From analog to digital
Two steps
- Sampling
- Quantization
Sampling Theorem
Let $s\in L^2(\mathbb{R})$ a band limited analogical signal un signal in the bandwidth $\nu\in[-\nu_0,\nu_0]$. Then $s$ can be reconstructed (ie, interpolated) without error from its samples $s(t_n)$ taken at times $t_n=\frac{n}{2\nu_0}$.
- $\nu_0$ is called the cutting frequency
- $\nu_s = 2\nu_0$ is called the sampling frequency
- if $\nu_s < 2\nu_0$ the reconstruction is not possible + aliasing
- if $\nu_s > 2\nu_0$ the reconstruction is possible
Quantization
Each values $s(t_n)$ must be mapped from a real value (infinite precision) to a decimal value (with finite precision).