144 C H A P T E R 2: Continuous-Time Systems
which can be written as
y(t)=
1
2 T
∫t
t−T
x(τ)dτ+
1
2 T
∫t+T
t
x(τ)dτ
At the present time t,y(t)consists of the average of a past and present values in[t−T,t]of the input, and
of the average of future values of the signal (i.e., the average of values x(t)for[t,t+T]). Thus, this system is
not causal.
An LTI system represented by its impulse responseh(t)iscausalif
h(t)= 0 fort< 0 (2.21)
The output of a causal LTI system with a causal inputx(t)(i.e.,x(t)= 0 fort< 0 ) is
y(t)=
∫t
0
x(τ)h(t−τ)dτ (2.22)
One can understand the above results by considering the following:
n The choice of the starting time ast=0 is for convenience. It is purely arbitrary as the system being
considered is time invariant, so that similar results are obtained for any other starting time.
n When computing the impulse responseh(t), the inputδ(t)only occurs att=0 and there are no
initial conditions. Thus,h(t)should be zero fort<0 since fort<0 there is no input and there
are no initial conditions.
n A causal LTI system is represented by the convolution integral
y(t)=
∫∞
−∞
x(τ)h(t−τ)dτ
=
∫t
−∞
x(τ)h(t−τ)dτ+
∫∞
t
x(τ)h(t−τ)dτ
where the second integral is zero according to the causality of the system (h(t−τ)=0 whenτ >t
since the argument ofh(.)becomes negative). Thus, we obtain
y(t)=
∫t
−∞
x(τ)h(t−τ)dτ