' $

Analogy between Signal Spaces and Vector Spaces

Consider two vectors V1 and V2 as shown in Fig. 1. If V1 is to be
represented in terms of V2

V1 = C12 V2 + Ve (1)

where Ve is the error.

V1

C12 V V
2 2

Figure 1: Representation in vector space

& %
' $
The error is minimum when V1 is projected perpendicularly onto
V2 . In this case, C12 is computed using dot product between V1
and V2 .

Component of V1 along V2 is

V1 .V2
= (2)
kV2 k

Similarly, component of V2 along V1 is

V1 .V2
= (3)
kV1 k

Using the above discussion, analogy can be drawn to signal spaces

& %
also.
' $

Let f1 (t) and f2 (t) be two real signals. Approximation of f1 (t) by

f2 (t) over a time interval t1 < t < t2 can be given by

fe (t) = f1 (t) C12 f2 (t) (4)

where fe (t) is the error function.

The goal is to find C12 such that fe (t) is minimum over the interval
considered. The energy of the error signal given by

Z t2
1
= [f1 (t) C12 f2 (t)]2 dt (5)
t2 t1 t1

To find C12 ,

& %
' $

=0 (6)
C12

Solving the above equation we get

Z t2
1
t2 t1 f1 (t).f2 (t) dt
t1
C12 = Z t2 (7)
1
t2 t1 f22 (t) dt
t1

The denominator is the energy of the signal f2 (t).

When f1 (t) and f2 (t) are orthogonal to each other C12 = 0.
Example: sin n0 t and sin m0 t be two signals where m and n are
integers. When m 6= n
& %
' $

Z
0
sin n0 t. sin m0 t dt = 0 (8)

0

Clearly sin n0 t and sin m0 t are orthogonal to each other.

& %