written 5.0 years ago by |
Let $x (n)$ be a discrete sequence.
Part 1: DTFT and $Z T$
By definition of $Z-$transform, $X(z)=Z\{x[n]\}=\sum_{n=-\infty}^{\infty} x[n] \cdot z^{-n}$ and
By definition of Discrete Time Fourier transform, $X(\omega)=D T F T\{x[n]\}=\sum_{n=-\infty}^{\infty} x[n] \cdot e^{-j \omega n}$
It is found that, $X(z)=D T F T\left\{x[n] r^{-n}\right\}$
If $X(z)$ is evaluated on a unit circle i.e. $z=r e^{j \omega}=e^{j \omega}$ then $X(z)=\operatorname{DTFT}\{x[n]\}$
Hence, Fourier Transform of a discrete signal is equal to $Z-$ Transform evaluated on a unit circle.
Part 2: DTFT and DFT
DFT can be seen as the sampled version (in Frequency Domain) of the DTFT output. i.e. $X(k)=\left.X(\omega)\right|_{\omega=2 \pi k / N}$
Part 3: DFT and ZT
From Part I and II, DFT of a discrete signal is equal to $Z-$Transform evaluated on a unit circle calculated at discrete instant of Frequency.
Part 4: DFT and DTFS
Let sequence $x_{p}(n)$ is a periodic repetition of sequence $x(n)$ . Let $\mathrm{N}$ be period of $x_{p}(n)$
By definition, Fourier Series coefficients $C_{k}=D T F S\left\{x_{p}(n)\right\}=\frac{1}{N} \sum_{N} x_{p}(n) \cdot e^{-j 2 \pi k n / N},$ where $k=0,1,2 \ldots \mathrm{N}-1$
By definition, $X(k)=D F T[x(n)]=\sum_{N} x(n) \cdot e^{-j 2 \pi k n/N}$
$\therefore C_{k}=\frac{1}{N} X(k)$ or $N C_{k}=X(k)$
Thus, the N-point DFT provides the exact line spectrum of a periodic scquence with fundamental period $N .$