Fourier Transform

Fourier Series

\begin{align} \text{Fourier series, }x(t)&=\int_{-\infty}^{\infty}c_{k}e^{j2\pi kF_{0}t}\\ \text{where }c_{k} &= \frac{1}{T_{p}}\int_{T_{p}}x(t)e^{-j2\pi kF_{0}t}\ dt \quad (1) \end{align}

Warning

Equation (1) is only true when x(t) is periodic and satisfies the Direchlet conditions.

Direchlet conditions

  1. x(t) has a finite number of discontinuities in any period
  2. x(t) has a finite number of maxima and minima in any period
  3. x(t) is absolutely integrable in any period, i.e.
\int_{T_{p}}|x(t)| \ dt < \infty\\

Note

All periodic signals of practical interest satisfy these conditions.

Further Simplification of Fourier Series

If x(t) is real, c_{k} and c_{-k} are complex conjugates. $$ c_{k} = |c_{k}|e^{j\theta_{k}}, c_{-k} = |c_{k}|e^{-j\theta_{k}} $$

\therefore \text{Fourier series, }x(t) = c_{0}+2\sum_{k=1}^{\infty}|c_{k}|\cos(2\pi kF_{0}t + \theta_{k})

We know that \cos(2\pi kF_{0}t + \theta_{k})=\cos(2\pi kF_{0}t)\cos\theta_{k} - \sin (2\pi kF_{0}t)\sin\theta_{k},

\begin{align} \therefore \text{Fourier series, }x(t) &= a_{0}+\sum^{\infty}_{k=1}(a_{k}\cos 2\pi kF_{0}t - b_{k}\cos 2\pi kF_{0}t)\\ \text{where }a_{0} &= c_{0}, \ a_{k} = 2|c_{k}|\cos\theta_{k}, \ b_{k}=2|c_{k}|\sin\theta_{k} \end{align}

Info

This is commonly expressed in many beginner texts on Fourier transform as:

\begin{align} \text{Fourier series, }f(t) = a_{0} + &a_{1}\cos(t)+a_{2}\cos(2t)+a_{3}\cos(3t)+...+a_{n}\cos(nt)\\ +&b_{1}\cos(t)+b_{2}\cos(2t)+b_{3}\cos(3t)+...+b_{n}\cos(nt)\\ \end{align}
\text{where }a_{0} = \frac{1}{2\pi}\int_{0}^{2\pi}f(t) \ dt,\ a_{t} = \frac{1}{\pi}\int_{0}^{2\pi}f(t)\cos(nt) \ dt,\ b_{t} = \frac{1}{\pi}\int_{0}^{2\pi}f(t)\sin(nt) \ dt

Equivalent Forms of Fourier Series

Fourier Series, x(t)

  1. x(t)=\int_{-\infty}^{\infty}c_{k}e^{j2\pi kF_{0}t}
  2. x(t) = c_{0}+2\sum_{k=1}^{\infty}|c_{k}|\cos(2\pi kF_{0}t + \theta_{k})
  3. x(t) = a_{0}+\sum^{\infty}_{k=1}(a_{k}\cos 2\pi kF_{0}t - b_{k}\cos 2\pi kF_{0}t),
    where a_{0}= c_{0}, \ a_{k} = 2|c_{k}|\cos\theta_{k}, \ b_{k}=2|c_{k}|\sin\theta_{k}

Fourier coefficient

  • c_{k} = \frac{1}{T_{p}}\int_{T_{p}}x(t)e^{-j2\pi kF_{0}t}\ dt
  • when x(t) is periodic and satisfies the Direchlet conditions

Useful Trigonometrical Identities

\begin{align*} \text{Sine: }\int_{0}^{2\pi} \sin(mt) \ dt &= \ 0 \text{ for any integer }m \\ \text{Cosine: }\int_{0}^{2\pi} \cos(mt) \ dt &= \ 0 \text{ for any non-zero integer }m \\ \\ \text{Product of sine and cosine: }\int_{0}^{2\pi} \sin(mt) \cos(nt) \ dt &= \ 0 \text{ for any integers }m, n \\ \\ \text{Product of sines: }\int_{0}^{2\pi} \sin(mt) \sin(nt) \ dt &= \ 0 \text{ when }m \neq n, m \neq -n \\ \int_{0}^{2\pi} \sin^{2}(mt) \ dt &= \pi \text{ for any non-zero integer }m \\ \\ \text{Product of cosines: }\int_{0}^{2\pi} \cos(mt) \cos(nt) \ dt &= \ 0 \text{ when }m \neq n, m \neq -n \\ \int_{0}^{2\pi} \cos^{2}(mt) \ dt &= \pi \text{ for any non-zero integer }m \\ \end{align*}

Fourier Transform Theorems and Properties

Linearity

a_{1}x_{1}(n)+a_{2}x_{2}(n)\xrightarrow{\mathscr{F}}a_{1}X_{1}(\omega)+a_{2}X_{2}(\omega).

Time shifting

x(n-k)\xrightarrow{\mathscr{F}}e^{-j\omega k}X(\omega).

  • Shifting signal in time domain by k changes phase by -\omega k.

Time reversal

x(-n)\xrightarrow{\mathscr{F}}X(-\omega)

  • When signal is folded about origin in time, phase spectrum undergoes phase reversal

Convolution

x_{1}(n) \ x_{2}(n) \xrightarrow{\mathscr{F}} X_{1}(\omega)X_{2}(\omega)

Wiener-Khintchine theorem

r_{xx}(l) \xrightarrow{\mathscr{F}} S_{xx}(\omega)

  • Energy spectral density of an energy signal S is the Fourier transform of its autocorrelation sequence R

Frequency shifting

e^{j\omega_{0}n}x(n) \xrightarrow{\mathscr{F}} X(\omega-\omega_{0})

  • Multiplying e^{j\omega_{0}n} to x(n) is equivalent to translating X(\omega) by \omega_{0}

Modulation

x(n)\cos \omega_{0}n \xrightarrow{\mathscr{F}} \frac{1}{2}X(\omega+\omega_{0})+\frac{1}{2}X(\omega-\omega_{0})

Multiplication (Windowing theorem)

x_{1}(n) \ x_{2}(n) \xrightarrow{\mathscr{F}} \frac{1}{2\pi}\int_{-\pi}^{\pi}X_{1}(\lambda) \ X_{2}(\omega-\lambda) \ d\lambda

Differentiation in the frequency domain

nx(n) \xrightarrow{\mathscr{F}} j\frac{dX(\omega)}{d\omega}

Conjugation

x^{*}(n)\xrightarrow{\mathscr{F}}X^{*}(-\omega)

Parseval's theorem

\int_{-\infty}^{\infty}x_{1}(n) \ x_{2}^{*}(n) = \frac{1}{2\pi}\int_{-\pi}^{\pi}X_{1}(\omega) \ X_{2}^{*}(\omega) \ d\omega