Classification of Discrete Time Systems

Discrete-time systems can be classified based on

several criteria, including their
• properties,
• behavior, and
• structure.
Next are some common classifications:
 Based on the Type of Input and Output

• Static versus Dynamic Systems

A discrete time system is called static or memory-less if its output at any
instant n depends at most on the input sample at the same time, but not
on the past or future samples of the input. In any other case, the system is
said to be dynamic or to have memory.
Examples: Dynamic System
Examples: Static System A swinging pendulum, a car in
A bridge under a constant load, motion, or a population model in
a building structure, or a still ecology.
fluid in a container.

Examples: y[n] = x2[n] is a memory-less system,

whereas the following are the dynamic systems:
(a) y[n] = x[n] + x[n-1] + x[n-2]
(b) y[n] = 2x[n] + 3x[n-4]
3
 Based on the Linearity
• Linear Systems: Follow the principles of superposition (additivity
and homogeneity).
• Nonlinear Systems: Do not stick on to the superposition principle.
A system H is linear if and only if
H[a1x1[n] + a2x2[n]] = a1H[x1[n]] + a2H[x2[n]]
for any arbitrary input sequences x1[n] and x2[n], and any arbitrary
constants a1 and a2.
a1
x1[n]
y1[n]
a2 + H
x2[n]
a1
x1[n] H
y2[n]
+
a2
x2[n] H

If y1[n] = y2[n], then H is linear. 4

 Example

Determine if the following systems are linear or nonlinear.

(a) y[n] = nx[n]
Solution:
To determine if the system is linear, we need to check two
properties of linearity: superposition and homogeneity.
Superposition: If x1​[n] produces y1​ Homogeneity: If an
[n] and x2​[n] produces y2​[n], then for input x[n] produces y[n], then for any
any scalars a and b, the input ax1​[n] scalar a, the input ax[n] should
+bx2​[n] should produce ay1​[n]+by2​[n]. produce ay[n].
For the input ax[n]:
Let y1​[n]=nx1​[n] and y2​[n]=nx2​[n].
y[n]=n(ax[n])=a(nx[n])=ay[n].
Then for the input ax1​[n]+bx2​[n]: The homogeneity property holds as well.
y[n]=n(ax1​[n]+bx2​[n])=nax1​[n]+nbx2​ Since both properties of linearity
[n]=a(nx1​[n])+b(nx2​[n])=ay1​[n]+by2​[n]. (superposition and homogeneity) hold,
The superposition property holds. the system y[n]=nx[n] is a linear system.

5
 Example

Determine if the following systems are linear or nonlinear.

(b) y[n] = Ax[n]+B
Solution: To determine whether the system described by the equation y[n]=y[n]=Ax[n]+B is
linear, need to check the two properties of linearity: superposition and homogeneity.

Superposition: For the system to be linear, if x1​ Homogeneity: For linear, the system should
[n] produces y1​[n] and x2​[n] produces y2[n], then for satisfy:
If x[n] produces y[n], then for any scalar a, the
any scalars a and b, the input
input ax[n] should produce ay[n].
ax1​[n]+bx2​[n] should produce ay1​[n]+by2​[n].
For the input ax[n]:
Let y1​[n]=Ax1​[n]+B and y2​[n]=Ax2​[n]+B.
y[n]=A(ax[n])+B = aAx[n]+B.
 For the input ax1​[n]+bx2​[n]:
For the output to be ay[n]:
y[n]=A(ax1​[n]+bx2​[n])+B =Aax1​[n]+Abx2​[n]+B. ay[n]=a(Ax[n]+B)= aAx[n]+ aB.
We need to compare this with ay1​[n]+by2​[n]: Again, these two results are not equal unless
•ay1​[n]+by2​[n]= B=0.
•a(Ax1​[n]+B)+b(Ax2​[n]+B)= aAx1​[n]+aB+bAx2​ Since the system fails to satisfy both the
[n]+bB =aAx1​[n]+bAx2​[n]+(a+b)B. superposition and homogeneity properties, we
conclude that the system y[n]=Ax[n]+B is
These two results are not equal
a nonlinear system unless B=0. If B=0, then it is
unless a+b=1, which is generally not true for
a linear system.
arbitrary scalars a and b. 6
 Based on Time-Invariance
Time-Invariant Systems: The system's behavior and characteristics do not change over time. If the input is
delayed, the output is delayed by the same amount.
Time-Varying Systems: The system's behavior changes over time. The output may not simply be a delayed
version of the input.
Time-Varying Systems Mathematical Definition
Time-Invariant Systems Mathematical Definition
y(t)=S[x(t)]
Let a system S such that: y(t)=S[x(t)] But the system operator S itself depends on time. Therefore, the time-
Where: y(t) output. x(t) input. varying system can be represented as: y(t)=S[t][x(t)]

S[⋅] operation of the system. Where S[t] indicates system's behavior can change
based on the time t.
System is time-invariant if for any time delay τ:
Example
S[x(t−τ)]=y(t−τ)
Consider a system defined by:
Example
Consider a simple linear system defined by:
y(t)=etx(t)
y(t)=ax(t) a is a constant gain. This means that the system applies a time-
dependent gain to the input signal. Now, let's apply a
If we apply a delayed input: delayed input:
y(t)=S[x(t−τ)] = ax(t−τ) If we take the delayed input x(t−τ):
This results in: y(t)=S[x(t−τ)]=etx(t−τ)
Now, compare the output with a delayed version of
S[x(t−τ)] = a⋅x(t−τ) = y(t−τ) the original output: y(t−τ)=et−τx(t−τ)
This shows that the system is time- Since et changes with time, etx(t−τ) is not equal to et−τ
x(t−τ) due to the varying gain et. Therefore, the
invariant, as the output is simply a
output of a time-varying system when the input is
delayed version of the input when we delayed does not match the delayed output of the
7
apply a delay. original system.
 Tutorial Based on Time-Invariance
Determine if the system shown in the figure is time invariant or time variant.
( A ) y[n] = x[n] – x[n-1]. ( B ) y[n] = nx[n] ( C ) y[n] = x[n]cosw 0n
Let's analyze each system one by one:
(a) y[n] = x[n] – x[n-1]. (b) y[n] = nx[n]
Original system: Original system: y[n]= nx[n]
y[n]= x[n] - x[n−1] Shift the input: Let x[n]→x[n−n0​].
Shift the input: Let x[n]→x[n−n0​], y[n]= nx[n−n0​]
where n0​is a time shift. Shift the output:
y[n] = −x[n−n0​] - x[n−n0​−1] y[n−n0​] = (n−n0​)x[n−n0​]
Shift the output: The output after shifting the input is not the same as
shifting the output because of the factor nn which
y[n−n0​]= x[n−n0​] - x[n−n0​−1]
depends on nn. Therefore, this system is
Since the output after shifting the input is the same time variant.
as shifting the output, this system is time invariant.
(c ) y[n] = x[n]cosw0n
Original system: y[n]= x[n]cos(ω0​n)
Shift the input: Let x[n]→x[n−n0​].
y[n]=x[n−n0​]cos(ω0​n)
Shift the output:
y[n−n0​]=x[n−n0​]cos(ω0​(n−n0​))= x[n−n0​]cos(ω0​n−ω0​n0​)
Since the output after shifting the input is not the same as shifting the output (due
8
 Based on Causality
Causal Systems: The output at any time depends only on current and past input
values (not future inputs).
Non-Causal Systems: The output depends on future input values.
n
y[n]   x[k ]
Determine if the systems described by the following input-output equations
k  are
causal or noncausal.
(a) y[n] = x[n] – x[n-1] (b) y[n] = ax[n] (c)
(d) y[n] = x[n] + 3x[n+4] (e) y[n] = x[n 2] (f) y[n] = x[-n]

A-(a) y[n]=x[n]−x[n−1] (d) y[n]=x[n]+3x[n+4]y[n]=x[n]+3x[n+4]

Causal: The output depends on Noncausal: The output depends on x[n+4],
x[n] (current input) and x[n−1] (previous input). which is a future input (since n+4>n).
(b) y[n] = y[n]=ax[n] (e) y[n]=x[n2]
n
Noncausal: The value of n2 can be greater than
Causal: y[nT]he  x[k ]depends only on the current
 output
n for n>1, which can lead to future inputs being
input x[n].
k 

used (future values of x for certain n).

(c) (f) y[n]=x[−n]
Noncausal: The output depends on x[−n],
Causal: The output depends on the where for positive values of n, this reference
goes to future negative indexed inputs.
current and past input.
9
 Based on Stability
Stable Systems: Bounded input leads to a bounded output(BIBO stability).
Unstable Systems: Bounded input may lead to an unbounded output
BIBO Stability (Bounded Input, Bounded Output): A digital system is BIBO
stable if every bounded input results in a bounded output.
Mathematically, this can be expressed as follows:
Let x[n] be a bounded input, such that ∣x[n]∣<M for all n( M:a finite
constant).
The system's output is y[n]=T(x[n]), where T is the system transformation.
The system is BIBO stable if there exists a constant K such that
∣y[n]∣<K for all n.
Asymptotic Stability: This is pertinent in systems where the system tends to
converge to a specific value over time, often seen in feedback systems.
Example of a Digital System
Consider a simple digital filter defined by the difference equation:
y[n]=0.5y[n−1]+x[n]

Stability Analysis: To check if this digital system is BIBO stable, we can

analyze its impulse response. 10
Impulse Response:
 Based on Stability
Impulse Response: First, we need to find the system's impulse response h[n]. For this
equation, when the input x[n] is an impulse δ[n]:
The output at n=0 will be:
y[0]=0+1=1
At n=1: y[1]=0.5y[0]+x[1]=0.5(1)+0=0.5
At n=2: y[2]=0.5y[1]+x[2]=0.5(0.5)+0=0.25

Continuing this, we see that: y[n]=0.5n for n≥0

Summing the Impulse Response: The system’s impulse response is:
h[n]=0.5nu[n] (where u[n] is the unit step function).
To check for BIBO stability, we compute:
 
1

k 0
h[n]  0.5 
k 0
n

1  0.5
2

The sum converges, leading to the conclusion that the system is BIBO stable.

Conclusion
In example, showed how the digital filter described by the difference equation is stable since
it does produce a bounded output for bounded inputs.
Mathematically, we verified its BIBO stability through the convergence of the impulse
response series. 11
 Based on the Structure
Digital Filters: Systems designed to process discrete signals, can be further classified into:
FIR (Finite Impulse Response) Filters: Have a finite duration impulse response.
IIR (Infinite Impulse Response) Filters: Have an infinite duration impulse response.
State-Space Systems: Represent systems using state variables and equations.
FIR Systems and IIR Systems. Each of these classes has different structural
characteristics, and we will analyze their features and behaviors while focusing on
their mathematical representations and implications for system performance.
1. Finite Impulse Response (FIR) Systems
Theoretical Overview
FIR systems have a finite duration of impact from an input signal, means output
depends only on a finite number of previous inputs.
FIR systems are inherently stable and can be designed to be linear phase, which is
advantageous for preserving the shape of signals.
Mathematical Structure M
FIR Structure define using Eq:
y[n]  bk x[n  k ]
k 0

where: y[n] is the output signal, x[n] is the input signal,

bk​are the coefficients (impulse response) of the FIR filter,
M is the order of the FIR filter. 12
 Based on the Structure
Properties
Stability: FIR filters are always stable if all coefficients bk​are finite.
Causality: FIR filters can be designed to be causal or non-causal.
A causal FIR filter only uses past and current input values.
Linear Phase: FIR filters can be designed to have a linear phase
response, preserving the waveform of signals.

Optimization
The design of FIR filters can be expedited using methods like
the Parks-McClellan algorithm, which uses the Remez exchange
algorithm to minimize the maximal ripple in the passband and
stopband frequency response, leading to an efficient and optimal filter
design.

M
y[n]  bk x[n  k ] 13
k 0
 Based on the Structure
2. Infinite Impulse Response (IIR) Systems
Theoretical Overview
IIR systems possess an infinite duration of impact from the input signal
due to feedback. Their outputs depend on both the current and past
inputs and past outputs, making them capable of achieving a specific
performance with fewer coefficients than FIR systems.
Mathematical Structure
An IIR system can be described using the following difference
equation: M N
y[n]  bk x[n  k ]  a j y[n  j ]
k 0 j 1

where:
aj​are the feedback coefficients, M is the order of the input,
N is the order of the output.

Properties 14
 Based on the Structure
2. IIR Systems
Properties
Stability: Stability criteria for IIR systems depend on the location of
the poles in the Z-domain; the system is stable if all poles are within
the unit circle.
Causality: Similar to FIR systems, IIR systems can also be causal or
non-causal.
Efficiency: IIR systems typically require fewer coefficients than FIR
systems to achieve a similar frequency response, leading to lower
computational load.

Optimization
Optimizing an IIR filter design can involve techniques such as:
Bilinear Transformation: Used to convert an analog filter prototype
into a digital IIR filter while preserving frequency characteristics.
Pole-Zero Placement: Strategic placement of poles and zeros can
achieve desired frequency response and stability. 15
 Based on the Structure
3. State-Space Representation
Both FIR and IIR systems can also be modeled using state-space
representation, which provides a comprehensive framework for
analyzing complex systems.
Theoretical Overview
In state-space models, the system is represented by a set of first-order
differential or difference equations.
Mathematical Structure
The state-space representation for a discrete-time system is given by:
​ x[n  1]  Ax[n]  Bu[n]
y[n] Cx[n]  Du[n]
where:
x[n] is the state vector, u[n] is the input,
y[n] is the output, A is the state matrix,
B is the input matrix, C is the output matrix,
D is the feedforward matrix. 16
 Based on the Structure
3. State-Space Representation
Properties
Controllability: Determines whether it's possible to move the system
state to a desired state using the input.

Observability: Determines whether the internal state can be inferred

from the output.

Optimization

State-space models can be optimized using techniques from control

theory, such as:

Linear Quadratic Regulator (LQR) design for optimal control

performance.
Kalman Filtering for efficient state estimation and control.
17
 Based on the Structure

Conclusion
By categorizing discrete-time systems based on their structures—
FIR, IIR, and state-space representation—we gain insights into their
theoretical properties and mathematical formulations. Each class
provides unique advantages for specific applications, and optimizing
their performance involves leveraging algorithms suited to their
structures. Understanding these classes aids in expediting the
analysis, design, and implementation of discrete-time systems in
various engineering applications.

 b x[n  k ]
k 0
k

18
 Based on the Nature of the System

Deterministic Systems: The output is precisely determined by the

input.
Stochastic Systems: The output is influenced by random variables,
leading to uncertainty in the output.

Based on Implementation
Analog Systems: Continuous-time systems that can be discretized.
Digital Systems: Systems that operate on discrete-time signals.

19
Summary
Understanding these classifications helps
in analyzing and designing discrete-time
systems effectively, as each category
has different implications for system
behavior and performance.

20