Convolution Representation of Linear Time - Invariant Continuous - Time Systems
Convolution Representation of Linear Time - Invariant Continuous - Time Systems
Convolution Representation of Linear Time - Invariant Continuous - Time Systems
Impulse Response
δ (t ) = 0 , t ≠ 0
∫ δ( λ) dλ = 1 .
h (t ) = δ ( t ) .
h( t )
x ( t ) = δ( t )
SYSTEM t
t
If x(t) is a causal function ( x(t) = 0 for all t < 0 ), by the sifting property,
∞
x (t ) = ∫ x (λ ) δ (t − λ ) dλ .
−
0
Therefore,
0 ,t <0
t
x ( t ) * v (t ) =
∫ x ( λ ) v ( t − λ ) d λ , t > 0
0
The integral exists if x(t) and v(t) are integrable. That is,
t
∫ x (λ )d λ < ∞
0
t
∫ v (λ )d λ < ∞ , t > 0.
0
Continuous Convolution 3
The following steps are recommended when performing convolutions of two functions x(t) and v(t).
Step 1. Graph x(λ) and v(−λ) (or vise-versa) as functions of λ. That is, fold v(λ) about the vertical
axis.
Step 2. Upon folding v(λ), denote the point where λ = 0 as t. All other points of v(−λ) are relative
to the reference point t. Remember, we're now in the λ-axis.
Step 3. Define an interval for which the product x(λ)v(t −λ) has the same analytical form (e.g. from t
= 0 to t = 1).
Step 4. Integrate the product x(λ)v(t −λ) as a function of λ over the integral defined in Step 3.
Step 5. Repeat Steps 3 and 4 as many times as necessary until x(λ)*v(t) is computed for all t >0.
EXAMPLE
v(t) x(t)
3 3
2 2
1 1
-2 -1 0 1 2 3 4 5 t -2 -1 0 1 2 3 4 5 t
Continuous Convolution 4
Choose to flip v(t) about the y-axis. This will then be the convolution at t = 0.
v( −λ) x( λ)
t=0
3 3
2 2
1 1
-2 -1 0 1 2 3 4 5 -2 -1 0 1 2 3 4 5
λ λ
t-1 t
The first region of integration is t<0 for which no overlapping area exists. We construct a table as
follows:
For ease of viewing the "overlap" intervals, the two functions are drawn on the same axes:
0<t<1
3
v(t −λ)
2
x( λ)
1
-2 -1 0 1 2 3 4 5 λ
t-1 t
The function v(t − λ) slides over the function x(λ). The second region region of overlap and function
to be integrated is 0 ≤ t < 1 with the lower limit of the overlapping area determined by x(λ) and the
upper limit t of the area is determined by v(t − λ) .
λ upper
∫
λ lower
(2)(1) d λ
0≤t<1 0 t
Continuous Convolution 6
The third region of overlap is 1 ≤ t < 2 as shown below. There is complete overlap of the two areas.
Therefore, the integration limit of the overlap area are from t − 1 to t defined by v(t − λ) in the region
1 ≤ t < 2.
1<t<2
3
2 v(t −λ)
x( λ)
1
-2 -1 0 1 2 3 4 5 λ
t-1 t
λ upper
∫
λ lower
(2)(1) d λ
0≤t<1 0 t
λ upper
∫
λ lower
(2)(1) d λ
1≤ t < 2 t−1 t
Continuous Convolution 7
The fourth region of overlap is 2 ≤ t < 3 over which the upper limit of integration is 2 as defined by x(λ)
and the lower limit is t − 1 as defined by v(t − λ) .
2<t<3
3
v(t −λ)
2
x( λ)
1
-2 -1 0 1 2 3 4 5 λ
t-1 t
λ upper
∫
λ lower
(2)(1) d λ
0≤t<1 0 t
λ upper
∫
λ lower
(2)(1) d λ
1≤ t < 2 t−1 t
λ upper
∫
λ lower
(2)(1) d λ
2≤t<3 t−1 2
Continuous Convolution 8
The last region is trivial in that there is no overlap between the two functions. The resultant area of
overlap is 0.
t >3
3
2 v(t −λ)
x( λ)
1
-2 -1 0 1 2 3 4 5 λ
t-1 t
λ upper
∫
λ lower
(2)(1) d λ
0≤t<1 0 t
λ upper
∫
λ lower
(2)(1) d λ
1≤ t < 2 t−1 t
λ upper
∫
λ lower
(2)(1) d λ
2≤t<3 t−1 2
x(t)*v(t)
-2 -1 0 1 2 3 4 5 t