Frequency and Phase Modulation

To generate an angle modulated signal, the amplitude of the modulated carrier is held
constant while either the phase or the time derivative of the phase is varied linearly with
the message signal m(t).

The expression for an angle modulated signal is:

𝑠(𝑡) = 𝐴𝑐 cos(𝜔𝑐 𝑡 + 𝜃(𝑡)) , 𝜔c is the carrier frequency.

The instantaneous frequency of s(t) is :

1 𝑑 1 𝑑𝜃(𝑡)
𝑓𝑖 (𝑡) = 2𝜋 𝑑𝑡 (𝜔𝑐 𝑡 + 𝜃(𝑡)) = 𝑓𝑐 + 2𝜋 𝑑𝑡

For phase modulation, the phase is directly proportional to the modulating signal :

𝜃(𝑡) = 𝑘𝑝 𝑚(𝑡), kp is the phase sensitivity measured in rad/volt.

The peak phase deviation is

∆𝜃 = 𝑘𝑝 × max (𝑚(𝑡)).

For frequency modulation, the frequency deviation of the carrier is proportional to the
modulating signal:

1 𝑑𝜃(𝑡)
= 𝑘𝑓 𝑚(𝑡) ⟹ 𝑓𝑖 = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡).
2𝜋 𝑑𝑡
The frequency deviation from the un-modulated carrier is
1 𝑑𝜃
𝑓𝑖 (𝑡) − 𝑓𝑐 = 2𝜋 𝑑𝑡

The peak frequency deviation is

1 𝑑𝜃
∆𝑓 = max {2𝜋 𝑑𝑡 }.

The time domain representation of a phase modulated signal is :

𝑠(𝑡) = 𝐴𝑐 cos (𝜔𝑐 𝑡 + 𝑘𝑝 𝑚(𝑡)).

The time domain representation of a frequency modulated signal is

𝑡
𝑠(𝑡) = 𝐴𝑐 cos (𝜔𝑐 𝑡 + 2𝜋𝑘𝑓 ∫−∞ 𝑚(𝛼)𝑑𝛼 ).

𝑡
where 𝜃(𝑡) = 2𝜋𝑘𝑓 ∫−∞ 𝑚(𝛼)𝑑𝛼

1
The average power in 𝑠(𝑡), for frequency modulation (FM) or phase modulation (PM) is:

(𝐴𝑐 )2
𝑝𝑎𝑣𝑎 = = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2

Example: Binary Frequency Shift Keying.

The periodic square signal m(t), shown below, frequency modulates the carrier
𝑐(𝑡) = 𝐴𝑐 cos(2𝜋100𝑡) to produce the signal 𝑠(𝑡) = 𝐴𝑐 cos ( (2𝜋100𝑡) +
2𝜋𝑘𝑓 ∫ 𝑚(𝛼)𝑑𝛼) where kf =10 HZ/V.

a. Find and plot the instantaneous frequency fi(t).

b. Find s(t).

Solution:
a) The instantaneous frequency is
𝑓𝑖 = 𝑓𝑐 + 𝑘𝑓 × 𝑚(𝑡)
𝑓𝑖 = 100+10 =110 Hz when 𝑚(𝑡) = +1

𝑓𝑖 = 100 -10 = 90Hz when 𝑚(𝑡) = −1

For 0 < 𝑡 ≤ 𝑇𝑏 , 𝑓𝑖 = 110 Hz

For 𝑇𝑏 ≤ 𝑡 ≤ 2𝑇𝑏 , 𝑓𝑖 = 90 Hz

The instantaneous frequency hops between the two values 110 Hz and 90 Hz as shown
below

2
 In digital transmission, we will see that a binary (1) may be represent by a signal of
frequency f1 for 0 ≤ 𝑡 ≤ 𝑇𝑏 and a binary (0) by a signal of frequency f2 for 0 ≤ 𝑡 ≤
𝑇𝑏 .

b) The two signals that represent the binary data are:

𝑠1 (𝑡) = Ą cos(2𝜋(110)𝑡 ), 𝑤ℎ𝑒𝑛 𝑚(𝑡) = +1
𝑠2 (𝑡) = Ą cos(2𝜋(90)𝑡 ), 𝑤ℎ𝑒𝑛 𝑚(𝑡) = −1

Exercise: Plot the transmitted signal 𝑠(𝑡) for 0 ≤ 𝑡 ≤ 4𝑇𝑏 assuming 𝑇𝑏 = 10𝑇𝑐 . You
should obtain a figure similar to this figure

Single Tone Frequency Modulation:

Assume that the message 𝑚(𝑡) = 𝐴𝑚 cos 𝜔𝑚 𝑡.

The instantaneous frequency is:

𝑓𝑖 = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡) = 𝑓𝑐 + 𝐴𝑚 𝑘𝑓 cos 2𝜋𝑓𝑚 𝑡.

This frequency is plotted in the figure.

The peak frequency deviation (from the un-modulated

carrier) is :

∆𝑓 = 𝑘𝑓 𝐴𝑚 .

The FM signal is:

𝑠(𝑡) = 𝐴𝑐 cos (𝜔𝑐 𝑡 + 𝛽 sin 2𝜋𝑓𝑚 𝑡).

Where 𝛽 is the FM modulation index:

3
 𝑘𝑓 𝐴𝑚 𝑝𝑒𝑎𝑘 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 ∆𝑓
𝛽= = =
𝑓𝑚 𝑚𝑒𝑠𝑠𝑎𝑔𝑒 𝑏𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑓𝑚

In the figure below, we plot an FM signal when fm=2;fc=20;Ac=1;beta=5;

Spectrum of a Single-Tone FM Signal

The objective is to find a meaningful definition of the bandwidth of an FM signal:

Let 𝑚(𝑡) = 𝐴𝑚 cos 2𝜋𝑓𝑚 𝑡 be the message signal, then the FM signal is:

𝑠(𝑡) = 𝐴𝑐 cos( 2𝜋𝑓𝑐 𝑡 + 𝛽 sin 2𝜋𝑓𝑚 𝑡 )

𝑠(𝑡) can be rewritten as:

𝑠(𝑡) = 𝑅𝑒{𝑒 𝑗(2𝜋𝑓𝑐𝑡+𝛽 sin 2𝜋𝑓𝑚 𝑡 }

= 𝑅𝑒{𝑒 𝑗(2𝜋𝑓𝑐𝑡) × 𝑒 𝑗(𝛽 sin 2𝜋𝑓𝑚𝑡) }

Remember that: 𝑒 𝑗𝜃 = 𝑐𝑜𝑠𝜃 + 𝑗𝑠𝑖𝑛 𝜃 and that 𝑐𝑜𝑠𝜃 = 𝑅𝑒{𝑒 𝑗𝜃 }

1
The function [ 𝛽 sin 2𝜋𝑓𝑚 𝑡 ] is “sinusoidal” and periodic with 𝑇𝑚 = 𝑓 . Therefore,
𝑚
𝑗(𝛽 sin 2𝜋𝑓𝑚 𝑡) 1
𝑒 is also periodic with 𝑇𝑚 = 𝑓 (but not sinusoidal)
𝑚

As we know, a periodic function g(t) can be expanded into a complex Fourier series as:

𝑔(𝑡) = ∑∞
−∞ 𝐶𝑛 𝑒
𝑗𝑛𝜔𝑚 𝑡
.

where,

1 𝑇𝑚
𝐶𝑛 = ∫ 𝑔(𝑡) 𝑒 −𝑗𝜔𝑚𝑛𝑡 𝑑𝑡
𝑇𝑚 0

⟹ If we let 𝑔(𝑡) = 𝑒 𝑗(𝛽 sin 2𝜋𝑓𝑚𝑡)

4
 1 𝑇
then, 𝐶𝑛 = 𝑇 ∫0 𝑚 𝑒 𝑗(𝛽 sin 2𝜋𝑓𝑚𝑡) × 𝑒 −𝑗𝜔𝑚𝑛𝑡 𝑑𝑡
𝑚

It turns out that ⟹ 𝐶𝑛 = 𝐽𝑛 (𝛽).

where 𝐽𝑛 (𝛽) is the Bessel function of the first kind of order n.

Hence, 𝑔(𝑡) = ∑∞
−∞ 𝐽𝑛 (𝛽) 𝑒
𝑗𝑛𝜔𝑚 𝑡

Substituting into 𝑠(𝑡), we get:

⟹ 𝑠(𝑡) = 𝐴𝑐 𝑅𝑒{𝑒 𝑗(2𝜋𝑓𝑐𝑡) × ∑∞

−∞ 𝐽𝑛 (𝛽) 𝑒
𝑗𝑛𝜔𝑚 𝑡
}

= 𝐴𝑐 𝑅𝑒{ ∑∞
−∞ 𝐽𝑛 (𝛽) × 𝑒
𝑗2𝜋(𝑓𝑐 +𝑛𝑓𝑚 )𝑡
}

= 𝐴𝑐 ∑∞
−∞ 𝐽𝑛 (𝛽) × cos(2𝜋(𝑓𝑐 + 𝑛𝑓𝑚 )𝑡)

Finally, the FM signal can be represented as

𝑠(𝑡) = 𝐴𝑐 ∑∞
−∞ 𝐽𝑛 (𝛽) × cos(2𝜋(𝑓𝑐 + 𝑛𝑓𝑚 )𝑡)

Bessel Functions:

The Bessel equation of order n is:

2
𝑑𝑦2 𝑑𝑦
𝑥 +𝑥 + (𝑥 2 − 𝑛2 )𝑦 = 0
𝑑𝑥 2 𝑑𝑥

This is a second order differential equation with variable coefficient. We can solve it by
the power series method, for example:

𝑑𝑦 𝑑𝑦2
Let 𝑦 = ∑∞
𝑛=0 𝐶𝑛 𝑥
𝑛
, = ∑∞
𝑛=1 𝑛 𝐶𝑛 𝑥
𝑛−1
, = ∑∞
𝑛=2 𝑛(𝑛 − 1)𝐶𝑛 𝑥
𝑛−2
.
𝑑𝑥 𝑑𝑥 2

𝑑𝑦 𝑑2 𝑦
Substituting 𝑦, 𝑑𝑥 𝑎𝑛𝑑 into the differential equation and equating terms of equal
𝑑𝑥 2
power results in:

∞ 1
(−1)𝑚 × (2 𝑥)𝑛+2𝑚
𝑦= ∑
𝑚! (𝑛 + 𝑚)!
𝑚=0

The solution for each value of n (see the D.E where n appears) is 𝐽𝑛 (𝑥), the Bessel
function of the first kind of order n. The figure, below, shows the first three Bessel
functions.

5
Some Properties of 𝑱𝒏 (𝒙):

1- 𝐽𝑛 (𝑥) = (−1)𝑛 𝐽−𝑛 (𝑥).

2- 𝐽𝑛 (𝑥) = (−1)𝑛 𝐽𝑛 (−𝑥).

3- Recurrence formula
2𝑛
𝐽𝑛−1 (𝑥) + 𝐽𝑛+1 (𝑥) = 𝐽𝑛 (𝑥) .
𝑥

𝑥𝑛
4- For small values of x: ⟹ 𝐽𝑛 (𝑥) ≅ 2𝑛 𝑛!
Therefore, 𝐽0 (𝑥) ≅ 1
𝑥
𝐽1 (𝑥) ≅ 2.
𝐽𝑛 (𝑥) ≅ 0 𝑓𝑜𝑟 𝑛 > 1
5- For large value of 𝑥:

2 𝜋 𝑛𝜋
𝐽𝑛 (𝑥) ≅ √ cos(𝑥 − − ) , 𝐽𝑛 (𝑥) behaves like a sine function with progressively
𝜋𝑥 4 2
decreasing amplitude.

6- For real 𝑥 and fixed, 𝐽𝑛 (𝑥) ⟶ 0 𝑎𝑠 𝑛 ⟶ ∞ .

7- ∑∞ 2
−∞(𝐽𝑛 (𝑥)) = 1 , 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑥.

6
 Table of Bessel Functions
β J0(β) J1(β) J2(β) J3(β) J4(β) J5(β) J6(β) J7(β) J8(β) J9(β) J10(β)
0 1 0 0 0 0 0 0 0 0 0 0
0.1 0.9975 0.0499 0.0012 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.2 0.9900 0.0995 0.0050 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.3 0.9776 0.1483 0.0112 0.0006 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.4 0.9604 0.1960 0.0197 0.0013 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.5 0.9385 0.2423 0.0306 0.0026 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.6 0.9120 0.2867 0.0437 0.0044 0.0003 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.7 0.8812 0.3290 0.0588 0.0069 0.0006 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.8 0.8463 0.3688 0.0758 0.0102 0.0010 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000
0.9 0.8075 0.4059 0.0946 0.0144 0.0016 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000
1 0.7652 0.4401 0.1149 0.0196 0.0025 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000
1.1 0.7196 0.4709 0.1366 0.0257 0.0036 0.0004 0.0000 0.0000 0.0000 0.0000 0.0000
1.2 0.6711 0.4983 0.1593 0.0329 0.0050 0.0006 0.0001 0.0000 0.0000 0.0000 0.0000
1.3 0.6201 0.5220 0.1830 0.0411 0.0068 0.0009 0.0001 0.0000 0.0000 0.0000 0.0000
1.4 0.5669 0.5419 0.2074 0.0505 0.0091 0.0013 0.0002 0.0000 0.0000 0.0000 0.0000
1.5 0.5118 0.5579 0.2321 0.0610 0.0118 0.0018 0.0002 0.0000 0.0000 0.0000 0.0000
1.6 0.4554 0.5699 0.2570 0.0725 0.0150 0.0025 0.0003 0.0000 0.0000 0.0000 0.0000
1.7 0.3980 0.5778 0.2817 0.0851 0.0188 0.0033 0.0005 0.0001 0.0000 0.0000 0.0000
1.8 0.3400 0.5815 0.3061 0.0988 0.0232 0.0043 0.0007 0.0001 0.0000 0.0000 0.0000
1.9 0.2818 0.5812 0.3299 0.1134 0.0283 0.0055 0.0009 0.0001 0.0000 0.0000 0.0000
2 0.2239 0.5767 0.3528 0.1289 0.0340 0.0070 0.0012 0.0002 0.0000 0.0000 0.0000
2.1 0.1666 0.5683 0.3746 0.1453 0.0405 0.0088 0.0016 0.0002 0.0000 0.0000 0.0000
2.2 0.1104 0.5560 0.3951 0.1623 0.0476 0.0109 0.0021 0.0003 0.0000 0.0000 0.0000
2.3 0.0555 0.5399 0.4139 0.1800 0.0556 0.0134 0.0027 0.0004 0.0001 0.0000 0.0000
2.4 0.0025 0.5202 0.4310 0.1981 0.0643 0.0162 0.0034 0.0006 0.0001 0.0000 0.0000
2.5 -0.0484 0.4971 0.4461 0.2166 0.0738 0.0195 0.0042 0.0008 0.0001 0.0000 0.0000
2.6 -0.0968 0.4708 0.4590 0.2353 0.0840 0.0232 0.0052 0.0010 0.0002 0.0000 0.0000
2.7 -0.1424 0.4416 0.4696 0.2540 0.0950 0.0274 0.0065 0.0013 0.0002 0.0000 0.0000
2.8 -0.1850 0.4097 0.4777 0.2727 0.1067 0.0321 0.0079 0.0016 0.0003 0.0000 0.0000
2.9 -0.2243 0.3754 0.4832 0.2911 0.1190 0.0373 0.0095 0.0020 0.0004 0.0001 0.0000
3 -0.2601 0.3391 0.4861 0.3091 0.1320 0.0430 0.0114 0.0025 0.0005 0.0001 0.0000
3.1 -0.2921 0.3009 0.4862 0.3264 0.1456 0.0493 0.0136 0.0031 0.0006 0.0001 0.0000
3.2 -0.3202 0.2613 0.4835 0.3431 0.1597 0.0562 0.0160 0.0038 0.0008 0.0001 0.0000
3.3 -0.3443 0.2207 0.4780 0.3588 0.1743 0.0637 0.0188 0.0047 0.0010 0.0002 0.0000
3.4 -0.3643 0.1792 0.4697 0.3734 0.1892 0.0718 0.0219 0.0056 0.0012 0.0002 0.0000
3.5 -0.3801 0.1374 0.4586 0.3868 0.2044 0.0804 0.0254 0.0067 0.0015 0.0003 0.0001
3.6 -0.3918 0.0955 0.4448 0.3988 0.2198 0.0897 0.0293 0.0080 0.0019 0.0004 0.0001
3.7 -0.3992 0.0538 0.4283 0.4092 0.2353 0.0995 0.0336 0.0095 0.0023 0.0005 0.0001
3.8 -0.4026 0.0128 0.4093 0.4180 0.2507 0.1098 0.0383 0.0112 0.0028 0.0006 0.0001
3.9 -0.4018 -0.0272 0.3879 0.4250 0.2661 0.1207 0.0435 0.0130 0.0034 0.0008 0.0002
4 -0.3971 -0.0660 0.3641 0.4302 0.2811 0.1321 0.0491 0.0152 0.0040 0.0009 0.0002
4.1 -0.3887 -0.1033 0.3383 0.4333 0.2958 0.1439 0.0552 0.0176 0.0048 0.0011 0.0002
4.2 -0.3766 -0.1386 0.3105 0.4344 0.3100 0.1561 0.0617 0.0202 0.0057 0.0014 0.0003
4.3 -0.3610 -0.1719 0.2811 0.4333 0.3236 0.1687 0.0688 0.0232 0.0067 0.0017 0.0004
4.4 -0.3423 -0.2028 0.2501 0.4301 0.3365 0.1816 0.0763 0.0264 0.0078 0.0020 0.0005
4.5 -0.3205 -0.2311 0.2178 0.4247 0.3484 0.1947 0.0843 0.0300 0.0091 0.0024 0.0006
4.6 -0.2961 -0.2566 0.1846 0.4171 0.3594 0.2080 0.0927 0.0340 0.0106 0.0029 0.0007
4.7 -0.2693 -0.2791 0.1506 0.4072 0.3693 0.2214 0.1017 0.0382 0.0122 0.0034 0.0008
4.8 -0.2404 -0.2985 0.1161 0.3952 0.3780 0.2347 0.1111 0.0429 0.0141 0.0040 0.0010
4.9 -0.2097 -0.3147 0.0813 0.3811 0.3853 0.2480 0.1209 0.0479 0.0161 0.0047 0.0012
5 -0.1776 -0.3276 0.0466 0.3648 0.3912 0.2611 0.1310 0.0534 0.0184 0.0055 0.0015

7
 The Fourier Series Representation of the FM Signal

We saw earlier that a single tone FM signal can be represented in a Fourier series as :
∞

𝑠(𝑡) = 𝐴𝑐 ∑ 𝐽𝑛 (𝛽) cos(2𝜋(𝑓𝑐 + 𝑛𝑓𝑚 )𝑡)

−∞
The first few terms in this expansion are:

𝑠(𝑡) = 𝐴𝑐 { 𝐽0 (𝛽) cos(2𝜋𝑓𝑐 𝑡) + 𝐽1 (𝛽) cos 2𝜋(𝑓𝑐 + 𝑓𝑚 )𝑡 + 𝐽−1 (𝛽) cos 2𝜋(𝑓𝑐 − 𝑓𝑚 )𝑡 +
𝐽2 (𝛽) cos 2𝜋(𝑓𝑐 + 2𝑓𝑚 )𝑡 + 𝐽−2 (𝛽) cos 2𝜋(𝑓𝑐 − 2𝑓𝑚 )𝑡 + ⋯ . .

The FM signal consists of an infinite number of spectral components concentrated around

𝑓𝑐 . Therefore, the theoretical bandwidth of the signal is infinity. That is to say, if we need
to recover the FM signal without any distortion, all spectral components must be
accommodated. This means that a channel with infinite bandwidth is needed. This is, of
course, not practical since the frequency spectrum is shared by many users.

In the following discussion we need to truncate the series so that say 99% of the total
average power is contained within a certain bandwidth. But first let us find the total
average power using the series approach.

Power in the spectral components of s(t)

Note that s(t) consists of an infinite number of Fourier terms, and the power in s(t) will be
equal the power in the respective Fourier components .

Any term in s (t) takes the form: 𝐴𝑐 𝐽𝑛 (𝛽) cos(2𝜋(𝑓𝑐 + 𝑛𝑓𝑚 )𝑡)

(𝐴𝑐 )2 (𝐽𝑛 (𝛽))2

The average power in this term is: 2

Hence the total power in s(t) is:

2 (𝑡)
𝐴𝐶 2 𝐽0 2 (𝛽) 𝐴𝐶 2 𝐽1 2 (𝛽) 𝐴𝐶 2 𝐽−1 2 (𝛽) 𝐴𝐶 2 𝐽2 2 (𝛽) 𝐴𝐶 2 𝐽−2 2 (𝛽)
<𝑆 > = + + + + +⋯
2 2 2 2 2

𝐴𝐶 2
= { 𝐽0 2 (𝛽) + 𝐽1 2 (𝛽)+𝐽−1 2 (𝛽)+𝐽2 2 (𝛽)+𝐽−2 2 (𝛽) + ⋯ . . }
2

𝐴𝐶 2
= 2
{ ∑∞ 2 ∞ 2
𝑛=−∞ 𝐽𝑛 (𝛽) } , where ∑𝑛=−∞ 𝐽𝑛 (𝛽) = 1, ( A property of Bessel
Functions).

8
 The average power becomes
𝐴𝐶 2
< 𝑆 2 (𝑡) > = .
2

Spectrum of an Fm Signal

Fourier transform of 𝑠(𝑡) = 𝐴𝑐 cos (𝜔𝑐 𝑡 + 𝛽 sin 2𝜋𝑓𝑚 𝑡) (only + ve frequencies

shown)

Note that in the figure above as 𝑓𝑚 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠, the spectral lines become closely
clustered about 𝑓𝑐 .

The power spectral density, which is a plot of |𝐶𝑛 |2 versus 𝑓, is shown below:

Example: 99% power bandwidth of an FM signal

Plot the FM spectrum and find the 99% power bandwidth when 𝛽 = 1 and 𝛽 = 0.2

Solution:

9
 𝑠(𝑡) = 𝐴𝑐 ∑∞
−∞ 𝐽𝑛 (𝛽) cos(2𝜋(𝑓𝑐 + 𝑛𝑓𝑚 )𝑡)

Case a: 𝜷 = 𝟏 (wideband FM)

The first five terms corresponding to 𝛽 = 1, are

𝐽0 (1) = 0.7652, 𝐽1 (1) = 0.4401, 𝐽2 (1) = 0.1149, 𝐽3 (1) = 0.01956, 𝐽4 (1) = 0.002477

𝐴𝐶 2
The power in s(t) is < 𝑆 2 (𝑡) > = 2

Let us try to find the average power in the terms at 𝑓𝑐 , 𝑓𝑐 + 𝑓𝑚 , 𝑓𝑐 − 𝑓𝑚 , 𝑓𝑐 + 2𝑓𝑚 , 𝑓𝑐 −

2𝑓𝑚

The average power in these five components can be calculated as:

𝐴𝐶 2 𝐽0 2 (𝛽)
1. 𝑓𝑐 : 2
𝐴𝐶 2 𝐽1 2 (𝛽)
2. 𝑓𝑐 + 𝑓𝑚 : 2
𝐴𝐶 2 𝐽−1 2 (𝛽)
3. 𝑓𝑐 − 𝑓𝑚 : 2
𝐴𝐶 2 𝐽2 2 (𝛽)
4. 𝑓𝑐 + 2𝑓𝑚 : 2
𝐴𝐶 2 𝐽−2 2 (𝛽)
5. 𝑓𝑐 − 2𝑓𝑚 : 2

The average power in the five spectral components is the sum:

𝐴2 𝑐
𝑃𝑎𝑣 = [𝐽0 2 (1) + 2𝐽1 2 (1) + 2𝐽2 2 (1)]
2

𝐴2 𝑐 𝐴2 𝑐
𝑃𝑎𝑣 = [(0.7652)2 + 2 ∗ (0.4401)2 + (0.1149)2 ] = 0.9993
2 2
So, these terms have 99.9 % of the total power.

Therefore, the 99.9 % power bandwidth is

𝐵𝑊 = (𝑓𝑐 + 2𝑓𝑚 ) − (𝑓𝑐 − 2𝑓𝑚 ) = 4𝑓𝑚

Case b: 𝛃 = 𝟎. 𝟐 (Narrowband FM)

For 𝛽 = 0.2, 𝐽0 (0.2) = 0.99, 𝐽1 (0.2) = 0.0995, 𝐽2 (0.2) = 0.00498335

The power in the carrier and the two sidebands (at 𝑓𝑐 , 𝑓𝑐 + 𝑓𝑚 , 𝑓𝑐 − 𝑓𝑚 ) is

𝐴2 𝑐
𝑃= 2
[𝐽0 2 (0.2) + 2𝐽1 2 (0.2)]

10
 𝐴2 𝑐
𝑃= [0.9999]
2

Therefore, 99.99% of the total power is found in the carrier and the two sidebands. The
99% bandwidth is

B.W= (𝑓𝑐 + 𝑓𝑚 ) − (𝑓𝑐 − 𝑓𝑚 ) = 2𝑓𝑚

Remark:

We observe that the spectrum of an FM signal when 𝛽 ≪ 1 ( called narrow band FM) is
“similar” to the spectrum of a normal AM signal, in the sense that it consists of a carrier
and two sidebands. The bandwidth of both signals is 2𝑓𝑚 .

Carson’s Rule

A 98% power B.W of an FM signal can be estimated using Carson’s rule:

𝐵𝑇 = 2(𝛽 + 1)𝑓𝑚

Generation of an FM Signal

First: Generation of a Narrowband FM Signal

Consider an angel modulated signal:

𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡 + 𝜃(𝑡))

When 𝑠(𝑡) is an FM signal, 𝜃(𝑡) = 2𝜋𝑘𝑓 ∫ 𝑚(𝑡)𝑑𝑡

𝑠(𝑡) can be expanded as:

𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) cos(𝜃(𝑡)) − 𝐴𝑐 sin(2𝜋𝑓𝑐 𝑡) sin(𝜃(𝑡))

When |𝜃(𝑡)| ≪ 1 , 𝑐𝑜𝑠 𝜃 ≅ 1 , sin(𝜃) ≅ 𝜃 and 𝑠(𝑡) , termed narrowband, can be

approximated as:

𝑠(𝑡) ≅ 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) − 𝐴𝑐 𝜃 sin(2𝜋𝑓𝑐 𝑡)

Using this expression, one can generate a narrowband FM or PM signals. This is

illustrated in the block diagram below:

11
 When 𝑚(𝑡) = 𝐴𝑚 cos(2𝜋𝑓𝑚 𝑡)

𝜃(𝑡) = 𝛽sin(2𝜋𝑓𝑚 𝑡)

the modulated signal takes the form

𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) − 𝐴𝑐 𝛽 sin(2𝜋𝑓𝑚 𝑡) sin(2𝜋𝑓𝑐 𝑡)

To generate a narrow band PM signal, we can use the scheme:

Spectrum of a single- tone NBFM:

For an FM signal, 𝜃(𝑡) = 𝛽sin(2𝜋𝑓𝑚 𝑡)

𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) − 𝐴𝑐 𝛽 sin(2𝜋𝑓𝑚 𝑡) sin(2𝜋𝑓𝑐 𝑡)

𝐴𝑐 𝛽
𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) − [cos(2𝜋(𝑓𝑐 − 𝑓𝑚 )𝑡) − cos(2𝜋(𝑓𝑐 + 𝑓𝑚 )𝑡)]
2

The spectrum of 𝑠(𝑡) is shown below:

12
The spectrum consists of a component at the carrier frequency 𝑓𝑐 , two components at
(𝑓𝑐 + 𝑓𝑚 ) and 𝑓𝑐 − 𝑓𝑚 ). Note the negative sign at the lower sideband.

The bandwidth of this signal is 2𝑓𝑚 .

Now, consider the normal AM signal with sinusoidal modulation.

𝑠(𝑡)𝐴𝑀 = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) + 𝐴𝑐 𝐴𝑚 cos (2𝜋𝑓𝑚 𝑡) cos(2𝜋𝑓𝑐 𝑡)

It can be represented as
𝐴𝑐 𝐴𝑚
𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡) − [cos(2𝜋(𝑓𝑐 − 𝑓𝑚 )𝑡) + cos(2𝜋(𝑓𝑐 + 𝑓𝑚 )𝑡)]
2

As we recall this signal consists of a term at the carrier and two terms at 𝑓𝑐 + 𝑓𝑚 and 𝑓𝑐 −
𝑓𝑚 .

13
Frequency multiplier
It is a device for which the frequency of the output signal is an integer multiple of the
frequency of the input signal. It is primarily a nonlinear characteristic followed by a band
pass filter. Now we illustrate the operation of this device.

The Square law device:

Let the input be an FM signal of the form:

𝑥(𝑡) = 𝐴𝑐 cos(2π𝑓 ′ 𝑐 + 𝛽′ sin 2π𝑓𝑚 𝑡) = 𝐴𝑐 cos(𝜙)

The output of the square law characteristic is:

𝐴𝑐2 𝐴𝑐2 𝐴𝑐2

𝑦(𝑡) = 𝑥(𝑡)2 = 𝐴𝑐2 𝑐𝑜𝑠 2 (𝜙) = [1 + cos(2𝜙)] = + cos(2𝜙)
2 2 2
𝐴𝑐2 𝐴𝑐2
= + 𝑐𝑜𝑠[2π(2𝑓 ′ 𝑐 ) + 2𝛽′sin(2π𝑓𝑚 𝑡)]
2 2
The bandpass filter

If 𝑦(𝑡) is passed through a BPF of center frequency 2𝑓𝑐 , then the DC term will be
suppressed and the filter output is:
𝐴𝑐2
𝑦 ′ (𝑡) = 𝑐𝑜𝑠[2π(2𝑓 ′ 𝑐 ) + 2𝛽′sin(2π𝑓𝑚 𝑡)]
2

𝐴𝑐2
𝑦 ′ (𝑡) = 𝑐𝑜𝑠[2π(𝑓𝑐 ) + 𝛽sin(2π𝑓𝑚 𝑡)]
2

As can be seen from this result, the output is a signal with twice the frequency of the
input signal and a modulation index twice that of the input. To get frequency
multiplication higher than two, a cascade of units, similar to what was described above,
can be formed with the number of stages that achieve the desired frequency.

Indirect Method for Generating a Wideband FM:

A wideband FM can be generated indirectly using the block diagram below (Armstrong
Method). First, a narrowband FM is generated, and then the wideband FM is obtained by
using frequency multiplication. Next, we analyze the operation of this modulator.

14
Let 𝑚(𝑡) = 𝐴𝑚 cos 2π𝑓𝑚 𝑡 be the baseband signal, then
𝑘𝑓 𝐴 𝑚
𝑠1 (𝑡) = 𝐴𝑐 cos(2π𝑓𝑐′ 𝑡 + 𝛽 ′ sin 2π𝑓𝑚 𝑡) ; 𝛽 ′ = 𝑓
𝑚

is a narrowband FM with 𝛽 ′ ≪ 1. The frequency of 𝑠1 (𝑡) is

𝑓 ′ 𝑖 = 𝑓𝑐′ + 𝑘𝑓 𝐴𝑚 cos 2π𝑓𝑚 𝑡

Multiplying 𝑓𝑖 by 𝑛 (through frequency multiplication), we get the frequency of 𝑠(𝑡) as

𝑓𝑖 = 𝑛𝑓𝑐′ + 𝑛𝑘𝑓 𝐴𝑚 𝑐𝑜𝑠2π𝑓𝑚 𝑡

This result is

𝑠(𝑡) = 𝐴𝑐 cos[2π(𝑛𝑓𝑐′ )𝑡 + 𝑛𝛽 ′ sin 2π𝑓𝑚 𝑡]

= 𝐴𝑐 cos[2π𝑓𝑐 𝑡 + 𝛽 sin 2π𝑓𝑚 𝑡]

Where 𝛽 = 𝑛𝛽′ is the desired modulation index of WBFM

𝑓𝑐 = 𝑛𝑓𝑐′ is the desired carrier frequency of WBFM

Direct method for generating an FM signal:

In a direct FM system, the instantaneous frequency of the carrier is varied in accordance
with a message signal by means of a voltage controlled oscillator (VCO). The voltage –
frequency characteristic of a VCO is given by

𝑓𝑖 = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡)

A schematic diagram of a VCO is shown in the figue

A realization of the CVO may be obtained by considering an oscillator (like the Hartley
oscillator) shown below in which a varactor ((voltage variable capacitor) is used. The
capacitance of the varactor varies in response to variations in the message signal. The
variation is linear when the variation in the message is too small.

15
The frequency of the oscillator is

1
𝑓𝑖 (𝑡) =
2𝜋√(𝐿1 + 𝐿2 )𝐶(𝑡)

Let 𝐶(𝑡) = 𝐶0 − 𝑘 𝑚(𝑡) (A diode

operating in the reverse bias region can act like
a variable capacitor)

k: is a constant,

When m(t) = 0, 𝐶(𝑡) = 𝐶0, and the

unmodulated frequency of oscillation is
1
𝑓𝑐 =
√(𝐿1 +𝐿2 )𝐶0

When m(t) has a finite value, the frequency of oscillation is Hartley Oscillator

1
𝑓𝑖 (𝑡) =
2𝜋√(𝐿1 + 𝐿2 )(𝐶0 − 𝑘 𝑚(𝑡))

−1⁄
2
𝑘 𝑚(𝑡)
= 𝑓𝑐 (1 − ) , [(1 + 𝑥)𝑛 ≅ 1 + 𝑛𝑥]
𝐶0

𝑘 𝑚(𝑡)
When ≪ 1 , we can make the approximation
𝐶0

𝑘 𝑚(𝑡)
𝑓𝑖 (𝑡) = 𝑓𝑐 (1 + ) = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡)
2𝐶0

Here it is clear that the instantaneous frequency varies linearly with the message signal.

Demodulation of the FM signal:

An FM signal may be demodulated by means of what is called a discriminator.

Let 𝑠(𝑡) = 𝐴𝑐 cos(𝜔𝑐 𝑡 + 𝜃(𝑡)) be an angle

Discriminator
𝑠(𝑡) 𝑦(𝑡)
modulated signal. The output of an ideal

discriminator is defined as:

1 𝑑𝜃
𝑦(𝑡) = 𝑘𝐷
2𝜋 𝑑𝑡

16
 𝑡 𝑑𝜃
When 𝜃 = 2𝜋𝑘𝑓 ∫−∞ 𝑚(𝛼)𝑑𝛼 , then = 2𝜋𝑘𝑓 𝑚(𝑡) and y(t) becomes
𝑑𝑡
𝑦(𝑡) = 𝑘𝐷 𝑘𝑓 𝑚(𝑡)

One practical realization of a discriminator is a differentiator followed by an envelope

detector.

The operation of this discriminator can be explained as follows:

Let 𝑠(𝑡) = 𝐴𝑐 cos(𝜔𝑐 𝑡 + 𝜃(𝑡))

𝑑𝑠(𝑡) 𝑑𝜃
= −𝐴𝑐 (𝜔𝑐 + ) sin(𝜔𝑐 𝑡 + 𝜃(𝑡))
𝑑𝑡 𝑑𝑡
𝑑𝜃
The output of the envelope detector is 𝐴𝑐 | (𝜔𝑐 + 𝑑𝑡 ) |

The capacitor blocks the DC term and so output is:

𝑑𝜃
𝑉0 = 𝐴𝑐 = 2𝜋𝑘𝑓 𝐴𝑐 𝑚(𝑡)
𝑑𝑡
A typical FM signal and its derivative are shown in the figure below.

17
We already know what an envelope detector (recall the material on the demodulation of a
normal AM signal). Now we explain how differentiation is accomplished.

18
From the properties of Fourier transform we know that if 𝐹{𝑔(𝑡)} = 𝐺(𝑓) , then
𝑑𝑔(𝑡)
𝐹{ } = 𝑗2𝜋𝑓 𝐺(𝑓)
𝑑𝑡

This means that multiplication by 𝑗2𝜋𝑓 in the frequency

domain amounts to differentiating the signal in the time-
domain. Hence, we need a circuit whose frequency response is
linear in 𝑓 to perform time differentiation. A circuit that
performs this task is a tuned circuit, provided that the signal
frequency falls within the linear part of the characteristic, i.e.,
between either (𝑓1 , 𝑓2 ) 𝑜𝑟 (𝑓3 , 𝑓4 ).

A balanced FM detector called balanced discriminator is such a

circuit.

Tuned circuit demodulator

To extend the dynamic range of the differentiating circuit, two tuned circuits with center
frequencies 𝑓𝑜1 𝑎𝑛𝑑 𝑓𝑜2 are used as shown in the figure

Balanced slope detector:

Two tuned circuits are tuned to two different frequencies
𝑓𝑜1 > 𝑓𝑜2 . The primary circuit is tuned to 𝑓𝑐 .

 This circuit has a wider range of linear frequency

𝑠(𝑡) Output
response.
 No DC blocking is necessary.

19
20
Phase shift discriminator (not required to ENCS students)
The quadrature detector: This demodulator converts frequency variations into phase
variation and detects the phase changes. The block diagram of the demodulator is shown
below

𝑡
Let 𝑠(𝑡) = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡 + 𝜑(𝑡)) ; 𝜑(𝑡) = 2𝜋𝐾𝑓 ∫0 𝑚(𝛼) 𝑑𝛼

𝑠(𝑡 − ∆𝑡) = 𝐴𝑐 cos[2𝜋𝑓𝑐 (𝑡 − ∆𝑡) + 𝜑(𝑡 − ∆𝑡)]

= 𝐴𝑐 cos[2𝜋𝑓𝑐 𝑡 − 2𝜋𝑓𝑐 ∆𝑡 + 𝜑(𝑡 − ∆𝑡)]

The delay ∆𝑡 is chosen such that 2𝜋𝑓𝑐 ∆𝑡 = 𝜋⁄2

Hence,
𝜋
𝑠(𝑡 − ∆𝑡) = 𝐴𝑐 cos [2𝜋𝑓𝑐 𝑡 − 2 + 𝜑(𝑡 − ∆𝑡)]

= 𝐴𝑐 sin[2𝜋𝑓𝑐 𝑡 + 𝜑(𝑡 − ∆𝑡)]

𝑉(𝑡) = 𝑠(𝑡)𝑠(𝑡 − ∆𝑡) = 𝐴2𝑐 sin[2𝜋𝑓𝑐 𝑡 + 𝜑(𝑡 − ∆𝑡)] 𝑐𝑜𝑠[2𝜋𝑓𝑐 𝑡 + 𝜑(𝑡)]

𝐴2𝑐 𝐴2𝑐
= sin[2𝜋(2𝑓𝑐 )𝑡 + 𝜑(𝑡) + 𝜑(𝑡 − ∆𝑡)] + sin[𝜑(𝑡) − 𝜑(𝑡 − ∆𝑡)]
2 2

The high frequency component is suppressed by the LPF. What remains is the second
term
𝐴2𝑐 𝐴2𝑐
sin[𝜑(𝑡) − 𝜑(𝑡 − ∆𝑡)] ≅ [𝜑(𝑡) − 𝜑(𝑡 − ∆𝑡)]
2 2
where ∆𝑡 is small to justify the approximation sin(𝑥) ≅ 𝑥. Hence,

𝐴2𝑐
𝑦(𝑡) = [𝜑(𝑡) − 𝜑(𝑡 − ∆𝑡)]
2

𝐴2𝑐 𝜑(𝑡)−𝜑(𝑡−∆𝑡)
𝑦(𝑡) = ∆𝑡
2 ∆𝑡

21
 𝑑𝜑(𝑡)
The second term is the derivative . The output then becomes
𝑑𝑡

𝐴2𝑐 𝑑𝜑
𝑦(𝑡) = ∆𝑡
2 𝑑𝑡

𝑡 𝑑
But 𝜑(𝑡) = 2𝜋𝑘𝑓 ∫0 𝑚(𝛼) 𝑑𝛼 and 𝜑(𝑡) = 2𝜋𝑘𝑓 𝑚(𝑡)
𝑑𝑡

𝐴2𝑐
𝑦(𝑡) = ∆𝑡2𝜋𝑘𝑓 𝑚(𝑡)
2

𝑦(𝑡) = 𝐾 𝑚(𝑡)
Therefore, 𝑚(𝑡) has been demodulated.

Transfer function of the delay:

From the Fourier transform properties

𝑔(𝑡) → 𝐺(𝑓)

𝑔(𝑡 − ∆𝑡) → 𝐺(𝑓)𝑒 −𝑗2𝜋𝑓∆𝑡

The transfer function of the time delay is

𝐻(𝑓) = 𝑒 −𝑗2𝜋𝑓∆𝑡

Therefore, a circuit whose phase characteristic is linear in 𝑓 can provide time delay of the
type that we need.

A circuit with linear phase characteristic is the network shown

𝑉𝑖 (𝑡) 𝑉𝑜 (𝑡)

1 𝑅
If 𝑓𝑜 = , 𝑓𝑏 = 2𝜋𝐿then it can be shown that arg (𝐻(𝑓)) for this circuit is
2𝜋√𝐿𝐶

𝜋 2𝑄 𝑓
arg(𝐻(𝑓)) = − 2 − (𝑓 − 𝑓𝑐 ) , 𝑄 = 𝑜
𝑓𝑜 𝑓 𝑏

22
 Θ(f) = 𝑎 − 𝑏𝑓

Remarks:

1. To perform time differentiation, we searched for a circuit whose amplitude

spectrum varies linearly with frequency
2. To perform time delay, we searched for a circuit with a linear phase spectrum.

The Super heterodyne Receiver:

Practically, all radio and TV receivers are made of the super heterodyne type. The
receiver performs the following functions :

o Carrier frequency tuning: The purpose of which is to select the desired

signal.
o Filtering: the desired signal is to be separated from other modulated
signals.
o Amplification: to compensate for the loss of signal power incurred in the
course of transmission.

The description of the receiver is summarized as follows:

o The incoming signal is picked up by the antenna and amplified in the RF

section that is tuned to the carrier frequency of the incoming signal.
o The incoming RF section is down converted to a fixed intermediate
frequency (IF). 𝑓𝐼𝐹 = 𝑓𝑙𝑂 − 𝑓𝑅𝐹
o The IF section provides most of the amplification and selectivity in the
receiver. The IF bandwidth corresponds to that required for the particular
type of modulation.
o The IF output is applied to a demodulator, the purpose of which is to
recover the baseband signal.

23
 o The final operation in the receiver is the power amplification of the
recovered signal.
o The basic difference between AM and FM super heterodyne lies in the
use of an FM demodulator such as a discriminator (differentiator
followed envelope detector)

Quadrature Carrier Multiplexing (QAM) (will be covered)

Quadrature Carrier Multiplexing: Modulation

This scheme enables two DSB-SC modulated signals to occupy the same transmission
B.W and yet allows for the separation of the message signals at the receiver.

𝑚1 (𝑡) and 𝑚2 (𝑡) are low pass signals each with a B.W = W Hz .

The composite signal is:

𝑠(𝑡) = 𝐴𝑐 𝑚1 (𝑡)𝑐𝑜𝑠2𝜋𝑓𝑐 𝑡 + 𝐴𝑐 𝑚2 (𝑡)𝑠𝑖𝑛2𝜋𝑓𝑐 𝑡

𝑠(𝑡) = 𝑠1 (𝑡) + 𝑠2 (𝑡)

where 𝑠1 (𝑡) and 𝑠2 (𝑡) are both DSB-SC signals.

B.W of 𝑠1 (𝑡) = 2W

B.W of 𝑠2 (𝑡) = 2W

B.W of 𝑠(𝑡) = 2W

This method provides bandwidth conservation. That is, two DSB-SC signals are
transmitted within the bandwidth of one DSB-SC signal. Therefore, this multiplexing
technique provides bandwidth reduction by one half.

24
Quadrature Carrier Multiplexing: Demodulation

Given 𝑠(𝑡), the objective is to recover 𝑚1 (𝑡) and 𝑚2 (𝑡) from 𝑠(𝑡) . Consider first the
in-phase channel

𝑥1 (𝑡) = 2𝑐𝑜𝑠2𝜋𝑓𝑐 𝑡 𝑠(𝑡)

= 2 cos 2𝜋𝑓𝑐 𝑡 (𝐴𝑐 𝑚1 (𝑡)𝑐𝑜𝑠2𝜋𝑓𝑐 𝑡 + 𝐴𝑐 𝑚2 (𝑡)𝑠𝑖𝑛2𝜋𝑓𝑐 𝑡)
= 2𝐴𝑐 𝑚1 (𝑡)𝑐𝑜𝑠 2 2𝜋𝑓𝑐 𝑡 + 2𝐴𝑐 𝑚2 (𝑡)𝑠𝑖𝑛𝜔𝑐 𝑡 𝑐𝑜𝑠𝜔𝑐 𝑡
1+𝑐𝑜𝑠2𝜔𝑐 𝑡
= 2𝐴𝑐 𝑚1 (𝑡) ( ) + 𝐴𝑐 𝑚2 (𝑡)𝑠𝑖𝑛2𝜔𝑐 𝑡
2

= 𝐴𝑐 𝑚1 (𝑡) + 𝐴𝑐 𝑚1 (𝑡)𝑐𝑜𝑠2𝜔𝑐 𝑡 + 𝐴𝑐 𝑚2 (𝑡)𝑠𝑖𝑛2𝜔𝑐 𝑡

After low pass filtering, the output of the in-phase channel is

𝑦1 (𝑡) = 𝐴𝑐 𝑚1 (𝑡).

Likewise, it can be shown that

𝑦2 (𝑡) = 𝐴𝑐 𝑚2 (𝑡) .

Note: Synchronization is a problem. That is to recover the message signals it is

important that the two carrier signals (the sine and the cosine functions) at the receiver
should have the same phase and frequency as the signals at the transmitting side. A phase
error or a frequency error will result is an interference type of distortion. That is, A
component of 𝑚2 (𝑡) will appear in the in-phase channel in addition to the desired signal
𝑚1 (𝑡) and a component of 𝑚1 (𝑡) will appear at the quadrature output.

25
 A multiple access channel is one where a set of users at one end want
to communicate with another set of users at the other end.
 Channel Allocation: The coordination of the usage of a single
channel among multiple source – destination pairs.
 The algorithm which implements the channel allocation are called
medium access control (MAC) or multiple access protocols.
 MAC protocols can be classified into
o Conflict free protocols:
o Random access protocols:

 A Conflict free protocols: Collisions are completely avoided by

allocating the channel access to sources in a predetermined manner.
Examples are TDMA, FDMA and CDMA. This is equivalent to
circuit switching and is inefficient for bursty type of loads.
 Random access protocols: These are classified as contention systems
where the stations compete to access the channel. The contention
could be completely random or controlled.
 Collisions can occur between transmitted packets of different users
trying to access the channel.
 A collided packet has to be transmitted until it is received properly at
the destination.

26
Frequency Division Multiplexing: (will be covered)
A number of independent signals can be combined into a composite signal suitable for
transmission over a common channel. The signals must be kept apart so that they do not
interfere with each other and thus they can be separated at the receiving end.

Transmitter:

Receiver:

27
Example: Single Sideband Frequency Division Multiplexed Signals

Example: Double Sideband Frequency Division Multiplexed Signals

Let 𝑚1 , 𝑚2 𝑎𝑛𝑑 𝑚3 be three baseband message signals each with a B.W = w.

The composite modulated signal 𝑠(𝑡) is

𝑠(𝑡) = 𝐴𝑐1 𝑚1 (𝑡)𝑐𝑜𝑠2𝜋𝑓1 𝑡 + 𝐴𝑐2 𝑚2 (𝑡)𝑐𝑜𝑠2𝜋𝑓2 𝑡 + 𝐴𝑐3 𝑚3 (𝑡)𝑐𝑜𝑠2𝜋𝑓3 𝑡

= 𝑠1 (𝑡) + 𝑠2 (𝑡) + 𝑠3 (𝑡)

𝑠1 , 𝑠2 𝑎𝑛𝑑 𝑠3 are DSB-SC signals with carrier frequencies 𝑓1 , 𝑓2 𝑎𝑛𝑑 𝑓3 , respectively. If

the spectrum of 𝑚1 (𝑡), 𝑚2 (𝑡) 𝑎𝑛𝑑 𝑚3 (𝑡) are as shown, the spectrum of 𝑠(𝑡) can be
found as shown below.

28
To prevent interference we demand that

𝑓2 − 𝑤 ≥ 𝑓1 + 𝑤 𝑜𝑟 𝑓2 − 𝑓1 ≥ 2𝑤

𝑓3 − 𝑤 ≥ 𝑓2 + 𝑤 𝑜𝑟 𝑓3 − 𝑓2 ≥ 2𝑤

The structure of the receiver is as follows:

29