Part 5 Analog Demodulation with

Noise
Impact of Additive Noise in Analog
Modulation Systems

Modulated x(t)
Bandpass Output
signal Demodulator
filter signal
s(t)
Noise w(t)

o To simplify the system analysis, we assume:

n ideal bandpass filter that is just wide enough to pass the
modulated signal s(t) without distortion,
n ideal demodulator,
n Gaussian distributed white noise process.
o So, the only source of imperfection is from the noise.
© Po-Ning [email protected] 5-2
Impact of Additive Noise in Analog
Modulation Systems

Modulated x(t)
Bandpass Output
signal Demodulator
filter signal
s(t)
Noise w(t)
BT

As a result, after passing through the ideal bandpass filter, s(t)

is unchanged but w(t) becomes a narrowband noise n(t).
Hence,
x(t) = s(t) + n(t),
where n(t ) = nI (t ) cos( 2pf ct ) - nQ (t ) sin(2pf ct ).
© Po-Ning [email protected] 5-3
Impact of Additive Noise in Analog
Modulation Systems
o Input signal-to-noise (power) ratio (SNRI)
n The ratio of the average power of the modulated
signal s(t) to the average power of the filtered noise
n(t).
o Output signal-to-noise (power) ratio (SNRO)
n The ratio of the average power of the demodulated
message signal to the average power of the noise,
measured at the receiver output.

© Po-Ning [email protected] 5-4

Impact of Additive Noise in Analog
Modulation Systems
o It is sometimes advantageous to look at the lowpass
equivalent model.
o Channel signal-to-noise (power) ratio (SNRC)
n The ratio of the average power of the modulated
signal s(t) to the average power of the channel noise
in the message bandwidth, measured at the receiver
input (as illustrated below).
Message signal Lowpass
with the same power filter of Output
as the modulated wave bandwidth W
Noise w(t)

© Po-Ning [email protected] 5-5

Impact of Additive Noise in Analog
Modulation Systems
o Notes
n SNRC has nothing to do with the receiver structure, but
depends on the channel characteristic and modulation
approach.
n SNRO is receiver-structure dependent.
o Finally, define the figure of merit for the receiver as:

SNRO
figure of merit =
SNRC

© Po-Ning [email protected] 5-6

Noise in Linear Coherent Receivers
o Recall that for demodulation of AM signal
n when the carrier is suppressed, linear coherent
detection is used.
n when the carrier is additionally transmitted,
nonlinear envelope detection is used.
o The noise analysis of the above two cases are
respectively addressed in the sequel.

© Po-Ning [email protected] 5-7

Noise in Linear Coherent Receivers

DSB-SC x(t) v(t)

Bandpass Lowpass
signal s(t) y(t)
filter filter

Noise w(t) Local

OSC cos(2⇡fc t)
<latexit sha1_base64="eSjgJReEuT2PinESHsAwqLAs/VE=">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</latexit>

Coherent detector

𝑠 𝑡 = 𝐴! 𝑚(𝑡) cos(2𝜋𝑓! 𝑡)

© Po-Ning [email protected] 5-8

Noise in Linear Coherent Receivers
o Average signal power Here, we assume m(t) stationary.

W
where P = E[m (t )] = ò SM ( f )df is the message power.
2
-W
© Po-Ning [email protected] 5-9
Noise in Linear Coherent Receivers
o Noise power in the message bandwidth
W W N0
ò -W
Sw ( f )df = ò
-W 2
df = WN 0

© Po-Ning [email protected] 5-10

Noise in Linear Coherent Receivers
Coherent detector

DSB-SC x(t) v(t)

Bandpass Lowpass
signal s(t) y(t)
filter filter

Noise w(t) Local

OSC cos(2⇡fc t)
<latexit sha1_base64="eSjgJReEuT2PinESHsAwqLAs/VE=">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</latexit>

o Channel SNR for DSB-SC

𝐴(' 𝑃/2 𝐴(' 𝑃
SNR !,#$%&$! = =
𝑊𝑁) 2𝑊𝑁)
o Next, we calculate the output SNR (observed at y(t)) under the condition
that the transmitter and the receiver are perfectly synchronized.
© Po-Ning [email protected] 5-11
 f
0 fc
<latexit sha1_base64="cJ1/p8R4ZEc11VLSXpt165g2BX4=">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</latexit>

© Po-Ning [email protected] 5-12

Þ Figure of merit for DSB - SC and coherent detection = 1.

Similar derivation on SSB and coherent detection yields the

same figure of merit.
o Conclusions
n Coherent detection for SSB performs the same as
coherent detection for DSB-SC.
n There is no SNR degradation for SSB and DSB-SC
coherent receivers. The only effect of these modulation
and demodulation processes is to translate the message
signal to a different frequency band to facilitate its
transmission over a band-pass channel.
n No trade-off between noise performance and bandwidth.
This may become a problem when high quality
transceiving is required.
© Po-Ning [email protected] 5-13
Noise in Envelope Detector for AM

AM signal x(t)
Bandpass Envelop Output
s(t) filter detector Signal y(t)

Noise w(t)

s(t ) = Ac [1 + ka m(t )] cos( 2pf c t )

1 T 2 ⎡
2 2⎤ 1 T 2
lim
T→∞ 2T
∫ −T E[s (t)]dt = A E ⎣
c(1+ k a m(t)) lim
⎦ T→∞
2T
∫ −T
cos (2π fc t)dt

Ac2
= (1+ ka2 P) (Assume m(t) is zero mean.)
2

© Po-Ning [email protected] 5-14

 W W N0
Also, ò-W Sw ( f )df = ò-W df = WN 0
2
o Hence, channel SNR for DSB-C is equal to:
Ac2 (1 + ka2 P)
Þ SNRC ,AM =
2WNo
o Next, we calculate the output SNR (observed at y(t)).

AM signal x(t)
Bandpass Envelop Output
s(t) filter detector Signal y(t)

Noise w(t)

© Po-Ning [email protected] 5-15

 Envelop detector

Square
Squarer Lowpass
x(t) rooter
( )2 filter
( )1/2

(Refer to Slides 5-20 and 5-22.)

© Po-Ning [email protected] 5-16

 SNRO ,AM Ac2 ka2 P (2WN 0 ) ka2 P
Þ » 2 = <1
SNRC ,AM Ac (1 + ka P ) (2WNo ) 1 + ka P
2 2

oConclusion
n Even if the noise power is small when it is compared
to the average carrier power at the envelope detector
output, the noise performance of a full AM (DSC-C)
receiver is inferior to that of a DSB-SC receiver due
to the wastage of transmitter power.

© Po-Ning [email protected] 5-17

 Noise in Envelope Detector for AM
o Assume m(t ) = Am cos( 2pf mt )
Þ s(t ) = Ac [1 + ka Am cos( 2pf mt )] cos( 2pf ct )
Hence,
1 1
ò [1 + k ]
T T

ò-T E[ s (t )]dt = A lim 2T p ( 2pf ct )dt

2 2 2 2
lim c T ®¥
A
a m
cos( 2 f m
t ) cos
T ®¥
2T -T

1
ò (cos (2pf t ) + 2k A
T
= Ac2 lim 2
c a m
cos( 2pf m
t ) cos 2
( 2pf c t )
T ®¥
2T -T

+ ka2 Am2 cos2 (2pf mt ) cos2 ( 2pf c t ) )dt

æ 1 k 2 2
A ö A2

= Ac2 ç + 0 + a m ÷ = c (1 + ka2 P )
è2 4 ø 2
© Po-Ning [email protected] 5-18
 2
1 T 2 A
where P = lim ò p =
2
Am cos ( 2 f m t ) dt m
.
T ®¥
T 0 2
Þ Following similar procedure as previous discussion,
SNRO ,AM ka2 P ka2 Am2 / 2
» = .
SNRC ,AM 1 + ka P 1 + ka Am / 2
2 2 2

So even if for 100% percent modulation (kaAm = 1), the figure of merit
= 1/3. This means that an AM system with envelope detection must
transmit three times as much average power as DSB-SC with coherent
detector to achieve the same quality of noise performance.

© Po-Ning [email protected] 5-19

Threshold Effect
o What if is violated in AM modulation with
envelope detection?

© Po-Ning [email protected] 5-20

 Assume

© Po-Ning [email protected] 5-21

 (See Slide 5-19.)

© Po-Ning [email protected] 5-22

Threshold Effect Ac2 / 2 Ac2
r= =
o Threshold effect 2WN0 4WN0
n For AM with envelope detection, there exists a carrier-
to-noise ratio r (namely, the power ratio between
unmodulated carrier Ac cos(2pfct) and the passband
noise n(t)) below which the noise performance of a
detector deteriorates rapidly.
8 2 2
> A k P
< c a = 2ka2 P ⇢, if Ac [1 + ka m(t)] |ñ(t)| i.e. ⇢ 1
SNRO,AM = 2W N0
2 2
: c ka P = k 2 P ⇢,
> A
if Ac [1 + ka m(t)] ⌧ |ñ(t)| i.e. ⇢ ⌧ 1
a
4W N0

© Po-Ning [email protected] 5-23

General SNRO in Envelope Detection
o For envelope detector, the noise is no longer additive; thus,
the original definition of SNRO (which is based on additive
noise) may not be applied.
o A new definition should be given:
n Definition. The (general) output signal-to-noise ratio
for an output y(t) due to a carrier input is defined as
so2
SNRO =
Var[ y (t )]
where so = E [ y (t )] - E [ yo (t )], and yo (t ) is equal to y (t )
in the presense of noise alone. Conceptually, y(t) = so + yo(t).

© Po-Ning [email protected] 5-24

General SNRO in Envelope Detection
o so is named the mean output signal.
o Var[y(t)] is named the mean output noise power.
o Example. y(t) = A + nI(t), where nI(t) is zero mean.

ìso = E[ A + nI (t )] - E[nI (t )] = A
í
î Var [ y ( t )] = Var[ n I ( t )] = E [ n 2
I ( t )]

A2
Þ SNRO =
E[nI2 (t )]
This shows the backward compatibility of the new definition.

© Po-Ning [email protected] 5-25

General SNRO in Envelope Detection
o Now, for an envelope detector, the output due to a carrier
input and additive Gaussian noise channel is given by:

y(t ) = ( A + nI (t ))2 + nQ2 (t )

I0( ) = modified Bessel function
of the first kind of zero order.
Þ y(t ) is Rician distributed with pdf
! ! " $%" %!
𝑓! " 𝑦 = " exp − " 𝐼' " for 𝑦 ≥ 0, where 𝜎(& = 𝐸 𝑛& 𝑡
#! &#! #!
= 2𝑊𝑁'
Þ yo (t ) = nI2 (t ) + nQ2 (t ) is Rayleigh distributed with pdf
! !"
𝑓!# " 𝑦 = " exp − " for 𝑦 ≥ 0, where 𝜎(& = 𝐸 𝑛& 𝑡 = 2𝑊𝑁'
#! &#!

© Po-Ning [email protected] 5-26

 Z 1
E[y(t)] = yfy(t) (y)dy
0
Z 1 2 ✓ 2 2
◆ ✓ ◆
y y +A Ay
= 2 exp 2 I 0 2 dy
0 N 2 N
Z 1 ✓ N2 ◆
N 2 u
= 3/2
exp( ⇢) u exp I0 (u)du,
(2⇢) 0 4⇢
2
by taking u = Ay/ N and ⇢ = A2 /(2 N 2
) = A2 /(4W N0 ).
Z 1 Z 1 2 ✓ 2
◆
y y
E[yo (t)] = yfyo (t) (y)dy = 2 exp 2 dy
0 0 2
Z 1 ZN 1 ⇣ N
⌘
2 z 2 /2 z 2 /2
= N z e dz = N z · ze dz
✓0 ⇣ ⌘1 Z 0
1⇣ ⌘ ◆
2 2
= N z· e z /2 e z /2 dz
0 0
Z 1 Z 1 r
z 2 /2
p 1 z 2 /2 ⇡
= N e dz = N 2⇡ p e dz = N
0 0 2⇡ 2

© Po-Ning [email protected] 5-27

 Confluent Hypergeometric Functions
o The Kummer confluent hypergeometric function is a
solution of Kummer’s equation
d2y dy
x 2 + (b - x) - ay = 0 for a, b complex
dx dx
with boundary conditions y(0) = 1 and y' (0) = a / b.
o For b ¹ 0, -1, -2, …., the Kummer confluent
hypergeometric function is equal to 1F1(a;b;x).
Generalized hypergeometric function
" " ¥
(a1 )k (a2 )k !(a p )k x k ì(a )k = a (a + 1) !(a + k - 1)
p Fq ( a ; b ; x ) = å × , where í .
k =0 (b1 ) k (b2 ) k !(bq ) k k! î(a )0 = 1
© Po-Ning [email protected] 5-28
Properties of
Confluent Hypergeometric Functions
a
1. 1 F1 ( a; b; x ) » 1 + x as x ® 0.
b
2. 1 F1 ( -1;1; x ) = 1 - x.
æ xö æ æxö æ x öö
3. 1 F1 ( -1 / 2;1;- x ) = exp ç - ÷ ´ çç (1 + x ) I 0 ç ÷ + xI 2 ç ÷ ÷÷
è 2ø è è2ø è 2 øø
x
»2 as x ® ¥.
p
¥ G( m / 2 ) é æ m 1 öù
ò exp( -b u ) I 0 (u )du =
m -1 2 2
4. u m ê 1 F1 ç ;1; 2 ÷ú
0 2b ë è 2 4b ø û
5. exp( -u ) ×1 F1 (a ; b ; u ) =1 F1 ( b - a ; b ;-u )
© Po-Ning [email protected] 5-29
General SNRO in Envelope Detection
o Hence,
sN ¥ æ u2 ö
E[ y (t )] = exp( - r ) ò u exp çç -
2
÷÷ I 0 (u )du
(2r ) 3/ 2 0
è 4r ø
sN G( 3 / 2 ) é æ 3 öù
= exp( - r ) F ;1; r ÷ú By Property 4
-3 / 2 ê 1 1 ç
(2r ) 3/ 2
2( 4 r ) ë è 2 øû
p é æ3 öù
= s N exp( - r ) ê 1 F1 ç ;1; r ÷ú
2 ë è2 øû
p é æ 1 öù
= s N ê 1 F1 ç - ;1;- r ÷ú By Property 5
2 ë è 2 øû

© Po-Ning [email protected] 5-30

General SNRO in Envelope Detection
o As a result,
é æ 1
p A2 ö ù
so = E[ y (t )] - E[ yo (t )] = s N ê 1 F1 çç - ;1;- 2 ÷÷ - 1ú.
2 ë è 2 2s N ø û
o Similarly, we can obtain:
é æ A2
ö pé æ 1 A 2
öù
2
ù
Var[ y (t )] = 2s N2 ê 1 F1 çç - 1;1;- 2 ÷÷ - ê 1 F1 çç - ;1;- 2 ÷÷ú ú
ê è 2s N ø 4ë è 2 2s N øû ú
ë û
é A2
p é æ 1 A2
öù
2
ù
= 2s N2 ê1 + - ê 1 F1 çç - ;1;- 2 ÷÷ú ú By Property 2
ê 2s N 4 ë è 2 2s N
2
øû ú
ë û
© Po-Ning [email protected] 5-31
General SNRO in Envelope Detection
o This concludes to:

2
[1 F1 ( 1/2; 1; ⇢) 1] A2
SNRO = 4 2 , where ⇢ = 2 2
⇡ (1 + ⇢) [1 F1 ( 1/2; 1; ⇢)] N
8 p
>
> 1]2
[2 ⇢/⇡
>
<4 p , as ⇢ ! 1 (Property 3)
(1 + ⇢) [2 ⇢/⇡] 2
⇡ ⇡
2
>
> [(1 + ⇢/2) 1]
>
:4 , as ⇢ ! 0 (Property 1)
2
⇡ (1 + ⇢) (1 + ⇢/2)

© Po-Ning [email protected] 5-32

(Continue from the previous slide.)

8 ⇡ p
<⇢ + 4⇡⇢, as ⇢ ! 1
= ⇡⇢2
: 2
, as ⇢ ! 0
16(1 + ⇢) ⇡(2 + ⇢)
(
⇢, as ⇢ ! 1
⇡ ⇡⇢2
16 4⇡ , as ⇢ ! 0
(
⇢, as ⇢ ! 1
=
0.91⇢2 , as ⇢ ! 0

© Po-Ning [email protected] 5-33

General SNRO in Envelope Detection
Curve of SNRO =
[1 F1 (- 1 / 2;1;- r ) - 1]2
4
(1 + r ) - [1 F1 (- 1 / 2;1;- r )]
2

p
and the two limiting approximates.
(See Slide 5-23)

SNRO ,AM = 2ka2 Pr

(SNR)O

(SNR)O
SNRO,AM = ka2 Pr

© Po-Ning [email protected] 5-34

General SNRO in Envelope Detection
+"
!, +"
!,
o Remarks Slide 5-12: SNR *,#$%&$! = (-. = "
/$
= 2𝑃𝜌
#

n For large carrier-to-noise ratio r, the envelope detector

behaves like a coherent detector in the sense that the output
SNR is proportional to r.
n For small carrier-to-noise ratio r, the (newly defined) output
signal-to-noise ratio of the envelope detector degrades faster
than a linear function of r (decrease at a rate of r2).
n From “threshold effect” and “general SNRO,” we can see that
the envelope detector favors a strong signal. This is
sometimes called “weak signal suppression.”

© Po-Ning [email protected] 5-35

 Impact of Noise in FM Receivers

FM Bandpass x(t) v(t) Baseband Output

Limiter Discriminator lowpass
signal s(t) filter signal
filter

Noise w(t)

o To simplify the system analysis, we assume:

n ideal band-pass filter that is just wide enough to pass the
modulated signal s(t) without distortion,
n ideal demodulator,
n Gaussian distributed white noise process.
o So, the only source of imperfection is from the noise.

© Po-Ning [email protected] 5-36

 FM Bandpass x(t) v(t) Baseband Output
Limiter Discriminator lowpass
signal s(t) filter signal
filter

Noise w(t) ìnI (t ) = r (t ) cos(y (t ))

í
înQ (t ) = r (t ) sin(y (t ))
t
s(t ) = Ac cos[2pf ct + f (t )], where f (t ) = 2pk f ò m(t )dt .
0

x (t ) = Ac cos[2pf ct + f (t )] + r (t ) cos[2pf ct + y (t )]
= Ac cos[2pf ct + f (t )] + r (t ) cos[2pf ct + f (t ) + y (t ) - f (t )]
= ( Ac + r (t ) cos[y (t ) - f (t )]) cos[2pf ct + f (t )]
- r (t ) sin[y (t ) - f (t )]sin[ 2pf ct + f (t )]
= ( Ac + r (t ) cos[y (t ) - f (t )])
2
+ r 2 (t ) sin 2 [y (t ) - f (t )] cos[2pf ct + q (t )]
ì r(t ) sin[y (t ) - f (t )] ü
where q (t ) = f (t ) + tan í -1
ý
A
î c + r ( t ) y
cos[ ( t ) - f ( t )] þ
© Po-Ning [email protected] 5-37
 FM Bandpass x(t) v(t) Baseband Output
Limiter Discriminator lowpass
signal s(t) filter signal
filter

Noise w(t)

x (t ) = (A c
+ r ( t ) cos[y ( t ) - f ( t
2
)]) + r 2
( t ) sin 2
[y (t ) - f (t )] cos[2pf ct + q (t )]
Limiter
® A × cos[2pf ct + q (t )]
Next, the signal will be passed through a Discriminator.
Recall on Slides 4-87 ~ 4-94, we have talked about the
Balanced Frequency Discriminator, whose input and output
satisfy:
Input s(t ) = A cos( 2pf ct + q (t )) Output ~
so (t ) = 2aAq ' (t )
Recall s(t) = Ac cos 2 fc t + 2 kf
t
0
m( )d ⇒ s!o (t) = 4π k f aAc m(t)

© Po-Ning [email protected] 5-38

Specifically, with ~ s (t ) = Aexp( jq (t )), we have :
é d~s (t ) ù
~
s1 (t ) = a ê + jpBT ~ s (t )ú = aAj[q ' (t ) + pBT ]exp[ jq (t )]
ë dt û
é d~s (t ) ù
~
s2 (t ) = - a ê - jpBT s (t )ú = - aAj[q ' (t ) - pBT ]exp[ jq (t )]
~
ë dt û
Þ~
so (t ) =| ~
s1 (t ) | - | ~
s2 (t ) |= 2aAq ' (t )

Thus, after passing through the discriminator

æ -1 ì r ( t ) sin[y ( t ) - f ( t )] üö
d çç f (t ) + tan í ý ÷÷
è î Ac + r (t ) cos[y (t ) - f (t )] þ ø
v (t ) = 2aAq ' (t ) = 2aA
dt

© Po-Ning [email protected] 5-39

Let ↵(t) = (t) (t).
1 r(t) sin( (t))
d (t) + tan Ac +r(t) cos( (t))
v(t) = 2aA (t) = 2aA
dt

© Po-Ning [email protected] 5-40

 ✓ ⇢ ◆
0 d 1 nQ (t)
v(t) = 2aA✓ (t) = 2aA (t) + tan
dt Ac + nI (t)
0
✓
B 1 n0Q (t)
= 2aA @ 0 (t) + ⇣ ⌘2 ⇥
Q n (t) Ac + nI (t)
1 + Ac +n I (t)

0
◆◆
nQ (t)nI (t)
(Ac + nI (t))2
Assumption 2: Ac r(t) with high probability.

So that Ac |nI (t)| and Ac |nQ (t)| imply Ac + nI (t) Ac .

0 1
✓ ◆
B 0 1 n0Q (t) nQ (t) n0I (t) C
v(t) ⇡ 2aA @ (t) + ⇣ ⌘2 ⇥ · A
nQ (t) Ac Ac Ac
1+ Ac
© Po-Ning [email protected] 5-41
 ✓ 0 ◆
0
nQ (t)
v(t) ⇡ 2aA (t) +
Ac
1
Assumption 3: 2aA = 2⇡ .

0 n0Q (t)
2⇡v(t) ⇡ (t) + Ac = 2⇡kf m(t) + 2⇡nd (t),

n0Q (t)
where nd (t) = 2⇡Ac . We then obtain the desired “additive” form.

d nQ (t ) j 2pf nd (t )
Table 6.2 : 8. g (t ) « j 2pfG ( f ) H( f ) =
dt 2pAc
© Po-Ning [email protected] 5-42
 ì S N ( f - f c ) + S N ( f + f c ),
ï
SN ( f ) = í
Q
for | f |< BT / 2
ï0, otherwise
î
vo (t ) » k f m(t ) + no (t )
FM Bandpass x(t) v(t) Baseband Output
Limiter Discriminator lowpass
signal s(t) filter signal
filter

Noise w(t)
Bandwidth W < BT/2
that is just enough
to pass m(t).

© Po-Ning [email protected] 5-43

SnQ (f )
<latexit sha1_base64="D5qtnII/7KIscVk08afL7PWI288=">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</latexit>

N0 W 2 N 0W 3
Þ E[n (t )] = 2 ò f df =
2 2
o
Ac -W 3 Ac2

BT BT
0
<latexit sha1_base64="Ljlo7/y2fX3fRFZipWyFck7iu6c=">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</latexit>
2 <latexit sha1_base64="HepjvNRVpOkMkG7hsfHf1ayo2xE=">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</latexit>
2
Observation from the above formula:
Snd (f )
<latexit sha1_base64="trSWW0ZAJ8Gag7Z2SL1nEjZsc7E=">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</latexit>

In an FM system, increasing carrier

power = Decreasing noise power.

BT BT
This is named the noise quieting
0
<latexit sha1_base64="Ljlo7/y2fX3fRFZipWyFck7iu6c=">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</latexit>
2 <latexit sha1_base64="HepjvNRVpOkMkG7hsfHf1ayo2xE=">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</latexit>
2
effect.
Sno (f ) <latexit sha1_base64="Ffi8vRzxW9jh4u8HHOdgfKvUTo4=">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</latexit>

<latexit sha1_base64="j0Ux1NLj7OdTtsxAr8//1qoDioE=">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</latexit>
W 0 W
<latexit sha1_base64="l3oyI1BWgb4ZdkLkM9i+278t6Zk=">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</latexit>
<latexit

© Po-Ning [email protected] 5-44

As vo (t ) » k f m(t ) + no (t ),

k 2f E[m 2 (t )] 3 Ac2 k 2f P
Þ SNRO ,FM = 3
= , provided Ac >> r(t ).
2 N 0W 2 N 0W 3

3 Ac2

We next turn to SNRC,FM.

t
s(t ) = Ac cos[2pf ct + f (t )], where f (t ) = 2pk f ò m(t )dt .
0

Þ average power in the modulated signal s(t ) is Ac2 / 2.

W N
Average noise power in the message bandwidth is ò 0
df = WN 0 .
-W 2

Ac2 / 2 Ac2
Þ SNRC ,FM = = .
N 0W 2 N 0W
© Po-Ning [email protected] 5-45
 3 Ac2 k 2f P
2
SNRO ,FM 2 N 0W 3 3k f P
Þ = 2
= .
SNRC ,FM Ac W 2

2WN0

SNRO ,FM
Remarks : For fixed W , increasing BT Û increasing .
SNRC ,FM
Df k f P1 / 2
1. Deviation ratio D = µ . Df = k f max | m(t ) |
W W
SNRO ,FM
Hence, µ D2. SNRO,FM
BT2
SNRC ,FM SNRC,FM

æ 1ö æ 1ö
2. BT ,Carson = 2Df ç1 + ÷ = 2 DW ç1 + ÷ = 2W ( D + 1)
è Dø è Dø
© Po-Ning [email protected] 5-46
 (From Slide 5-19, 2𝑃 = 𝐴"! )
#! $" #! "&
(From Slide 4-60, 𝐷 = 𝛽 = %"
= '
)

Summary

o Specifically,
n for high carrier-to-noise ratio r (equivalent to the
assumption made in Assumption 2), an increase in
transmission bandwidth BT provides a corresponding
quadratic increase in figure of merit of a FM system.
o So, there is a tradeoff between BT and figure of merit.
o Notably, figure of merit for an AM system has
nothing to do with BT.
© Po-Ning [email protected] 5-47
Single-Tone FM Signal with Noise
o m(t) = Am cos(2pfmt)
o Then we can represent the figure of merit in terms of
modulation index (or deviation ratio) b as:

o In order to make the figure of metric for an FM system to

be superior to that for an AM system with 100%
modulation, it requires:

1 1
BT,Carson = 2 f 1 + = 2 fm 1 + = 2( + 1)fm

© Po-Ning [email protected] 5-48

Capture Effect
o Recall that in Assumption 2, we assume Ac >> r(t).
o This somehow hints that the noise suppression of an FM
modulation works well when the noise (or other unwanted
modulated signal that cannot be filtered out by either bandpass
or lowpass filters) is weaker than the desired FM signal.
o What if the unwanted FM signal is stronger than the desired FM
signal.
n The FM receiver will capture the unwanted FM signal!
o What if the unwanted FM signal has nearly equal strength as the
desired FM signal.
n The FM receiver will fluctuate back and forth between them!

© Po-Ning [email protected] 5-49

FM Threshold Effect
o Recall that in Assumption 2, we assume Ac >> r(t)
(equivalently, a high carrier-to-noise ratio) to simplify q(t)
so that the next formula holds.

o However, a further decrease of carrier-to-noise ratio will

break the FM receiver (from a clicking sound down to a
crackling sound).
o As the same as the AM modulation, this is also named the
threshold effect.

© Po-Ning [email protected] 5-50

FM Threshold Effect
o Consider a simplified case with m(t) = 0 (no message
signal). From Slide 5-41, we have

To facilitate the understanding of “clicking” sound effect,

we let r(t) = l Ac, a constant ratio of Ac.

© Po-Ning [email protected] 5-51

FM Threshold Effect
cos[y (t )] + l
Þ 2pv(t ) = q ' (t ) = ly ' (t )
1 + 2l cos[y (t )] + l2

Then at the time, say,y (t ) » p , and l > 1 but l » 1

cos[y (t )] + l l
Þ 2pv(t ) = q ' (t ) = ly ' (t ) » y ' (t )
1 + 2l cos[y (t )] + l l - 1
2

Thus, a sign change in y’(t) will cause a spike!

Notably, when l = 0 (no noise), the output equals m(t) = 0 as

desired.
© Po-Ning [email protected] 5-52
FM Threshold Effect
cos[p sin(t )] + l
y (t ) = p sin(t ) Þ 2pv(t ) = q ' (t ) = lp cos(t )
1 + 2l cos[p sin(t )] + l2

l = 1.05
l =5
l = 0.05

© Po-Ning [email protected] 5-53

How to Avoid “Clicking” Sound?
Fix modulation index (or deviation ratio) b and message
signal bandwidth W:
1. Determine BT by either Carson’s rule or Universal curve.
2. For a specified noise level N0, select Ac to satisfy:
æ Ac2 ö Ac2
10 log10 çç ÷÷ ³ 13 dB or equvalently, ³ 20.
è 2 BT N 0 ø 2 BT N 0

Experiments found that occasional clicks are heard at r

around 13 dB.

© Po-Ning [email protected] 5-54

Threshold Reduction
o After our learning that FM modulation has threshold
effect, the next question is naturally on “how to reduce
the threshold?”

© Po-Ning [email protected] 5-55

Threshold Reduction
o Threshold reduction in FM receivers may be
achieved by
1. negative feedback (commonly referred to as an
FMFB demodulator), or
2. phase-locked loop demodulator.
o Why PLL can reduce threshold effect is not covered in
this course.

© Po-Ning [email protected] 5-56

 Received Bandpass x(t ) Limiter v(t ) Baseband
+ lowpass
FM wave s(t) filter
Discriminator filter

Assume noise-free. svco(t) Voltage-

controlled
oscillator
t
s(t ) = Ac cos[2pf ct + f (t )], where f (t ) = 2pk f ò m(t )dt .
0
t
svco (t ) = 2 cos[2pf vcot + fvco (t )], where fvco (t ) = 2pak f ò m(t )dt .
0

s(t ) svco (t ) = 2 Ac cos[2pf c t + f (t )] cos[2pf vco t + fvco (t )]

Bandpass
® Ac cos[2p ( f c - f vco )t + (1 - a )f (t )]
The new frequency deviation Df new = (1 - a )Df original .
Thus, the bandpass filter can conceptual ly have a smaller passband
as wide as (1 - a ) BT , centered at ( f c - f vco ).
© Po-Ning [email protected] 5-57
 n(t)svco (t) = 2n(t) cos[2 fvco t + vco (t)]

The noise at the Mixer output can be treated white with the same
noise level as the input white noise.

x(t) = Ac cos[2⇡fc0 t + (1 ↵) (t)] + r(t) cos[2⇡fc0 t + (t)]

Limiter
! cos[2⇡fc0 t + ✓(t)]

where E[n2I (t)] = E[n2Q (t)] = E[n2 (t)] = (1 ↵)BT N0 ,

⇢
1 r(t) sin[ (t) (1 ↵) (t)]
and ✓(t) = (1 ↵) (t) + tan .
Ac + r(t) cos[ (t) (1 ↵) (t)]

Since E[n2I (t)] = E[n2Q (t)] is smaller, and Ac remains the same,
the condition Ac r(t) = n2I (t) + n2Q (t) holds with higher probability.

Experiments show that an FMFB receiver is capable of realizing a

threshold extension on the order of 5~7 dB.
© Po-Ning [email protected] 5-58
Threshold Reduction of an FMFB Receiver
o To sum up:
n An FMFB demodulator is essentially a tracking
filter that can track only the slowly varying
frequency of a FM signal.

© Po-Ning [email protected] 5-59

 Pre-Emphasis and De-emphasis for FM
o Recall that the noise PSD at the output shapes like a bowel.
o If we can “equalize” the signal-to-noise power ratios over the
entire message band, a better noise performance should result.

FM Bandpass x(t) v(t) Baseband Output

Limiter Discriminator lowpass
signal s(t) filter signal
filter

Noise w(t)

Sno (f )
<latexit sha1_base64="Ffi8vRzxW9jh4u8HHOdgfKvUTo4=">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</latexit>

<latexit sha1_base64="j0Ux1NLj7OdTtsxAr8//1qoDioE=">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</latexit>
W 0 W
<latexit sha1_base64="l3oyI1BWgb4ZdkLkM9i+278t6Zk=">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</latexit>
<latexit

© Po-Ning [email protected] 5-60

Pre-Emphasis and De-emphasis for FM

Pre-emphasis De-emphasis
m(t) filter Hpe(f) TX RX filter Hde(f)

Noise w(t)

o Now instead of change/equalize the signal PSD, we produce an

undistorted version of the original message at the receiver output
with
H pe ( f ) H de ( f ) = 1 for - W £ f £ W .
o This relation guarantees the intactness of the message power.
o Next, we need to find Hde(f) such that the noise power is
optimally suppressed.
© Po-Ning [email protected] 5-61
Pre-Emphasis and De-emphasis for FM

Pre-emphasis De-emphasis
m(t) filter Hpe(f) TX RX filter Hde(f)

Noise w(t)

o Under the assumption of high carrier-to-noise ratio, the

noise PSD at the de-emphasis filter output is given by:

W N0 f 2
Þ Average noise power = ò 2
| H de ( f ) |2
df
-W Ac
© Po-Ning [email protected] 5-62
 Pre-Emphasis and De-emphasis for FM
o Since the message power remains the same, the
improvement factor of the output signal-to-noise ratio after
and before pre/de-emphasis is:
W N0 f 2
ò-W Ac2
df
ò
W
f 2 df 2W 3
I= 2
= W
-W
= W
N0 f
ò 3ò
W 2 2
f 2 | H de ( f ) |2 df
ò f | H de ( f ) | df
2
2
| H de ( f ) | df -W -W
-W Ac

© Po-Ning [email protected] 5-63

Pre-Emphasis and De-emphasis for FM

With W = 15 KHz, we plot I as

a function of f0.
I (dB)

f0 (KHz)

© Po-Ning [email protected] 5-64

Pre-Emphasis and De-emphasis for FM
o Final remarks:
n In the previous trial, we simply use a first order
linear filter to improve the system.
n Nonlinear pre-emphasis and de-emphasis filters
have been applied to applications like tape
recording. These techniques, known as Dolby-A,
Dolby-B, and DBX systems, use a combination of
filtering and dynamic range compression to reduce
the effects of noise.

© Po-Ning [email protected] 5-65

Phase-Locked Loop
o Phase-locked loop
t
s(t ) = Ac sin[2pf ct + f1 (t )], where f1 (t ) = 2pk f ò m(t )dt .
0
t
r(t ) = Av cos[2pf ct + f2 (t )], where f2 (t ) = 2pkv ò v(t )dt .
0

km Error
FM wave e(t) Loop Output
s(t) filter v(t)
Feedback
signal
r(t)
VCO

© Po-Ning [email protected] 5-66

Loop filter = lowpass filter + filter h(t).
e( t ) = k m s ( t ) r ( t )
= km Ac sin[2pf c t + f1 (t )] × Av cos[ 2pf c t + f2 (t )]
km Ac Av
= (sin[4pf ct + f1 (t ) + f2 (t )] + sin[f1 (t ) - f2 (t )])
2
Low Pass k A A
® m c v sin[fe (t )], where fe (t ) = f1 (t ) - f2 (t ).
2
¥
Also, v (t ) = ò-¥ {e(t )}LowPass h(t - t )dt .
km Error
FM wave e(t) Loop Output
s(t) filter v(t)
Feedback
signal
r(t)
VCO

© Po-Ning [email protected] 5-67

where k0 = km kv Ac Av / 2.

© Po-Ning [email protected] 5-68

The previous formula suggests an equivalent analytical model for PLL.
2⇡KO
<latexit sha1_base64="CSQbnv9xIUwo/8CCqaG3OJvivnM=">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</latexit>

1 (t) e (t)
h(t)
<latexit sha1_base64="fX7Y3KnezlUp1WHPhcF4O6FG2Uc=">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</latexit> <latexit sha1_base64="/1aOO70bxuKXpNZ4g6XHcl0qhm4=">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</latexit>

+ sin(·)
– <latexit sha1_base64="qaHqfAQLM965htxAzt+EVwArdgk=">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</latexit>

<latexit sha1_base64="443O2KTwcw0HKj4diyU74PIt8fA=">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</latexit>
2 (t)
Z t
dt
<latexit sha1_base64="k9szK0peT5YyxPrmDKhq2M9+DBg=">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</latexit>
0

When fe(t) = 0, the system is said to be in phase-lock.

In this case, f1(t) = f2(t) or equivalently, kvv(t) = kfm(t).

When fe(t) is small (< 0.5 radians), the system is said to be nearly
phase-locked.
In this case, we can approximate sin[fe(t)] by fe(t); hence, a linear
approximate model is resulted.
© Po-Ning [email protected] 5-69
 Linearization approximation model for PLL.
2⇡KO
<latexit sha1_base64="CSQbnv9xIUwo/8CCqaG3OJvivnM=">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</latexit>

1 (t) e (t)
h(t)
<latexit sha1_base64="fX7Y3KnezlUp1WHPhcF4O6FG2Uc=">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</latexit> <latexit sha1_base64="/1aOO70bxuKXpNZ4g6XHcl0qhm4=">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</latexit>

+
–

<latexit sha1_base64="443O2KTwcw0HKj4diyU74PIt8fA=">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</latexit>
2 (t)
Z t
dt
<latexit sha1_base64="k9szK0peT5YyxPrmDKhq2M9+DBg=">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</latexit>
0

We can transform the above time-domain system to its

equivalent frequency domain system to facilitate its analysis.

© Po-Ning [email protected] 5-70

 Linearization approximation model for PLL.
2⇡KO
<latexit sha1_base64="CSQbnv9xIUwo/8CCqaG3OJvivnM=">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</latexit>

F1 ( f ) Fe ( f )
+
– H( f )
F2 ( f )

1
G( f ) =
j 2pf

Fe ( f ) Fe ( f )
=
F1 ( f ) [F1 ( f ) - F 2 ( f )] + F 2 ( f )
Fe ( f )
=
F e ( f ) + 2pk0F e ( f ) H ( f )G ( f )
1 jf
= =
1 + 2pk0 H ( f )G ( f ) jf + k0 H ( f )

© Po-Ning [email protected] 5-71

First-Order PLL
H ( f ) = 1.
Fe ( f ) j ( f / k0 )
=
F1 ( f ) 1 + j ( f / k0 )

A parameter k0 controls both the loop gain and bandwidth of the

filter. In other words, it is impossible to adjust the loop gain
without changing the filter bandwidth.

© Po-Ning [email protected] 5-72

Second-Order PLL
H(f ) = 1 + a/(jf ) and using linear PLL model.

Fe ( f ) jf jf ( jf ) 2
= = =
F1 ( f ) jf + k0 H ( f ) jf + k0 (1 + a /( jf )) ( jf ) 2 + k0 ( jf ) + k0a
( jf / f n ) 2
=
1 + 2z ( jf / f n ) + ( jf / f n ) 2

where natural frequency f n = ak0 and damping factor z = k0 /( 4a ) .

© Po-Ning [email protected] 5-73

Second-Order PLL
Fast response but require
longer time to stabilize

impulse input at the origin Slow response but quick

stabilization.
The output due to an

© Po-Ning [email protected] 5-74

Summary
o Notes
1. SSB modulation is optimum in noise performance
and bandwidth conservation in AM family.
2. FM improves the noise performance of AM family
at the expense of an excessive transmission
bandwidth.
3. FM offers the tradeoff between transmission
bandwidth and noise performance.

© Po-Ning [email protected] 5-75