LTI System and Convolution

3.1 Introduction to System

An entity that manipulates one or more signals to accomplish a function thereby
yielding a new signal is called a system.
In other words, any physical device that performs an operation on a signal is
called a system.
A process in which input signals are transformed by the combination of
components (device) or cause the device to respond in some way resulting in another
signal as output.

The schematic diagram of LTI System is shown in figure below :

The time invariant system which holds the linearity property in its input - output
relationship is called linear time invariant (LTI) system. A LTI system can be
characterized by its impulse response. Impulse response of a LTI system is the response
of the system when it is excited by the delta function.
LSI stands for linear shift invariant system. It is also called the LTI system.
Properties of the LTI system are : Linearity, Causality, Stability, etc.
A time invariant system which holds the linearity property in its input-output
relationship is called linear time invariant (LTI) System. A LTI system can be

Prepared By : Er. Bigyan Bhattarai 1

LTI System and Convolution

characterized by its impulse response. Impulse response of a LTI system is the response
of the system when it is excited by delta function.

Mathematically,
y(t)=T{x(t)}
= x(t).h(t)
If x(t) is delta function then y(t)= δ(𝑡)h(t)
y(t) = h(t)
y[n]=T{x[n]}
= x[n].h[n]
If x[n] is delta function the y[n]=δ[𝑛]ℎ[𝑛]
y[n] = h[n]

3.2 Types of system

1. Continuous time system
The associated signals are continuous time signals i.e. input and output
signals both are continuous time signals then the system is called continuous
time system.
The schematic diagram of continuous time system is shown in figure below :

Figure: Continuous time system

Example : audio amplifier, video amplifier, power amplifier, etc.

2. Discrete time system

The associated signals are discrete time signals i.e. input and output
signals both are discrete time signals then the system is called discrete time
system. Here, the LTI system is one of the systems of signal analysis.
The schematic diagram of continuous time system is shown in figure below :

Prepared By : Er. Bigyan Bhattarai 2

LTI System and Convolution

Figure: Discrete time system

Example : Registers, Counters, Microprocessors,etc.

Definition
A discrete time system is a physical device (or an algorithm) which
performs required operation on a discrete time signal.
A discrete time system is represented as shown in figure:

Here, x (n) is an input discrete time signal applied to the system. It is also
called excitation. The system operates on discrete time signals. This is called as
processing of input signal, x (n). The output of the system is denoted by y (n). It is
also called a response system.

Notation
When input signal is passed through the system then it is represented by
following notation:
T
x(n) →y (n)
y(n) = T[x (n)]
Here T is called a transformation operation.

Illustrative Example
A filter is a good example of a system. A signal containing noise is applied
at the input of the filter. This is an input signal to the system (filter). The filter
cancels or attenuates noise signals. This is called the processing of signals. A
noise free signal obtained at the output of a filter is called as a response of the
system.

Prepared By : Er. Bigyan Bhattarai 3

LTI System and Convolution

3.3 SYMBOLS USED TO REPRESENT A DISCRETE-TIME SYSTEM

The operation of a discrete-time system may be described simply by drawing a
block schematic. We use different building blocks to form a complete block schematic.
These building blocks can be listed as under.
1. Adder
As the name indicates, an adder block is used to perform the
symbol of an adder has been shown in figure below Here, 𝑥1 (𝑛) and 𝑥2 (𝑛)

are two input sequences. Adder block adds 𝑥1 (𝑛) and 𝑥2 (𝑛). The
corresponding output is y(n).

Therefore y(n) + and 𝑥2 (𝑛)

Sometimes, instead of addition sign (‘ +’) sigma (𝚺) sign is used
to represent the addition. The subtraction operation that means, y(n) =
𝑥1 (𝑛) - 𝑥2 (𝑛) is represented as shown in figure below.
In case of addition (or subtraction) operation, it is not required to
store any one of the inputs in the memory. Hence, this is a memoryless
operation.

Figure : A subtraction operation

Prepared By : Er. Bigyan Bhattarai 4

LTI System and Convolution

2. Constant Multiplier
A constant multiplier operation is used to multiply input sequence
by some constant a (say).
That means this operation is used to change the amplitude of the
input sample. This is also a memoryless operation. Figure below shows a
constant multiplier operation.

Figure : A constant multiplier

3. Signal Multiplier
As the name indicates this block is used to multiply two input
sequences. Here, multiplication takes place on a sample to sample basis.
Further, this is also a memoryless operation. Figure below shows a signal
multiplier symbol.

Figure: Signal multiplier

4. Unit Delay
An unit delay block is used to delay input sequence by one
sample. A unit delay can be represented as shown in figure 1.6

Figure: Unit delay

Prepared By : Er. Bigyan Bhattarai 5

LTI System and Convolution

Hence, input sequence x(n) is delayed by one sample and the

output is y(n) = x(n - 1) This indicates that the calculation of output at a
particular instant, needs to be the previous input. Here, x(n) is the present
input and x(n -1) is the previous input.
For example, let us say we want to calculate output at n = 0 Then,
putting n = 0 in equation of y(n) we get,
y(0) = x(0 - 1)
Therefore, y(0) = x(- 1)
Hence, to calculate output at n = 0 i.e. y (0), we do not require
input at n = 0 i.e. x(0). But, we want x(- 1) Here, x(- 1) is the previous
input compared to x (0). Therefore, in this operation, every time, it is
required to store previous input in the memory. It is not a memoryless
operation.
If we want to delay input sequence by two samples, then, two unit
delay blocks should be connected in series as shown in figure figure
below:

Figure: Delay by two samples

5. Unit Advance
An unit advance block is used to advance input sequence by one
sample. This block can be represented as shown in figure below:

Figure Unit advance

The output equation is

y(n) = x(n + 1)
In this operation, to calculate output y(n), a future input sample is needed.
To calculate the output at n = 0, let us put n=0 in the equation of
y(n).
Therefore y(0) = x(1)
This means that to calculate present output y(0), we are not
making use of present input x(0). But , we want input x (1). Here, x (1) is
future input corresponding to x(0). In this operation we calculate output

Prepared By : Er. Bigyan Bhattarai 6

LTI System and Convolution

y(n), a future input sample is required. This is not practically possible.

Hence operation is not realizable in practice.

3.4 Properties of Discrete Time System

1. Static and Dynamic System (Dynamicity )
Static System :
Definition
It is a system in which output at any instant of time depends on the input
sample at the same time.
Illustrative Examples
(i) y(n)=5x(n)
Here, 5 is constant which multiplies input x (n). But, output at nth
𝑡ℎ
instant that means y(n) depends on input at the same (𝑛 ) time instant x
(n). Hence, this is a static system.
2
(ii) y(n)=𝑥 (n)+5x(n)+10
𝑡ℎ 𝑡ℎ
Here, also output at(𝑛 ) instant, y (n) depends on the input at 𝑛
instant. Hence, this is a static system.

Significance
Let us observe the input-output relations of a static system. Output does
not depend on delayed [x(n-k)] or advanced [x(n+k)] input signals. It depends
𝑡ℎ
only on the present (𝑛 ) input signal. If output depends on delayed input signals,
then such signals should be stored in memory to calculate the output at nth
instant. This is not required in static systems. Thus, for static systems, memory is
not required. Hence, static systems are memoryless systems.

Dynamic System
Definition
It is a system in which output at any instant of time depends on the input
sample at the same time as well as at other times. Here, other times means,
other than the present time instant. It may be past time or future time. Here, it
may be noted that if x(n) represents input signal at present instant, then,
(a) x(n-k) means a delayed input signal is called a past signal.
(b) x(n+k) means an advanced input signal is called a future signal.
Thus, in dynamic systems, output depends on present input as well as
past or future inputs.

Illustrative Examples
(i) y(n)=x(n)+5x(n-1)

Prepared By : Er. Bigyan Bhattarai 7

LTI System and Convolution

𝑡ℎ 𝑡ℎ
Here, output at 𝑛 instant depends on input at 𝑛 instant, x(n) as well as
𝑡ℎ
(n-1) instant x(n-1) is the previous (past) sample. Hence, the system is
dynamic.

(ii) y (n)=3x(n+2)+x(n)
Here, x (n+2) indicates an advanced version of the input sample that
means it is a future sample. So, this is a dynamic system.

Significance
Let us observe input-output relations of a dynamic system. Since output
depends on past or future input samples, we need a memory to store such
samples. Thus, a dynamic system has a memory.

2. Time variant and time invariant system :

Definition
The system is time invariant if its input - output characteristics do
not change with time.
The system in which input / output characteristics do not change
with time is called time invariant / shift invariant system.
Explanation
A time invariant system means its input - output characteristics are
not changing with time shifting. Let us consider that we are applying input
signal x(n) to the system as shown in figure below :
It produces output y(n).
Now, we delay input by ‘k’ samples. This means that the input
becomes x(n-k). Now, we apply this delayed input to the same system as
shown in figure below : Let us say the system now gives output y(n-k).
Then, this system is called a time shift or shift invariant system.

Figure Time Invariant System

Here, it may be observed that initially output is y(n) with input x(n).
When input is delayed by ‘k’ samples, then the output is also delayed by
same ‘k’ samples. Thus, the input-output characteristics are not changed.
Therefore, the system is time invariant.

Prepared By : Er. Bigyan Bhattarai 8

LTI System and Convolution

Theorem :
A system is time invariant if and only if
T T
x(n) →y (n) implies that x(n-k) →y (n-k)

Also,
If y(t) = T{x(t)}
Then y(t-𝑡0) = T{x(t-𝑡0)}
If y[n] = T{x[n]}
Then y[n-𝑛0] = T{x[n-𝑛0]

How can we determine whether the system is time invariant or not?

To check whether the system is time invariant (TIV) or time variant (TV),
we follow the following steps:
(i) First, we delay the input x(n) by 'k' units, that means x(n - k).
We denote the corresponding output by y(n, k)
This means that,
T
x(n-k) →y (n, k)

(ii) In the given expression of system y (n), we replace 'n' by ‘n - k'

throughout. This output is y(n - k).
(iii) If y (n, k) = y (n - k) then, the system is time invariant (TIV) and
if y (n,k) ≠ y(n - k) then the system is time variant (TV).

3. Linear and Non Linear System (Linearity Property)

Definition
A linear system is a system which follows the superposition principle. A
linear system may be defined as one whose response of the sum of weighted
inputs is equal to the sum of the weighted responses.
Let us consider a system as shown in figure below whose response is T.
Its input is x(n) and it produces output y(n).

Figure : A system with response T

T : Response of T
x[n] : Input Signal
y[n] : output signal

Prepared By : Er. Bigyan Bhattarai 9

LTI System and Convolution

Let, 𝑥1(𝑛) produces output 𝑦1(𝑛) and input 𝑥2(𝑛) produces output 𝑦2(𝑛) .
Let us consider two arbitrary constants 𝑎1and 𝑎2. Then simply we can say that
input 𝑎1𝑥1(𝑛) produces output 𝑎1𝑦1(𝑛) and input 𝑎2𝑥2(𝑛) produces output 𝑎2𝑦2(𝑛).

Theorem,
A system is said to be linear if the combined response of 𝑎1𝑥1(𝑛) and
𝑎2𝑥2(𝑛) is equal to the addition of individual responses.
T[𝑎1𝑥1[𝑛]+ 𝑎2𝑥2[𝑛] ] = 𝑎1T[𝑥1[𝑛]]+𝑎2T(𝑥2[𝑛]) For discrete time system
and
T[𝑎1𝑥1(𝑡)+ 𝑎2𝑥2(𝑡) ] = 𝑎1T[𝑥1(𝑡)]+𝑎2T(𝑥2(𝑡)) For continuous time system

The schematic diagram of linear system is shown in figure below:

If a system is a nonlinear system, then it does not hold the principle of

superposition.

Important Characteristics
An important characteristic of a linear system is that when input is zero
then it produces zero output. But, if zero input gives some value at the output
(non-zero), then it is a nonlinear system. If this condition is satisfied, then we
apply the superposition theorem to check whether the system is linear or not

Prepared By : Er. Bigyan Bhattarai 10

LTI System and Convolution

How to decide whether the system is linear or not?

1. First, we apply zero input and check the output. If output is zero then the system
is linear. If this step is satisfied then follow the remaining steps.
2. Then, we apply individual input to the system and note down corresponding
'
outputs. Then, we add all the outputs. We denote this addition by 𝑦 (𝑛). This RHS
of the equation as shown above.
''
3. Next, we combine all inputs, apply it to the system and find out output 𝑦 (𝑛). This
is the LHS of the equation as shown in the figure.
' ''
4. If 𝑦 (𝑛)=𝑦 (𝑛), then the system is linear, otherwise it is a nonlinear system.

4. Causal and non causal

Causal System
Definition
If the output of any system depends only on the present and past inputs
but not future inputs of the system is called a causal system. All the real time
systems are causal systems. Since, the systems cannot know the future values
of the input signals. I.e. Causal systems are physically realizable.
Causal : output value depends on : present and past value.
Non causal : output value does not depends on : present and past values (i.e.
depends on future values)

Illustrative Example :
Causal Signal :
y(t) = 2ax(t) (for continuous)
y[n]=2ax[n] (for discrete)
Non Causal Signal :
y(t) = x(t+1) (for continuous)
y[n]=x[n+1] (for discrete)

Significance
Since a causal system does not include future input samples, such a system is
practically realizable. This means that such a system can be implemented
practically. Generally all real time systems are causal systems, because in real
time applications only present and past samples are present. Since future
samples are not present, the causal system is a memoryless system.

Anticausal or non causal system

Definition
A system is said to be an anticausal system if its output depends not only
on present and past inputs but also on future inputs.

Prepared By : Er. Bigyan Bhattarai 11

LTI System and Convolution

Illustrative Examples
In this case,output y(n) is the function of x(n), x(n-1), x(n-2)..... etc. as well as it is
the function of x(n+1), x(n+2), … etc. Some example of non - causal systems are:

(i) y(n) = x(n) + x(n+1)

(ii) y(n) = x(n+2)
(iii) y(n) = x(n) + nx(n+1)

Significance
Since, a non-causal system contains future samples, a non-causal system is
practically not realizable. That means in practical cases it is not possible to
implement a non-causal system.
But, if the signals are stored in the memory and at a later time they are used by a
system then such signals are treated as advanced or future signals, because
such signals are already present, before the system has started its operation. In
such cases it is possible to implement a non causal system.

Some practical examples of non causal systems are as follows :

(i) Population Growth
(ii) Weather forecasting
(iii) Planning commission

Causality of LTI System Continuous time signal

In LTI system causality can be translated in the condition of impulse
response for this, let us consider a LTI system having output at t=𝑡0. given by the
convolution formula,
∞
y(𝑡0) = ∫ ℎ(𝛕)𝑥(𝑡0 − 𝛕) 𝑑𝛕
−∞
0 ∞
y(𝑡0) = ∫ ℎ(𝛕)𝑥(𝑡0 − 𝛕) 𝑑𝛕 + ∫ ℎ(𝛕)𝑥(𝑡0 − 𝛕) 𝑑𝛕 ……………… (i)
−∞ 0
(I) (II)

In the Ist terms, the integration involves future components only and in the IInd
term, the integration involves past and present values of inputs.
Therefore, it is clearly such that for a system to be causal, the impulse response
must satisfy the condition.
h(t)=0, t<0………………..(2)
Which is both a necessary and sufficient condition for causality.

Similarly, the convolution integration is given as,

Prepared By : Er. Bigyan Bhattarai 12

LTI System and Convolution

∞
y(t) = ∫ ℎ(𝛕)𝑑(𝑡 − 𝛕) 𝑑𝛕…………….(3)
0
∞
If y(t) = ∫ 𝑥(𝛕)ℎ(𝑡 − 𝛕) 𝑑𝛕 then we can write the expression for convolution of
−∞
causal LTI system as,
𝑡
y(𝑡) = ∫ 𝑥(𝛕)ℎ(𝑡 − 𝛕) 𝑑𝛕
−∞

Causality of LTI System Discrete time signal

5. Stable and unstable

Basic Aspects
To define stability of a system, we will use the term ‘BIBO’. It stands for
Bounded Input Bounded Output. The meaning of the word ‘bounded’ is some
finite value. Hence, bounded input means input signal is having some finite value
i.e. input signal is not infinite. Similarly, Bounded output means, the output signal
attains some finite value i.e. the output is not reaching to infinite.

Definition
An initially relaxed system is BIBO stable if and only if every bounded
input produces bounded output. Here a relaxed system means when input to the
system is zero then the output of the system is also zero.

Mathematical Representation
Let us consider some finite number 𝑀𝑥 whose value is less than infinity.
That means 𝑀𝑥<∞, so, it's a finite value. Then, if input is bounded, we can write.
|x(n)| < 𝑀𝑥< ∞

Definition of unstable system

An initially relaxed system is said to be unstable if bounded input
produces unbounded (infinite) output.

Significance
i. Unstable system shows erratic and extreme behavior.
ii. When an unstable system is practically implemented, then it causes overflow.

Prepared By : Er. Bigyan Bhattarai 13

LTI System and Convolution

A system is said to be stable / BIBO stable if and only if every bounded input
produces a bounded output.

I.e. |f[t]|≤M Where, M is the positive real finite number (output)

|f[n]|≤M
The output of a stable system does not grow unreasonably large, if an input and
output signal are bounded and translated mathematically to mean that there
exists some finite numbers, say 𝑀𝑥 𝑎𝑛𝑑 𝑀𝑦.
I.e. |x(t)|≤𝑀𝑋<∞ |x[n]|≤𝑀𝑋<∞
|y(t)|≤𝑀𝑦<∞ |x[n]|≤𝑀𝑋<∞
For all t for all n
If for some bounded input signal, if the output is unbounded (infinite) then the
system is said to be unstable.
Proof :
Now, if x(t) is bounded input to a LTI system and we have,
∞
y(𝑡) = ∫ ℎ(𝛕)𝑥(𝑡 − 𝛕) 𝑑𝛕
−∞
∞
|y(𝑡)| = | ∫ ℎ(𝛕)𝑥(𝑡 − 𝛕) 𝑑𝛕|
−∞
∞
But, |y(t)| ≤ ∫ |ℎ(𝛕)||𝑥(𝑡 − 𝛕)| 𝑑𝛕
−∞
∞
y(𝑡) ≤ 𝑀𝑋 ∫ ℎ(𝛕) 𝑑𝛕
−∞
Output of the bounded if the impulse response of the system is absolutely
∞
integrable i.e. if ∫ |ℎ(𝛕)| 𝑑𝛕<∞
−∞
Every bounded input produces a bounded output
∞
i.e. |y[n]| = ∑ |ℎ(𝑘). 𝑥(𝑛 − 𝑘)|
𝑘=−∞
∞
= ∑ ||ℎ(𝑘)|. |𝑥(𝑛 − 𝑘)||
𝑘=−∞
Hence, the magnitude values are equal for input and output.

Stability property of LTI System

Prepared By : Er. Bigyan Bhattarai 14

LTI System and Convolution

3. Memory (dynamic) and Memoryless (static)

A memoryless system does not have memory to store any input values because
it just operates on the current input. If a system is not memoryless, we can call it a
system with memory.
A system is said to be memoryless if its output signal value depends on the
present value of the input signal value.
I.e. Memoryless : output value depends : present value
Memory : output value depends : present and past value, present and future
value, future value, past value, future value and past value.

Mathematically,
Memoryless system i.e. depends on present value only
y(t)=x(t)
y[n]=x[n]

Mathematically,
Memory system i.e. depends on past and future value
𝑡
1
1. y(t) = 𝐶
∫ 𝑥(𝛕)𝑑𝛕 : Capacitor (past values)
−∞
∞
2. y[n] = ∑ 𝑥(𝑡) : Summer / accumulator (future values)
𝑘=−∞

Proof :
A continuous time LTI system is memoryless if,
h(t)=0, t≠0 ………..(i)
∞
y(t) = ∫ 𝑥(𝛕)ℎ(𝑡 − 𝛕) 𝑑𝛕
−∞

Prepared By : Er. Bigyan Bhattarai 15

LTI System and Convolution

A memoryless system has a form,

y(t)=kx(t).................(ii)
For some constant k, and has impulse response,
h(t)=k.δ(t).............(iii)
For k=1, then the system is called identity system,

For convolution integral formula as,

∞
y(t) = ∫ 𝑥(𝛕)𝑑(𝑡 − 𝛕) 𝑑𝛕
−∞
= x(t) * d(t) (also called shifting property)
A system is said to be memory when,
h(t)≠0 h≠0

7. Vertibility and Invertibility

A system is said to be invertible if distinct input leads to distinct output.
I.e. a system is said to be an invertible system if the input of the signal can be
recovered from the output signal.
The block diagram is shown in figure below :

Prepared By : Er. Bigyan Bhattarai 16

LTI System and Convolution

If the system is invertible then there exists an inverse system that was
cascaded with the original system which yields an output signal {z[n] or z(t)}
equal to the input signal {z[n] or z(t)}.

The system which produces zero output sequence {y[n]=0 or y(t)=0} from
2 2
any input signal {y(t)= 𝑥 (𝑡) 𝑜𝑟 𝑦[𝑛] = 𝑥 [𝑛], in which we cannot determine the
sign of the input from the knowledge of the output are non-invertible systems.

Invertibility of LTI system:

Consider, h(t) be the impulse response of the LTI system. Then the LTI
system is invertible only when an inverse system exists that when cascaded with
the original system, produces an output equal to the input to the first system.

3.5 Impulse response of LTI System

In this article, we will show that the LTI system can be completely characterized
by its impulse response. We shall use the following notations related to discrete time
systems.
x(n) = Input Sequence
h(n) = Impulse response of system
y(n) = Output sequence

Let impulse response h(n) is a delta function 𝞭(n). 𝞭(n) is delayed by k samples,
then we can write,
𝞭(n-k) = 1 for n=k
= 0 otherwise

Composition of input sequence x(n)

Prepared By : Er. Bigyan Bhattarai 17

LTI System and Convolution

∞
x(n) = ∑ 𝑥(𝑘) 𝞭(𝑛 − 𝑘)
𝑘=−∞
If we transpose the delta function we will get an impulse response i.e. if we
transpose the composite delta signal we will get a convolution sum (linear
convolution).

3.6 Convolution Theorem

1. Continuous time signal
(Convolution Integral)
From the definition of convolution
∞
x(t) = ∫ 𝑥(𝛕). 𝛅(𝑡 − 𝛕) 𝑑𝛕…………….(i)
−∞
Since, the system is linear, the response of y(t) of the system to an
arbitrary input x(t) can be expressed as,
∞
y(t) = T{x(t)} = T{ ∫ 𝑥(𝛕)ℎ(𝑡 − 𝛕) 𝑑𝛕
−∞
Since, the system is linear
∞
= ∫ 𝑥(𝛕) . 𝑇{𝛅(𝑡 − 𝛕)} 𝑑𝛕………….(ii)
−∞
Since, the system is time invariant we have,
h(t-𝛕) = T{𝛅(𝑡 − 𝛕)}
h(t)=𝑇{𝛅(𝑡)} ………….. (iii)
From equation (ii) and (iii) we get,

∞
y(t) = ∫ 𝑥(𝛕) . ℎ(𝑡 − 𝛕) 𝑑𝛕
−∞
y(t) = x(t) * h(t)
Hence, this is the required expression of convolution property.

2. Discrete time system

(Convolution series)
It is interesting to note that the proof of linear convolution or Proof
of LTI system is completely characterized by unit impulse response h(n).
Let us consider the relaxed LSI (LTI) system. A relaxed system
means if input x(n) is zero then output y(n) is zero. Let us say unit impulse
δ(n) is applied to this system then its output is denoted by h(n). Here, h(n)
is called the impulse response of the system.
T
i. δ(n) → y(n) = h(n)

Prepared By : Er. Bigyan Bhattarai 18

LTI System and Convolution

Since the system is time invariant, if we delay input by ‘k’ samples, then
output should be delayed by the same amount.
T
ii. δ(n-k) → y(n) = h(n-k)

Now, multiplying both sides by x(k) we get,

T
iii. x(k)δ(n-k) → y(n) = x(k) h(n-k)
Since the system is linear, we can apply the superposition
theorem. Hence, taking summation at both sides, we get

T
∞ ∞
∑ 𝑥(𝑘). δ(𝑛 − 𝑘) → y(n) = ∑ 𝑥(𝑘). ℎ(𝑛 − 𝑘)
𝑘=−∞ 𝑘=−∞
Thus we have,
∞
y(n) = ∑ 𝑥(𝑘). ℎ(𝑛 − 𝑘)
𝑘=−∞
This summation is called convolution sum. In this case output y(n) is
obtained by taking the convolution of x(n) and h(n). This is called as linear
convolution and it is denoted by symbol ‘*’
Therefore,
∞
x(n) * h(n) = ∑ 𝑥(𝑘). ℎ(𝑛 − 𝑘)
𝑘=−∞
In this case, we can say that the behavior of the LTI system is completely
characterized by its impulse response.

3.7 Convolution properties

1. Commutative Property
Basic Property of convolution
I.e. 𝑦(𝑡) = 𝑥(𝑡) * ℎ(𝑡) = ℎ(𝑡) * 𝑥(𝑡)
𝑦[𝑛] = 𝑥[𝑛] * ℎ[𝑛] = ℎ[𝑛] * 𝑥[𝑛]

Prepared By : Er. Bigyan Bhattarai 19

LTI System and Convolution

2. Associative Property
Basic Properties of Convolution
x(t)*[ℎ1(𝑡)* ℎ2(𝑡)] = [x(t)*ℎ1(𝑡)]*ℎ2(𝑡) = y(t)
x[n]*[ℎ1[𝑛]* ℎ2[𝑛]] = [x[n]*ℎ1[𝑛]]*ℎ2[𝑛] = y[n]

3. Distributive Property
This is basic property of Convolution
i.e. x(t)*[ℎ1(𝑡)+ ℎ2(𝑡)] = x(t)*ℎ1(𝑡)+ x(t)*ℎ2(𝑡) = y(t)
x[n]*{ℎ1[𝑛]+ ℎ2[𝑛]} = x[n]*ℎ1[𝑛]+ x[n]*ℎ2[𝑛] = y[n]
The block diagram is shown below :

3.8

Prepared By : Er. Bigyan Bhattarai 20

LTI System and Convolution

3.9 Linear Constant coefficient Differential Equation

Until now, we have studied different characteristics of the linear time
invariant system (LTI) system. We have also discussed that output of discrete
time system can be calculated using equation of linear convolution is
∞
y(n)= ∑ 𝑥(𝑘). ℎ(𝑛 − 𝑘) …………….. (i)
𝑘=−∞

Here, h(n) is the impulse response of the system. By the use of h(n) we
can calculate output y (n) for any input x(n) using equation (i)
Now in this section we will discuss the different equations which are
useful to implement (realize) the discrete time system. These different equations
are especially used for designing the digital filter.
We discussed that impulse response h(n) is used to calculate output of
discrete time systems. Now, if the system is causal, then impulse response h(n)
is zero for negative values of n. Hence, for the causal LTI system equation of
convolution becomes.
∞
y[n]= ∑ 𝑥(𝑘). ℎ(𝑛 − 𝑘) ………… (ii)
𝑘=0

Equation (ii) can be written as

∞
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘) ………… (ii)
𝑘=0

Since output depends on impulse response h(n), the number of samples

present in h(n) plays an important role in determining output y(n). The
number of samples present in h(n) is called length of impulse response.

Depending upon the length of impulse response, there are two types of
discrete time systems:
1. Finite impulse response (FIR) systems.
2. Infinite impulse response (IIR) systems.

1. Finite impulse response systems (FIR)

As the name indicates; impulse response h(n) is having a
finite length. That means h(n) contains some finite number of
samples. Let the range of h(n) is from - M to + M. Thus, equation
of convolution for FIR system becomes,

Prepared By : Er. Bigyan Bhattarai 21

LTI System and Convolution

𝑀
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘) (non causal) ………….. (iii)
𝑘=−𝑀

Here, it may be noted that in equation (iii), negative terms are

present, so, this equation is for the non-causal system. Now, for the
causal FIR system, we will consider only positive samples of h(n). Thus,
the range of h(n) becomes 0 to M instead - M to M. Equation (iii)
becomes,
𝑀
y[n]= ∑ ℎ(𝑘.)𝑥(𝑛 − 𝑘) ………….. (iv)
𝑘=0

Here, it may be noted that the summation is taken from 0 to M.

Hence, there will be total M + 1 terms in h(n) or h(k). These terms are as
under:
h(0), h(1), h(2) …….. h(M) Number of terms in h(n). Now,
expanding equation (iv), we get,
y(n) = h(0) * x(n) + h(1) * x(n - 1) + h(2) * x(n - 2) +...+h(M)(n - M) … (v)

In equation (v), x(n) is the present input and x(n - 1), x(n - 2) etc.
are past inputs. Every input is multiplied by the corresponding value of
h(n). Thus, equation (v) indicates that output y(n) is the weighted sum of
inputs.
Here, it may be noted that in this case, the system has to view
input samples only from x(n) to x(n - M) and not all input samples. These
input samples are the most recent input samples. Thus, this system has
to store x(n - 1), x(n-2)......x(n-M) , past samples in the memory. Since
these are a finite number of samples, FIR has limited or finite memory
requirements.

2. Infinite impulse response (IIR) systems

As the name indicates, impulse response h(n) contains ∞
samples. That means the length of h(n) is infinite. Now, the
equation of convolution for non-causal IIR system becomes
∞
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘) (Non causal) …………. (vi)
𝑘=−∞

For the causal system, limits of summation will be from 0 to ∞

Thus,
∞
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘) (causal) ………. (vii)
𝑘=0
Here, the range of h(n) = h(k) is as follows:

Prepared By : Er. Bigyan Bhattarai 22

LTI System and Convolution

h(0), h(1), h(2)........ h(∞). So, this is an infinite impulse

response system.

Now, expanding equation (vii), we get,

y (n) = h (0) x (n) + h (1) x (n-1)+h (2) x (n-2)+ ……… +h (∞) x (n-∞) …… (viii)
In this case, the IIR system has to store all past samples from x(n-1) to
x(n-∞). Thus, the IIR system needs infinite memory. This is practically not
possible.

Important Point: We discussed that the FIR systems need limited

memory. So, it is possible to design FIR systems using convolution equations.
But IIR systems cannot be designed as a convolution equation. This is because
of the infinite memory requirement. So, to implement such IIR systems:
difference equations are used:

Now depending upon length of h(n), we have classified the system as FIR
and IIR. Similarly, depending upon output of system; the discrete time systems
are classified as under:
(i) Non-recursive systems
(ii) Recursive systems.

Why is this type of classification required?

Equation of convolution indicates that for causal systems: the output
depends on present and past input as well as it depends on the impulse
response, h(n). But this is not always the case. There are some systems whose
present output depends on present and past inputs as well as past (previous)
output. For such systems we should take into account the previous output to
calculate present output. This type of classification gives the idea about the
requirement of previous output to calculate present output.

(i) Non-recursive system

Non-recursive systems do not require any past output sample to calculate
the present output. Consider the equation of FIR causal system (equation iv) This
equation is,

𝑀
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘)
𝑘=0

Prepared By : Er. Bigyan Bhattarai 23

LTI System and Convolution

To calculate present output we need present and past inputs only as

shown by equation (v). We will rewrite equation (v) as follows:
y(n)=h (0) x (n) + h (1) x (n-1)+....+h (M) x (n-M)

This equation does not contain any previous output term like y (n-1), y
(n-2)...etc. So this is the equation of a non-recursive system. Thus we can
conclude that the causal FIR system is non-recursive. In other words we can say
that non-recursive implementation of FIR causal systems is possible.
Now, for a non-causal FIR system; the limits of summation will be from -M
to M. Thus these systems contain present, past and future inputs. But it does not
contain future terms. So this is also a non-recursive system.
In case of non-recursive systems, past output terms are not required.
Past (previous) output terms are obtained by taking feedback from output to the
input. Thus, non-recursive systems do not require any feedback.

Is it possible to implement a non-recursive IIR system?

We know that,
Non-recursive ↔ No feedback or no past output term present and
IIR ↔ Infinite Impulse Response.

The equation of convolution for causal IIR system is,

𝑀
y[n]= ∑ ℎ(𝑘). 𝑥(𝑛 − 𝑘)
𝑘=𝑜

Expanding the summation, we get,

y(n) = h(0) x(n) + h(1) x(n-1)+h(2) x(n-2)....+h(∞)x(-∞)

We have discussed earlier that such systems require storing all

past inputs. That means it is required to store infinite input terms in the
memory. So, infinite memory is required. That is not possible practically.
Thus, non-recursive implementation of a causal IIR system is not
possible.

The representation of non-recursive system is shown in figure below:

Fig. . Non-recursive system

Prepared By : Er. Bigyan Bhattarai 24

LTI System and Convolution

(ii) Recursive System

A discrete time system in which output y(n) depends on present
input, past inputs as well as previous outputs; is called a recursive
system.
Let us consider one example. Let us consider a system
represented by block schematic as shown in figure

First, we will write the equation for the system.

(i) As shown in figure , a connection is drawn from output y(n) to the input
side. This is feedback.
−1
(ii) This feedback signal, y(n) is passed through an unit delay (𝑧 ) block.
Hence, the output of the delay block is y(n-1). This is past output
compared to y(n).
(iii) The signal y(n-1) is multiplied by constant multiplier ‘a’ to obtain a
y(n-1)
(iv) An adder is used to add input x (n) and signal ay (n - 1)
(v) Now, the addition of x(n) and a y(n - 1) produces the output

Thus the output of system is expressed as,

y(n) = x(n) + ay(n - 1)
At instant n = 0 we get,
y(0) = x(0) + ay(- 1)
y (0) is present output
x(0) is present input
y(-1) is past output
Similarly at n=1 x(1) + ay(0)
Thus, y(0) is past output.
Hence, at any instant present, output depends on past output.
Therefore, it is a re cursive system. The representation of non
recursive system is shown in figure below:

Prepared By : Er. Bigyan Bhattarai 25