Classification of Systems, Impulse Response, and Transfer Function on Mac

4. Classification of Systems, Impulse Response, and Transfer

Function

A system is a mathematical model that relates the output signal to the input signal of a
physical process.

Figure 4.1 Representation of a system.

Classification of Systems [1]

1. Linear and non-linear systems

Let x i (t) and y i (t), i > 1, be input and output signals of a system, respectively. A
system is called a linear system if the input x 1 (t) + x 2 (t) + ... + x i (t) + ...
produces a response y 1 (t) + y 2 (t) + ... + y i (t) + ..., and a x i (t) produces
ay i (t) for all input signals x i (t) and scalar a. This is known as the superposition
theorem and a linear system obeys this principle.

Figure 4.2 Linear system.

In practice, it may be found that a system is only linear over a limited range of input
signals.

A non-linear system does not obey the superposition theorem.

2. Causal and non-causal systems

Let x(t) and y(t) be the input and output signals of a system. A causal (physically
realisable) system produces an output response at time t1 for an input at time t0, where
t0 > 0 and t0 ≤ t1.

Figure 4.3 Signals associated with causal system.

In other words, a causal system is one whose response does not begin before the input
signal is applied.

A non-causal system response will begin before the input signal is applied. It can be
made realisable by introducing a positive time delay into the system.

3. Time-invariant and time-varying systems

If the input x(t - t 0 ) produces a response y(t - t 0 ) where t 0 is any real constant,

4.1
 Classification of Systems, Impulse Response, and Transfer Function on Mac

the system is called a time-invariant system.

Figure 4.4 Time-invariant system.

If the above condition is not satisfied, the system is called a time-varying system.

A system is called a linear time-invariant (LTI) system if the system is linear and time-
invariant.

Classification of signals and systems will help us in finding a suitable mathematical model
for a given physical process that is to be analysed.

Impulse Response

The impulse response h(t) of an LTI system is defined as the response of the system
when the input signal x(t) is a delta function δ (t). The output y(t) of an LTI system
can be expressed as the convolution of the input signal x(t) and the impulse response
h(t) of the system, i.e.,

∞
y(t) = x(t) * h(t) = ∫ x(λ) h(t-λ) dλ (4.1)
−∞
∞
y(t) = h(t) * x(t) = ∫ h(λ) x(t-λ) dλ (4.2)
−∞

For a causal LTI system,

∞ ∞
y(t) = ∫ x(λ) h(t-λ) dλ = ∫ h(λ) x(t-λ) dλ (4.3)
0 0

Transfer Function

In the frequency domain, the Fourier transform of y(t) = h(t) * x(t) is

Y(f) = H(f) X(f) (4.4)

where H(f) is the Fourier transform of h(t). H(f) is called the transfer function or
frequency response of the LTI system.

4.2
 Classification of Systems, Impulse Response, and Transfer Function on Mac

Example 4.1 An RC circuit is shown in Figure 4.5. Using Kirchhoff’s voltage law,
we get

x(t) = i(t) R + y(t)

dy
Since i(t) = C , we can write
dt

dy
RC + y(t) = x(t)
dt

and

RC(j2π f)Y(f) + Y(f) = X(f)

1
The transfer function of the RC circuit is H(f) = and the impulse
1 + j 2π RC f
 − t
 1 e RC , t ≥ 0.
response is h(t) =  RC
0, t<0


Figure 4.5 Characteristics of an RC circuit.

Consider the input signal x(t) = δ(t). The Fourier transform of x(t) is

∞ -j 2π ft
X(f) = F[δ(t)] = ∫ δ(t) e dt = 1 (4.5)
−∞

and equation (4.4) becomes Y(f) = H (f). X(f) = 1 implies equal amplitude at all
frequencies. It is equivalent to exciting the system with all frequencies simultaneously.
Rather than applying a sinusoidal signal and varying its frequency continuously to obtain
H(f), a technique to measure H(f) is as follows:

1. Excite the system with x(t) = δ(t).

2. Measure y(t) = h(t).
∞ -j 2π ft
3. Find H(f) = ∫ h(t) e dt.
−∞

In general, H (f) is a complex function of frequency. In polar form, H (f) can be

expressed as

H(f) = |H(f)| ejθ(f) (4.6)

4.3
 Classification of Systems, Impulse Response, and Transfer Function on Mac

From our earlier study of the properties of Fourier transform, we have seen that if h(t) is
real,

H(-f) = H*(f) (4.7)

|H(-f)| = |H(f)| (4.8)
θ(-f) = -θ(f) (4.9)

where H * (f) is the complex conjugate of H (f). If x(t) and y(t) are real and H (f)
is the transfer function of a LTI system, we can obtain the following results.

Let x(t) and y(t) be the input and output energy signals of a LTI system. The energy
spectral densities of x(t) and y(t) are E xx (f) = |X (f)| 2 and E yy (f) = |Y (f)| 2 ,
respectively. Since Y(f) = H(f) X(f), we have

E yy(f) = |H(f)|2 E xx(f) (4.10)

E yy(f) = H * (f)H(f) E xx(f) (4.11)

Let x(t) and y(t) be the input and output power signals of a LTI system. Also, let
xT(t) and yT(t) be the truncated signals of x(t) and y(t), respectively, where

−T / 2 < t < T / 2
 x ( t ),
xT(t) = 
0, elsewhere
( )
= x(t)Π T
t
(4.12)

and

−T / 2 < t < T / 2
 y( t ),
yT(t) = 
0, elsewhere
( )
= y(t)Π T
t
(4.13)

xT (t) and yT (t) are energy signals as long as T is finite. The power spectral densities
 |X (f )| 2
 T 
of x ( t ) and y ( t ) are P x x ( f ) = lim   and P y y ( f ) =
T →∞ T
 |Y (f )| 2
 
lim  T  , respectively. Since Y(f) = H(f) X(f), we have
T →∞ T

P yy(f) = |H(f)|2 P xx(f) (4.14)

P yy(f) = H * (f)H(f) P xx(f) (4.15)

4.4
 Classification of Systems, Impulse Response, and Transfer Function on Mac

We can also obtain the relationship between the input and output auto-correlation functions
of a LTI system. Since the energy/power spectral density and the auto-correlation function
are Fourier transform pairs, the inverse Fourier transform of Eyy(f) or Pyy(f) is [2]

Ryy(τ) = h(- τ) * h(τ) * Rxx(τ) (4.16)

where R xx ( τ ) and R yy ( τ ) are the auto-correlation functions of x(t) and y(t) ,

respectively.

Figure 4.6 Input and output relationships of linear system.

Distortionless Transmission [1]

In communication systems, a distortionless transmission is often desired. This implies that

the output signal y(t) is given by

y(t) = K x(t - td) (4.17)

where K is a constant and td is a time delay. The Fourier transform of y(t) is

Y(f)
π ft
= K e-j2 d X(f) (4.18)
= H(f) X(f) (4.19)

where H(f) = K e -j2 π ftd . Figure 4.7 shows the waveforms and spectra associated
with distortionless transmission.

Figure 4.7 Waveforms and spectra associated with distortionless transmission.

Classification of Filters [1]

1. Ideal Low-Pass Filter.

The transfer function of an ideal low-pass filter is defined by

 −j 2π f td for |f | ≤ fc
H LPF (f) = e (4.20)
 0 elsewhere

Figure 4.8 Frequency response of an ideal low-pass filter.

The bandwidth of an ideal low-pass filter is equal to fc.

4.5
 Classification of Systems, Impulse Response, and Transfer Function on Mac

2. Ideal High-Pass Filter.

The transfer function of an ideal high-pass filter is defined by

 −j 2π f td for |f | ≥fc
H HPF (f) = e (4.21)
 0 elsewhere

or

H HPF (f) = e-j2π ftd - H LPF (f) (4.22)

Figure 4.9 Frequency response of an ideal high-pass filter.

The bandwidth of an ideal high-pass filter is not defined, or infinite.

3. Ideal Bandpass Filter.

The transfer function of an ideal bandpass filter is defined by

 −j 2π f td for f
c1
≤ |f | ≤ f
c2
H BPF (f) = e (4.23)
 0 elsewhere

Figure 4.10 Frequency response of an ideal bandpass filter.

The bandwidth of an ideal bandpass filter is equal to fc2 - fc1.

4. Ideal Bandstop Filter.

The transfer function of an ideal bandstop filter is defined by

 −j 2π f td for 0≤ |f | ≤ f and |f | ≥f
c 1 c 2
H BSF (f) = e (4.24)
 0 elsewhere

or

H BSF (f) = e-j2π ftd - H BPF (f) (4.25)

Figure 4.11 Frequency response of an ideal bandstop filter.

4.6
 Classification of Systems, Impulse Response, and Transfer Function on Mac

The bandwidth of an ideal bandstop filter is not defined, or infinite.

Concept of Non-Ideal or Practical Filter (System) Bandwidth [1]

A common definition of a non-ideal filter’s bandwidth is the 3dB bandwidth.

For low-pass filters, the bandwidth is defined as the positive frequency at which the
amplitude spectrum |H(f)| drops to a value equal to |H(f)|/ 2 .

For bandpass filters, the bandwidth is defined as the difference between the frequencies at
which the amplitude spectrum |H(f)| drops to a value equal to |H(f0 )|/ 2 , where
H(f0) is the peak value of |H(f)|.

Figure 4.12 Non-ideal filter bandwidth.

References

[1] H. P. Hsu, Analog and Digital Communications, McGraw-Hill, 1993.

[2] L. W. Couch II, Digital and Analog Communication Systems, 5/e, Prentice Hall,
1997.

4.7
 Classification of Systems, Impulse Response, and Transfer Function on Mac

x (t ) Functional Functional y (t )
...
block block

system

Figure 4.1 Representation of a system.

x (t ) y (t ) x (t ) y (t )
1 1 2 2

x ( t ) + x ( t ) + ... y ( t ) + y ( t ) + ...
1 2 1 2
a x (t ) a y i (t )
i

Figure 4.2 Linear system.

x ( t ), y ( t ) Input x ( t ) Output y ( t )

Time
0 t 0 t1
Figure 4.3 Signals associated with causal system.

x (t - t 0 ) y (t - t 0 )

Figure 4.4 Time-invariant system.

4.8
 Classification of Systems, Impulse Response, and Transfer Function on Mac

R i (t )

x (t ) C y (t )

(a)

h (t )
1
RC

0.37RC
t
0 RC
(b)

|H ( f )| 2

1 1
f 0= π
2 RC
0.5

f
-f 0 0 f 0
(c)

Figure 4.5 Characteristics of an RC circuit.

4.9
 Classification of Systems, Impulse Response, and Transfer Function on Mac

x (t ) y ( t ) =h ( t ) * x ( t )
h (t )
Auto-correlation R xx ( τ ) R yy ( τ ) = h (- τ ) * h ( τ ) * R xx ( τ )
function
Time domain

X (f ) Y ( f ) =H ( f ) X ( f )
H (f )
ESD E xx ( f ) E yy ( f ) = |H ( f ) | 2 E ( f )
xx
PSD P xx ( f ) P yy ( f ) = |H ( f ) | 2 P xx ( f )
Frequency domain

Figure 4.6 Input and output relationships of linear system.

|H (f )|
x (t )
K
A

t f
0 t1 0
(a) (c)

y (t ) θ (f )

Slope = -t
d
KA
t f
0 td t + t 0
1 d
(b) (d)

Figure 4.7 Waveforms and spectra associated with distortionless transmission.

4.10
 Classification of Systems, Impulse Response, and Transfer Function on Mac

|HLPF (f )| θ (f )
1 Slope = -t
d
f f
-f c 0 fc -f c 0 fc
Figure 4.8 Frequency response of an ideal LPF.

|HHPF (f )| θ (f )
1 Slope = -t
d
f f
-f c 0 fc -f c 0 fc

Figure 4.9 Frequency response of an ideal HPF.

|HBPF (f )|
θ (f )
1 fc 1fc 2
f f
-f c 2 -f c 1 0 f c 1 f c 2 -f c 2 -f c 1 0
Slope = -t
d
Figure 4.10 Frequency response of an ideal BPF.

|HBSF (f )| θ (f )
1
fc 1 fc 2
f f
-f c 2 -f c 1 0 f c 1 f c 2 -f c 2 -f c 1 0
Slope = -t
d
Figure 4.11 Frequency response of an ideal BSF.

4.11
 Classification of Systems, Impulse Response, and Transfer Function on Mac

|H (f )|

A
A / 2

f
-f c 0 fc
(a) Low-pass filter

|H (f )|

A
A / 2

f
-f c 2 -f 0 -f c 1 fc 1 f
0 fc2
(b) Bandpass filter

Figure 4.12 Filter bandwidth.

4.12