Quantization Noise & Error Study Guide

ECE 426 Digital Communications. Department of Electrical & Communication Engineering

Interactive Simulation
SQNR Calculator
01. Fundamentals

What is Quantization?

Quantization is the process of mapping input values from a large set (often continuous) to output values in a smaller countable set. It is the core mechanism of Analog-to-Digital Conversion (ADC).

Key Definitions

  • L Number of quantization levels (e.g., 8 levels = 3 bits).
  • Δ Step size (Uniform Quantization). Δ = (Vmax - Vmin) / L
  • e_q Quantization Error: e_q = x_q(nT_s) - x(nT_s)
VISUALIZATION: STEP FUNCTION
Analog Digital Error
02. Methodology

Uniform Quantization Types

The placement of reconstruction levels determines the type of uniform quantizer and affects the maximum possible error.

Mid-Rise Quantizer

The origin (zero) lies in the middle of a rising part of the staircase. It does not have a zero output level.

Input Output
Error Range -Δ/2 ≤ e ≤ +Δ/2
Zero Level No explicit zero (Dead zone around 0)

Mid-Tread Quantizer

The origin lies in the middle of a tread (the flat part). It has a zero output level.

Input Output
Error Range -Δ/2 ≤ e ≤ +Δ/2
Zero Level Exists (Good for voice signals)

03 Interactive Quantization Lab

Visualize the effect of bit depth on signal quality and calculate theoretical SQNR.

1 bit 3 bits 8 bits

Calculated Parameters

Levels (L) 8
Step Size (Δ) 0.25
SQNR (dB) 19.08 dB
Theory: 6.02n + 1.76 dB

Insight

At 3 bits, the signal is barely recognizable. The error is large because the step size is significant relative to the signal amplitude.

Time Domain Analysis

Original
Quantized
Error

Quantization Error Signal

0.00

Formula Used

SQNR = 10 log₁₀(P_signal / P_noise)
For Full Scale Sine Wave: 1.76 + 6.02n dB

Noise Analysis

Quantization error is typically modeled as a random variable with uniform distribution over the interval [-Δ/2, Δ/2].

Mean (Average Error)

E[e_q] = 0

The error is unbiased for uniform quantizers.

Variance (Noise Power)

σ² = Δ² / 12

This represents the average quantization noise power.

Probability Density Function (PDF)

Uniform Dist
-Δ/2
+Δ/2
1/Δ

The error is equally likely to be any value between -Δ/2 and +Δ/2.