Quantization: ADC Levels, Error Models, and Quantizer Design

All flashcards

What is quantization?

Mapping values from a large input set to a smaller, countable set of output values.

What does quantization error measure?

The difference between an input value and its quantized approximation.

How many levels can an N-bit fixed-width representation distinguish?

Up to $2^N$ levels.

How does dither affect quantization?

It adds noise before quantization to reduce signal-dependent distortion, usually at the cost of a higher noise floor.

What determines a Lloyd–Max reconstruction value?

It is the conditional mean of source values within that quantizer interval.

Quiz questions

1. In a common analog-to-digital conversion model, what does sampling do before quantization?Answer: It converts a time-varying signal into discrete-time, real-valued samples.
2. What distinguishes a mid-riser quantizer at zero?Answer: It has a classification threshold at zero and no zero reconstruction level.
3. What does a dead-zone quantizer do to inputs within its symmetric region around zero?Answer: It maps them to the zero reconstruction value.
4. What is a principal purpose of adding dither before quantization?Answer: To reduce coherent, signal-dependent distortion by making error less input-dependent.
5. In Lloyd–Max quantization, where is a decision threshold placed relative to its neighboring reconstruction values?Answer: Halfway between the two neighboring reconstruction values.