lecture 7: signal processing iv

27
Lecture 7: Signal Processing IV EEN 112: Introduction to Electrical and Computer Engineering Professor Eric Rozier, 2/27/13

Upload: keira

Post on 24-Feb-2016

44 views

Category:

Documents


4 download

DESCRIPTION

Lecture 7: Signal Processing IV. EEN 112: Introduction to Electrical and Computer Engineering. Professor Eric Rozier, 2/ 27/ 13. SCHEDULE. Schedule. QUANTIZATION. Recall the types of functions. Surjective. Injective. Classification and Reconstruction. 0 0.1 0.15762 0.2 0.333333 - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Lecture  7:  Signal Processing  IV

Lecture 7: Signal Processing IV

EEN 112: Introduction to Electrical and Computer Engineering

Professor Eric Rozier, 2/27/13

Page 2: Lecture  7:  Signal Processing  IV

SCHEDULE

Page 3: Lecture  7:  Signal Processing  IV

Schedule

Date M W

2/25 Signals Signals (Quiz)

3/4 Wrap up signals NASA Guest Speaker

3/11 Spring Break Spring Break

3/18 Computer Engineering Computer Engineering

3/25 Computer Engineering Midterm II

4/1 EE/Circuits EE/Circuits

4/8 EE/Circuits EE/Circuits

4/15 EE/Circuits EE/Circuits

4/22 Course Synthesis Course Synthesis

Page 4: Lecture  7:  Signal Processing  IV

QUANTIZATION

Page 5: Lecture  7:  Signal Processing  IV

Recall the types of functions

Surjective Injective

Page 6: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 7: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 8: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 9: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 10: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 11: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 12: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 13: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 14: Lecture  7:  Signal Processing  IV

Classification and Reconstruction

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

00 (0)01 (1)10 (2)11 (3)

00.1

0.157620.2

0.3333330.447

0.6666660.91.0

Page 15: Lecture  7:  Signal Processing  IV

Quantization Error

• Sampling is error free when we follow the Nyquist

• Quantization always has some error.

Page 16: Lecture  7:  Signal Processing  IV

Quantization Error

• Let’s look at the error of quantizing the numbers 1-100 using various numbers of bits…

Page 17: Lecture  7:  Signal Processing  IV

2-bit Quantization

Bit vector k x

00 0 0, 1, 5, 10, 29 1

01 1 35, 50, 61 34

10 2 75, 99 67

11 3 100 100

Page 18: Lecture  7:  Signal Processing  IV

3-bit QuantizationBit vector k x

000 0 0, 1, 5, 10 1

001 1 29 15.143

010 2 35 29.285

011 3 50 43.429

100 4 61 57.571

101 5 75 71.714

110 6 99 85.857

111 7 100 100

99/7 = 14.1429…

Page 19: Lecture  7:  Signal Processing  IV

4-bit QuantizationBit vector k x

0000 0 0, 1, 5 1

0001 1 10 7.6

… … … …

99/15= 6.6

Page 20: Lecture  7:  Signal Processing  IV

5-bit QuantizationBit vector k x

00000 0 0, 1, 1

00001 1 5 4.194

… … … …

99/31 = 3.194…

Page 21: Lecture  7:  Signal Processing  IV

6-bit QuantizationBit vector k x

000000 0 0, 1 1

000001 1 2.571

… … … …

99/63 = 1.571…

Page 22: Lecture  7:  Signal Processing  IV

Quantization Error

• The error introduced when reconstructing a signal

• Given an N-bit quantization over a range, [a,b], what is the maximum error? Hint, think in terms of

Page 23: Lecture  7:  Signal Processing  IV

Quantization Error over [1,100]

Number of bits Maximum quantization error

2 <33

3 <14.14

4 <6.6

5 <1.597

6 <1.572

7 0.78

8 0.388

16 0.00152

32 0.0000000230

64 0.00000000000000000536

Page 24: Lecture  7:  Signal Processing  IV

Linear vs. Non-linear Quantization

• So far we’ve dealt with linear quantization

• There are other ways we might quantize data

Page 25: Lecture  7:  Signal Processing  IV

Non-linear Quantization

Page 26: Lecture  7:  Signal Processing  IV

Non-linear Quantization

Page 27: Lecture  7:  Signal Processing  IV

Non-linear Quantization

• How should we change our classifier and our reconstruction rule?

• Hint: