ME 5224: Signals and Systems

Spring: 2022
Instructor: Dr. I-Fan Lin

Homework 5 –Orthonormal Transformations

Spring 2022

Exercise 1. Orthogonality and Normalized Functions

The inner product of two functions over the interval t ∈ (a, b) is defined as
Z b
hf (t), g(t)i = f (t)g ∗ (t)dt
a

where g∗ denotes the complex conjugate of g(t).

The function f (t) and g(t) are called orthogonal if hf (t), g(t)i = 0. A function f (t) is said to be normalized
if Z b
hf (t), f (t)i = |f (t)|2 dt = 1
a
(a) Are the functions sin(mωt) and sin(nω0 t) orthogonal over the interval t ∈ (a, T ) where T = 2π/ω0
and m 6= n?
(b) Let x(t) be an arbitrary real-valued signal, and let
x(t) − x(−t)
x0 (t) =
2

and
x(t) + x(−t)
xe (t) =
2

be the odd and even parts of x(t), respectively. Show that x0 (t) and xe (t) are orthogonal over the
interval t ∈ (−T, T ) for any T .
(c) Find the value of a so that the functions f (t) = t and g(t − at2 ) are orthogonal over the interval
t ∈ (0, 1). Determine the real-valued constants c1 and c2 so that fc(t)
1
and fc(t)
2
are both normalized.

Exercise 2. Orthonormal Set of Functions

A set of functions {φk (t)}, where k is any integer, is called an orthonormal set if i)φk (t) and φm (t) are
orthogonal for k 6= m and ii) all functions in {φk (t)} are normalized.
For each of the following problems, check if the given set of functions form an orthonormal set over the
specified interval. If the functions are not normalized determine the constant C so that the set {φk (t)/C} is
orthonormal.
ejkω0 t 2π
(a) φk (t) = √ for ω0 = and t ∈ (0, T ).
T T
2π
(b) φk (t) = cos(kω0 t) for ω0 = and t ∈ (0, T ).
T

Exercise 3. Function Expansion Using Orthonormal Functions

Given a complete orthonormal basis {φk (t)}∞
k=−∞ over the interval t ∈ (a, b), then we can express a function
x(t) on the interval (a, b) as
∞
X
x(t) = ak φk (t) (1)
k=−∞
Show that the coefficients, ak , in the above expression can be determined using the formula
Z b
am = x(t)φ∗m (t)dt
a

(Hint: Multiply both sides of Equations (1) with φ∗m (t) and integrate both sides.)

© 2022 I-Fan Lin. All Rights Reserved. 1

 (a) The commutative property of DT convolution, that is, xn ∗ yn = yn ∗ xn .
(b) The associative property of DT convolution, that is, (xn ∗ yn ) ∗ zn = xn ∗ (yn ∗ zn ).

(c) The distributive property of DT convolution, that is, xn ∗ (yn + zn ) = xn ∗ yn + xn ∗ zn .

(d) Let hn be the impulse response of a DT system. Then the system is causal if and only if hn = 0 for
n < 0.

Exercise 4. Fourier Series Expansion

jkω0 t
As you have shown in Exercise 2, the set e √T is orthonormal over the interval t ∈ (0, T ). Using the result
of Exercise 3 we see that we can expand a given function x(t), which is periodic with period T using this set
as
∞
1 X
x(t) = √ ak ejkω0 t
T k=−∞

This representation is called the orthonormal Fourier series representation of x(t).

Consider the periodic function with period T = 2
∞
X
x(t) = (t − kT )[u(t − kT ) − u(t − kT − 2)]
k=−∞

(e.g. for t ≤ t ≤ 2, x(t) = t). Compute the values of the coefficients, ak , for all integers k.

Exercise 5. Parseval’s Formula

For a signal expressed using equation (1) show that
Z b ∞
X
|x(t)|2 dt = |am |2
a m=−∞

This important result is known as tje Parseval’s formula. Note that the left side is the energy in x(t).

© 2022 I-Fan Lin. All Rights Reserved. 2