Communication

Systems

Chapter 3
Signal Transmission and Filtering

Dr. Le Dang Quang

Department of Telecommunications (113B3)
Ho Chi Minh City University of Technology
Email: [email protected]

Telecomm. Dept. CS-2016

1
Faculty of EEE HCMUT
 Chapter Outline
3.1 Response of LTI Systems
3.2 Signal Distortion in Transmission
3.3 Transmission Loss and Decibels
3.4 Filters and Filtering
3.5 Correlation and Spectral Density
3.6 Probability and Random Variables
3.7 Random Signals and Noise

Telecomm. Dept. CS-2016

2
Faculty of EEE HCMUT
3.1 Response of LTI Systems

Telecomm. Dept. CS-2016

3
Faculty of EEE HCMUT
 Response of LTI Systems

Signal transmission is the process whereby an electrical

waveform gets from one location to another.
In the ideal case there is no distortion in the transmission.
In practice, there is some distortion in the channel which should
be controlled and compensated.
Usually, the channel is modeled by a linear time-invariant (LTI)
system. On the other hand, linear time-invariant filters are used in
different stages of the transmission system to modify the
transmitted signal.
Telecomm. Dept. CS-2016
4
Faculty of EEE HCMUT
 Response of LTI Systems

 Impulse response of linear time-invariant (LTI) system

The input signals can be, e.g., current or voltage signals.

The system is linear if

and time-invariant if

Such system is usually called as "filter". Analog filters consist of

inductors, capacitors, and resistors and they can be described by
differential equations.
Telecomm. Dept. CS-2016
5
Faculty of EEE HCMUT
 Response of LTI Systems

In this course we do not discuss about design or implementation of

the filters but they are considered as a black box having a certain
impulse response.

The impulse response h(t) is the response of the system to the

unit impulse:

The response of the system for an arbitrary input signal can be

calculated using the convolution integral (superposition integral):

Telecomm. Dept. CS-2016

6
Faculty of EEE HCMUT
 Response of LTI Systems

 Transfer function
The convolution integral, which determines the response of the
system in the time domain is difficult to calculate and it is intuitively
difficult concept.

The transfer function determines the response of system in the

frequency domain and it is easy to calculate and understand.

The transfer function of the linear time-invariant system is

determined by the Fourier transform of the impulse response:

In practice, it is assumed that systems are stable which means

that their impulse responses converge with time.

Telecomm. Dept. CS-2016

7
Faculty of EEE HCMUT
 Response of LTI Systems

Usually, the impulse response of the system is real function.

Therefore, it has the following properties:

According to the convolution theorem, the linear time-invariant

system can be characterize in the frequency domain by

The output spectrum equals the input spectrum multiplied by the

transfer function.

Telecomm. Dept. CS-2016

8
Faculty of EEE HCMUT
 Response of LTI Systems
 The 'steady-state' response of the filter
Consider the response of the filter to the complex exponential:

The 'steady-state' response of the system (i.e., the response

after the transients) for the above-mentioned signal is

The output signal is a complex exponential whose frequency is the

same as in the input but the amplitude and phase depend on the
transfer function of the system:

Telecomm. Dept. CS-2016

9
Faculty of EEE HCMUT
 Response of LTI Systems

For the real sinusoidal signals, this can be given by

where Ay and y are as before.

According to the superposition and the results above, there are
only those frequencies in the output signal which are also in
the input signal.

 The frequency response of the system

H(f0) is the response of the system for the sinusoidal signal whose
frequency is f0.
H(f) is also called the frequency response of the system which is
usually illustrated by two functions in the frequency domain:
amplitude response: | H(f)|
phase response: arg H(f)

Telecomm. Dept. CS-2016

10
Faculty of EEE HCMUT
 Response of LTI Systems

Example: First order RC filter

The differential equation for the filter is:

The step response is:

and the impulse response is

Telecomm. Dept. CS-2016

11
Faculty of EEE HCMUT
 Response of LTI Systems

The response of this filter to the rectangular pulse is:

Telecomm. Dept. CS-2016

12
Faculty of EEE HCMUT
 Response of LTI Systems
The circuit of the filter can be illustrated in the frequency domain by
using the impedances ZR = R and ZC = 1/ jωC. We can write:

and the transfer function is:

This is a low-pass filter. The parameter

B is the bandwidth of the filter (3 dB
cutoff frequency). At this frequency:

Telecomm. Dept. CS-2016

13
Faculty of EEE HCMUT
 Response of LTI Systems
The effect of the filter to the input signal can be studied by considering both
the frequency response of the system and the spectrum of the input signal:

a. The bandwidth of the filter B is

much higher than the bandwidth W
of the signal, so there is no distortion.
b. B = W → The high frequencies of
the signal are attenuated.
c. The bandwidth of the filter B is
much lower than the bandwidth W of
the signal → the output signal now
looks like the filter’s impulse response
→ we can reasonably model the input
signal as an impulse.

Telecomm. Dept. CS-2016

14
Faculty of EEE HCMUT
 Response of LTI Systems

 Block-diagram analysis
Telecommunication systems consists of many blocks which are
connected together. In the case of the
LTI system, the transfer
function of the overall
system can be calculated
from the transfer functions
of the subsystems.

In the figure nearby are the

overall transfer functions of
the blocks which are
connected in:
a) parallel,
b) cascade, and
c) feedback:
Telecomm. Dept. CS-2016
15
Faculty of EEE HCMUT
 Response of LTI Systems
Example: Zero-order Hold circuit

The hold circuit if widely used in telecommunication applications.

The following time and frequency-domain block diagrams can be
given for this circuit:

Telecomm. Dept. CS-2016

16
Faculty of EEE HCMUT
 Response of LTI Systems

The transfer function of the system is:

The amplitude response of the system is sinc-function!

The impulse response is:

The impulse response is a rectangular pulse, which could be also

obtained from the Fourier transform of the above transfer function.

Telecomm. Dept. CS-2016

17
Faculty of EEE HCMUT
3.2 Signal Distortion in
Transmission

Telecomm. Dept. CS-2016

18
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Distortionless channel:
Distortionless transmission means that the output signal has the
same “shape” as the input, but there can be some attenuation.
The received signal is thus

where K are td constants, and x(t) is the transmitted signal. The

spectrum of the received signal is

and therefore, the transfer function of a distortionless channel is

The amplitude and phase responses are

Telecomm. Dept. CS-2016

19
Faculty of EEE HCMUT
 Signal Distortion in Transmission
The amplitude response of the distortionless system is a
constant and the phase response is a linear function of the
frequency.

These conditions have to be met only in those frequency bands

where the transmitted signal has components.

Example: The spectrum of a voice signal

Typical energy spectral density for voice signal:

Telecomm. Dept. CS-2016

20
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Distortions
The are three major types of distortion in transmission:
 Amplitude distortion:

 Phase distortion:

 Nonlinear distortion: Caused by nonlinear elements of the

system.

The first two types are linear distortions, and they can be
analyzed by using the transfer function of the system. Nonlinear
system does not have transfer function, so the analysis of this kind
of system has to be done using some other methods.

Telecomm. Dept. CS-2016

21
Faculty of EEE HCMUT
 Signal Distortion in Transmission
Example

The frequency response of the system:

This system is distortionless in the band 20≤ f ≤30 kHz.

Telecomm. Dept. CS-2016

22
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Amplitude distortion
In amplitude distortion, some frequency
bands of the signal are attenuated or
amplified. Typical example is the
attenuation / amplification of the low
or high frequencies. In the time domain,
the attenuation of high frequencies can be
seen as a smoothing of the rapid changes.
In the upper figure we have the fundamental
Low frequencies attenuated
frequency and two harmonics, together with
the resultant signal (thick line).

In the lower figures, there are some

examples of the effect of the amplitude
distortion to the rectangular pulse train. High frequencies attenuated
Telecomm. Dept. CS-2016
23
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Phase distortion
Phase distortion causes, e.g., different delays for the different harmonics
of the signal, and therefore, destroys the shape of the signal.

The phase delay of the distortion-less system is a

constant.

It should be pointed out that this is different from the constant phase shift
which in general causes distortion of the signal.

In the example of the previous page,

90o constant phase shift leads to the
result in the figure. This looks more
like triangular pulse than rectangular.

Note also that phase distortion can cause the maximum of the signal to be
much higher than before the distortion.
Telecomm. Dept. CS-2016
24
Faculty of EEE HCMUT
 Signal Distortion in Transmission

 Equalization
If the transfer function of the channel that causes linear distortion is
known, it can be compensated by the inverse transfer function.

If the transfer functions of the channel and equalizer are HC(f) and Heq(f),
respectively, then the transfer function of the overall transmission
system is
H(f) = HC(f) Heq(f)

Telecomm. Dept. CS-2016

25
Faculty of EEE HCMUT
 Signal Distortion in Transmission

When the goal is to obtain an ideal channel, the transfer function of the
equalizer is chosen to be

The limitation is the fact that in those frequencies where the

attenuation of the channel is high, equalization does not work. If the
signal is covered by higher level noise, it can not be compensated.

Telecomm. Dept. CS-2016

26
Faculty of EEE HCMUT
 Signal Distortion in Transmission

 Transversal filter as equalizer

Transversal filter consists of tapped-delay-line. The output is a weighted
average of delayed signals:

The output in this simple case is:

Telecomm. Dept. CS-2016

27
Faculty of EEE HCMUT
 Signal Distortion in Transmission
Therefore the transfer function of the equalizer is

Generally, the transfer function of transversal filter can be given by

This is a Fourier series whose period in the frequency domain is 1 / Δ.

In most of the cases, the tap gains are readjusted to compensate the
changes in the channel response. Such a system is called as adaptive
equalizer.

Transversal filters are usually implemented digitally, however, there are also
some analog implementations (CCD, SAW).

Telecomm. Dept. CS-2016

28
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Multipath distortion
Radio systems sometimes suffer from multipath distortion caused by two
or more propagation paths between transmitter and receiver. The same
phenomenon can occur in cables caused by impedance mismatch.

For two multipath components, the received signal is

The transfer function of the channel is

where t2 > t1, k = K2 / K1 and t0 = t2 − t1.

The transfer function of the equalizer is

Only a finite number of terms are needed from this series, and this number
depends on k. Equalization can be implemented using transversal filter.
Telecomm. Dept. CS-2016
29
Faculty of EEE HCMUT
 Signal Distortion in Transmission

 Nonlinear distortion
Here we consider nonlinear functions: y(t) = T[x(t)] which are
memoryless, i.e., the response at time t does not depend on the
previous signal values (nor on the future values). A nonlinear
function can be illustrated by a curve. Typical example:

Under small-signal input conditions: piece-wise linear

approximation.

Telecomm. Dept. CS-2016

30
Faculty of EEE HCMUT
 Signal Distortion in Transmission
More general: polynomial approximation.
Any non-linear function can be approximated by the polynomial:
y(t) = a1x(t) + a2x2(t) + a3x3(t) + ...
From this polynomial, the spectrum of the output is (using
convolution theorem):
Y(f) = a1X(f) + a2X X(f) + a3X X X(f) +...
If the bandwidth of the input signal is W, then the bandwidth of the
output signal of any linear system contains no frequencies
beyond W.
In a nonlinear system, there can be also some other frequency
components which are not present in the input. In the above
equation, the bandwidth of the term X X (f) is 2W and, in general,
the bandwidth of the nth term is nW. The frequency components
caused by nonlinearity can be filtered out if they are situated outside
the frequency band of the input signal.
However, if, for example, X X (f) contains components for | f | < W,
these components are overlapping with those of the original input
signal X(f). They form the nonlinear distortion.
Telecomm. Dept. CS-2016
31
Faculty of EEE HCMUT
 Signal Distortion in Transmission
If the input signal is a pure sinusoidal signal, then nonlinearity
produces its harmonics. According to the equations in the previous
page, we can write, if the input is a cosine wave:

This is called as harmonic distortion. Harmonic distortion can be

determined by the amplitude ratio of the harmonics and the
fundamental (i.e., the coefficients of cosω0t term). Most significant
are usually the second-harmonic distortion terms.

Now, if the input signal consists of the sinusoidals having the

frequencies of f1 and f2, then the output consists of the harmonics
as well as the linear combinations of these frequencies:

Telecomm. Dept. CS-2016

32
Faculty of EEE HCMUT
 Signal Distortion in Transmission
 Companding
Companding = Compressing + Expanding
It is needed to deal with the non-linearities in the system (very
useful concept later on when discussing the exponential
modulations).

Sometimes nonlinear blocks are used in telecommunication

systems. For example, high signal levels can be decreased by
compressing them using the following nonlinear function:

Compressing decreases the effect of

nonlinearities of the channel. The
compressor has a higher amplification
at low signal levels than at high signal
levels.

Telecomm. Dept. CS-2016

33
Faculty of EEE HCMUT
 Signal Distortion in Transmission

The inverse operation, expanding, is done in the receiver.

The use of companding allows signals with a large dynamic range

to be transmitted over facilities that have a smaller dynamic range
capability. Companding reduces the noise and crosstalk levels at
the receiver.

Telecomm. Dept. CS-2016

34
Faculty of EEE HCMUT