How can I prove that power of a periodic signal is equal to sum of all magnitude squared fourier series coefficients without using Parseval's theorem?

99 Views Asked by At

If $x(t)$ is periodic with period $T$ and it satisfies dirichlet conditions we can represent it as:

$x(t)=\sum_{n=-\infty}^{\infty}c_n e^{J\frac{2\pi nt}{T}}$

Power of $x$ is defined as: $P=\frac{1}{T}\int_{-T/2}^{T/2}{|x(t)|^2}dt$

So how does the equation $P$ = $\sum_{n=-\infty}^{\infty}|c_n|^2$ hold true?

1

There are 1 best solutions below

0
On

recall $x(t)=\sum_{n=-\infty}^{\infty}c_n e^{J\frac{2\pi nt}{T}}$

before pluging $x(t)$ on to the definition of power: $P=\frac{1}{T}\int_{-T/2}^{T/2}{|x(t)|^2 dt}$

represent ${|x(t)|}^2$ as $x(t){x(t)}^* $

$x(t)=\sum_{n=-\infty}^{\infty}c_n e^{J\frac{2\pi nt}{T}}$ ==>

${x(t)}^*=\sum_{n=-\infty}^{\infty}c_n^* e^{-J\frac{2\pi nt}{T}}$

change variable $n$ with $r$ on summation of ${x(t)}^*$:

${x(t)}^*=\sum_{r=-\infty}^{\infty}c_r^* e^{-J\frac{2\pi rt}{T}}$

$ {|x(t)|}^2 = x(t){x(t)}^∗ =$

$ [\sum_{n=-\infty}^{\infty}c_n e^{J\frac{2\pi nt}{T}}]\times [\sum_{r=-\infty}^{\infty}c_r^* e^{-J\frac{2\pi rt}{T}}]$

according to distributive law of multiplication we can write:

$ {|x(t)|}^2 = \sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty}[c_n e^{J\frac{2\pi nt}{T}}c_r^* e^{-J\frac{2\pi rt}{T}}]$

=>

$ {|x(t)|}^2 = \sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty}[c_n c_r^* e^{J\frac{2\pi (n-r)t}{T}} ]$

instead of plugging $x(t)$ on the definition of power, plug $ |x(t)|^2$:

$P=\frac{1}{T}\int_{-T/2}^{T/2}{|x(t)|^2 dt} =$

$ \frac{1}{T}\int_{-T/2}^{T/2}{[\sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty}[c_n c_r^* e^{J\frac{2\pi (n-r)t}{T}} ]] dt}$

Take the integral inside as $c_n$ and $c_r$ are independent of $t$:

$P=\frac{1}{T}{\sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty}[\int_{-T/2}^{T/2}c_n c_r^* e^{J\frac{2\pi (n-r)t}{T}} dt}]$

call the Integral as $I$:

$I=\int_{-T/2}^{T/2}c_n c_r^* e^{J\frac{2\pi (n-r)t}{T}} dt$

$I$ is a piecewise function of integers n and r:

$I(n,r) = \left\{ \begin{array}{lr} T |c_n|^2, & \text{if } n = r\\ \frac{Tc_n {c_r^*}\sin(\pi(n-r))}{\pi(n-r)}, & \text{if } n\neq r \end{array} \right\}$

As $n$ and $r$ are both integers $(n-r)$ is also an integer.

So, sine of integer multiple of $pi$ radians gives always zero no matter whether it's zero, positive or negative. ==>

$I(n,r) = \left\{ \begin{array}{lr} T |c_n|^2, & \text{if } n = r\\ \frac{Tc_n {c_r^*}\times 0}{\pi(n-r)}, & \text{if } n\neq r \end{array} \right\}$

Finally:

$I(n,r) = \left\{ \begin{array}{lr} {T |c_n|^2}, & \text{if } n = r\\ {0}, & \text{if } n\neq r \end{array} \right\}$

recall $P=\frac{1}{T}{\sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty}[\int_{-T/2}^{T/2}c_n c_r^* e^{J\frac{2\pi (n-r)t}{T}} dt}]$

replace the integral with $I(n,r)$ :

$P=\frac{1}{T} {\sum_{n=-\infty}^{\infty}\sum_{r=-\infty}^{\infty} I(n,r) }$

As $I$ is zero, for $r\neq n$:

$P=\frac{1}{T} {\sum_{n=-\infty}^{\infty}\sum_{r=n}^{n} I(n,r) }$

can be simplified into:

$P=\frac{1}{T} {\sum_{n=-\infty}^{\infty}I(n,n) }$

$P=\frac{1}{T} {\sum_{n=-\infty}^{\infty}T |c_n|^2}$

$P={\sum_{n=-\infty}^{\infty}|c_n|^2}$

done!