Digital Signal Analysis & Processing Chapter 3

Chapter 3: Analysis of LTI System in Frequency Domain

3.1. Frequency response of LTI system, response to complex exponential
3.2. Linear constant co-efficient difference equation and corresponding system function
3.3. Relationship of frequency response to pole-zero of system
3.4. Linear phase of LTI system and its relationship to causality.
Introduction:
In this chapter, we develop the characterization of LTI systems in the frequency domain. The
basic excitation of signals in this development are the complex exponentials and sinusoidal
functions. The characteristics of the system are described by a function of the frequency
variable ω called the frequency response, which is the Fourier transform of the impulse
response h[n] of the system.
The System Function of an LTI System:
We know, for an LTI system
x[n] h[n] y[n]

𝑦[𝑛] = 𝑥[𝑛] ∗ ℎ[𝑛]

Taking z-transform, Y(z) = X(z)H(z) => H(z) = Y(z)/X(z)
If we know the x[n] and we observe the output y[n] of the system, we can determine the unit
sample response. It is clear that H(z) represents the z-domain characterization of a system,
whereas h[n] is the corresponding time-domain characterization of the system. The transform
function H(z) is called the system function.
When the system is described by an LCCD equation,
𝑁 𝑀

𝑦[𝑛] = − ∑ 𝑎𝑘 𝑦[𝑛 − 𝑘] + ∑ 𝑏𝑘 𝑥[𝑛 − 𝑘]

𝑘=1 𝑘=0
Taking z-transform
𝑁 𝑀

𝑌(𝑧) = − ∑ 𝑎𝑘 𝑧 −𝑘 𝑌(𝑧) + ∑ 𝑏𝑘 𝑧 −𝑘 𝑋(𝑧)

𝑘=1 𝑘=0
𝑌(𝑧) ∑𝑀𝑘=0 𝑏𝑘 𝑧
−𝑘
𝑆𝑖𝑚𝑝𝑙𝑖𝑓𝑦𝑖𝑛𝑔 𝑤𝑒 𝑔𝑒𝑡, 𝐻(𝑧) = =
𝑋(𝑧) 1 + ∑𝑁 𝑘=1 𝑎𝑘 𝑧
−𝑘

Hence, and LTI system described by a LCCD equation has a rational system function.
Case I: If ak = 0 for 1 ≤ k ≤ N
𝑀 𝑀
−𝑘
1
𝐻(𝑧) = ∑ 𝑏𝑘 𝑧 = 𝑀 ∑ 𝑏𝑘 𝑧 𝑀−𝑘
𝑧
𝑘=0 𝑘=0
In this case, H(z) contains M zeros, whose values are determined by system
parameters {bk} & an Mth order pole at z=0 (trivial poles). Hence the system is called
all-zero system. Such a system has finite duration impulse response & is called FIR
system.
Case II: If bk = 0 for 1 ≤ k ≤ M,
𝑏0 𝑏0 𝑧 𝑁
𝐻(𝑧) = ∑𝑁 = ∑𝑁 , 𝑎0 ≌ 1
𝑘=1 𝑎𝑘 𝑧 −𝑘 𝑘=1 𝑎𝑘 𝑧
𝑁−𝑘

Page 1 of 6
Compiled by: Rupesh D. Shrestha, nec
Digital Signal Analysis & Processing Chapter 3

In this case H(z) consists of N poles, whose values are determined by the system
parameters {ak} & and Nth order zero at z = 0. The corresponding system is called all-pole
system. Due to presence of poles, the impulse response of such a system is infinite in duration
& hence it is an IIR system.

Response of Complex Exponential and Sinusoidal Signals

The response of any relaxed LTI system to an arbitrary input signal x[n] is given by the
convolution sum as
∞

𝑦[𝑛] = ∑ ℎ[𝑘]𝑥[𝑛 − 𝑘]
𝑘=−∞
In this input-output relationship, the system is characterized in the time domain by its unit
sample response h[n],
To develop a frequency-domain characterization of the system, let us excite the system with
the complex exponential
𝑥[𝑛] = 𝐴𝑒 𝑗𝜔𝑛 , −∞ < 𝑛 < ∞
Where A is the amplitude and ω is any arbitrary frequency confined to the frequency interval
[-π, π]. Substituting the value of x[n], we obtain the response
∞ ∞
𝑗𝜔(𝑛−𝑘)
𝑦[𝑛] = ∑ ℎ[𝑘][𝐴𝑒 ] = 𝐴 [ ∑ ℎ[𝑘]𝑒 −𝑗𝜔𝑘 ] 𝑒 𝑗𝜔𝑛
𝑘=−∞ 𝑘=−∞
We obverse that the term in brackets in above equation is a function of the frequency variable
ω. In fact, this term is the Fourier transform of the unit sample response h[k] of the system.
Hence, we denote this function as
∞
𝑗𝜔
𝐻(𝑒 ) = 𝐻(𝜔) = ∑ ℎ[𝑘]𝑒 −𝑗𝜔𝑘
𝑘=−∞
Clearly, the function H(ω) exists if the system is BIBO stable, that is, if
∞

∑ |ℎ[𝑛]| < ∞
𝑛=−∞
Therefore, the response of the system to the complex exponential is given as
𝑦[𝑛] = 𝐴𝐻(𝜔)𝑒 𝑗𝜔𝑛
We note that the response is also in the form of a complex exponential with the same frequency
as the input, but altered by the multiplicative factor H(ω).
As a result of this characteristic behavior, the exponential signal is called an eigenfunction of
the system. In other words, an eigenfunction of a system is an input signal that produces an
output that differs from the input by a constant multiplicative factor, H(ω) (called eigenvalue
of the system).
# Determine the output sequence of the system with impulse response
𝟏 𝒏
𝒉[𝒏] = ( ) 𝒖[𝒏]
𝟐
when the input is the complex exponential sequence
𝒙[𝒏] = 𝑨𝒆𝒋𝝅𝒏/𝟐 , −∞ < 𝒏 < ∞
Solution: First we evaluate the Fourier transform of the impulse response h[n],
Page 2 of 6
Compiled by: Rupesh D. Shrestha, nec
Digital Signal Analysis & Processing Chapter 3
∞
1
𝐻(𝜔) = ∑ ℎ[𝑛]𝑒 −𝑗𝜔𝑛 =
1
𝑛=−∞ 1 − 2 𝑒 −𝑗𝜔
𝜋 𝜋 1 1 2 −𝑗26.60
𝑎𝑡 𝜔 = , 𝐻( ) = = = 𝑒
2 2 1 −2 𝑗𝜋 1 √5
1− 2𝑒 1 + 𝑗 2
2 0
𝑇ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒, 𝑡ℎ𝑒 𝑜𝑢𝑡𝑝𝑢𝑡 𝑖𝑠 (𝜔 = 𝜋/2), 𝑦[𝑛] = 𝐴𝐻(𝜋)𝑒 𝑗𝜋𝑛/2 = 𝐴 ( 𝑒 −𝑗26.6 ) 𝑒 𝑗𝜋𝑛/2
√5
2 0
𝑦[𝑛] = 𝐴𝑒 𝑗(𝜋𝑛/2−26.6 ) , −∞ < 𝑛 < ∞
√5
This example clearly illustrates that the only effect of the system on the input signal is to scale
the amplitude by 2/√5 and shift the phase by -26.60. Thus, the output is also a complex
exponential of frequency π/2, amplitude 2A/√5, and phase -26.60.

In general, H(ω) is a complex-valued function of the frequency variable ω. Hence it can be

expressed in polar form as
𝐻(𝜔) = |𝐻(𝜔)|𝑒 𝑗∅(𝜔)
Where |H(ω)| is the magnitude and φ(ω) is the phase shift imparted on the input by the system
at the frequency ω.
For a linear time-invariant system with a real-valued impulse response, the magnitude and
phase functions possess symmetry properties. The magnitude |H(ω)| is an even function of ω
and the phase φ(ω) is an odd function of ω. Hence if we know |H(ω)| and φ(ω) for 0 ≤ ω ≤ π,
we also know these functions for -π ≤ ω ≤ π.
Response of LTI system to sinusoid:
We know, if the input is
𝑥1 [𝑛] = 𝐴𝑒 𝑗𝜔𝑛 , 𝑡ℎ𝑒 𝑜𝑢𝑡𝑝𝑢𝑡 𝑖𝑠, 𝑦1 [𝑛] = 𝐴|𝐻(𝜔)|𝑒 𝑗∅(𝜔) 𝑒 𝑗𝜔𝑛
𝑠𝑖𝑚𝑖𝑙𝑎𝑟𝑙𝑦, 𝑥2 [𝑛] = 𝐴𝑒 −𝑗𝜔𝑛 , 𝑡ℎ𝑒 𝑜𝑢𝑡𝑝𝑢𝑡 𝑖𝑠, 𝑦2 [𝑛] = 𝐴|𝐻(−𝜔)|𝑒 𝑗∅(−𝜔) 𝑒 −𝑗𝜔𝑛
Applying symmetry properties
𝑦2 [𝑛] = 𝐴|𝐻(𝜔)|𝑒 −𝑗∅(𝜔) 𝑒 −𝑗𝜔𝑛
Now, by applying the superposition property of the LTI system, we find the response of the
system to the input
1
𝑥[𝑛] = {𝑥1 [𝑛] + 𝑥2 [𝑛]} = 𝐴 cos 𝜔𝑛 , 𝑖𝑠
2
1
𝑦[𝑛] = {𝑦1 [𝑛] + 𝑦2 [𝑛]} = 𝐴|𝐻(𝜔)| cos[𝜔𝑛 + ∅(𝜔)]
2
Similarly, if
1
𝑥[𝑛] = {𝑥1 [𝑛] − 𝑥2 [𝑛]} = 𝐴 sin 𝜔𝑛,
𝑗2
1
𝑦[𝑛] = {𝑦1 [𝑛] − 𝑦2 [𝑛]} = 𝐴|𝐻(𝜔)| sin[𝜔𝑛 + ∅(𝜔)]
𝑗2
It is apparent from this discussion that H(ω) completely characterize the effect of the system
on a sinusoidal input signal of any arbitrary frequency. By knowing H(ω), we are able to
determine the response of the system to any sinusoidal input signal. Since H(ω) specifies the
response of the system in the frequency domain, it is called frequency response of the system.
Page 3 of 6
Compiled by: Rupesh D. Shrestha, nec
Digital Signal Analysis & Processing Chapter 3

Correspondingly, |H(ω)| is called the magnitude response and φ(ω) is called the phase response
of the system. If the input to the system consists of more than one sinusoid, the superposition
property of the linear system can be used to determine the response.

The Frequency Response Function:

We know that if the system function H(z) converges on the unit circle, we can obtain the
frequency response of the system by evaluating H(z) on the unit circle. Thus
∞

𝐻(𝜔) = 𝐻(𝑧)|𝑧=𝑒 𝑗𝜔 = ∑ ℎ[𝑛]𝑒 −𝑗𝜔𝑛

𝑛=−∞
In the case where H(z) is a rational function of the form H(z)=B(z)/A(z), we have,
∑𝑀𝑘=0 𝑏𝑘 𝑧
−𝑘
𝐻(𝑧) =
1 + ∑𝑁 𝑘=1 𝑎𝑘 𝑧
−𝑘
𝑁 𝑁
𝑁−𝑀
𝐻(𝑧) = 𝑏0 𝑧 (∏(𝑧 − 𝑧𝑙 )) / (∏(𝑧 − 𝑝𝑘 ))
𝑙=1 𝑘=1
jω
If the ROC of H(z) includes a unit circle (z = e ), then we can evaluate H(z) on unit circle,
resulting in a frequency response function or transfer function H(ejω)
𝑁 𝑁
𝑗𝜔 𝑗(𝑁−𝑀)𝜔 𝑗𝜔
𝐻(𝑒 ) = 𝑏0 𝑒 (∏(𝑒 − 𝑧𝑙 )) / (∏(𝑒 𝑗𝜔 − 𝑝𝑘 ))
𝑙=1 𝑘=1
Here magnitude response function,
𝑗𝜔
|𝑒 𝑗𝜔 − 𝑧1 ||𝑒 𝑗𝜔 − 𝑧2 | … . |𝑒 𝑗𝜔 − 𝑧𝑀 |
|𝐻(𝑒 )| = |𝑏0 | × 𝑗𝜔
|𝑒 − 𝑝1 ||𝑒 𝑗𝜔 − 𝑝2 | … . |𝑒 𝑗𝜔 − 𝑝𝑁 |
The phase response function,
𝑀 𝑁
𝑗𝜔 𝑗𝜔
𝐴𝑟𝑔 (𝐻(𝑒 )) = {0 𝑜𝑟 𝜋} + (𝑁 − 𝑀)𝜔 + ∑ 𝐴𝑟𝑔( 𝑒 − 𝑧𝑘 ) − ∑ 𝐴𝑟𝑔( 𝑒 𝑗𝜔 − 𝑝𝑘 )
𝑘=1 𝑘=1

# Evaluate the frequency response of the system described by the system function
1 𝑧
𝐻(𝑧) = −1
=
1 − 0.8𝑧 𝑧 − 0.8
Solution: Clearly, H(z) has a zero at z = 0 and a pole at p =0.8. Hence, the frequency response
of the system is
1
𝐻(𝑧)|𝑧=𝑒 𝑗𝜔 =
1 − 0.8𝑒 −𝑗𝜔
The magnitude response,
1 1 1
|𝐻(𝜔)| = | | = =
1 − 0.8𝑒 −𝑗𝜔 √(1 − 0.8𝑐𝑜𝑠𝜔)2 + (0.8𝑠𝑖𝑛𝜔)2 √1.64 − 1.6𝑐𝑜𝑠𝜔
And the phase response is
0.8𝑠𝑖𝑛𝜔
∅(𝜔) = 0 − tan−1 ( )
1 − 0.8𝑐𝑜𝑠𝜔
ω 0 π/5 2π/5 3π/5 4π/5 π
|𝑯(𝝎)| 5 1.7011 0.9343 0.6845 0.5838 0.5556
∅(𝝎) 0 -0.9271 -0.7907 -0.5478 -0.2781 -0
Page 4 of 6
Compiled by: Rupesh D. Shrestha, nec
Digital Signal Analysis & Processing Chapter 3

Linear phase of LTI system and its relationship to causality.

In the context of Linear Time-Invariant (LTI) systems, the concept of linear phase is crucial
for understanding how signals are processed without distortion. Linear phase is characteristic
of an ideal filter.

Let us assume that a signal sequence {x[n]} with frequency components confined to the
frequency range ω1 < ω < ω2 is passed through a filter with frequency response

𝐶𝑒 −𝑗𝜔𝑛0 , 𝜔1 < 𝜔 < 𝜔2

𝐻(𝜔) = {
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Where C and n0 are constants. The signal at the output of the filter has a spectrum

𝑌(𝜔) = 𝑋(𝜔)𝐻(𝜔) = 𝐶𝑋(𝜔)𝑒 −𝑗𝜔𝑛0 , 𝜔1 < 𝜔 < 𝜔2

By applying the scaling and time-shifting properties of the Fourier transform, we obtain the
time-domain output

𝑦[𝑛] = 𝐶𝑥[𝑛 − 𝑛0 ]

Consequently, the filter output is simply a delayed and amplitude-scaled version of the input
signal. A pure delay is usually tolerable and is not considered a distortion o the signal.
Neither is amplitude scaling. Therefore, ideal filters have a linear phase characteristic within
their passband, that is,

∅(𝜔) = −𝜔𝑛0

The term −ωn0 indicates a linear phase shift, which means that all frequency components of
the input signal are delayed by the same amount of time, preserving the waveform shape.

Page 5 of 6
Compiled by: Rupesh D. Shrestha, nec
Digital Signal Analysis & Processing Chapter 3

The derivative of the phase with respect to frequency has the units of delay. Hence, we can
define the signal delay as a function of frequency as

𝑑𝜑(𝜔)
𝜏𝑔 (𝜔) = −
𝑑𝜔

τg(ω) is usually called the envelope delay or the group delay of the filter. We interpret τg(ω) as
the time delay that a signal component of frequency ω undergoes as it passes from the input to
the output of the system. Note that φ(ω) is linear, τg(ω) = n0 = constant. In this case all frequency
components of the input signal undergo the same time delay.

In conclusion, ideal filters have a constant magnitude characteristic and a linear phase
characteristic within their passband. In all cases, such filters are not physically realizable but
serve as a mathematical idealization of practical filters.

Relationship to Causality

Causality is a fundamental property of physical systems, meaning that the output of the
system at any time depends only on past and present inputs, not future ones. For an LTI
system to be causal, its impulse response h[n] must be zero for n < 0.

The relationship between linear phase and causality can be summarized as follows:

• Linear Phase and Symmetry: For an LTI system to have a linear phase, its impulse
response h[n] must be symmetric or anti-symmetric. This symmetry ensures that the
phase response is linear.
• Causality and Symmetry: A causal system cannot have a symmetric impulse
response unless it is non-zero only for n≥0. This constraint means that achieving both
perfect linear phase and causality is challenging.

In practice, achieving an exact linear phase while maintaining causality often requires
compromises, such as using finite impulse response (FIR) filters, which can be designed to
approximate linear phase while being causal.

Reference:
1. J. G. Proakis, D. G. Manolakis, “Digital Signal Processing, Principles, Algorithms and
Applications”, 3rd Edition, Prentice-hall, 2000. Chapter 3.

Page 6 of 6
Compiled by: Rupesh D. Shrestha, nec