Recall two useful properties of Fourier Transforms

If 𝑔 𝑡 ⟺𝐺 𝑓 ,

1. Time-shifting property:

𝑔(𝑡 − 𝑡0 ) ⇔ 𝐺(𝑓)𝑒 −𝑗2𝜋𝑓𝑡0

2. Frequency-shifting property:

𝑔(𝑡)𝑒 𝑗2𝜋𝑓0 𝑡 ⇔ 𝐺(𝑓 − 𝑓0 )

I Sem 2020-21 ECE/EEE F311 Communication Systems 1

 Parseval’s Theorem

Parseval's theorem usually refers to the result that the sum (or
integral) of the square of a function is equal to the sum (or integral)
of the square of its transform.

For Periodic signals:

∞ ∞
1 𝑇0
𝑃𝑔 = 𝑔(𝑡) 2 𝑑𝑡 = 𝐷𝑛 2 = 𝐷0 2 + 2 𝐷𝑛 2
𝑇0 0
𝑛=−∞ 𝑛=1

This allows us to determine the signal power from either the time-domain
or the frequency-domain specification of the same signal.

I Sem 2020-21 ECE/EEE F311 Communication Systems 2

 Parseval’s Theorem

For energy signals,

+∞ +∞

𝐸𝑔 = 𝑔(𝑡) 2 𝑑𝑡 = 𝐺(𝑓) 2 𝑑𝑓
−∞ −∞

This allows us to determine the signal energy from either the time-domain
specification 𝑔(𝑡) or the frequency-domain specification 𝐺(𝑓) of the
same signal.

𝐺(𝑓) 2 is the Energy Spectral Density

I Sem 2020-21 ECE/EEE F311 Communication Systems 3

Power Spectral Density (PSD)

Consider a power signal

𝑔(𝑡). Then, the truncated
signal 𝑔𝑇 (𝑡) is an energy
signal.

From Parseval’s theorem,

∞
𝐸𝑔𝑇 = 𝐺𝑇 (𝑓) 2 𝑑𝑓
−∞

1 ∞
Hence, the power of 𝑔(𝑡) is: 𝑃𝑔 = lim 𝐺𝑇 (𝑓) 2 𝑑𝑓
𝑇→∞ 𝑇 −∞

EEE F311 Comm_Systems; I Semester 2016-17 4

Power Spectral Density (PSD)

The power spectral density is defined as

𝐺𝑇 (𝑓) 2
𝑆𝑔 𝑓 = lim
𝑇→∞ 𝑇

Then,
∞

𝑃𝑔 = 𝑆𝑔 𝑓 𝑑𝑓
−∞

EEE F311 Comm_Systems; I Semester 2016-17 5

 FT of autocorrelation functions

For energy signals,

The Energy spectral density is the Fourier Transform of the

Autocorrelation Function.

ℱ ℛ𝑔 𝜏 = 𝐺(𝑓) 2

For a power signal,

the Power spectral density 𝑆𝑔 𝑓 is the Fourier Transform of the

Autocorrelation Function ℛ𝑔 (𝜏).

ℱ ℛ𝑔 𝜏 = 𝑆𝑔 𝑓

EEE F311 Comm_Systems; I Semester 2016-17 6

Signal transmission through a linear time-invariant system.

Consider the following LTI system model.

Let ℎ 𝑡 be the system response to a unit impulse input.

Input, 𝑥 𝑡 Output, 𝑦 𝑡
LTI system

If 𝑥 𝑡 is a unit impulse, 𝑦 𝑡 = ℎ t .

In general, the system response to a bounded input signal 𝑥 𝑡 follows

the convolutional relationship:

𝑦 𝑡 = ℎ 𝑡 ∗ 𝑥(𝑡)

EEE F311 Comm_Systems; I Semester 2016-17 7

 Convolution Theorem

Time convolution property:

𝑔1 𝑡 ∗ 𝑔2 𝑡 ↔ 𝐺1 𝑓 𝐺2 𝑓

Frequency convolution property:

𝑔1 𝑡 𝑔2 𝑡 ↔ 𝐺1 𝑓 ∗ 𝐺2 𝑓

I Sem 2020-21 ECE/EEE F311 Communication Systems 8

Signal transmission through a linear time-invariant system.

We know,
ℎ 𝑡 ↔ 𝐻(𝑓)

where 𝐻(𝑓) is the Transfer function (or the frequency response) of the system.

EEE F311 Comm_Systems; I Semester 2016-17 9

Example 1

Let us consider a ‘distortionless Transmission’ (i.e, where the input and

the output have identical wave shapes within a multiplicative constant)

𝑦 𝑡 = 𝑘. 𝑥 𝑡 − 𝑡𝑑

Taking the Fourier Transform, 𝑌 𝑓 = 𝑘𝑋(𝑓)𝑒 −𝑗2𝜋𝑓𝑡𝑑

Therefore, 𝑯 𝒇 = 𝒌𝒆−𝒋𝟐𝝅𝒇𝒕𝒅

EEE F311 Comm_Systems; I Semester 2016-17 10

Example 2

Consider a LTI system which

has a Transfer function as
shown in the figure.

If a input signal is band-limited to 𝐵 Hz, the output is input delayed by 𝑡𝑑 .

𝑓
𝐻 𝑓 =Π 𝑒 −𝑗2𝜋𝑓𝑡𝑑
2𝐵

The impulse response is the inverse Fourier Transform of 𝐻 𝑓 .

Note: The system is an ‘ideal low pass filter’

EEE F311 Comm_Systems; I Semester 2016-17 11

Impulse response of an Ideal low-pass filter .

ℎ(𝑡)
impulse

Impulse response

Realisation of the above filter is physically impossible. Why ?

EEE F311 Comm_Systems; I Semester 2016-17 12

Approximate realization of an ideal low-pass filter

by truncating its impulse response for 𝑡 < 0.

If 𝑡𝑑 is sufficiently large, this will be a close approximation of the

impulse response of a ideal filter.

EEE F311 Comm_Systems; I Semester 2016-17 13

 Example 3

Consider the case where 𝐻(𝑓) is not equal to a constant.

Let,
𝐻 𝑓 = (1 + 𝑘 cos 2𝜋𝑓𝑇)𝑒 −𝑗2𝜋𝑓𝑡𝑑 for 𝑓 < 𝐵;
zero elsewhere.

EEE F311 Comm_Systems; I Semester 2016-17 14

Consider an input signal 𝑔 𝑡 which is Band-limited to B Hz.

𝑦 𝑡

Find output 𝑦 𝑡 .

EEE F311 Comm_Systems; I Semester 2016-17 15

We can write
𝑓
𝐻 𝑓 =∏ . (1 + 𝑘 cos 2𝜋𝑓𝑇)𝑒 −𝑗2𝜋𝑓𝑡𝑑
2𝐵

Then,
𝑌 𝑓 =𝐺 𝑓 𝐻 𝑓
𝑓
= 𝐺(𝑓)∏ . (1 + 𝑘 cos 2𝜋𝑓𝑇)𝑒 −𝑗2𝜋𝑓𝑡𝑑
2𝐵

𝑓
Since 𝑔 𝑡 is bandlimited to B, 𝐺 𝑓 ∏ =𝐺 𝑓 .
2𝐵

Therefore,
𝑌 𝑓 = 𝐺 𝑓 . (1 + 𝑘 cos 2𝜋𝑓𝑇)𝑒 −𝑗2𝜋𝑓𝑡𝑑

EEE F311 Comm_Systems; I Semester 2016-17 16

𝑌 𝑓 = 𝐺 𝑓 . (1 + 𝑘 cos 2𝜋𝑓𝑇)𝑒 −𝑗2𝜋𝑓𝑡𝑑

=𝐺 𝑓 . 𝑒 −𝑗2𝜋𝑓𝑡𝑑 + 𝐺 𝑓 𝑘 cos 2𝜋𝑓𝑇. 𝑒 −𝑗2𝜋𝑓𝑡𝑑

𝒌
𝒚 𝒕 = 𝒈 𝒕 − 𝒕𝒅 + 𝟐 [𝒈 𝒕 − 𝒕𝒅 − 𝑻 + 𝒈 𝒕 − 𝒕𝒅 + 𝑻

H(𝑓)

Note the ‘spreading’ or ‘dispersion’ of the pulse.

EEE F311 Comm_Systems; I Semester 2016-17 17

 Distortionless
transmission

Linear
distortion

I Sem 2020-21 ECE/EEE F311 Communication Systems 18

Non-linear distortion

y 𝑡 = 𝑥 𝑡 + (158)10−6 𝑥 2 (𝑡)
𝑥(𝑡)
𝐻(𝑓)
Nonlinear channel

If 𝑥 𝑡 = 2000 𝑠𝑖𝑛𝑐 2000𝜋𝑡 , find 𝑦 𝑡 .

𝑦 𝑡 = 2000 𝑠𝑖𝑛𝑐 2000𝜋𝑡 + 0.316 2000 𝑠𝑖𝑛𝑐 2 (2000𝜋𝑡)

𝑓 𝑓
𝑌 𝑓 =Π + 0.316 Δ( )
2000 2000

I Sem 2020-21 ECE/EEE F311 Communication Systems 19