Discrete Fourier Transform Continuous: Discrete:
Bandwidth Limited Transforms 1 Inverse Let For
Bandwidth Limited Transforms 2 Discrete t: Coefficients cn of the exponential form of the FS for F(2x) for
Bandwidth Limited Transforms 3 This set of points in the time domain determines F() completely F() determines f(t) for all time Must have W1W so that all the bandwidth for nonzero F() is used (otherwise lose information)
Sampling Frequency Sampling frequency: Since W1W (Nyquist frequency) Can sample at a higher frequency but inefficient If we sample at less than Nyquist rate? f(t) not completely determined Can get ‘aliassing’
Aliassing Sampling at Nyquist frequency gives correct signal • Less than Nyquist rate eg - Gives incorrect wavelength (aliassing)
Aliassing Example • • • • • • • • • • • • • • • Example
Inverse Discrete FT Numerical rule: Inverse Discrete FT
Forward DFT Consider (for some integer m) But what is m? DFT
DFT - Summary Inverse Discrete Fourier Transform
Example1 : Find DFT of fn=n, with N=3
Example 2: F is the 12-point DFT of a real signal f of length 12.