ P( E3) = I~ "" t 71 .
4j) "" 903 ways..983.5)IO =~.\.s)IO = 120 " 5 45 120 ) 849 P(atleastfourO's)l. We Cart choose S numbers of our 6 in (f) = 6 ways and can choose one number out of 43 in(4?) '"' 43 ways.1·5 Let xo\ be the event that the flJ'St chip is oscillator and x 1\ be the event that the fU'St chip is PLL.10.4Sx 10S 13983816 (c) To match exactly 4 numbers means we pick 4 out of the chosen 6 number in(:) =15ways and choose 2 out of the remaining 43 numbers in ( numbers out of 6 and • 15)( 903 P (matchmg exactly 4 numbers) '"' 13983816 Hence.16 Using the notation in the solution ofProb.. we have 43 x 6cornbinations in which exactly 5 numbers match..+ .s)IO P (three 0'5) '"' =~ 1024 1024 (1f"'O.: _5_ j\ P(twoO's) = (If''o. respectively.1·7 (a) We can have(1~) ways of getting two I's and eight O's in 10 digits 10! (10)= 2!8! 45 2 P (two 1's and eight 0'5) = 45(Ojt(O. 10... Prob(matching all 6 numbers) 1 ~.+ .~) = P(XO\ )p(x Pj Ixo\)+ p(x.. 6 P(matching exactly 5 numbers) = 43 x =IJ. let x02 and x Pj represent events that the second chip drawn is an oscillator and a PLL. =(IP"O. ~ 1024 (b) peat least four O's) = p(one 0) j\ l[P(exactlyone 0)]+[ P(exactly two O's)]+[ P(exactly three O's)] 1024 512.~t . we find: (a> p(xO:z Ix 1\) =~ (b) p(xO:z Ixo l ) =~ 10.512 1024 1024 1024 10.( . 45(Ojt _ 2 4~ .15. Also. Thus.. Then P{I osc and 1 PLL) 0: 1'(xo\ .. Hence.18 <a> Total ways of drawing 6 balls out of 49 are 49! ( ~) =6!43! =13. Thus there are 1S x 903 ways of picking exactly 4 =9.\ )p(x~ IXI\) (lx~)+(~xl)= ~ 5 4 5 4 5 10.816 13983816 (b) To match exactly 5 number means we pick S of the chosen 6 numbers and the last number can be picked from the remainina43 numbers.686 x 10 4 72 .xPj) + p(x.
(d) Similarly, we can pick three numbers exactly in(~X4J) = 20x 12341 = 246820 ways. Hence,
P (matchmg exactly 3 numbers) =
.
246820
13983816
• 0.0 176S
10.19
(a) Let f represent the system failure. Then
P(i)= (1_0.01)10 =0.90438 P(f) =1 P(i) =0.0956
(b)
p(i):: 0.99 and P(/) "" 0.01
If the probability of failure of a subsystem $j is p, then
P(i). (1 p)ID or 0.99. (1 pt =p. 0.0010045
10.110
Iff represents the system failure andfu andfL represent the failure of the upper and the lower paths, respectively, in the system, then:
(a)
P(/) = p(/ufd = p(/u)P(fd =[P(fu)t P(fu) =1 p(ill) =1(1_0.01}10 =0.0956
and
P(f) =(0.0956)2
Reliability is
(b)
=0.009143
P{i) =1 P(/) =0.9908 P(i) =0.999 P(f) =1 0.999 = 0.00 1 P(fu ) = .JO.OOI =0.0316 PUll) =(1 p)10 =10.0316 ~ P =0.003206
10.111 Let P be the probability of failure ofa subsystem ($1 or $2) . For the system in Fig. a: The system fails if the upper and lower branches ran simultaneously. 'I1!e probability of any branch not failing is
(1 P)(I p) = (1 p)2. Hence, the probability of any branch failing is 1(1 p)2.
Clearly, PI' the probability of the system failure is p/ • [1(1 p)2 1(1 p)2];; 4p2
1
P« 1
For the system in Fig. b: We may consider this system as a cascade of two subsystems Xl and .12 , where.ll is the parallel combination of $1 and $1 and Then
.12
is the parallel combination of $2 and $2' Let p/ (Xi) be the probability of failure of.%1 .
P/(XI):: P/(X2} = p2 The system functions ifneither Xl nor .12 fails. Hence, the probability of system not failing is (1 p2)( 1 p2). Therefore, the probability of system failing is
p/
=1(1 p2X1 p2) =2p2 _ p4 ;;:2p2
P« I
Hence the system in Fig. a has twice the probability offailure of tile system in Fig. b.
73
10.112 There are (Sf) ... 2598960 ways of getting 5 cards out of 52 cards. N~ber of ways of drawing 5 cards of the same suit (of 13 cards) is(1l) ... 1287. There are 4 suits. Hence there are 4 x 1287 ways of getting a tlush. Therefore, P{tlush} =4)( 1287 2598960
=
1.980ax 103
10.113 Sum of 4 can be obtained as (1,3), (2,2) and (3,1). The two dice outcomes are independent. Letxt be the outcome of the regular die and Xl be the outcome of irregular die.
PXJ%2 {1,3} =pot\ (1}P%2 (3) ='6)( '3 =Is P %2 {2,2} ... P.q (2)P%2 (2) = '6 x 0 =0 ott PXJ %2 {3,1) ... pot\ (3)P%2 (I) ... '6)( '6 =36
Therefore ~(4) =is + 36 ... 12 Similarly, P'(S) p.,•• (1,4) + p.". (2,3) + p.", (3,2) + p.'" (4,1)
xO+x+xO+x= 6 6 3 6 6 6 12
I III III
1
1
1
1
1
1
1
1
1
1
P{B) ... P{A)P{B\A) + p( AC)P( ~AC)
=(;6X5\)+(~~X;1)= ;6
P{AIB) a P(AB) _ P{B)
(~H) =~ J..
51
26 10.115 CI) Two 1'5 and three 0'5 in a sequence of5 digits can occur in(n =10ways. The probability one such sequence is
p .. (0.8)2(0.2)3 ... 0.00512 Since the event can occur in 10 ways, its probability is
(b)
=0.0512 Three 1'5 occur with probabilityUX0.8)3(0.2)2 =0.2048
lOx 0.00512
Four I's occur with probability (
1X (0.2) 1 ... 0.4096 0.8)4
=03277
Five 1'5 occur with probability ( ~XO.8)S(O.2)O
Hence, the probability of at least three 1'5 occuring is P =0.2048 + 0.4096 + 03277 =0.9421 10.116 Prob(no more than 3 error) = p(no error) + P{1 error) + P(2 error) ~ P(3 error)
... (1 p')I00 +(lOO)P,(I p')99 +OOO}p, 2(1_ p,)98 +OOO)p'3{1 P, )97
.. (1IOOPe )+ lOOPe (199Pe )+4950Pe2(1 .... 98Pe )+ 161 JOOP/(197 Pel
74
10.1·17 Error can occur in 10 ways. Consider case of error over the first link pc ( correct detection over every link) (11\ )(1 ~)... (1 Plo)
=
P£ =1P c
=1(II\XIF,z)... (Il\o) =1[1 (1\ + ~ +...+ 1\0) + higher order tenDs]
== PI + ~+.. ·+I']o
10.1·18 p(e) 
Pt «I
j=3
±e
)Pe'(l p.)5j =10P.3(1 p,)2 + SP, 4(1_ p.)+ P, S
10.1·19 (a) P(success in 1 trial)
=1~ =0.1 =1 Pit Ph Pf3 Pi. PIs
(b) P(success in S trials):= 1 P(failure in all S trials)
PI) = Prob(failure in I" trial)., 9/10 Ph
=Prob(failure in 2l1li
trial) = 8/9
Similarly, Ph =718, PI. =6/7, andP" =5/6 Hence, P(success in S trials) =1 (.!.X!X7..X!X~) =11.. =0.5 10 9 8 7 6 10
10.1·10 Let x be the event of drawing the short straw and the Pt (x) denote the event that ith person in the sequence
draws the short straw. Now, 1'1 (x) =0.1
~(x) ... Prob(111 person does not draw the short straw»)( Prob(2nd person draws the short straw)
=
[1I\(X)]t .. (l~Xt) =
0.1
Similarly, P3(x) ... Prob(neither I" nor 2l1li person draws the short straw»)( Prob(3n1 penon draws the short straw)
 [1
~(x) ~(x)]i ( I~Xi}= 0.1
Similarly. P4(X)  ps(x} = I']o(x) =0.1 ...
10.1·21 All digits are generated independently
(a) p(alllO digits are 0) ... (OJ)IO (b) There areC~)ways ofamnging eight 1'5 and two 0'5. Hence, p(cight 1'5 and two 0'5)'" C~XO.7)I(03)2 (c) P(at least five O's).p(five O's)+p(six O's)+ ... +p(ten 0'5) 7 = o.7)s(OJ)S •7 O )4 (0.3)6 + CY)(0.7)3(03) +( 18Xo.7)2(03)1 +
(lfX
+cgX
(J8XO.7)(03)9 + (03)10
10.1·1
Py(O) = Pxy(l,O) + Pxy(O,O) =Px(I)Pyjx(O!I)+ Px(O)Pyjx(ojO) ... 0.6)( 0.1 +0.4[1 Py/x(IIO)]  0.06+032 =038
Py(l) =1 Py(O) =0.62
7S
.22 (a) pxjy(lll)Py(l) =(lQ)P..25 10. Py/x(l. 4$ei132 and rectify y (all negative ofy multipJed by I). Hence.1856 o Fil· SI0. SI0. (e) Fie. 76 . YaSsumes only positive values..25 x is a gaussian r...SI0.p(y < 0) =O..)Q (b) Pxly(O\I)'" 1. 4 y (b) Hence. with mean 4 and U x . (a) (b) i· Hence..23 (a) p(x ~ 1) =r.Pe)Q 10.. + xe % dx 1 2 2 3 2 1 (e) Prob(xS2).09176= 0..l)Px(l) (1. i P(X~4)=~4.[ xe%dx. (I) p(x ~ 1) =2P(y ~ 1) =2q~) =OJ026 (il) P(I < x S 2) = 2P(1 <y S 2) { ~~)~~)]= 0.24 Since this is a halfwave rectifier.9083 P(x ~ 2) = ~ 2 4) . Fy(O+) .9773 3 ".Q(2) 10.!xe%dx =! 1 2 e 1 1 '" (b) Prob ( 1 < x:so 2) = xe%.v.s P(x ~ 0) = ~0.. So .. it is a unilateral (rec:tified) version of Gaussian PDF.4)_Q(0)=0. we can use the expression of Gaussian r.3 Hence.10) 10.v. Hence. For a gaussian r.2' (a) From the sketch it is obvious that x is not gaussian..Pxjy{I/I) (note that Py(l) and Py(O) are derived in Example 10. However.+(IP. with a multiplier of 2.26 (e) If we take a Gaussian random variable y Py(y) . Fy(Y)'" 0 (fory < 0) and P(y < 0+) 10.0227S = 0.4)= l~i)= 10.1.v. 1. 2 2e 2 f 1: 1 3 =1 e 2e2 ..24 o Fil.. p y (y)= I 4& e%2132 withu . 10. the resulting variable is the desired random variable x.
28 <a> Similarly. 2(Iy) OSy:S1 1 1% {2(1%) OS. Px(x) .y}dy y J 2 A =2 But y =x + 1and the limits on y are 0 to 1. 10.ylx) = { o odIerwisc Clearly x and y are not independent. 2 Px(x) = P"Y(x. it is obvious that x and y are independent.10.27 Similarly. J~..y) must be unity. SI0.t 2dy =2 Py(y) = Therefore.2·9 Therefore 77 . 10..17 The volume V under pxy(x.. and (b) From results in <I).y) Fi.t:S4 • 0 o oIhawisc Similarly.x. V =!(J x I)A =~ =I.y(y) = 2(1. { o odIcrwisc 2 { 1I1y OSy:S1 PxlY(xly) = . PYlx(.y)'" 0 ochcJvisc IIIx OSx:S1 P (x.
r(O) Fit..r2[I:eiX)'~~ =1 But f: eyl~dy = Jie. er x =3. KJiJ:e3.211 P.' 4 I ..) = 1 Because ofsymmetry of pxy(x.r(I}+ P(£jO)P:r(O} If the optimum threshold isa...rl/4 and.Y  pxy(x. then P(til)=l~a~:p )=~A:~a) p( ti0) = P.Px(X)Py(y}..r2/4ea =KJ'ii( fl. Px x =:E='e 78 ..e I {~'• .) _ Pxly( xj.r(O)] P. e(Apa)l/2ts"lp. .r(O)] =0 Hence.e(Ap+a)2/2ts/lp.y) Pylx y = Px(X) .\)'+y2)~ =KI:e.1 Since P"y(X'Y) .r(I)+~ 2trer" dP.z (0) And a = er" 2 2Ap In[ P.rl+..10. da =L[e (Apa)l/2ts.r2)2/ 18 3'1/2tr 10.y) with respect to x andy...3i / v 4. SI0...y) _1 () = ..31 _ x=2.r2+~+yl} 4" and (xl) p.210 Kf:f: e _(.(x. e Py y vtr { .2·11 p = qA:~a )P. x and y are not independent 10.r(I)e (Ap+4/2~ P. Py(y) = /3e. ~a: : p ) a:: )P. = P(£jl)P.? P.r(l} ( ) 1 (...r (1).
. SIO..10.31 px(x)"'!lxle1zI 2 Because of even symmetry of Px (x). SI0.px(x. x2 =2 J.35 . ..[!!+ 27 _! +.] .!!_ 25 =! x 399 3 Fig.. .+ ) 8 \ 8( 4 3_\ 3 2 ..36 36 36 36 36 36 36 5 4 3 2 1 256 (8)+(9)+(10)+(11)+(12) =. Hence K =:! and Px (x) ..2 x'" rx. .( . x Px(x}dr =2J: x i xe2 2 x =0 and Xdt =fax 3e.) \ x 8 Jo 8 3 2 3 1 64 16 5 =() .34 The area of the triangle must be 1. 8 3 2 3 o 3 4 "2 1 x 1 1 x4 x 3 x (x+I}dr ..83 36 36 36 36 36 .33 10. ..=7 36 36 36 36 36 36 2 12 1 2 3 4 5 6 x = rx? Px(x.!..=2 36 36 36 36 36 36 5 4 3 2 1 (64) +(81) + (100)+ (121)+ (144) =54. (x + 1) 2 8 ISxs3 x=J _ 1 1 y3 y2 xp (x)dt=rty(yI)dy=.' 4 5 6 10. !! 8 4 3 4 3 3 u 2 :x2 _(x)2 .!.31 10..) =(4)+(9)+(16)+(25)+(36)+(49)+ .) =(2)+(3)+(4)+(5)+(6)+(7)+ 12 1 2 79 .!.Xdt =3! =6 x 2 =u x 2 +x 2 =u~ =3!=6 Fig.
Q)t5(x + I). we fmd from tables _ XII = {(l)(3)(S)_(III~ o .66(x3) 1 2/1 e . = 6 i = 1.the integrand is an odd tUnction ofx..41 10. Then. GO For n even..Q)t5(y+2} 80 . 10 Fil·510. _ 1+2+3+4+5+6 x·.r2/2a:dr. Fornodd. Therefore x" =0.42 >tI2" lW2" 10.37 x" = ax 21r Jr. .IGOx"e." even ~ " 7 =2 10.P)(l. .Q)t5(y+ 1)]·[ Pt5(yl)+{1 .1I Pn{n) = 2S and pyy=~e () 1 _y2/1 +~e 3 {y3)2/1 Fia·510.38 Let Xi be the outcomes of the ith die. Pn{n) =P6(nl)+(IQ)6(n+ 1) Py(y) =[Qt5(yI)+{I.2.r} =Qt5{x 1)+(1.p)t5(y+ I)l = (P+ Q2PQ)t5(y) + PQ6(y2} +(1.41 Px{x) =0.46(x)+0.43 px(..10.
Ro3 =0. the SNR improvement is 6 2 1010. 0 and x. sin 20 = . =027S3P".y) = (~x~x)~(K)x+K2K)XK~) =K)tT~. Hence. Hence. 10. a3'" 0. and tT xy Pxy I tT xy lSI or tTxtTy IPI S 1 Hence.I)' :: K)tT.. 1.1025.. Pxy 10. a2 From Eq.. 2atTxy ~ O. we have Taking inverse Fourier transfonn we get Pz (Z) = I ~2n{tT~ + tTn e {Z{I+y)]2/2(~+~) It is clear that 10..44 pz(Z) =Px(x)* Py(y) Taking Fourier transfonn of both sides.4tT~tT~ SO. Hence. But x Hence.!. .82Sa) +0.I)' cosO dO =0 Similarly.!. L021r sin 20 p( O}iO = _1 jo~Jr sin 20 dO = 0 In this case Rl\ = R21 R33 = P'" R)2 = R2) .308P". If K) is negative.308a3)]P". y =0 2 2 4" 2 + y2 = 1..• 10.R31 =Ro2 = 0562P".{ P'" ) =5.86) yields: a) . y:: K)x + K2 = (xx)(y.0779 =[1(O... . we obtain = ml =02883. the discriminant of this quadratic in a must be nonpositive. = tT xy tTxtTy :: KltT~ / K)tT~ =1 if K) is positive. Substituting these values in Eq..52 When y = K)x + K2 tT. But tT x and tTy are both positive. R)3 .63 dB. that is. tT xy = X Y.6J =Xy . JO. Ro) =OJ2SP". [a(x.53 =1 x = r21r cosOp(O}iO= _1 r21r Jo 2"JO tT.027S3P". R23 =R32 .KftT.51 z=x + y and C7! • C7! + ~ ::2 . For any real a. 81 . or a2tT~ +tT. that is: 4tT~ .562a2 +0. (10. is negative.. y are uncorrelated. tT.. cos Osin 0 =. Hence. Hence.. x and y are not independent. (10.x)(yy)] ~ 0.87).
Fia. SI1. (\00 =~ Jjro1oo cos( at + 01Jtl). the amplitudes of all sample functions is b.12 Ensemble statistics varies with I. f'la.14 A LA tta ::3cos(~\ +8)COs(tJJt2 +8) A2 83 . This can be seen by fmding  x(t) =Acos(at+O) =A Jo cos(~+O)P(tl))dtl) This is 1 function of t.Chapter 11 11.. lOOt Hence..I3 11.. This is not the case at other values of t. Hence. Fla.2 COs(U\ +0)COs(U2 +0). att =0.11 This is clearly a nonstationary process..12 11.14 x(t) = acos(~ + 0) x(t) == acos(~ + 0) ~ acos(~ + 0) =cos(ox + 0) tAaP1(a)da =[cos(~ + 0)/2AJ I~A ada =0 Rx (t\.11 11. the statistics clearly depend on t. SU. ~. COs(U\ +0)Cos(~2 +0)12 2 =cos(ox\ + 0)Cos(tJJt2 + 0) A a A Fia·511. ~+==::. . amplitudes of all sample ftmctions are zero at same instants (one is shown with 1 dotted lind). 511.13 This is clearly a nonstationary process since its statistics depend on t.. For example. I !~  e:=. the process is nonstationary.t2)·. For example.
Since x(t) = and Rx (tlo t2) =1. 84 .1' t~ ° t'+ not equal to the ensemble mean (x ".s. %(t)".: Fla. • I PC"') t IS~ . 511..S. Rx (tltt2) =KK =K2 = JI K2 P{K~'" lJI K2dK". 11.17 (b) x(t) (c) =K = ° j.j~~I _ _ _ _ _ _ _ _ ___________ .' acos(mc' +8)". 511.. cos(a1ct+8) =0 a (c) Rx(tl.S.3 (b) x(t)'.. 3 (e) The process is not ergodic since the time mean of each sample function is different from that of the other and it is (d) ~ ~. a = L2 a 2 p{a)da.a2 cos(mctl +8)COs(Ct)ct2 +8) =i{cosmC(tlt2)+co~mC(tl +/2)+28]} 1\ . =iCOSa1c(/lt2)+ 2~f:1f co~mC(tl t2)+28}i8 ".lcOSCt)c(tl/2) 3 (d) The process is W.S.0) (f) 2 • x =Rx O=3 1 11.. But a'" 0 2 11.18 x(t) =acos(Ct)ct + 8) 2'1 a ". a=O.16 Hence.11. l I 2 c3 The process is W.t2) ./' "'==" Fig.15 . b 3 2' 2 1a 4 Also.S. aI +b =It +b.18 A ~ t..12 4 3 2 3 Rx(llh) =(all +b)(at2 +b) =a 2tlt2 +a(t lb+t2b)+b =a 2tlt2 +~(tl +t2)+b 2 =i3 tlt2 +b 2 x(t)".
.(O) rect(. Prob«n+ l)nodes] Prob{state chanle at even nmnber of nodes) + PrOb(1I nodes) Prob(State changes at eveen nmnber of nodes) The number of ways in which changes at K nodes out of N nodes cx:cur is (~).). I) .Prob«II + 1)nodes] =(II + 1) .(O) sinc (2/rBr) andS" (ClJ) = R. 6. Xl and x(t + r) = x2 Then. 2B 4/rB toB Hz.. I) .. R...P"21xl (III))J ..I) P"I"2(I.(r) = lim x(t)x{t+r) f+OO Asr + co. R.22 (XI±X2) =Xl +x2 +2Xlx2:ii!O..4)"+' . Prcx:esses in (b).[( S+'XO6)O(O. x(t) is a white process bandlimited .P"I X (1.Px2 1 I(III)] x =2Px1 (l)[ P"21xl (111) (1.... etc. 8S ..(O) Hence.. (0.. and (g) violate th~ conditions.(r~ (b) R.1)] =2P"1 (1)[ P"21xl (111).Px1x2 (I. Hence.(r) and x\2 ZX22 =R. 11.( 1') = R.. In this case.. +(~+'XO.P"I"2(I.1) But because of symmetry of 1 and 0.24 R" (t) = PXiX2 (I. Hence.P"I"2 (I. Hence. nonnegative and even function ofQ}. 2(I.(O)="3 11. R.. 0. there are at least"nodcs and a possibility of (n+ I) • nodes Prob«n+ I)nodes] =rIITb =. 1) + P"I"2 (1.!.PX1X2 (1.. 2 t (f) x ..23 R.x2 = R. f+ao lim R. Time means of e&l:h sample function is different and is not equal to the ensemble mean..4)"1 + ](~ II)+ .21 (a). 2B Since Rx(~)= 2B °for aUn exceptn . x.. 7b The event (x2 = llx I =1) can cx:cur if there are N nodes and no state change at any node or state change at only 2 nodes or state change at only 4 nodes.. I).. Rz ( 1') is a wavefonn bandlimited to B Hz and according to Eq. PSD is always a real.!!..( r) =0 for r = ±~ and its Fourier transfonnSz(Q}) is bandlimited to B Hz.:ao 2B :E R.(e) The process is not ergodic.(r) =x(t)x(t + r). (c).2P"21xl (111)1 Consider the casen16 <I~ «n+ 1)16. 1). x{t) andx(t+r) become independent.II lb lb Prob(n nodes) .!.1.1) ... Hence.I) and P"I X 2 2 and Rx (r) =2[ P"I X (I. 11.(O):ii!IR.. (a) Let x(t) ..1 Ob Rx(r) = . P"llxl (111) . P"21x1 (111) . so lim Rx(r) =x(t)x(t+r) = (XXx) = x f+ao 2 11. (d)...6)2(O.(~)SinC (2/rBrn). and (e) are valid PSDs. Xl +x2 :ii!± 2x lx2 222 22 But. Others are not valid PSDs.
. a¥. From the timedifferentiation property.44+ 024 11 =0. f' .aI' .2.6)2(0. 16 <1~<3lb (n=2) • .~) M 1b 1b 2lb I~<lb 1b <I~ < 2lb (n=O) (n·.4)" +(~XO.4)"2 +..[(3)(0. 86 ..I) =0. SI1.x / dr 2 is a sequence of impulses as shown in Fig..'" 'ta.6)°(0....24..J andRx (r)=2Px1 Ix l(*)1 Thisyields Rx (r)=1.. I _ •.0481:i and soon . 1·2 d't: •• 2~ I 0. SII..l4 The PSD can be found by differentiating Rx (r) twice...J360. .• . The second derivative d 2 R..a.12 (n+ 1.
) . Hence. Rm(T) =Rm(O} sinc (2Bt) and Rm(2~)=0.I'+1 (ak a. Ro .lhl (1..028~ eflAl7b +e.ahl (1)(1) a Because of duobinary rule.29.. Hence. 4 4 2 2 RI :.5 ak :..15 Because Sm(cu) is a white process bandlimited to B.. Similarly.4 + 1.T) (:) e...44(efAl7b + e.576cos2Ct116 + 0. we have _tU2Sx (CtI):.jAlT . (I)Pah211..2 Rm(O):.. the neiihboring pulses must have the same polarities.. Pa• ah2 (1.1 =a. (III) . all Nyquist sample are uncorrelated.!. al x Rn Sy(w) = 11.!. I.42JicostU16 1.. PI •a. +l) == Pat (I)PI••11a.+(I). recalling that 6(1'. Pa..[2.. by duobinary rule.] Tb and Sx(tU) =~[2.. Thus.!. .j2A17b )+..COS2tU16 +.1152 cos3Ct1Tb+. Hence. o 4 4 '\2 Ro =al =(1)2 .1) =116 87 . =0 n ~ 1 and whereak is the kthNyquist sample. akak+1 ak ahl Because ai and ak+l each can take 3 values (0.: . I). ... + 02(..b [2. buu. Hence.. n=±l. (t)(I)PI • I .+1 (1.!.. ±2..:(I)1.. Ip(cut [Ro + fRm cosnCtlo16j Tb "..16 . Pa. i Hence. + (_1)2. (~1) '"' Similarly.) = . the double sum on the righthand side of the above equation has 9 terms out of which only 4 are nonzero. ±3. PI • a.ahIPI.] =.jAl7b ) . RI = i i(i)= 1~ Also R2 . andak+2 are the pulse strengths separated by one time slot.+ 2(1.:1'22 Rx(T) ++ (jtU)2 Sx(tU) = tU 2Sx(tU).1) = i iG) ..1) + (I)(I)P • ahl (1)(1) + (1)(I)Pa• a.!.!..ak+. R) ..1) =L La.+ 1(1.!. 2BRm(O)jP(w)2j since Tb = 2~ For duobinary Pat (1) =Pat (1) =025 and Pat (0) =0.. we have the same four terms as before.. This shows thatx(t>{t + 2~)== Rm(2~) = 0 Sy(tU) '"' Thus.1) .+1(IXl)(l)(l)PI.=2 =Rm(O).4 +2.+J. from Eq.!.)] 16tU '\ S 25 125 11.p~). 11. Now. akak+2 In this case.cos3tUTb __I_COS4tUTb+ .88costU16 0....
pulses occuPYa& fraction of time.( "':..28 . we can show that p•••• +2 (1. In the limit as &+ 0. 1!.29) and noting that R" is Hence. PX21x I (h/h) =Prob(XI lie in the hatched reaion) . ahPx2 lx) (hjh). a{ I_I. Sl1. Rx (r) becomes an impulse of strength a. h2a&PX2 1xI (hjh). Rx( r) . !:.ahl . 0 for n ~ 2. Sy(w) =Ip(w)1 11.. consider the rangel~ < &. .h) 5 h2 PX (h)PX1Ixl (hjh) ) Since there are a pulses/second.) For halfwidth rectangular pulse 11.28a) Rx(r) =rr%I.@(r)+a 2 88 Fig.".) 1. PX2 (h) =a& Rx( r) =a1h& '"' a 2 Hence.0 (Fig.. i k i k +1 == (2Q_l)2 2 7b nel Hence. This means XI lies on one of the impulses. '~~'G(I + <os ... Px2 1xI (hjh).28b).28 [1+ 2(2Ql)2(fcosnw7b)il J Approximate impulses by rectangular pulses each of height h and width &such thath& =1 and &. If XI lies in the hatched interval. out of which only one is nonzero (corresponding lOxl =h.'k+2 (1. x2 falls on the same pulse. Hence.. we _in Sy( .. G & and Rx(r)=an(I.)] '~t ...1) . Rn =a. Xl and x2 become independent....R2"' 0 Using a similar procedure. . Now.. Hence.. When r > G. Hence..: )andSy(tv) . from Eq.In a similar way... there will only be 4 tenns in the summation.. SI1. I~ Thus.... we can show that R" an even function or. Mark off an interval ofr from the edge of this impulse (see fig.. ~ SinC 2( W1 )cos2( w: ) =(I)Q+(I)(1Q)=2Q1 Ro'" ai _(1)2 Q+(_1)2{I_Q)= I Because all digits are independent.) Since Rx( r) is an even function of r. t Sinc (tv.!.1) =P'..l"2PxtX (Xl%2) l XI X2 Since x I and x2 can take only two values hand 0. Rx(r)=a4(r) I~=O... Rx(r) .27 ak . ) p(tv) .r. Rx( r) = h2 PXtX1 (h. (11. given that xI = h.. x2'" h). PXt (h) = a& and Rx( r) . Px2 1xI (hjh) is the Prob(X2) =h.. Hence. S 11.
or there may be an mutually exclusive possibilities. Hence PX2(X21xI =xI)=eftrpX 2(X2!XI =xI. r9l0andR. %1)dcldr2 =%1). either there may be no amplitude shift(X2 ampJitude shift(X2 9l XI)' We can therefore express PX2 (%2/XI = %1) as PX2 (%2/XI = %1) =PX2 (x2lxl = x\t no amplitude shift)p(no amplitude shift) + PX2 (x2lxl = Xl. x2 amplitudes before and after a shift are independent.210 The process in this problem represents the model for the thermal noise in conductors. The average number of changes or shifts in amplitudes are fJper second...po(r) =l_eftr . SI1.28. It can be shown that this process is stationary of order 2. 511.210 The number of amplitude shifts are given to have Poisson distribution. Hence. PX2 (x2lxl I: (3) ¢ x I' Moreover. and the number of changes are Poissondistributed. no amplitude shift) =4(X2 XI) whenever there are one or more shifts involved. Hence. p(%. the product x(t )x(t + r) is equally likely to be h and . PX2(X2IxI =xI. x(t + r) are both nonzero. The probability of" shifts in r seconds is given by Pi(r)  (P:t eftr where there are on the average fJ shifts per second. x2 = XI and the probability density ofx2 is concentrated at the single value XI' This is obviously an impulse located at x2 =XI' Thus.210.h Hence.. The f1l'Storder probability density of the process is p(%. where po(r)_eftr The probability of amplitude shiftl. amplitude shift) (2) when there is no shift. t) can be expressed as p( %).29 In this case the autoc:OlTelation function at r Rx(r)O.11. 2 But for r > 0 whenever 2 • x(t). in general. 11. t).(r)=a6(r) =0 remain same as in Prob 11. we are given that the (4) X\' amplitude shift) =PX2 (X2) =p(x) 89 . We have . I: f: %1%2PxtZl (%It%2)taldr2 (1) %1%2Px\ (%I)PX2 (%2/%1 . amplitude shift)P(amplitude shift) t Fla. J:J: To calculate PX2 (%2IXI Rx( r) . The probability of no shifts is obviously Po ( r) . A typical sample fUnction of this process is shown in Fig. The amplitude after a shift is independent of the amplitude prior to the shift. The signalx(t) changes abruptly in amplitude at random instants. . we observe that in r seconds (interval between XI and x2)' there are two =XI). no amplitude shift) + (teftr )PX 2(X2/XI =xI.
0(X2 XI)+(1ePr )PX 2(X2) aepr{0(X2 xl)+(ePr 1) Px 2(X2)] Substituting this equation in Eq.31 For any real numbero.where PX2 (X2) is the fU'Storder probability density of the process. 11. This is obviously p( x).20xy ~ 0 Therefore the discriminant of the quadratic in 0 must be nonpositive .. (Q. Rxy( r) = ABcos(mot +. ~~~~ Rv( r) =[2x(t) + 3y(t)I2x(t + r)+3y(t + r)] .y2 or (iY)2x2 y2 < Now. (5) becomes x=0 and Rx(r) === x2e. Rx(r)+ Ry(r) + Rxy(r) +Ryx(r) .. Substituting Eqs.y)2 ~ 0 + y2 .]) 90 .. Hence.. Rx(r) + Ry(r) since x( t) and y(t) are independent. epr[J:I:xlxlPX\ (XI)6(X2 XI)ctrldr2 + I:I:xlx2(eP == epr[J: x? Px\ (XI)dr1 +(eP t)PX\ (XI)PX2 (Xl)dr\dr2] (5) r l)C xlPx\ (XI)drIJ: x2PX2 (X2)dr2] =epr [x2 +(e pr t)x2] where xand xl are the mean and the meansquare value of the process. we have Rx(r)=x2e~rl and 11.( r} "" u(t)u(t + r). we get PX2 (x2lxl = XI)· e.. [x(t) + y(t}Ix(t + r}+y(t + r)] .]+co~nMo(1 + r)Mol +(nl).. 4Rx(r)+9Ry(r) since Rxy(r) = Ryx(r) =0 Ruv( r) .] =~B {CO~MOI +nMo(t + r)+(n+ t). (2). [x(t) + y(t)I2x(t + r)+ 3y(t + r)] =2Rx( r)+ 3Ry( r) Ruv( r) === Ruv( r} =2Rx( r) +3Ry( r) 11. (1).32 1\.. For a thermal noise Eq.33 ~.)co~nmo(t + r) +n.(2XY) 2 <4xl . identify x with x(t\) and y with y( t2) . and the result follows. 0 2 xl . we act pr Rx{r) =ePr I:I:xlx2PX\ (XI)[8(Xl xl)+(e pr t)PX2(Xl)}ir1dr2 r . (3) and (4) in Eq.pr r >0 Since autocorrelation is an even function of r.
R2 2c 2RlRI+(RJ+R2)2' RJ+R2 =a> =aJ2Rl~?C2+(Rl+R2)2 2KTR. we observe that all harmonics are incoherent. the principle of superposition applies to the PSD.R2 +(R.] =0 and Rxy( r) . + R2 Svo = RJ+R2 joCR.41 (a) SI(01) =2KTRI and S2(aJ) =2KTR2 Since the two sources are incoherent. R jt»R.. Hence follows the result.. 11. 0 11. O1ob =2nb is a r. uniformly distributed in the 16 range (0. Tb)... .v.. 91 .. Similarly.. =0 2"~ ~ .)d.R2 jt:£ R.co~naJo(1 + r) aJOI +(nl).34 X(I) =Co + l:CllcOSn01o(lb)+Oll 11=1 CD CD . 2"). If SOJ (aJ) and SOl (01) are the PSD's at the outputterminals due toSl(01) and S2(aJ) respectively.33.J'01ot+ n010(1 + r)+(n+ I). Sl1..41 H2(aJ)'" Rl R2(jt»RIC+I)+RI .] =_I (211' c.IR2C+RI +R2 S (01)= 2 KTRIRI and S (a»2KTR2Rf l l OJ a>2RlRJC2+(Rl+R2)l Ol O1 RlRJC +(RJ+R2)2 2KTR1R2{R J + Rl) Svo(m)=SOJ(aJ)+SOl(01).v.1 1 1 2 2 aJ RJ R2 C + (RJ + R2 ) (b) H(m)= lIjaC _1_+ R.. +R2)2 2KTR.. +Rl) =IH(01)112KTRIR2 RI+R2 (R.R2(RJ +R2) which is the same as that found in part (a). Using the argument in problem 11.CO~0101 +n010(1 + r)+(n+ 1). unifonnly distributed in the range (0. Co + LCII(naJolnaJob+Oll) .. Hence the autocorrelation function of Rx( 1') is the sum of autocorrelation function of each term. then (01t SI(aJ) andSo2(aJ) = 2(aJ)1 S2(aJ) SOJ (aJ) = 1 IH IH 2 where c vo c (c) (a) (b) Similarly. Fi.1 Since b is a r.
and Rx(t) =a2q2 + I+CI>2 t 2 xU) (fll[ q2 ]_ a 2q2 + aq2 ..' 4 Sn(td) . x(t}I: h{a}x(t + :a)da  = I: h{a)x(t)x(t + t a)ia . Also.( . 2KTR e"rlRCue r) 11...5t 92 . a +2mz 2c5(CI» Hence. I jDJRC+I andSlIVo (td) =2KTR l(jtdRC + I) and Rnvo (:) ..... Sy(a» IH(mt[a+2Ira 2c5(CI») . and is given by =n~ 11K 2[ SOQO.a)da .. 11.. Fig....) is the area under Snc (tV).. I I: h{a)Rx(:. .43 11..All x 5000] =1.1. 1 jCl>+: /H(a>t = q2 .h(:) (b) h{t) =!l.5t n(t) == "c(t)COSCI>ct + ns(t)sina>ct The PSD of"c (t) and. H(CI» R+joC j~ .and is given by n2 _ 2[.11..eI1rll(t}.41 y(t) . nl is the area under '..25 x 104 . ac5(:) + a 2 .13.!)] =1.. 11......and Sx(m). r r.AI n~ or n~ n~ '2 ...43 (a) We have found Rx(:) of impulse noise in Prob.5·1..2S x 104 . H(CI»'" 1. S 11. H{a»Sx{CI» In Fig.. SIl. 2lra 2lH(ot 3(CI»+a/H(tdt and Ry(:) =~l[Sy(m)]'" a2IH(O~2 + all{ :). Rx( r) and Sxy (CI» .( t) are identical. I: h(a)x(t ..AI x 104 + 10 2· 2 (~.. They are shown in Fig.A1 2 2 2".eWr I+CI>2:2 2r t ttt t tt tt mt 1 H(w) y(t) h(l) c Fl•.a)da Rxy(:) =x(t)y(t +:) .. h{ r).. Stt.5P..28 Rx{r) .\J + .
Jtc. viz.n1er ..1) is the same.54 <a) Hop () W =Sm () +Sn () =.5.& No =J10 "" 1. In all the three cases.1 2 +10 9+ 0.. For the 3 center ftequencics Sflc (It) ~or S. fHa* ~. Thus in all 3 cases n~ = n~ = 1. 11.52 We follow a procedure similar to that of the solution of Prob.52 q5 It.53 11.. the area under SIIc (0.10 (d) Noise power at the output of the filter is N o =21r 1 fao Sm(O))Sn(O. f .949 sec...t except that the center frequencies are different.r _ 6 6 641 +60 2 = 0. a reasonable value of timedelay required to make this filter . 1.~. f"'~ . (It)) ] are shown in Fig.eJiO~1 2.21r Jao It)2 + 10 tv:: 21rJiO To 0) lao _ aD  110 3 The signal power at the output and the input are identical 6 2 dO) "" 1 SI :: So s .52..r 5k lo~ 'OK 15 K 'tJ~ ".I) '" _ 1 roo 6 d 6 tanI ao Sm (0) + Sn (0.25)( 104 ~ .10 (e) The time constant is ~. Fil.1 2 (b) h Of' (I) =::b.10 realizable is ~ :: 0. .~· /0 E k.. Sl1....OK (~V\t.054 3 93 .25)( 104 ~..+ 6 6 0) 0) Sm(w) ?:. Fll·511.11. S t t . Hence.1) tv .1 lao 21r aD 9 + It) SNR:: .
0..55 94 ..r:. = 1 =1.67) 1 S. A reasonable value of timedelay required to make this fiher realizable is 3 x 0306 .838 No 0. (d) Noise power at the output of the filter is No =_I Sm(t»)Sn(t») dt» = _I 00 32 dt» ... =So"'00 4 ~t»=1 211' 4+t» Sn ¥.67 (b) "'(/)'" !6(t)+ 8.. + 64 t»2 +64 1 [I 53. 9' + t»2 + 10.4 32 t»l..l66jtj (e) The time constant of the filter is 0. 0.. 9t»2 +96 .4 11 5 5 •  () • H ( ) Sm{t») op t» =Sm(t»)+Sn(t») =_4_+ 'oJ2 + 4 . SI1..306 sec.33] .918 sec..544 211' Sm(t»)+Sn(t») 211' ~t»2+10.544 r. 1 The signal power is 1 Fig..163e 3.
.96 4a Ie 6 a a Sr = .8 . Fla.11 5 no (IV) =5 n (IV)IHd (IVt = 10.. . 8000.. 1 Ir 106rec:{~) \l Hc(e»/1 dOJ"\1 1 Ir t Sm(e»II a Ie 2 J10 ra 26.. ~B ..8'" 26...96 x 51 . .68 x 104 26. ..10 (IV +a tIV 2 2 a2 r = 102 a =>50 1r a (e»3 + a2IV ..:::2.11 5 .1°(4000) . 4)( 107 (c) 5.. .8 .7 x 109 Ir GO Hence. r = No JJB = 5· 4 =>5.96 x 10 a = 68.96 x 106dOJ'" 26.10 ( No""!" IV2a~a2) 10 a = 80001r Ir 0 r 1O.11 ~lW) H+w.65 x 109 Jo 11.. 10.. 1000 .37 x 103 => m2(t}. r Sm (iI1 )dOJ .. 4)( 10 10. Hc(e») =103 • Hence.::.' )( 4000 ~ 51 =0. 5r = IHc(e»)! 5r =_1 . . a 50 =.8 [2 )( 800011''1 =8 x 104 ~ . r 26.. 21S..7 x 109 a 106 . (12.11 S 51 (a) 30 dB .!..7).37 x 103 3 103 But So(/)" m(t}.08 5. 10 211' 11.. SI1.. No ... .2 m2(t} =3. m2 . IHc(IVt Sr and 1O8 5 r =4 x 104 => 5r =4 x 10" 95 . 2 = 8 x 10" Also.. _1 21r fao {Jdil1'" pa .  Also... 21S..96 1 e» tan air a 0 Ir Jo la = 2.96 x 106( OJ +a 10 2 6} 2 iI1 26. 32 x 107 3 0 3 l 35dB=3162=~= No 50 32 x 107 =3.=r o =JJ~' 5· 5· r = 1000 = 2 x 10.. Hence.10 )(4000 ' (b) From Eq..Chapter 12 11.96 x 106 and5m(iI1) =26.!.. .
Hence.11.PF~9 ~"'l.13 Let the signals ml(t) and ml(l) be transmitted over the same band by carriers of the same trequency (t» c). !!!. V\.:::L:r=L (e) For tone moduiatioDK // "'p 2 No 2+1 3 96 . but in phase quadrature. the output of the lower demodulator is ml (t) +"* ns (I) These are similar to the outputs obtained for DSBSC on page 535.. 512..10 x 8000 =4 x 107 (e) Sj =IHc(mt ST =108 ST =4 x 104 ~ ST =4 x 104 11. the received signal is [Jim) (I) + nc(I)]cosmcl + [Jiml (I) + ns(I)]sinl1l cl Eliminating the high frequency tenns.. 11.13 The bandpass noise over the channel is nc(l)cOSl1lcl + ns(l) sin l1I cl. Hence.PF Fic. we get the output of the upper lowpass filter as ml(t)+ *nc(l) Similarly.21 (a) So No =1000 = Sj = ~ Sj 1010 x 4000 ~S I =4 x 104 (b) No =JJB = 10.. we have ~ =r for both QAM No channels. The two transmitted signals are J2[ ml (t) cost» cl + ml (I) sin 11I cl ] Milot: ) ... A A 1 S I· 1 = m =2 andforp=l• .{ot:) L.e.14 (a) p = [m(t)]min ..
42 m2 .§..42(I+p)2.sz.O"~ 2 2 8 Hence.42 S Sj =and mp =pA.~=.. These results are derived for tone modulation and for p = 1(the case most favorable for AM). letm(t) .24..4(1 + p cos m". 0 ) Prob(E ~ A) = .J. S No = (05)2 9+(05)2 .:...SSB(t) =m{t}cost»el + m.01 160"2 +~ 17 m m =.)I S S 2... Hence.A2 . Sp = p2.1 sin OJet = pA cOs(OJe . pAcost»".!.15 (a> From Prob..04 2 =I).S ..AlB 2. 0.79dB . p2A2/2 Sj Hence.4(1+ p)and Sp = .. 12. and the camer amplitude A'"' mp =40"m (for p m2 .2+.JL 10 =05.42 = p2. 40"".:2'= .L 2 No .42 For AM . ..42 + p2.fipA)2=2p2.... O"~ (assuminl iii .AlB 2~ 97 .A O"n 0"2 O"n /ltrIl 2 ::.B 2JJB p2.t ·cosOJet = *[COs(OJe + 4I".1..60S)=9. and SSBSC has 9dB superiority over AM.UA cosm". ~=8 ~ 4. .(/) sin t»el ::.42 and mp = ~+~ ='VJ.= . 2 2 2 2 mp =.A 2 4 4 2 ~2 ~2 No Hence.AIB 4 ForSSBSC . mp ...i 2 ."". ie..2. = 2 No K +p and when p (b) When p >= I. for p ::.41".. 11..Sp 4.42+{p2.. S 17 0"2 17(40"2) 17 rThresh =_1 =.)t + cos(me .AM(t) =. the peak power Sp=(.. Sp(2+p2) Sj= 4(I+p)2 So ~ and No'"" . ...t ) cos OJet .. . /ipAcosOJ".42/2)·JJB =\2+p2~4(I+P)2. 12. 1. =.~ =(4. !R. =~ No I(UJB Hence.16 For tone modulation.. .AlB Sp WhereSj=.17 For 40" loading.42 p2.. For BSBSC.AIB = Therefore.42 +m2 2 8.42and~=r"'~= ..t COSOJci + pA sin m". for a given peak power (liven Sp) DSBSC has 6dB superiority. Ii En2eEII212~dElI '"' e.. For Gaussian m(/).::2.. =..AIB (p2 i Sp(2+p2) 1 Under best condition.t.OJ"....)1 J S.42 5=+=+ .~ t:: .60Sand Sj'" 20"n .42 and ~'"'r=j = E . =L No 9+1 . DSB(t) . •L 36 11.
m(t) =(at CtJ \ sinCtJlt +a2 CtJ 2 sinw2t). =al +a2 =alCtJl +a2CtJ2 2" (50 /N o)PM (5o/No)FM = (21d1}2m.y 631 =. y (a) Also.34 m(t) =al coswlt + a2 c05w2'.0199 5 631 12.4197 x 103 kIm / {30'm) =p ::> 2 = k30. This yieldsc05 2 wot = 2sin2 CtJot ·2 2. 1l. w~(al +a2)2 2 3(alWt+ a2CtJ2) .  ~W 21dJ 50 =a2k}m2(t)=a2k}a~ =(104) (20. mp mj.!L = 28dB = 631.OOOn 2trB 2 (b) p . Hence.ooon)2 .ooon ::>"10'm =20.=0.12x9 =473.25 dJB =~=>5i =yvfB =473.32 mp = B. 1 (2 or 1SID CtJo'= SID CtJ ot::>SIDCtJ ot='J3' COSCtJ ot=V3 and 12. 5 No Therefore.2CtJ 0 cosCtJ ot sin 2 CtJot] For a maximum 0' m(t) =O.2 . 4n2 (c) No = R..25x2 x 10.31 ...12.10 x 15000 = 1.33 m(t) = cos3 CtJol and mp = 1 m(t) =3CtJ 0 C052 CtJ otsin CtJ ot and m(t) = 3CtJo[CtJ o C052 CtJ cosCtJot . 3mj.
the normalized PSD's is.2Jo ~oh2ul dM . B=0. I + (II10)2k ro 1 dl = 31ao .42a) } 12. There should be 41r2 in the denominator (see below)..1:41r21 2Sm (2 tt/)cU' From Eq... B =03420' x =0.12...(w) =W2S m(M)..l«11_1_dx + x 2k ao 0 2 1.0. =1.4SO'.!!!.. C[m(t)t tit . B =039SO' ~:? =0.' sm(~) Ir tr 0 • (31r) sm 2k J 2ksin(2:) The definite integrals are found from integral tables..1 So'.. 1: Sria(2Jf)cU' ..e_W2/2~ ].9S => W =2.9 =W =2. + I. S. 1+ (III..f1..  =(21C8)2 "" 2 ('II ~c» 12rr d(j) = 20'2.060'2 > W2 => PM superior 2 99 .99 w 3 =3. 'm 0 sin(31r/2k) 0 (3tr/2k) 3 0 12. Jo 20'  3 1 . (12.and p(co) =le P(W) W 2 /2 2 tI . Hence.. Jo 20'2 p(co)2.eGl2/2a2 20'2 If W =21C8..x2 10 «I1+x dx ri J... c • [ 2hm(*) /.37 S (M) m  ~eoh2u2 . If P(W) is the power within the p(W) =2 rW .0 )2k J.::eoh2t11 dw =2[1.36 2 8 m= ro 12 cU' J..35 Error in this problem.030'. 2 reo 2 2 0' 0' Hence.m2 .99~W=3. 2 2 sin(Ir/2k) 2 ( 1r/2k) 1 2 As k + <Xl B =I. then W2 bandW to W..4820' ~:? =0.
.!.=2. 0.2. Sincemp =30'..8. 00 /111 a? +a1 2 2 at +af Smce B =h.000 or(Y17l20) 212)(10 = 3 20 Y17I => y.!.9 12.) => P=0. and Sm(l) ..310 From Eq.40) p2 .540'2 < W2 => FM superior 3 w . . Y'Tlucsh '"' 20(P+ 1) So PThrcsh .5 164 =1.000) 7 So 12 = P y No 3 1 =... p2y =1.1.1 1 =721 20 ~ ==1.12)+6(1 + 12}] m 2 a2 a2 = J: Sm(l)4f == (a? + ai) 12 B~ =(fIX) f 2Sm(l)dl Q)= _2_[arll + alIi] =a?f?+a1fl.000'2 =W2 => PM and FM equal 2 m(t) =al COSCi)I' +a2 COS Ci)2t. (.2z 1 1 2 =3/J y=p y. (12.39 <a> 2 So 2 m 1 2 ..56 =0.8=3 P2Y =3(2) r 2 No mp 3 r= 218. 3 p2 = (1. (1) 3 1+ 1/9 3 1+ 1/16 where For tone modulation. )] (1) Tone modulation IP =. 1...8 x 3 . (30") 9 2 100 . 1 ·f(ada2}2(1t112)2+1 1 2 <(al/a2) + 1 3 ·fl 2 1 l+x2 or 1 + X Y < 3 11.11)+6(1 + 11))+1[6(1 . m2 For Gaussian modulation with 30' . m2 mp =~ =1..47 1+0. 0.PM is superior to FM ifff > that · .5 Required increase in Y = 242.x =0. 218.38 3 W2 ...[ 2 3 1+ m2 mp . Also. 242.5 => fJ .loading '!'(L) p2 .95 x=0.7 dB 12.(721)2(164. ..!. +[6(1 .1} =2844 =34.4dB218.(Y17l20)2 Y17l = 10..164. 4 =. 2 (2) Gaussian with 30' . ~a?/?+alfl) al +a2 2 2 ' IS.479 =1. .53 dB No 3 3 (b)  (40 dB == 10.loading.loading (3) Gaussian with 40' .547 => P. m 0'2 and23. 164..5 2 mp.
. m For GaUSSlan moduI ' WI·th 4 a  Ioad'mg, 2 atlon
2
mp
1 =a 2 ==6 
2
(4a)
I
12.311 Let us fU'St analyze the L+R channel. In this case, the demodulator output signal, when passed through the 015 kHz (Iowpass) filter, is given by(L + R)' + no (I), whereS n (OJ) ="OJ [see Eq. (12.33)]. o A2 When this signal is passed through the deemphasis filter Hd(QJ) = OJ) and the output noise power
N~
=: If
,
"
2
jQJ+tV)
,the signal is restored to (L+R)
N~ is
given by
rWI 12 ..AI rW QJttV2 Jo Hd(tV) Stlo (m)dQJ ==  2 Jo 2 2dtV =..AIQJt [ W OJ) tan 1 2 1rIf QJ + tV I 1rIf QJ 1 Let us now consider the (LR) channel. LetQJ, = 2". x 38,000 andQJI = 2". x2100. The received signal is FM demodulated (Fig. 5.19c). The PSD of the noise at the output of the FM demodulator isSno (tV) == ..AItV 2 J A2 [see Eq. (12.33)] The output of the FM demodulator is separated
1
w]
into ( L + R)' over 015 kHz and (L  R)' cos 01 ,t over the band 38± 1S or 23 kHz 10 S3 kHz. Let US consider the signal over this passband, where the noise can be expressed as nc(/)cOSOJ,t + ns(/)sinQJcl. The signal is (L  R)' cosm,'. Hence, the received signal is[(L  R)' + nc(/)]cOSOJ c + ns(/)sinQJcl. This signal is multiplied by2cosmcl and then lowpassfiltered to yield the output(L Sfic (01) == Sn(m+OJc)+Sn(mQJ,) ==
R)' + nc(/).
But
~ [(01+01,)2 +(mOJ c )2]
jOJ+ml
When this signal is passed through deemphasis filter Hd(m) = the output noise power N:; is given by
. 4)1
,the signal is restored to (LR) and
" 1 rWI ~ (W[ NO==Jo Hd(m)12Sn (OJ)dm=2Jo (QJ+QJc) 2+(0101,) 2J
". ; 1rIf
2
mf
01 +011
2dQJ
2..A1mr mof2
[w + lO~

01)
tVr tan 1 
w]
W = 2". x 15,000
01)
Hence, the (LR) channel is noisier than (L+R) channel by faaor!jf, liven by
o
N~ =
Substituting B '"' 15,000,
N"
2(W + OJ: +QJt tan)~)
01 1
2 2(B+ f,2 + 1.
W 01) tan
t(
lOl
Ii
;;
W)
""'
B  Jl tan '
t( II
tanI..!) II
B)
I, == 38.000, II == 2100 in this equation yields:
~ N~
Nil
== 166.16 = 222 dB.
101
12.41
L = M n => n =IOiM L
!..£. = 3L2 m2(t)
No
m2 p
3M"(:': )
11.41
No For uniform distribution
 2 m 1 =I"'Pm2dm 2m
p
III
~ =55 dB = 316200
P
=mp 3
(0) 316200
I
o =3{2)'.[:,: )
=3(2)2n(i)=2 2n
Fig. SJ1.41
2n =18.27 Since n must be an integer, choose n = 10 and L = 1024
3 B PCM =2nB = 90 MHz (assuming bipolar signaling) (c) To increase the SNR by 6 dB, increase n byl, that isn =II. Then the new bandwidth of transmission is 22 x 4.5 = 99 MHz.
No
(b)
~ =3(2)20 1. =1.048576)( 106 ~ 60.17 dB.
12.43
=2 BnEp , Ep = 2)( 105, B =4000, Sj =2 )( 4000)( 8 )( 2 x 105 = 128 S 128 A2 r = _I = =2.56)( hi 7
Sj
n
=8
~ t;) Q.m = 7.569 )( 109
B",
..)JB
2)(625)(10 )(4000
=nB =8)( 8000 =64 kHz (assuming bipolar line code)
(0)
"ff,;= 1+4(22n1)~t;) ;;;
S
3(2)2n
( m2
}
where
So.
rr =~2Ep,. V; S (1) it; 1)Q(J32) 9"
..)J
2x2xl0 =J2S6 =.ffi 2 x62Sx 107 8 ;:21845= 43.4 dB.
5
3(2)16 = 1+4(216
(b) Ifpower is reduced by 10 dB, theny
=25.6, Q(J3.2) = Q(1.79) =0.0367 and
~=
No
3(2)16 (1.);:227.3.56dB. 1+4(216_ 1)Q(m) 9 102
The table below gives SNR for various values ofn under the reduced power.
(d)~~~~rr~~r~~~~
Hence, n =3 yields the optimum SNR. The bandwidth in this case is Bm 12.44
1 PE = P (correct detection over all K links) + smaller order terms == (1 p,)KI(I_ p~) == [I(K 1)p,Il p~] == 1 P~ (K 1)P,
(b)
=3)( 8000 =24 kHz.
So PE = p~ +(K 1)Pe r" 2S dB= 316.2,
r = 23 dB= 199.5
P, = Q(~3162/8) = Q(6.287) = 1.6)( 1010 p~ .. Q(~199.5/8) =Q(4.994) ... 3)( 107
PE .. 99)( 1.6)( 10 10 +3)( 107 .. 3.16)( 107
=P~
12.45
12.5~1
As noted on Pg. (570), the optimum tilters for OSBSC and SSB·SC can be obtained from Eqs. (12.83a) and (l2.83b), provided we SUbstitutei[Sm(t» + Q),,) + Sm(Q)  Q),,)] for S.(Q1} in these equations. Let
Sm(Q)) ... irSm(Q) + Q)t:)+Sm(Q) Q)t:)]
~(.,+.,:;, +a' + +a'] (.,.,:;,
a 2(Q)2 +Q)~ +a 2 )
.. alOGO",
(Q)2
+Q)~ +a 2 ) 4t»2Q)~
2
(1)
"c ..2",xIOS
We shall also require the power ofSm(t»).
I:; _I roo Sm(Q)):/Q) 2n Joo
We can simplify the evaluation of this integral by recognizing that the power of the modulated signal m(l)cosQ)"t is half the power ofm(t). Hence,
1 1 a tV a 1= Joo Sm(tV):/Q) =  1 tao a 2dQ) = tan _I  lao =2 2n 00 2nQ)2 + a 211' a 0 4 We shall use the POE system shown in Fig. 12.19
2
(2)
103
(a) For this system IHp(t» t =__S~TLS. (12.3 Hence. (12.83b) \/1 IHd(m" (b) The output signal is Gm(t) = Gl J: JSm(m)df 10 4 ST .. Sn(OJ)dOJ 1 f: £.. JSm(OJ)dOJ tr/ JSm(t») (4) .1.S. (1)../i/Sm{m) a l m2(t) =3 10 $.51 12.% Similar to Prob. we have \/2 STJ"""I/S"m(m) _ lol/~ ~ 1 f't. (4) above.. Hence. 2. AI50from Eq. r S.~~)J. where n~ '" n 3tr 2  2 =. Hence ala (10 So =2= 2)2 (3ooo/r) 2 3/r =20 To fmd the output noise power No. (3) Because H c(t») and Sn (t») are constants..jSm(m) = • Also.I"::\ IHp (mil '"' J: "/Sm(t»)df .(e»)de» lOr S. Sn{t») the noise PSD at the output of Hd(OJ) is Sn(t») =Sn(m)IHd(m~ A 2 = 2xlO.. the improvement ratio is 104 .S. 12.if:"Sm(m)dt» wbereSm(w) is found in Eq. . Hence. JSm(m)dm /r ..16 J: . So" We have already found the powerofm{/) to be2(a/4) .(e»)de» 3tr/20 12.n(t»~)=/ S=m.(/)/Ii and No and 12 [see Eq. = N.=(t» ) (\Ir«' Sm(t»)Sn(t») df IHe t»I1J_ IHe(t»)I... the output noise power is n(. we observe that the noise signal with PSD Sn( OJ) =2 x 109 passes throu&h the deemphasis filter Hd(OJ) in Eq.
..b) 1b t tc! ) Fig. b and and p2 = p.N and lOS . ~ Po (I) =A(l_e.... the output is the integral of p(t).. Po(T.. (p. t Ap ..) =A(I_e~/RC) 2 0" n =2". ~Tp f'0 lt:~ outf't t ofWt ~LJ) ~I+U Clf' eutpllt of''*'''f. .. It" r ft\ 4IItt.. the output h(t) =u(t) . Hence.) =AT . 13. ..Nrc dO) ..u{t 7j.11 "Lt:.) .(t) vVlb/2 V¥ p This is exactly the value of p2 for the matched filter. For the integrate and dwnp filter (1&0)./ RC ) t . att =T".. '2 Jco 1+ 0)2 R2 C 2 = 4iiC l.. I>T".Chapter 13 13..12 The output Po{t) of this RC filter is p'lt:) ~ . Hence.) C. S13... which occurs atTb: ..11 If we apply 6{t} at the input of this filter. ATb ~ 1'. E n.. A(1_e~/RC~{tn)/RC The maximwn value ofPo{/) is Ap...) A Tb ~n~l4t t:.htd tc PLt:) t. . .Po(T.(Tb ) = A2Tl =2.
~ Observe that for the matched filter. Letting x =1bIRC. ~ PmlX " " . ~ (1_e. or 2 =RC 1j.22 Let CI be the cost of error when I is transmitted.2Epq The energy of p( t) is 1j.We now maximize p2 with respect to RC.22.26 Hence. times the pow~r of p( t) . we ha~e 2 4. Hence. 1 1.Ap~ao) Co:: COl P (Elm = 0) =COl Q ( .: : 13. Hence.r)2 x and This gives or and x == 1.21 P~IX =~f:[P(t)_q(t)]2dt= Ep+~/. SI3.Ap(T:ao) 106 .A 2 16 p=_. 2 2Ep ~ 2A21j.A 1j. and 13. Let the optimum threshold beao in Fig. and Co be the cost of error when 0 is transmitted. Similarly. . Then: q ::qoP(E1ml)qoQ( .816) 2.26 p~ :: (0.
SI3.(l} qo ]= laOAp =>ao u~ = u~ 2Ap In [P". The only difference is Pm{l} and Pm (0) are not 0.The average cost of an error is (1) For optimum threshold dC/doo =O.5.P (1) q Pm (0) Co =Pm(l) qo Q ( Pm(O) COl Q ( Apu: ) m + Apu~QO )+ and Hence.(I) CIO 107 .22.23 We follow the procedure in the solution ofProb. ao C .22 Hence. Hence. 13.(O) COl] P". to computedC/doo. In [P". Hence. we observe that P~ po and "r 4c Ap Y" Fig. 13. and But Hence. Hence.(O) COI P".
p( Q»))e jeur" ~ 2 p( Q))e:' jc»r" and 11(/) ..(O) COl ] 4 P (I) CIO . 2P(7b I) Qo =i[E3P Ep] =i[9Ep. Qo =c.52 Here.51 tf.2)~~J:~ ) p. Hence. This is shown in Fig.AlEp 2 The thresholds are ± E p /2 and p(elmo)=2~ E. SI3. But multiplication ofII(t) by a constant does not affect the performance.= 2 c.}2~~) p(e~d = p(el"... Hence we shall choose II( t ) to be p( 1b .. [3 p( Q)) .. respe~vely. H( Q») .52.iH~)+~~)+~~)] a1~~) 13. Also.t) rather than 2P( 7b I). This will also halve the threshold to Q o SI3. 108 ..But Hence. p( I) and q( t) arc identified with 3p( I) and p( I).Alln [P".51 s r Fia. 13.I)~ E.Ep]" 4Ep =2Ep..
53 For M = 2.069W BT =256.7 x 107 )( 256. But amplitude modulation doubles the bandwidth.000 =0.. =EbRb .J.2 x 108 This case is identical to MASK for M 7 10 =~J2~b )~ Eb" 2. 13.000 "" 1.4W =2 13.L ) 32 1023c.. =EbRb =2.7)( 107 S.and The energylbit is Eb . 9E +E P2 p =5Ep Hence.43 x 106 S. 5.I 2(31)~~OE =~ 1012 32"" Sx 107 =. c.Al.AI =2 x 108 For 256.7 x 107 x 256. 4.7 x 107 Sj For M =16 =Eb~ =2. Hence BT =256kHz 107 =~ J2~b )~ Eb =2.719 x 105 x 256...069W This yields Eb .719 x 105 Sj "" Eb~ =1. 5.000 bps the baseband transmission requires a minimum bandwidth 128 kHz.39W B T For M=32 = 256.000 "" 512 kHz 1012 32 P..43 x 106 x 256.000 )( 2 = 256 kHz 2 109 .000 =0.000 .A1 This yields Eb =1.54 For M '" 2 andc.
4275W In MPSK. B T = 256.000 =134W B T = 2'6.000 s 1012 16 64 kHz For M =32 Sj :: EbRb :: 5. the minimum bandwidth is equal to the number ofMary pulses/second.000 =0.67 x 106 x 256.For M =16 S. =EbRb =1.000 ='12 kHz 1012 32 110 . Hence.24 x 106 x 256.
1."'r . FiC·SI4._ ..11 The following signals represent 2 sets of Smutually orthogonal signals.. SI4..Il 111 Z"1: ..L I ~ ~ _.. e ve _.L 2.T tJ t.. I I ~ Fig.. t. \J'f: . \ffc.T ~ ~ ~ ~ ~ '1.:.Chapter 14 14..fTe '" :2.11 14.11 itl !l .
14.l.13 I) (1.O)isb[l+~sin"'oI] "To "'0 21r = To :l/vro ~/~~~r~ 1/v=e Fig.13 112 . SI4.
(ot: ) 'fII! It .4f) /) ...~) f.2 / '. I)~J ". Rx (1) =.2) ""0 (1..f~ ic) 14. ft... xl. e x2 e 1 =0'''3''1 .. Hence. R. ') ..a. tfFo FiR.5") b) The energy of each signal is: £3 = 9 + 4 + 9 + 16 + 1 10 =39 To c) F3· F4 14. R" (1) .. ~ Si +* \lTD wi r..14 ~ ~ .(O) == 1). ) "0 . "3 (1"1"2 (12 "2 (1 "3"2 The covariance matrix is: 2 K =[U (1"2"1 (1"3"1 "I U. X(l + 1) =x2 We wish to detennine = x(t + 2) =xl Since the process x(t) is Gaussian. a.::2.i...!. xlx3 = x(t) x(t +2) =R.." s x(t) x(t + 1) . ....(2) = ~ e so 113 . 1 (1"1"3 (12 "3 (1 "1"2 == 0'''2''1 =i.. x(t + 1) x(t + 2) .. Let X(I) xI. " J'bt t 'b 't'" "'t (3. 13(1) andf4(t) are orthogonal.... x. Also (1 "2"3 (1"1"3 =0'''3''2 . la.. SI4..L.. xl arejointl)' Qaussian with identical variance ( (12 == (12 "I "2 =(12 .11 =(68 +6+8 +0) =O.1 '3.14 (::2) "I) lJ 2 / .) I) it) "f If 1\ ~ . .X3 .X.. ft ...
. SJ4.31 PrO{nl < i) and p(CjmM) =PrO{nJ > ~a) Hence and Hence 114 . }_... • ..a.31 J i • t S...1 e 1 K= e J e2 e2 1 e and e 14.. 5"" FiR... P(CjmJ) = • ~ at .
52c) 14./2 J • c. ~..32 p( qmd =p( C. ) p( qmd =~ n 1 < i. t .M H [1.2~~)] [1. ...$ c.) [I~~)][I~iv)H~~)r P(C\m ~Inll i. S14. ..m .~Jh)] p( C\m J=t[ ~ Jb )] [ 2) Jh)] = 1 2 3q =1P(C)=t~Jb)[53~iN)] .+ r.. n2 > ...ms) =p(Clms) 4) p( Cl'"2) = p( C\m)) = p( C/m6) = p(C\m.) + P..~ltte. 2) ~a) = < 82 > and P( C) = p( qm....r4tc>r ~ r.1~t I~~t Fie.. (13....r.Jl 115 ...p.1.Hence the average pulse energy E is Also Hence (M2 1)a 2 Eb = Iog 2 M = 1210g2 M E Which agrees with the result in Eq.~ \. S.s&2. p( C.~ ..&.
.34 Because of symmetry.. has the smallest angle... the detector is a phase comparator. "s2. and Let E) = E2 =.3Q(JJ)] assummg Q I = ( ~) 25Q (~) «1 This perfonnance is considerably better than MASK in Prob..·. 14.= E 4 = • 5)= d d d) (2' 2' .. 2...'·sM . Since r· s" =. which yields P. d 2 or2 d • Nd 2 j • • . a" a" 14.·s)..fir cosO" is the angle between' and Sic . '2 • 1• • Then Fit. it is clear that we are to pick that signal Sic with which .34 116 . We now compare . the optimum receiver is the same as that in Fig... M. It chooses that signal which is at the smallest angle with ..8 with tenns 's omitted.....33 In this case.. Hence... In short. SI4.. iQ( rn [S. p(qmd =P(q m2)=·····= p(qmM) where M =2N 5·· !) =: M :. constants 's are same for k =1.31. . EM .M =1.The average pulse energy E is E=i({(i)' + {(i)' +(~)'])=~' (i)} and P. . N:.7SQ (~02~Eb ) for M =8 14.. 14.
l < p)= 12~ ~..M =1.35 J P(Cimo)= Prob an. TJq11ln(K/B) Jl...\I J o e[<qld)l+qU""]tiq dq 2 I ql tIn(K/B) 117 ....!.) + ~ 2~E/.and p(C} =P(C\ml) P../2 J P{ C) =~ p( qmo) + p( q~) + p(C\mI) ± ± _H2~*J]2~Pi2J Also Hence 1 P. M = 2 N • Hence.p)l = I~ .[n..AI 2E+1n2 14.35 . SI4. and 14..) =!'rob.u =IP(C) =1[12~J~ Jr EO Here.P(C) p(C) =~(p(qml)+ p(q'"2)] P(C\ml) __ lJ e[<qldY+tll'''''tiqltiq2 1 JI...M=1P(C}=2 ~ ~ 2~E/.AI 2E_1n2 )~ .36 P. > (d .~ Fli.\j RI 3d =.AI P(mo) d P =.) ~ p(cjm.I n . I ~ =~1n2+! 2d 2 p(cjm.. each symbol carries the infonnation 1082 M Hence N bits .+ 2d p(md 2 ~ .
~ Jtt }I Note that d d I. Fla.\ "2. .1.2' 5) = "2.2' d d 118 . . 514. 51 ="2..'" • Fig.. • Sa t.:a. .) c1.lt) ~4 lb 1 • .36 14.\ "2..and 't.. ~ ~ tI>. I t. 514.37 (Q rr. ( /.37.
4d =02d JJ v\I 2 2 O'~ ~=JSE 20'" 4JJ and d 2JiO'II =JSE &.(F) I 0 =i" ~~ ~ 0 $. 0 !Iz ~ "Ii.(. C2) ~ 2~ P( noise lie in AI) = p( noise lie within outer square) .. ('1:) ..37b (c) ~  c i r" ~l2 20'" t.AI' Thus 1'(Cf"'2) =noise originating fi'om s2 lie in R2 =p( noise lie in 1st quadrmt) . 20'11 ) E.JTj.. CI) . T. the average energy is E =~( 4~2 ) = O. S14. t.p( noise lie within inner square) J(See ±[ Jr p( noise lie in A I ) 1'(noise oriainatina fiom" lie hi AI) =i[{Inll. Fia. CI./i ') ~ . 1b i::a. "1b . ~ ~ 0 ~(~) 5.4d 2 and E = O.b assuming and J :Ii ) « 1 p{£!ms)=4 J We also observe that l2.. Plq.. R2 can be expressed as the ftrst quadrant (horizontally hatched area in Fig.) "4~ J J ~ The decision region R2 for "'2 is shown in Fig..$Sl!J t:. _d . In21<~)] ±{[12~2~Jr [12~~)f} 119 ....9. in21<i){lnll.[1But Fig.AI Therefo.• and again in Fig..
"' .• .. . S14. Gi>~~ .! inner square of side d 4 4 120 .4 1[ ~ d) ~ d )] 4 +4 20'n 2J2O'n and and Fig. "..A . outer square of side dJ2 +. .."'4 • • • "'3 ~. ):: [ P(qm1) + P{Q'"2) + P(qm3) + P(C!m 4)] 4 4 1 The decision region R1 for ml (see Figure) can be expressed as R1 =outer square of side d.iL .>''b\P ( . .~ (outer square .fi ..inner square of side d) :..~. by symmetry P(~m2):: P(~ml):: P(ti m =P(~m4) 3) Hence 14. 'R..!.: 'e ~ .=.~ 4 • . • Fig.38 p(c) = 1 16 ' ..' . . S14.•. 1 I: 4 P(C/m..38 .37c Moreover. . ""t ..
..e:..ln21<d)+i1Inll... p( n lie in outer square) + ! P( n lie in inner square) 4 4 =iil~< ~)+~{Inl<i) =~[12~~)r +~[I2~~)r Similarly R2.. t I .!...4)+ P(noise in Rs) . ~ " '. . The decision region R3 for m) can be expressed as R3 =R.inner square of side d) 2 =1.4 + Rs .. . "'..' .ln21<~)] =~[12~&)H[I2~h)r ~{[1~2~h)r [12~~)]'} =~[12~gd)H[I2~~)r +~[I2~~)J p... :.<{'...J .fi . 1>· ..:'t. inner square of side d 2 2 and p(<=1"'2) = noise originating from "'2 lie in R2 ·iil~<*)+i~lnl<~) =i[I2~h)r +i[I2~Jh)r r \.fi .P(noise in Rc) = p(nl >O....1 f •••• '..r.... (outer square .. In21<*)~lnll.. · ....'.!..~ 2A ~ ~A121 .Now ) p(qml) = Prob( noise originating from ml lies in R1 =....••.. ..... · ..ln21<*)~[ 11n11. outer square of side d.  • •' ....!.... .'· 1 . . . ' Zl _.. the decision region for "'2 (see figure above) can be expressed as R2 = outer square of side d. d . P(noise in R. ~ "'4~: . .. .:.:".."._...Rc and p( C1m3) =Prob(noise originating from m) lie in R3) ".
)« 1.Jj Moreover Hence And 122 . E .a(h) Now Therefore And so that Therefore E) =d 2 .~{POntl: ~Inll' In21 < 12)} +{d8)]' ±[J2{d8)]' +±[J2~h)r For any practical scheme Q(. we have p{qml)==1~h)3~~) P(qm2) == 12~h )2q J:~) == 1.. £3 =4d2 .12kQ(') Using this approximation. n2 > d) .{ d p(qm3)==12{d8)~h)a(~) p(qm4) Hence 8). i(d 2 +2d 2 +4d2 +&12 ) _ ~ d 2 Eb"=108216 4~ E E 4 !L .. E =~d2 . and £4 = &1 2 .. £2 = 2d 2 . can be expressed as I4=R A RB In21 < d) P(qm4) = P(n) > d.The decision region and 14 for m.AI 16. and we can express [lkQof s.
... Hence.. ok are all independent gaussian random variables each with variance JJ/2... POnkl < nl + Ii)] (IIJ+JE) \.M=IP(qml)=I~ .P(C) =~Q(J IS Eb) 8 e 2 JJ Comparison of this result with that in Example 14.[nl > JE..[i nl or nl bl >~ implies E +a+1i nl > a+IE n2 bl >b_2 implies E +a+JE 01 Hence Similarly Hence P(Cjmt}=Prob.S7)] shows that this configuration requires approximately 1.5 times the power of the system in Example 14. 123 . c) have identical error probability.M =1. If 51 is transmitted. b2.. ~ Also 14.. b). onoff (Fig. Inkl < (nl + JE)] Since nl.(14. P(q ml) = [p(nl > Ii) POn21 < nl +JE)P(ln 31 < n1 +1£) . The set in Fig.2tr I _~ { 1U)2}[12Q(Y)] Je' Y~:.41 E b = log2 2N E The onoff signal set and its minimum energy equivalent set are shown in Figs. n2 .3 to achieve the same performance.~JJ/2 ~=== n1 +/i P.J N 1 . In21 < (nl + JE). = = L ct Y  .[i nl b2 ~ =a=1i n2 aIi n2 and Note that hi> b_ 1 implies E+ a+1i nl > E +a.J1VJ JE J1VJ JE I 1 j enlIJJ[r+JE enlIJJdntldn j I j enlIJJ[I2) yJE)J JJ/2 GO N 1 . (e) is also given by the set in Fig. ·· bA. b_k) >aIi n2 >Ii or "2 < Ii +nl or "2 > (JE +n1) =prob. In 31 < (nl + IE) .dnl .P. (b) is polar with half the energy of on off or orthogonal signals. a) and orthogonal (Fig. (b). The minimum energy eqUivalent set of orthogonal signal set in Fig.. .3[Eq.{bt >hlo b2.39 ~=E+a+linl b_ l =E+a. (a) and (b). b_3 . we have 14.. respectively.
we have II Hence the new minimum energy set is II =1. i4 =~ Ji.."Ji fSJ +~.rs =l" 4>2 '2 =15 4>1 13 =15 4> II .... · · .43 To fmd the minimum energy set. 4 ....42 Here Therefore SI{I) =50 ... · CIt d 0 (A....2(1) S2(1) ..') i: d c..I{I) S3(1) = 15 .41 14.... =sl~ .. 124 ..3 =13 fSJ .(11 +12 +$3 +14) = tJi .. 12 =tJi +Ji~.""'"0 ..... SI4..1(1) =..42 14. Fig.)  FigSI4.
..44 ]25 . s. ortJ. s Fia... . 514... and (i4 i 1) are orthogonal. .J ".._ _ _. S......... (i2 i 3).s~t ~Tht: MiniMuM C._ $. .3f ) c .. This set is shown in Fig.16)J2 :: I'sb 103 14. 514..pee) ::1(tQ(3. (i) i 4)...Note that all the four signals fonn vertices ofa square because (il i 2).M'f.'D ¢:a..44 IOV'. "t9nQ1 ../2) =[I~~Jr =['~iv)r +~~)r =f!(X116)j' P'M =1. (t.3 t.".43a..0jcnal .:.J2 and n2 > .!y«t I .o i~ I~rs va) The.43 Observing symmetry we obtain P(C) =P(qml) =P(CI"'2):: P(qm P(qm4) 3):: =1'(nl > ..lr' e. bqsi~ Set CD) TFie. The distance between these signal pairs is always 2J2.. s~ s~ s'r. _+_~o.....) . SI4.If o tFig... __ r I J I t6.
P(Clm3) == P(qml)=
~
J J nl <~)= I_ ,. vl05 l 20.yS
j
y#.)=
1Q(7.07)
p(Clm == 11nl < 20rs) =1 2Q(7 .07) 2)
and
P( C)  j[2 P( Ciml) + P( C\"'2)]  [2  2Q(7.07) + ( 2Q(7.07))
=:
4 13'Q(7.07)
P,M
=1 P(C) == ~Q(7.07) == 1.03 x 1012
Also
E1
=£3:0: =
(s.ro
r
== 4 x 10
3
E2
LoJwf +Lol.or
=2xl0
3
E=j(EI+E2+E3>:o:ix102
Mean energy of the minimum energy set:
Emin == j(2 x 103 +0+2 x 103) =~ x 103
14.45
The use ofEq. (14.76) and signal rotaiion shows that the minimum energy set in this case is identical to that in Prob. 14.44. Hence the minimum energy set is as shown in Fig. SI4.44c. this situation is identical to that in Prob. 14.35 with d == ~. From the results in the solution of Prob. 14.3S, we have 10.y5
E ==~==103
2
Also, we are given SII("') =
.JJ == 105 • Hence, ..AI == 2 x 105 •
2
2
(a) From the solution ofProb. 14.35
P'M == tQ(7.02) + Q(7.l2) == 1.09 x 10 12
(b) and (c) identical to those in Prob. 14.44
14.46
(a) The center of gravity of the signal set is (II +ll)/2 Hence, the minimum energy signal set is 2 The minimum energy signals are
%1(1)
xI = 51
(11+12)
1112 (11+ 12) ..   cl Xl =12 
h51
=.....l..
2
2
2
=0'sO.707sin OJot}
2
"'0
%2(1) =0.707 sin OJ;'  O,s
(b)
EXI
=20001r
=
OoOOI(
!
OJ0 O'sO.707sin 2
,)2dl=0.4984xlO
5
E%2 == EXI
Ij, =
=0.4984 x 105 .
We arc given ..AI =Sx 106
~ J2~~ )= Q(4.46S) • OA Ix 105
126
(c:) We use GramSc:hmidt orthogonaJization procedure in appendix C to obtain
S,it)
$:/ll.1:.)
.i
......
~
y,~)
o·cC!
<::
O'ODI
t,.
.
t:. ....
~ .•.••••.•.•..••....•...•.•..•.. tit s~
"
.
.
St
Fia. S'4.46
127
Chapter 15
151.1 PI:: 0.4,
~
=0.3, PJ =0.2 and P4 :: 0.1
H(m) =(PI log PI + ~ log ~ + PJ log I'] + P4 log P ) 4 =1.846 bits (source entropy)
There are 10' symbols/so Hence. the rate of infonnation generation is 1.846 x 10' bits/so Infonnation/element == log2 10 =3.32 bits. Infonnation/picture frame" 3.32 x 300,000:: 9.96 x lOS bits.
15.12
15.13
Infonnationlword == log2 10000 '" 13.3 bits. Infonnation content of 1000 words ... 13.3 x 1000 =13,300 bits. The infonnation per picture frame was found in Problem 15.1·2 to be 9.96 x lOS bits. Obviously, it is not possible to describe a picture completely by 1000 words, in general. Hence, a picture is worth 1000 words is very much an underrating or unde~tating the reality. (a> Both options are equally likely. Hence,
15.14
/ =log{ts) =I bit
(b) P{2 lanterns)
=0.1
/(2 lanterns) = log210 =3.322 bits
15.15
Ca) All 27 symbols equiprobable and P(Xj) =~7'
HI(x) :: 27(f, log2 27):: 4.755 bitst symbol
(b) Using the probability table, we compute 27 Hw(x) = rp(x;)log P(X/) =4.127 bits/symbol
1.1
(e) Using Zipfs law, we compute entropy/word Hw(x). 8727 Hw(x) =Per) log Per)
,=1
r
= '" jlJ.log(jIJ.):: 9.1353 bits/word. ~ , ,
,=1
HIIetter =11/8215.52.14 bits/symbol. Entropy obtained by Zipfs law is much closer to the real value than HI (x) or H2 (x).
8717
128
.1 32 bits 5. H(m) L 1.. 112 112 0 1 L = LP. s. m.. 11110 1/32 "" 111110r 1132 111111 0 112 114 10 110 118 1110 1116 11110r1l16 11111 112 0 1/4 10 118 110 1110rll8 1111 0 112 10 114 110]'+114 111 lor 0 11 5. m. = 1..(4)+_1 (5)+1.4442 1. =1(1)+1(2)+1(3)+1. 1 7 1 1 1 1 1 3 3 9 9 27 13 .21 H(m) .. 2289 =1.L. 1. ~ Message m.. m1 Probability 112 1/4 118 1/16 1/32 1164 1164 Code 0 10 110 1110 112 114 118 1116 5..  L P.. Jog P.ary units Message Probability Code m.4442 3 . . '9 3 ary d" 19lts  .15.L... (1.TJ) =()o1o 15.TJ)loo .=..=1 7 ... m6 m7 1/3 t/3 o I 20 1/3 113 1/9 o 1 113 113 o I 2 1/9 1/9 1127 1127 1127 21 1/9 2 2 0 r }/9 221 20j1l3 21 22 222 L . 0% 129 . m2 m3 m..x 100 =100% Redundancy .4442 log2 3 3 .(6)+_1 (6) I 2 4 8 16 32 64 64 63 • 32 b'mary d" 19lts Efficiency TJ = H(m) x 100 =100% L Redundancy r =(100.11 H(m) 7 63 =L P.ary digits EffiCiency TJ =. 2289 bits . =(1)+(1)+(2)+(2)+3(3) . LP. m2 m3 m. log P..4442 r :.
L x 100= .5 101. 0. = 1.24 =(1.1 0.S 0.69 x 100 =992% Redundancy r =(1.5 0.1(2) + 0.8% s. 0. Thus.5(1) + 03(2) + 0..1(2) =12 3 ary digits H(m) =1. m2 m] O.2 21 2 22 L =0. log P..69 bits = 1.0663 12 log2 3 1.0. m2 m3 m.0663 3ary units EffiCIency .x 100= 88.. m2 m3 m.14% Message m. H(m) = :~ bits .= .LI~ 3arydigits From Problem 15....15..5 I1J o 1 =r P.1 0.3 0.23 H(m) =  r4 P..)100 =537% 130 .3 For ternary coding. 1J o 0.69 bits ..5 10 0. H(m) 1242 )( 100 =x 100 '"' 94. 1242 3ary units EffiCIency '1" Redundancy ..86% Redundancy r 15.2 11 L rn. = 0.)100 =0. m.1 o 0. Message m.. ~ Probability 112 114 118 1/16 1/32 1/64 1/64 Code o 1 112 1/4 o 1 20 21 112 114 1I4 o 1 20 21 rn.1(3) =1. H(m) .1(3) + 0.)100 =11.1 0 20[j0.3 0.5(1) + 03(1) + 0.7 binary digits L 1. we need one dummy message of probability O.21..69 =1. 22U221 222 118 1/16 1/16 22 J 16 2 L=rp.7 Efficiency" = H(m) x 100 =1.3 110. L.5 Code o 1 o 1 0.=1 Code Message Probability m.....63% L 13125 r =(1. ms Probability 0.
667 Redundancy r . 5.585 = . • :~ .x 100 =95.(1) + (2).289 =(11])100 =4.92% 15. m2 m3 Probability 113 Code 51 12 53 1 00 113 113 I 1/3 m. 2.585 bits (b) Ternary Code Message Probability 113 Code 0 1 2 m. Probability 113 1/3 1/3 Code 00 01 m2 m3 I~213 113 0 1 L =.x 100= x 100 =9'08% L 2.(2) 3 3 .:: .4074 binary digits H(m) =2289 bits (See Problem 15. Efficiency 1]..585 =~ =1. ml 113 1/3 m3 H(m) .16 (a) H(m) = 3(1 log3)1.15 Message m.x 100 =.667 binary digits 3 EffiCiency 1] =.!.585 bits 1... 2.15. H(m) 1.4074 Redundancy r . (11])100 =4.2). ms 114 m7 113 119 119 1127 1127 1127 011 0100 01010 0101 010111 lOr H(m) 119 1/9 119 1/9 0100 1127 OIOlOr 119 0101 2127 01011 00 011 0100 113 OIlJf2i9 00 113 113 119 1 00 010 011 113 113 1/3 00 01 11213 113 0 1 L l: P..08% L 1.!.= 1 3aryunlt 10g2 3 Efficiency 1] = H(m) x 100 =100% L Redundancy (c) Binary Code Message ml r = (11])100 =OOA.L.2 .92% 131 . =1.
001 0000 0001 110 III mlm.611 MasalC m.41 (b) H(x) '" P(xl)log+ P(x2)log 1 1 P(xI) P(x2} SFig.x 100 =98. m2m2 Frob V9 119 119 119 119 119 119 119 119 a.391'A.mz m. H(m) 1.585 bits EfficIency" =.918 bits 332 To compute H(xly).m. m2m. we find 132 .m) 100 101 010 011 219 119 119 119 119 119 119 119 '1 01 0000 0001 110 III 100 101 OO'J~ 2i9 119 119 119 119 119 52 01 10 0000 0001 110 2i9 219 119 119 . 00.m. m. SI5.41 10 = 1..585 =1. So 219 219 219 01 1/3 I~SI9 00 419 01 51 0 I 10 219 II III 15.. m.Iog2 3 + !.x 100 '" .m: m. L 1. SIS.m. 01 10 II 2J9 219 219 Ie 119 oo.'" 5..log2 l =0. m.(d) Second extension .~~ 001]f219 0000 119 0001 10 Jl 000 001 01/'.41 (a) The channel matrix can be represented as shown in Fig.binary code H{m) .
1_0p] o 1.8)]=O. 32 Also.P)IOg_I_J+ J(1.42 P(x.[O(P)+(IPXIJog..!!(0..!! log 45 + 32 log 45 =0..P) H(xly) =I:I:P(Yj)P(%iIYj)logI 1 j P(%.779)+E.p)log_1_+ PIOg. P Ip' Ip P I(xly) =H(x) .!.) matrix as Yj [~ .P) ..P)IOg(l~ P)J = O(P) +(1.8673 bits/ symbol I P(y.. I(xly) =H(x) .8)] =0 log P log(1. H(ylx) =H(y)/(xlY) = 0.618 bits/ symbol j".PXIIog.2493:: 0.P)O(p) =O(P)+(i. .6687 =0.P){1..P)(1log.624):: 0.Plogl . • .ly.) %i to obtain Fia·515. )P(yjlx.P) H(x) = rp(Xi)Jog~ =PlogP2QlogQ P(x.and H(xly) =P(Yl)H(xlYl)+ P(Y2)H(%IY2) .P)O(p) +rIplog. (p) X .6687 45 45 Thus.IYj) . This means dP ~[PIOgP+(l.] .(1.+(I.86730.H(xly) =O(P) + (1.) 45 13 4S . 0 + 2QO(p) =(1.(0.!.!.pX1O(p)] Letting P= 20(P) or O(p) = 10gP I(xly) = O{P)+(1.H(xly) 0.P)log(l. eXI I 0%1 0] Yj 0 P [o Ip Ip P Now we use P(xjIYj) =~ i P (eR) (a() P(YjIXj)P(%/) ~ P(x.) =[ Plog P+(l..918 0.p P with (2Q= 1.P) + [1logp] =0 133 .8) ~J{xIY)"'O dP dP or ..24893 bits/ binit H(y) = =~ P(Yi) 10g_I_::.
then from the results derived in part (b). This is valid for any It. 3 134 Now .P 2 so :::> p=L. We have already proved it for k =2.+Iog+ (IlogfJ) =log/J+2 /J fJ+2 2 1+2' • ' fJ fJ+2 2 fJ+2 loa.P1)(11!z)+ l\1!z = 1.. 2 +4P.l\ 1!z 2l\~ Pyfx (011) =(1.p.43. we have. the channel matrix of the k cascaded channels is M k =M kI M.P£ = =(1.7.o Hence. +6P.! /J 15.l o g .3 and P£ =3p. From the results in Example 10.....p. for a cascade of 3 channels 1.. (b) We have already shown that the channer matrix of two cascaded channels is MI M2 .3 +3P. 6p. In this case Pyfx(lll) =(1.2l\ P2 1 .P..~)1!z + 1\ (11!z) = l\ + 1!z .)l +3P/(1.2 3P.~~~. the channel matrix of the cascade is [ 1Pl~2l\~ 1\+~2P1P l\+1!z21\1!z J=[I1\ 11\i!z21\P 1\ This result will prove everything in this problem. (a) With ~ =i!z =Pe.~~~.43 Consider the cascade of2 BSCS shown in Fig. We can verify these results from the development in Example 10.2 _4P.L (a) . SI5.. If Mt _ 1 is the channel matrix of the first k 1 channels in cascade.Therefore log  P IP =1 + log fJ Note: 1 + 10&2 fJ =log2 2 + log2 fJ =10&2 fJ 2 P fJ =1. (e) Consider a cascade of k identical channels broken up as k 1 channel cascaded with the ktll channel. that M2 Ml. and I_P=_2_ fJ+2 fJ+2 fJ fJ+2 2 C = MAX l(xly) =.) = 13Pe +3Pe2 .7. from the above result it foUows that the channel matrix is indeed M2. Using the process of induction it is clear that Mk = Mk..3 =13P.
=I[P£ 1081+(1.7 for k =3.25 C.. (d) From Equation 15.7 that P£ =1[(1. 15..x.M 3 :: [IPe p.6J!(k.. =1[kPeIOgI+(1_kP. O P(yz) .. .J)! I: .P£ )Iog~] P£ IP£ where PE is the error probability of cascade of k identical channel. If kPe «I. p Let q =Ip xI =0 YI =0 o q p Also. p/(l.)IOg~J kP..6P. :: P(YI) ql2 P(x21Yt) = P(Ytl x 2)P(x2) P(Yt) =0 P(XtIY2)= P(YZIXI)P(XI) .. Clearly P£ . P(Ytlxl)P(xt) ql2 1 . and c. 2 +4p'3 which continns the results in Example 10. P( xI IYI ) . 2 Yl P(yzlxz)P(xz) = P(yz) ql2 = 1 ql2 p 2 2 P(xIIY3) = P(Y3I Xt)P(Xt) = p/2 s! P(Y3) P(x2IY3) = PCnlxl)P(x2) = pl2 =! P(Y3) p 135 . q Yz 0 y..Pe)ki ] j_2. 3p.44 The channel matrix is Y...P(x I ) ..4. lkP.Pe)k + k! . We have shown in Example 10. .
it follows that H(xlz) .) = P(zkIYj) By Bayes' rule FiC. H(x) =P(X2)P(Y3Ix2) =1!. Therefore.Y2) = P(x2)P(Yllx2) = P(xl.x.P(XloYl) = P(Xl)P(Yllxl) = P(Xl.!.H(xly) ~ 0 From the relationship for l(xly) and l(xlz).q )(O+. p = p 2 2 2 2 l(xly) = H(x) . the output z depends only on y.45 Note that for cascaded channel.Y3) = P(Xl)P(Y3I x l) ='2 P(X2.Y2) 'q 2 = P(xl)P(Yllxl) =0 p P(Xl.45 if • and It can be shown that the summation over x of the tenn inside the bracket is nonnegative.lYJ) =1(0)+0+.Y3) f Therefore.H(xly) =1.P bits I symbols 15. p+O+. Hence.!..Yj) log . it immediately follows that l{xly) ~ l(xlz) 136 . 515.P(xl) log P(xl) 1 1 =+= 1 2 2 H(xly) =I: I: P(Xj. 2 P{xl) log P(xl) =.!. j ° 1 P(x. P{zkIYj.Yl) = P(X2)P(Yllx2) =0 P(x2.
we have or and Hence. 2M H(x)= M 1 JM 1 J MP(x)logdx= .2(x.Mlog2Mdx=log2M P(X) 2M 15.(x.2 =1 .p)=plogpand Op =(l+logp) . F(x. plogpdx.and e" I =al = " A . we have and Hence.37. al 1 a .p) . ealI =_1_ andp(x) =_1_ 2M Also. =. of =plogp and op =(1 + logp) op =x .p)=pand 0.51 We have H(x) =J.IS so I v e 7.51 We have H(x) Thus.. M 1 =LMPIOgr= JM _Mplogpdx and J~pdr = 1 F(x.15..37.4 p(x)= : x~O { x<O 137 To obtain H(x) .P) of =p and 0..1 =1 op (1+logp)+al =0= p=e al .I(X.I Substituting these quantities in Equation 15.p) = px and 0.1 op Substituting these quantities in Equation 15.
000 =SO db) B= 1... 15. Under this condition. C=Jhlog [Ss(tlI)+Sn(tlI)]df=JhIOg [ 1 II Sn(tlI) II aSn(tlI) Jdf =(12 . we obtain (} I og [Ss +Sn] +a OS. 2J~2 Ss(Q) df =S (a constant) Using Equation IS.) log (.988xIO' bits/sec.I.(OJ)]dl .(OJ)A/ The maxi. (See Problem 105. = 0 OSs Sn OSs or 1 S + Sn = .96 x lOs bits. xp(x)dx • log A + loge = 10g(eA) 15.S a (a constant) Thus. p(x) log p(x)dx =J.000 N Hence..J~210g 138 [S.~). {log A .54 Consider a narrowband AI where A/ + 0 so that we may consider both signal noise power density to be constant (bandlimited white) over the interval !if .fllum channel capacity over this band 4f is given by Sn(tlI) The capacity of the channel over the entire band (/1. For 30 picture frames per second. The signal and noise power Sand N respectively are given by S =2Ss (OJ)/if and N = 2S. we need a channel with capacity C given by C=30x9.(tlI)]df II Sn(tlI) We now wish to maximize C where the constraint is that the signal power is constant.96xI05 =2.~ lOge] dx =10gAl.8 MHz We are given. But for a white Gaussian noise C B Jog (1 + ! ) = (Note: 100.37. h) is given by N C4I =A/log [~]/ifIOg [Ss(OJ)+Sn(tlI)] C =Jh log [Ss(tlI) + S.1). This shows that to attain the maximum channel capacity.!:: SO db = 100.. p(x)dx+ I~ef. the signal power density + noise power density must be a constant (white).H(x) =J.53 Information per picture frame =9.
c= !1log [Ss(~)+Sn(CII)].J!t21og [Sn(4I)]eIf constant We shall now show that the integral J!t210g [Sn(CII)] elf is maximum when Sn(4I) = constant if the noise is constrained to have a given mean square value (power). we use the results of ProbJem IS. This shows that white Guassian noise is the worst possible kind of noise.37. we have a asn a (logSn)+a. we have shown that for a noise with a given power.=0 Sn aSn or +a=O Sn ) and Sn(CII) =1 a (a constant) Thus. 139 . Under the best possible conditions. Thus.54. we wish to maximize under the constraint 2 iR log (Sn (CII)] elf = N (a constant) Using Equation 15. the integral i!t210g [Sn(CII)]d/ is maximized when the noise is white.15.55 In this problem.
11 (a) There are (j) ways in which j positions can be chosen from n.. But for a ternery code..6) code for t = 2 t (j)21 + 3.+ . J3 For (18.11 j=o L (~) =(~)+(P)+(P)+(P) 3 2048 ~ 1+ 23 + 23 x 11 + 23 x 77 =2048 16. Ford =0 c=O[1 1 1][000] Ford::: 1 c =1 [1 1 I] =[11 IJ 140 . Hence the number of possible errors in j places is k (jX31)" or 3" ~ 3 (b) (11.' =1+18+153+816=988 17! 2! 16! 3! 15! 211 ~ I: 3 Hence 211>I:eB) j=O 3 Clearly.7) code to correct up to 3 errors (}B) or 211 ~(bB)+(l8)+U8)+OB) i·O 18' 18' 18' ::: 1+' + .7) code.: ~ l' " Chapter 16 211 ~ 16.11 GH T = [Ik p] [~J =pep 0 16.' . 16. there exists a possibility of 3 error correcting (18.11 C::: dGwhere d is a single digit (0 or I).k ~ t (j)2i 1=0 jO This is satisfied exactly. a digit can be mistaken for two other digits. 16. this code could correct some 4 error patterns in addition to all patterns with up to 3 errors. Since the Hamming bound is oversatisfied..
16.16. '..15 (8) 100 . 1m and 141 . [~.. P p=[] Note that m . ·01 010·· ·01 000·· f. == 0 and . Therefore this code can detect a single error.... Such a code can correct up to two errors using majority rule for detection.13 e =dG where d is a single digit (0 or 1). 1 (b) o 1 1 100 1 0 1 110 111 Data word 000 o 0 1 010 o o o o 1 1 Codeword 000 0 1 1 1 0 1 1 0 0 1 0 I 0 1 0 1 1 1 1 0 1 0 1 (c) This is a parity check code..)N T =cN T etH T If there is no error .14 0 is transmitted by [000] and 1 is transmitted by [11 1] (8) This is clearly a systematic code with G[ll1] 16. If a single error occurs anywhere in the code word. T Now and rNT ={ee.R =[1] Also rN T =. the parity is violated.... But since m =1.n T =0 NT ... (d) Equation (16.J. ·11 . Ford =0 e=O [11111]=[00000] Ford =1 e =1[1 1 1 I 1] = [1 1 1 1 1] Hence in this code a digit repeats S times.9a) in the text shows that cN T =O.
Hence 16. Hence .28 H T is a IS x 4 matrix with all distinct rows.17 0 0 0 0 1 1 I 1 =3. e has a single 1 element with all other elements being O.If there is a single error in the received word r. One possible H Tis: 111I 1 1 10 I I01 1 100 10 1 1 )0)0 100 I HT = 001 1 o1 I 1 =[~J o I 10 0101 1000 0100 0010 0001 142 .16 Data word 0 0 0 0 1 Codeword 0 1 1 0 0 J 1 0 1 1 0 0 1 1 0 0 1 0 1 0 1 0 1 0 1 1 1 0 1 1 0 1 0 0 1 0 0 1 1 J 1 0 0 0 0 0 0 1 1 I 1 0 0 I 1 0 0 1 1 0 1 0 1 1 0 1 0 dmin From this code we see that the distance between any two code words is at least 3.H T = eH T = 1 (for single error) 16. Data word 0 0 0 1 1 ) 0 0 1 1 0 1 0 1 0 I 0 0 0 0 1 1 1 1 0 0 I I 0 0 1 1 Codeword 0 0 0 J I 0 1 0 1 0 1 1 0 0 I I 1 1 0 0 0 0 I I 0 I I 0 I Observe that d min =3 0 0 16.
100000000001111 010000000001110 001000000001101 000100000001100 000010000001011 G=[l k p]= 000001000001010 000000100001001 000000010000011 000000001000 III 000000000100110 000000000010101 For d =1 0 1 1 1 0 1 0 1 0 1 c =dG =[101 t lOt 010 t] G 10111010101 1110 111 1I0 101 NT= 100 010 001 16. Since there are 6 single errors and 7 syndromes. Cd) e 1 0 0 0 0 0 1 • :dl T s 0 0 0 1 0 0 1 0 0 0 0 1 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 1 1 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 1 1 143 . Hence. Q..OJ III UU p (b) Data word 0 0 0 0 0 1 0 1 0 0 1 1 1 0 0 1 0 I 1 1 0 1 1 1 0 0 0 0 1 1 1 1 0 0 1 1 0 0 1 1 Code word 0 0 0 1 1 0 0 1 1 1 0 1 0 1 1 1 0 1 0 0 0 1 1 0 0 1 0 1 1 0 1 0 Ce) The minimum distance between any two code words is 3. this is a single error correcting code. we can correct all single errors and one double error .19 Ca) 100111 G= 010110 &.
111 100000 010000 001000 000100 000010 000001 100100 G=[/k P ]= 0010011 [1000101] 0100111 0001110 cdG 144 .110 <a> done in Prob.27 o1 1 (b) HT = I0 J 1 10 100 010 001 t s 011 101 110 100 010 001 111 six single errors 1 double error 16. 16._ r 101100 000110 101010 S • C 110 110 000 010000 010000 00000o 111100 010110 101010 d 111 010 101 16.(e) _ _ _ _ _ _ _ _ _ _ _ _ _.
40000 0001 0010 0011 0100 0101 0110 0111 1000 1001 1010 1011 1100 11 01 1J 10 1111 ~ 0000000 0001110 0010011 0011101 0100111 0101001 0110100 0111010 1000101 1001011 1010110 1011000 1100010 1101100 1110001 1I111JJ 101 111 o1 1 HT = 110 100 010 001 ! 0000001 0000010 0000100 0001000 0010000 0100000 1000000 ! 001 010 100 110 011 111 101 s = . on P. Since the decoding table specifies. or 001001 have the same syndrome namely 111.737. = 100010 and the error is not corrected. = received code c. If for example. ! 010100 111 = = 16. This gives a (12. =111 for. Ifwe wish to correct the 2error pattern 010100 (along with six single error patterns). the new decoding table is identical to that in Table 16. 8) code. 010100. Choose n =12.H T where.ee c = corrected code 16. The other two patterns will not be corrected. Hence.8~n+l' n8~log2(~+I) This is satisfied for n ~ 12. it will be comcted. s . =010100 occurs. for a simple error correcting code 2"k~n+l or 2.111 and we shall read ftom the decoding table. HT is chosen to have 12 distinct rows of four elements with the last 4 rows fonning an identity matrix.211 We observe that the syndrome for all the three 2error patterns 100010.213 From Eq. 145 . 100010 whenever e 100010 occurs.3 except for the last entry which s}lould be ..
1 =1S. it can correct all Ierror patterns. ! 0000 0011 0101 0110 0111 1001 1010 101 1 1100 1000 0100 0010 0001 1111 1 1 10 1101 16.o1 1 1 1001 HT "" 001 1 0101 01 10 p G= 1010 10 1 1 1 10 0 1000 0100 0010 0001 100000000011 010000000101 001000000110 000100000111 000010001001 000001001010 000000101011 000000011100 The number of nonzero syndromes =16 .214 ! 000000000000 100000000000 010000000000 001000000000 000100000000 000010000000 000001000000 000000100000 000000010000 000000001000 000000000100 000000000010 000000000001 100000010000 001000001000 000000010001 Codeword 000000 011011 101110 110101 Data word 00 01 10 11 The minimum distance between any two code words is dmin = 4. There are 12 single error patterns. 146 . Since the code oversatisfies Hamming bound it can also correc:t some 2error and possibly some 3error patterns. Hence we may be able to correct 3 doubleerror patterns. Therefore.
Since the code oversatisfies Hamming bound it can also correc:t some 2error and possibly some 3error patterns. Hence we may be able to correct 3 doubleerror patterns. 146 .214 ! 000000000000 100000000000 010000000000 001000000000 000100000000 000010000000 000001000000 000000100000 000000010000 000000001000 000000000100 000000000010 000000000001 100000010000 001000001000 000000010001 Codeword 000000 011011 101110 110101 Data word 00 01 10 11 The minimum distance between any two code words is dmin = 4. Therefore. ! 0000 0011 0101 0110 0111 1001 1010 101 1 1100 1000 0100 0010 0001 1111 1 1 10 1101 16. There are 12 single error patterns.o1 1 1 1001 HT "" 001 1 0101 01 10 p G= 1010 10 1 1 1 10 0 1000 0100 0010 0001 100000000011 010000000101 001000000110 000100000111 000010001001 000001001010 000000101011 000000011100 The number of nonzero syndromes =16 . it can correct all Ierror patterns.1 =1S.
31 Systematic (7.(b) 1 1 10 10 1 1 H T = 0 10 0 0010 0001 1000 and s = eH T 6 siDgle error _ _ { 100000 010000 001000 000100 000010 000001 IJOOOO 7 doubleerror patterns 2 tripleerror patterns { 16. 4) cyclic code g{x} 101000 100100 100010 100001 011000 010010 000111 001101 s 1110 1011 1000 0100 0010 0001 0101 0110 1010 1100 I 1 11 0011 1001 0111 1101 =x 3 +x+ 1 1) =x 6 +X S +x4 +x3 x 3 +x2 + 1 For data 1111 X 3(X 3 +x2 +x+ x 3 +x+ l)x6 +x S +x4 +x3 x6 + x4 + x 3 xS x S +x3 +x2 x 3 +x 2 x 3 +x+ 1 x 2 +x+ 1 c(x} '" (x 3 +x+ lXx 3 +x+ 1) = x 6 +x S +x4 +x3 +x2 +x+ 1 The code word is 11111111 147 .
x 3 +X+ I)X6 +x S +X4 X6 xS xS +X4 +X3 +x3 +X3 +X2 X 2 The code word is 1110100 A similar procedure is used to fmd the remaining codes (see Table 1). =1 0 0 0 0 0 0 x 3 +x+ l)x6 x6+ x 4+x3 x4 +x3 x4 +x 2 +x x 3 +x2 +x x3 x s=101 2 +x+ I +1 148 . (b) From Table I it can be seen that the minimum distance between any two codes is 3. Hence this is a single error correcting code. d c 1111111 1110100 1101001 1100010 1011000 1010011 1001110 1000101 0111010 0110001 0101100 0100111 0011101 0010110 0001011 0000000 1 111 1110 11 01 1100 101 1 1010 1001 1000 0111 0110 0101 0100 0011 0010 0001 0000 Table 1 (e) There are seven possible nonzero syndromes. for .
.1'22 + xiS +x17 +%15 +.1'3 +.0101100 Hence d =0 1 0 1 16...1'2 x" +. d l =000011110000.1''' +.1'3 x! +..=110IIOO$1000000. d l {x).1'6+.1'6 +.X 7 +x6+ x 5+ x 4 d 2 = 1 01 0 1 0 1 0 J 0 I 0.32 g(x) =xii + x 9 + x 7 + x 6 + x 5 + x + 1 c{x) =d{x)g(x) 1.1'3+. xli + x 9 + x 7 + x 5 + .1'2 .=1000000 c=r$..1'5 .1'3 x4 2 s{x)=x +1 +.1'2 +x and C2 .1'3+. 1 0 0 0 1 1 0 1 0 1 0 0 0 0 1 0 0 1 1 1 1 1 0 149 . 1000000 0100000 0010000 0001000 0000100 0000010 0000001 s 101 1J1 110 011 100 010 001 Table 2 (d) The received data is 1101100 r{x) =x 6 +x5 +.The remaining syndromes are shown in Table 2.1'2 5 x +.1' x3 + x2 + x x3 x+l .1'2+.1'13 +xS +%5 +%4 +x3 +.1' 7 +.1'3 +.1'2 +.1'3 + x cJ{x) =dJ{x)g(x) =xiS +.1''' +. d2(x) ..1'17 + .1'9 +x S +.=101 From Table 2 .1'4 and CJ=00001100011101110010000 C2(X) =d2(X)g(x) =.1'+1).1'13 +x12 +xll +.
Hence x S +x4 +x2 + I =(x+ IXx4 +x+ 1) 16. This means it cannot have a third order factor either. The 2ndorder prime factors not divisible by x + 1 are x 2 and x 2 + x + 1.34 X+l)x5+ x4+x2+1 x S +x4 x 2 +1 x 2 +x x+l x+1 o Now try dividing x4 + x + 1 by x + I. This also yields a remainder I.33 x+ I)X3 +X2 +x+ 1 X 3 +x 2 x+1 x+1 o Hence x 3 +x2 +x+l.16. Hence (x + 1) is not a factor of (x4 +x + I).(x+ IXx 2 + I) =(x+ 1)(x+ IXx+I)= (x+ 1)3 16. we try dividing by (x 2 + x + I).3·4 Try dividing x 7 + I by x + I x 6 +x5 +x" +x3 +x2 +x + 1 X+l)x7 +1 x 7 +x6 x 6 +1 x 6 +x S 5 x +I x 5 +x 4 x4 +1 x4 +xl x3 + 1 x 3 +x 2 x2 + I x 2 +x x+ 1 x+l o ISO . Since (x4 + X + I) is not divisible by x 2 . Hence x 4 + x + 1 does not have either a fll'st or a second order factor. we get a remainder I.
001101 . It does not dividc.Now try dividing (X 6 + x S + X4 + X3 + X2 + x + 1) by (x + 1). Ncxt try dividing by (x 3 + I).:]= 0001101 0001101 0000000 c=[OOOl] [E:.!. It docs not divide. So try dividing by (x 2 + 1). It does ~ot dividc. it must bc a x 7 + 1. Wc fmd (x 6 +x S +x4 +x3 +x2 +x+ I) =(x 3 +x+ IXx 3 +x2 + 1) 1) is not divisible by x or x + 1 (thc only two firstordcr primc factors).36 For a single error correcting (7.dG' c[OOOO] [:i::r. Hencc IXx 2 +x+ lXx 2 +x2 + I) 16. Now try dividing by (x 3 +x+ 1). It docs not dividc cithcr.1 g(x) x k. Try dividing by (x 2 + x+ 1). It divides. 4) cyclic code with a generator polynomial g(x) .(x + Since (x 3 +x2 + thirdordcr prime factor.2 g(x) ~ x 3 g(x) x 2 g(x) == g(x) == x 6 +x S +x3 x S +x4 +x2 x4+x3+ x x 3 +x2 + 1 x g(x) g(x) Hence 1 101000] 0110100 G'= 0011010 [ 0001101 Each codc word is found by matrix multiplication c .:]== 0. 151 .x 3 +x2 +1 k:!:4 n=7 Xk.
. pO 0 0 .. . ~ o I 0 0 .. .. I )( 1 == 1 cIS == [1 I IIJ 1000110] 0100011 0010 t 11 [11 l1111J [ = 0001101 cl4 =[11 1 0] G = [111 00 I 0] and so on. ~k h3k . I )( 0 . 0 hl3 ~3 h33 . ~ o0 1 0 . G'= ..X m) c == dG Proceeding with matrix multiplication...The remaining codes are found in a similar manner. 0 and 0)( 0 = 0. 0.. . 0. and noting that o+ 0 . . (b k) The code is found by using (k.. Ohl1 ~I ~I . .. . 1 hl/e I. See table below. . 152 ... .. h". d 0000 0001 0010 0011 0100 0101 0110 0111 1000 1001 1010 1011 1100 1101 1110 111 1 c 0000000 0001101 0011010 0010111 0110100 0111001 0101110 0100001 1101000 1100101 1110010 1111111 1011100 1010001 1000110 1001011 The desired fonn is 1000· . 0 )( 1 . ~ . I + I . . . ~'~~ ... .. . . 0 hl2 hn h32 . . we get 0 + 1 =I + 0 == 1..
g2 = I. [ 0011010] 0001101 The I is eliminated by adding row 4 to row 3.hl For this case gl =I. 153 . g4 =1 Since hlk =g2. use row 4 with one left shift.5 16. to get row 3.. [ 0010111] 0001101 Next for row 2. so far we have Next..k1 + "'g. h]k = g4' the fowth row is immediately found. 83 = 0. use row 3 with 1 left shift. ~k = g3.d 1111 11 10 1101 1100 1011 1010 1001 1000 0111 0110 0101 0100 0011 0010 0001 0000 1111111 1110010 1101000 1100101 1011100 1010001 1001011 1000110 0111001 0110100 0101110 0100011 0011010 0010111 0001101 0000000 These results agree with those of Table 16. c=dG' In general g{x} =glx.. Thus.k + g2 x.38 (a) 101 1000] 0101100 G'= 0010110 [ 0001011 (b) The code is found by matrix multiplication.
Hence this is a single error correcting code. Hence row 4 is 000 1 0 1 1. Row 3 is left shift of row 4.row I .row 3 +.:] 0001011 Row 4 is ok.0101110] 0010111 [ 0001101 The 1 is eliminated by adding row 4 to row 2. use row 2 with 1 left shift. left shift row 2. And add row 1 to obtain row 2. ~:.39 g{x) =x 3 + x+ I. For row 1. For row 2. 1000110] 0100011 0010111 [ 0001101 'Ibis is the desired fonn. G'[:!.row4 +. left shift row 3. 0001011 0010110 0101100 0100111 1001110 1000101 154 +. 0100011] 0010111 [ 0001101 Next for row 1. d c 0000 0000000 0001 0001011 0010 0010110 0011 0011101 0100 0101100 0101 0100111 0110 0111010 0111 0110001 1000 1011000 1001 1010011 1010 1001110 1011 1000101 1100 1110100 1101 1111111 1 J 10 1100010 1 1 1 1 1101001 (c) All code words are at a minimum diSllnce of3 units.row 2 +. 16. And add row 1 to obtain row I.
... PEII 16. =6.b/~) =12Q(J2 x 9. By checking which bit in the augmented segment violates the parity. it will violate the parity in that segment Hence we locate the segment where the error exists.51 16.IJ . If in any segment of b digits a single erTOr occurs. This erTOr will also cause parity violation in the augmented segment. 155 .18 (nearly doubled).12) =9.18 . .8227 x 109 and J2£.41 The burst (of length 5) detection ability is obvious. It can also correct any 3 random errors in each code word.AI This means Eb /~ must be increased from 9. .03 => Eb =18.7·1 =kQ(J2E.825 x 106 P£c =(l' H r J2~.IJ.12 to 18.G= 0100111 1000101] [ 0010110 0001011 16.872 x 109 for PEN' we need new value Eb/~ say Eb/. we can locate the wrong bit position exactly. Then 9. ) 9 =(l'XQ(J9516S )]' ".872 x 10 =k~ J2~b )= 12~ i2~b ) Hence ~~2~b ) = 0. The single emr correcting ability can be dem onstrated as follows.12 x 10" To achieve a value 9. The code can correct any 3 bursts of length 10 or less.
This action might not be possible to undo. Are you sure you want to continue?