EE6640 Wmu
EE6640 Wmu
EE6640 Wmu
Digital Communications
• Syllabus
• Personal Intro.
• Textbook/Materials Used
• Additional Reading
• ID and Acknowledgment of Policies
• Textbook
• Chapter 1
ECE 6640 2
Syllabus
• Everything useful for this class can be found on Dr. Bazuin’s web site!
– http://homepages.wmich.edu/~bazuinb/
• The syllabus …
– http://homepages.wmich.edu/~bazuinb/ECE6640/Syl_6640.pdf
ECE 6640 3
Who am I?
ECE 6640 5
Required Textbook/Materials
ECE 6640 6
Supplemental Books and Materials
• John G. Proakis and Masoud Salehi, “Digital Communications, 5th
ed.,” McGraw Hill, Fifth Edition, 2008. ISBN: 978-0-07-295716-7.
• John G. Proakis and Masoud Salehi, “Communication Systems
Engineering, 2nd ed.”, Prentice Hall, 2002. ISBN: 0-13-061793-8.
• A. Bruce Carlson, P.B. Crilly, “Communication Systems, 5th ed.”,
McGraw-Hill, 2010. ISBN: 978-0-07-338040-7.
• Leon W. Couch II, “Digital and Analog Communication Systems, 7th
ed.”, Prentice Hall, 2007. ISBN: 0-13-142492-0.
• Stephen G. Wilson, “Digital Modulation and Coding, ” Prentice-Hall,
1996. ISBN: 0-13-210071-1.
• Ezio Biglieri, D. Divsalar, P.J. McLane, M.K. Simon, “Introduction
to Trellis-Coded Modulation with Applications”, Macmillan, 1991.
ISBN: 0-02-309965-8.
ECE 6640 7
Identification and Acknowledgement
ECE 6640 8
Course/Text Overview
1. Signals and Spectra.
ECE 6640 9
Course/Text Overview (2)
3. Baseband Demodulation/Detection.
Exam #1
ECE 6640 10
Course/Text Overview (3)
5. Communications Link Analysis.
What the System Link Budget Tells the System Engineer. The
Channel. Received Signal Power and Noise Power. Link Budget
Analysis. Noise Figure, Noise Temperature, and System
Temperature. Sample Link Analysis. Satellite Repeaters. System
Trade-Offs.
ECE 6640 11
Course/Text Overview (4)
6. Channel Coding: Part 1.
Exam #2
ECE 6640 12
Course/Text Overview (5)
8. Channel Coding: Part 3.
Final Exam
ECE 6640 13
Course/Text Overview (6)
Advanced Topics (as time permits)
Final Exam
ECE 6640 14
Text Appendices
A. A Review of Fourier Techniques.
D. Often-Used Identities.
F. List of Symbols.
ECE 6640 16
Chapter 1
1. Signals and Spectra.
1.1 Digital Communication Signal Processing.
1.2 Classification of Signals.
1.3 Spectral Density.
1.4 Autocorrelation.
1.5 Random Signals.
1.6 Signal Transmission through Linear Systems.
1.7 Bandwidth of Digital Data.
ECE 6640 17
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 18
Prentice Hall PTR, Second Edition, 2001.
Simplified Communications System
• Format: making the message compatible with digital processing
• Source Coding: efficient descriptions of information sources
• Channel Coding: signal transformation enabling improved reception
performance after expected channel impairments
• Modulation: formation of the baseband waveform
• RF Mixing: frequency domain translation of baseband signal
• Transmit/Receive: RF Amplifiers and Filters
RF Signal
Noise
Noise
Interference
Transmitting Receiving
Antenna Antenna
RF Communication Channel
ECE 6640 21
Why Digital?
1. Noise, Interference, Path Loss, and Channel Impairments
(signal environment)
2. Cost
3. Inherent Availability
4. Reliability and Reconfigurability
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 22
Prentice Hall PTR, Second Edition, 2001.
Terminology
• Information Source
• Textual Message
• Character
• Binary Digit (Bit)
• Bit Stream
• Symbol
• Digital Waveform
• Data Rate
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 23
Prentice Hall PTR, Second Edition, 2001.
Signal Processing Functions
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 24
Prentice Hall PTR, Second Edition, 2001.
Classification of Signals
ECE 6640 25
SKLAR DSP Tutorial
X f X f X f
*
x t dt
1
PX
2
T0 T0
2
1 *
G X f lim X T f X T f
T T
ECE 6640 27
Autocorrelation
• of an Energy Signal
R XX xt x t dt
• Properties:
1. Energy
R XX 0 E X 2 X 2
2. Symmetry R XX R XX
3. Maximum R XX R XX 0
ECE 6640 28
Autocorrelation
• of a Power Signal
T
1 2
XX lim x t x t dt
T T
T
2
• Properties: T0
2
XX 0 x t
1
dt
2
1. Energy T0
T0
2
ECE 6640 29
Random Signals
1 Distribution Functions
Probability Distribution Function (PDF) or
Cumulative Distribution Function (CDF) [preferred]
0 FX x 1, for x
FX 0 and FX 1
FX is non-decreasing as x increases
Pr x1 X x 2 FX x 2 FX x1
For discrete events
For continuous events
ECE 6640 30
Random Signals
2. Density Functions
Probability Density Function (pdf) Probability Mass Function (pmf)
f X x 0, for x f X x 0, for x
f x dx 1
X f x dx 1
X
x x
FX
f X u du FX
f X u du
x2
Pr x1 X x 2 f x dx
x2
Pr x1 X x 2 f x dx
X
X
x1 x1
ECE 6640 31
Random Signals
Mean Values and Moments
1st, general, nth Moments
X EX
x f X x dx or X E X x Pr X x
x
E g X
gX f
X x dx or E g X g X Pr X x
x
E X x E X x
f X x dx or X Pr X x
n n n n n n
X
x
Central Moments
x X
X X n
E XX
n n
f X x dx
x X
X X n
E XX
n n
Pr X x
x
Variance and Standard Deviation
x X
X X
2
2
E X X
2 2
f X x dx
x X
2 X X
2
E XX
2 2
Pr X x
x
ECE 6640 32
Random Signals
The Gaussian Random Variable
f X x
1
exp
x X 2
, for x
2 2
2
where X is the mean and is the variance
x
v X 2
FX x
exp dv
1
2 2 2
v
Unit Normal
x
u2
x
1
exp du
2 2
u
x 1 x
x X x X
FX x or FX x 1
The Q-function is the complement of the normal function, :
(Appendix B)
u2
Q x
1
exp du
2 2
ECE 6640
ux 33
Random Processes
5. Random Processes
5.1. Introduction
Ensemble
From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and
ECE 6640 System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9 34
Random Processes: Continuous,
Discrete and Mixed
Continuous and Discrete Random Processes
A continuous random process is one in which the random variables, such as X t1 , X t 2 , X t n ,
can assume any value within the specified range of possible values. A more precise definition for a
continuous random process also requires that the cumulative distribution function be continuous.
A discrete random process is one in which the random variables, such as X t1 , X t 2 , X t n ,
can assume any certain values (though possibly an infinite number of values). A more precise
definition for a discrete random process also requires that the cumulative distribution function
consist of numerous discontinuities or steps. Alternately, the probability density function is better
defined as a probability mass function … the pdf is composed of delta functions.
A mixed random process consists of both continuous and discrete components. The probability
distribution function consists of both continuous regions and steps. The pdf has both continuous
regions and delta functions.
ECE 6640 From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and 35
System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9
Random Processes: Deterministic and
Nondeterministic
Deterministic and Nondeterministic Random Processes
A nondeterministic random process is one where future values of the ensemble cannot be predicted
from previously observed values.
A deterministic random process is one where one or more observed samples allow all future values
of the sample function to be predicted (or pre-determined). For these processes, a single random
variable may exist for the entire ensemble. Once it is determined (one or more measurements) the
sample function is known for all t.
ECE 6640 From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and 36
System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9
Random Processes: Stationary and
Nonstationary (1)
Stationary and Nonstationary Random Processes
The probability density function for random variables in time as been discussed, but what is the
dependence of the density function on the value of time, t, when it is taken?
If all marginal and joint density functions of a process do not depend upon the choice of the time
origin, the process is said to be stationary (that is it doesn’t change with time). All the mean values
and moments are constants and not functions of time!
For nonstationary processes, the probability density functions change based on the time origin or in
time. For these processes, the mean values and moments are functions of time.
ECE 6640 37
From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and
System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9
Random Processes: Stationary and
Nonstationary (2)
Stationary and Nonstationary Random Processes
The requirement that all marginal and joint density functions be independent of the choice of time
origin is frequently more stringent (tighter) than is necessary for system analysis.
A more relaxed requirement is called stationary in the wide sense: where the mean value of any
random variable is independent of the choice of time, t, and that the correlation of two random
variables depends only upon the time difference between them.
That is
E X t X X and
E X t1 X t 2 E X 0 X t 2 t1 X 0 X R XX for t 2 t1
ECE 6640 38
From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and
System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9
Random Processes: Ergodicity
Ergodic and Nonergodic Random Processes
Ergodicity deals with the problem of determining the statistics of an ensemble based on
measurements from a sample function of the ensemble.
For ergodic processes, all the statistics can be determined from a single function of the process.
This may also be stated based on the time averages. For an ergodic process, the time averages
(expected values) equal the ensemble averages (expected values).
That is to say,
T
x n f x dx lim
X t dt
1
Xn n
T 2T
T
From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and
ECE 6640 System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9 39
Random Processes
The power spectral density is the Fourier Transform of the autocorrelation:
S XX w R XX
EX t X t exp iw d
For an ergodic process,
T
XX lim
xt xt dt xt xt
1
T 2T
T
T
XX E X t X t
xt xt dt exp iw d
1
lim
T 2T
T
T
XX lim
xt exp iwt xt exp iwt d dt
1
T 2T
T
T
XX lim
xt exp iwt X w dt
1
T 2T
T
T
XX X w lim
xt exp i wt dt
1
T 2T
T
XX X w X w X w
2
ECE 6640 40
From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and System Analysis, 3rd ed.,Oxford
University Press Inc., 1999. ISBN: 0-19-512354-9
Binary Sequence, Low Bit Rate
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 41
Prentice Hall PTR, Second Edition, 2001.
Binary Autocorrelation and PSD
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 42
Prentice Hall PTR, Second Edition, 2001.
Bandwidth Consideration
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 44
Prentice Hall PTR, Second Edition, 2001.
White Noise
Noise is inherently defined as a random process.
You may be familiar with “thermal” noise, based on the energy of an atom and the mean-free path
that it can travel.
As a random process, whenever “white noise” is measured, the values are uncorrelated with each
other, not matter how close together the samples are taken in time.
Further, we envision “white noise” as containing all spectral content, with no explicit peaks or
valleys in the power spectral density.
S XX w S 0
This is an approximation or simplification because the area of the power spectral density is infinite!
ECE 6640 From: George R. Cooper and Clare D. McGillem, Probabilistic Methods of Signal and 45
System Analysis, 3rd ed.,Oxford University Press Inc., 1999. ISBN: 0-19-512354-9
Band Limited White Noise
Thermal noise at the input of a receiver is defined in terms of kT, Boltzmann’s constant times
absolute temperature, in terms of Watts/Hz. Thus there is kT Watts of noise power in every Hz of
bandwidth.
W
E X 2 R XX 0
S 0 dw 2 W S 0
W
How much noise power, in dBm, would I say that there is in a 1 MHz bandwidth?
dBkTB dBkT dBB 174 60 114 dBm
ECE 6640 46
White Noise in Comm.
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 47
Prentice Hall PTR, Second Edition, 2001.
Noise as A Gaussian Random Process
A Gaussian Random Variable
f X x
1
exp
x X 2
, for x
2 2 2
where X is the mean and is the variance
x
v X 2
FX x
exp dv
1
2 2 2
v
Convolution Integrals
y t
xt h d
0
or
t
y t
ht x d
ImHf
f tan 1
ReHf
x t A cos2 f 0 t
yt h t x t
ECE 6640 50
Filtering a Random Process
RYY d1 d2 h1 h2 R XX 1 2
0 0
SYY w RYY d1 d2 h1 h2
R XX 1 2 exp iw d
0 0
ECE 6640 52
Ideal Filter (1)
exp j2 f t t 0
fu
h t
j2 t t 0 f u
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 55
Prentice Hall PTR, Second Edition, 2001.
Realizable Filters, RC Network
1st order
Butterworth
Filter
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 56
Prentice Hall PTR, Second Edition, 2001.
White Noise in an RC Filter
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 58
Prentice Hall PTR, Second Edition, 2001.
Signal Filtering in the Real World (2)
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 59
Prentice Hall PTR, Second Edition, 2001.
Bandwidth Considerations, Easy
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 60
Prentice Hall PTR, Second Edition, 2001.
Bandwidth Considerations, Harder
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 63
Prentice Hall PTR, Second Edition, 2001.
Bandwidth Definitions
(a) Half-power bandwidth. This is the interval between frequencies at
which Gx(f ) has dropped to half-power, or 3 dB below the peak value.
(b) Equivalent rectangular or noise equivalent bandwidth. The noise
equivalent bandwidth was originally conceived to permit rapid
computation of output noise power from an amplifier with a wideband
noise input; the concept can similarly be applied to a signal bandwidth.
The noise equivalent bandwidth WN of a signal is defined by the
relationship WN = Px/Gx(fc), where Px is the total signal power over
all frequencies and Gx(fc) is the value of Gx(f ) at the band center
(assumed to be the maximum value over all frequencies).
(c) Null-to-null bandwidth. The most popular measure of bandwidth for
digital communications is the width of the main spectral lobe, where
most of the signal power is contained. This criterion lacks complete
generality since some modulation formats lack well-defined lobes.
ECE 6640 64
Bandwidth Definitions (2)
(d) Fractional power containment bandwidth. This bandwidth criterion
has been adopted by the Federal Communications Commission (FCC
Rules and Regulations Section 2.202) and states that the occupied
bandwidth is the band that leaves exactly 0.5% of the signal power
above the upper band limit and exactly 0.5% of the signal power below
the lower band limit. Thus 99% of the signal power is inside the
occupied band.
(e) Bounded power spectral density. A popular method of specifying
bandwidth is to state that everywhere outside the specified band, Gx(f )
must have fallen at least to a certain stated level below that found at
the band center. Typical attenuation levels might be 35 or 50 dB.
(f) Absolute bandwidth. This is the interval between frequencies, outside
of which the spectrum is zero. This is a useful abstraction. However,
for all realizable waveforms, the absolute bandwidth is infinite.
ECE 6640 65
Spectrum and Time Domain of a
Band-limited Bandpass Signal
ECE 6640 Notes and figures are based on or taken from materials in the course textbook: 66
Bernard Sklar, Digital Communications, Fundamentals and Applications,
Prentice Hall PTR, Second Edition, 2001.
Summary
• Communication must consider a number of aspects
– Time and Frequency Domain Signals
– Discrete and Continuous Time Signal Constructs
– Deterministic and Random Signal Properties
– Models of Signal Propagation
• Simple time and amplitude changes
• Complex channel impairments
– Models of Other Signals in the Environment
• Noise (white, Gaussian, or more complex)
• Interference
• Multipath
• To successfully model and analyze modern communication
systems, there is a lot of prerequisite knowledge required.
ECE 6640 67
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Signal Processing Functions
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 4
Prentice Hall PTR, Second Edition, 2001.
Formatting
ECE 6640 5
Baseband Signaling
ECE 6640 6
Formatting and Transmission
Digital info.
Textual Format
source info.
Pulse
Analog Transmit
Sample Quantize Encode modulate
info.
Pulse
Bit stream waveforms Channel
Format
Analog
info. Low-pass
Decode Demodulate/
filter Receive
Textual Detect
sink
info.
Digital info.
ECE 6640 7
Textual Data
• Code may be sent serially with start, parity and stop bits
• Code may be structures as words/symbols
ECE 6640 8
Data Format for Asynchronous
Data Communication
• Data is transmitted character by character bit-serially.
• A character consists of
– one start bit (0-level)
– 7 to 8 data bits (often, an ASCII character plus a parity bit)
– an optional parity bit
– one, or one and a half, or two stop bits (1-level)
– least significant bit is transmitted first
– most significant bit is transmitted last
• We may not discuss or simulate all the “real world” effects even in this
class.
ECE 6640 12
Analog to Digital Conversion
• Sampling 1 ws
fs
– Sampling Theorem T 2
• Nyquist rate fs>=2*fmax
1
– Sample Rate T
– Sample Period
fs
• Impulse Sampling Function
x t t n T s
k
x s t x t x t x t t n T
s
n
x s t x n T t n T
s s
ECE 6640 n 13
Fourier Domain – Replicated Spectra
1
X f f n f s
Ts n
X s f Xf X f
X s f X X f d
1
X s f n f s Xf d
T
s k
1
X s f n f s Xf d
Ts k
1
ECE 6640 X s f Xf n f s 14
Ts k
Fourier Domain
ECE 6640 15
Sampling Pulses and Filters
1
X s f ck X f n f s
ECE 6640 Ts k 16
Sample and Hold, Zero Order Hold
ECE 6640 17
ZOF Spectral Domain
1
X sp f Pf X f n f s
Ts k
1
X sp f Ts sincf Ts X f n f s
Ts k
X sp f sincf Ts Xf n f
s
k
ECE 6640 19
Filter Terminology
• Passband
– Frequencies where signal is meant to
pass
• Stopband
– Frequencies where some defined
level of attenuation is desired
BWPB
• Transition-band
BWSB
– The transitions frequencies between
the passband and the stopband
• Filter Shape Factor
– The ratio of the stopband bandwidth
to the passband bandwidth
BWSB
SF See FilterNotes and FIR_Filter_DSPNotes
BWPB
or
ECE 6640 MRSP Chap 4 Nyquist/Raised Cosine Filter 20
Reducing the Sample Rate
Dynamic
nearer to the passband edge.
Range
ECE 6640 21
Oversampling
• Without Oversampling
– High performance LPF
– Nyquist Sampling still includes some aliased components
• With Oversampling
– Lower performance LPF
– Aliased components can be significantly reduced
– High performance digital filter may be employed
– Identical or similar data rate can be achieved
ECE 6640 22
2.5 Sources of Corruption
• Quantization Noise
– Saturation
– Timing Jitter
ECE 6640 23
Quantization Noise
• Round-off Error
– +/- one half of the LSB q
2
– Uniform error distribution about the pe de
2 2
e
quantized value q
2
– Error mean 0 q
1 2 2
– Error Variance q^2/12 e de
q q
2
q
• Truncation Error 1 e 3 2
1 q3 q3
q 3 q 24 24
– 0 to +1 LSB q
2
qL
rmsVp
values
ECE 6640 25
Quantized Peak SNR
• Web
– InterSymbol Interference (ISI)
• http://en.wikipedia.org/wiki/Intersymbol_interference
– Nyquist ISI Criterion
• http://en.wikipedia.org/wiki/Nyquist_ISI_criterion
– Inter Symbol Interference (ISI) and Raised cosine filtering
• http://complextoreal.com/wp-content/uploads/2013/01/isi.pdf
• From C. Langton “Complex to Real” web site
ECE 6640 30
Nonuniform Quantization
ECE 6640 31
Baseband Signaling
ECE 6640 32
PCM Transmission
ECE 6640 34
PCM Types
AMI-Bipolar Encoding
Biphase Mark Code
(Alternate Mark Inversion)
ECE 6640 35
PCM Type Selection
• Spectral characteristics
(power spectral density and bandwidth efficiency)
• Bit synchronization capability
• Error detection capability
• Interference and noise immunity
• Implementation cost and complexity
ECE 6640 36
Spectral Attributes of PCM
ECE 6640 37
M-ary Pulse-Modulation Waveforms
ECE 6640 38
Correlative Codes
ECE 6640 39
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Chapter Goals
ECE 6640 4
Demodulation and Detection
6
Receiver Block Diagram
• Chapter 9: Noise
• Chapter 10: Noise in analog modulated signals
• Chapter 11: Baseband Digital Transmission
ECE 6640 8
Noise Approximation
• Uniform Noise Spectral Density
– Resistor description (Thevenin Model)
G vv f 2 R T
• Available Power from the “noise source”
– Source output power into a matched load
v sout
R
vs Psout
v
sout
2 2
v 1 v
s s
2
2R R 2 R 4R
G vv f 2 R T T N 0
G ss f
4R 4R 2 2
R ss
N0
2 9
System Noise
• Since the noise power spectrum is uniform, a systems
noise power is the product of the noise power and the
integral of the filter power.
R NN 0 Hf df N 0 Hf df
N0 2 2
2 0
10
Noise Equivalent Bandwidth
• If we want the total noise power after the filter, we can
integrate the PSD for all frequencies or use the Filtered
noise autocorrelation function at zero.
– Both of these approaches may be difficult
– Could we great a more simple “noise equivalent bandwidth for
filters” that is rectangular?
11
Noise Equivalent Bandwidth
R NN 0 0 Hf df 0 2 Hf df
N 2 N 2
2 2 0
H rect _ mod el f GainDC _ Power f
rect Gain DC _ Power H0
2
2 B
EQN
0 0
Hf df
2
BEQN 0
H0
2
12
Noise Equivalent Bandwidth
• Low pass filter
coherent _ gain H0 Gain DC _ Power H0
2
Hf df
2
B EQN 0
H0
2
PN R NN 0
N0
2 BEQN N 0 BEQN
2 13
Model of Received Signal
with Noise
At
Pr eDt cos2 f c t t nt ht
x t At cos2 f c t t L
At At
xR t cos2 f c t t st cos2 f c t t nt
L L
xc t
nt
y t xc t nt
Noise
Transmitting Receiving
Antenna Antenna
RF Communication Channel
16
Signal-to-Noise Ratio
SNRR
E xER t
2
E xER t
2
SR
E nER t
2
N 0 BEQN N 0 BEQN
can be used to
• Equivalent destination SNR represent receiver
noise figure
E xPr eD t
2 contributions
SD SD
SNRR
E nPr eD t
2
N D N 0 BEQN 18
Increase in SNR with filtering
• If a filter matched to the input signal is applied, the noise
power would be reduced to the smallest equivalent noise
bandwidth that is allowed.
– Filter to minimize noise power
– Importance of the IF filter in a super-het receiver!
19
Typical Transmission Requirements
Signal Type Freq. Range SNR (dB)
Intelligible Voice 500 Hz to 2 kHz 5-10
Telephone Quality 200 Hz to 3.2 kHz 25-35
AM Broadcast Audio 100 Hz to 5 kHz 40-50
High-fidelity Audio 20 Hz to 20 kHz 55-65
Video 60 Hz to 4.2 MHz 45-55
Spectrum Analyzer 100 kHz-1.8 GHz 65-75
20
CW Communication with Noise
Model of a CW communication system with noise: Figure 10.1-1
At
x t At cos2 f c t t vt cos2 f c t t n t
L
A t At
x c t cos2 f c t t PreDt cos2 f c t t nt hR t
L L
21
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Signal and Noise Power
• What are the signal and noise powers at the receiver?
T
1 2
Ps lim x t x t dt
T T
T
2
Pn E n t
2
• What is the receiver input power
T T
2
1 2
Pv lim E vt vt dt lim E x c t n t x c t n t dt
1
T T T T
T T
2 2
22
Receiver Signal plus Noise Power
• What is the receiver input power
T T
2
1 2
Pv lim E vt vt dt lim E x c t n t x c t n t dt
1
T T T T
T T
2 2
T
1 2
Pv lim E x c t 2 x c t n t n t dt
T T
2 2
T
2
T
1 2
Pv lim x c t 2 x c t En t E n t dt
T T
2 2
T
2
T T T
1
2 2 2
Pv lim xc t dt 2 xc t 0 dt E nt dt
2 2
T T
T 2 T
2
T
2
T
1 2
Pv lim x c t dt E n t Ps Pn
T T
2 2
T
2
23
Signal-to-Noise Ratio (SNR)
• The SNR is a measure of the signal power to the noise
power at a point in the receiver.
– Typically described in dB
– The above computation was performed at the input
Ps
SNR
Pn
• Matlab SNR example: SNR_AM_Example.m
– Pre-D “AM” SNR based in filter Beqn
• Effective BEQN RF due to sampling spectrum (B=Fis/2)
• AM signal power based on carrier plus signal
24
Noise Equivalent Bandwidth
• Since the noise power spectrum is uniform, a systems
average noise power is the product of the noise power and
the integral of the filter power.
R NN 0 Hf df N 0 Hf df
N0 2 2
2 0
25
Noise Equivalent Bandwidth
R NN 0 0 Hf df 0 2 Hf df
N 2 N 2
2 2 0
H rect _ mod el f GainDC _ Power Gain DC _ Power H0
f
rect 2
2 B
EQN
0 0
H f df
2
BEQN 0
H 0
2
26
Noise Equivalent Bandwidth
• Low pass filter
coherent _ gain H0 Gain DC _ Power H0
2
Hf df
2
BEQN 0
H0
2
PN R NN 0
N0
2 BEQN N 0 BEQN
2 27
Filtering
• What happens if the receiver input is filtered?
v f t x c t n t h1 t h 2 t
v f t x c t h 1 t h 2 t n t h 1 t h 2 t
Ps
SNR
Pn
28
Filtering
• What effect does the filter have on the noise?
– Normally you would expect for two filters
Ps
PN _ Post _ Filter1 N 0 BEQN _ Filter1 SNRPost _ Filter1
N 0 BEQN _ Filter1
– Assume that the filters follow each other and that the first filter is
narrower than the second filter
Ps
SNRPost _ Filter 2
N 0 BEQN _ Filter1 29
Filters Provide SNR “Gain”
• If filter 2 Beq < filter 1 Beq:
Ps BEQN _ Filter1
SNR Post _ Filter 2 SNR Post _ Filter1
N 0 BEQN _ Filter 2 B
EQN _ Filter 2
fc fc
f c BT f c BT
2
2
E n t E n i t E n q t
2
N0
2
• What about ?
E nt E ni t cos2 fc t nq t sin2 fc t
2 2
32
Quadrature Noise (2)
• Noise in a quadrature process
E n t E n i t cos2 f c t n q t sin 2 f c t
2 2
n t 2 cos 2 2 f t
i c
n q t sin 2 f c t
2
2
2 1 2 1
E nt E ni t cos2 2 f c t 2 nq t cos2 2 f c t 2
2
2 2
2 1
2
2 1
2
E nt E ni t E cos2 2 f c t 2 E nq t E cos2 2 f c t 2
2
E nt
2 1
2
2
1
2
E ni t E nq t 0
2 N
2
33
Mixing Noise (1)
• Think of the two noise bands as
1. The band of interest
2. The image band
nt cos2 f LO t hIF t n1i t cos2 f LO f IF t n1q t sin 2 f LO f IF t cos2 f LO t hIF t
n2i t cos2 f LO f IF t n2 q t sin 2 f LO f IF t cos2 f LO t hIF t
nt cos2 f LO t hIF t n1i t cos2 f IF t n1q t sin 2 f IF t hIF t
1
2
n2i t cos2 f IF t n2 q t sin 2 f IF t hIF t
1
2 34
Mixing Noise (2)
• Defining the equivalent IF noise
nt cos2 f LO t hIF t n1i t cos2 f IF t n1q t sin 2 f IF t hIF t
1
2
n2i t cos2 f IF t n2 q t sin 2 f IF t hIF t
1
2
1
2 1i n t
1
n2i t cos 2 f IF t
2
nt cos2 f LO t hIF t hIF t
1 n t 1 n t sin 2 f t
2 1i 2
2i
IF
E ni t
2
E n1i t E n2i t 0
1 2 1 2 N
E nq t
2
1
2
2
1
2
E n1q t E n2 q t 0
2 N
2
2 2 2
• But this is the same as quadrature noise
nt hIF t ni t cos2 fIF t nq t sin2 fIF t hIF t
E nt
2
carrier
N0
2
2 BT N 0 BT 2 N 0 W
E n1 t
2
AboveC
E n2 t
2
BelowC
N0
2
2
BT
2
N 0
BT
2
N 0 W
E nt
2
E n1 t E n2 t N 0 BT 2 N 0 W
2 2
Mix to IF
0 or 1
BT
N0BT 37
Why did we do these derivations?
• The past derivations were all about mixing and filtering.
– Quadrature noise is the noise that gets mixed to the intermediate. The
bandwidth and noise power do not change
E nt
2 N0
2
2 BT N 0 BT 2 N 0 W
BT
W
E nt N 0 BT 2 N 0 W
2
38
Complex Noise
• Noise in a complex process
nt ni t j nq t
• Noise power is related as
E n t E n i t j n q t n i t j n q t
2
H
E n t E n t j n t n t j n t n t n t
2 2 2
i i q q i q
E n t E n i t E n q t
2 2 2
E n t E n t
N
E n t
N 2 2 0
0
2
i q
2 4
MATLAB : n randn ( m , n )
MATLAB : n randn ( m , n ) i randn ( m , n ) sqrt 2
39
Noise Envelope and Phase (1)
• Noise as a magnitude and phase
n t A n t cos2 f c t n t
n i A n cos n n q A n sin n
EA n
NR
2
E An 2 NR
2
40
Noise Envelope and Phase (2)
• Probability of An exceeding a
P An a exp a
2
2 N R
• Phase Distribution
NR
E An
2
p n
1
0 n 2
2
for
E An 2 N R
2
• Noise Power
E nt E An t E cos2 f c t n t
2 2 2
E nt 2 N N
2 1 N 0
R R
2 2
41
Noise Characteristics
• The noise power does not change based on the
representation, the center frequency, or due to mixing.
42
CW Communication with Noise
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
At
x t At cos2 f c t t vt cos2 f c t t n t
L
A t At
x c t cos2 f c t t Pr eDt cos2 f c t t nt hR t
L L
43
Chapter 11
ECE 6640 44
Digital Pulse-Amplitude Modulation
(PAM)
• Also referred to as pulse-code modulation (PCM)
• The amplitude of pulse take on discrete number of
waveforms and/or levels within a pulse period T.
x t a k pt kT
k
• p(t) takes on many different forms, a rect for example
1 0tT
pt
0 else
45
Digital Signaling Rate
• For symbols of period T,
the symbol rate is 1/T=R
46
Transmission
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
yt a k ~
p t t d kT n t
k
47
Transmission
y t a k ~
p t t d kT n t
k
p t fn pthe
• Recovering or Regenerating
~
c t
t hsignal may not be trivial
yˆ mT t d am ak ~
p mT kT nmT t d
k m
48
ABC Binary PAM formats
49
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
PAM Power Spectral Density:
Polar NRZ
t Td k Tb
vt a k rect
k Tb
Ea n 0, E a n 2
2
pTd ,
1
Tb
0 Td Tb
E a j a k 0, for j k
Rvv E vt vt 1 ,
2
Tb Tb
Tb
S vv w E vt vt 2 Tb sinc 2 f Tb
A2 f
sinc 2
rb rb 50
PAM Power Spectral Density:
Arbitrary Pulse
t Td k D pTd
1
vt a k p , 0 Td D
k D D
E an ma , E an a ma
2 2 2
S vv f P f Ra n exp j 2 f D
1 2
D n
a 2 ma 2 , n0 Tb D, rb
1
Ra n 2 D
ma , n0
rb is symbol rate
• Using Poisson’s sum formula
2
a
2
ma n n
2
S vv f P f P f
2
D D n D D
S vv f a rb P f ma rb Pn rb f n rb
2 2 2 2
n 51
Power spectrum of
Unipolar, binary RZ signal
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 11.1-5
t
pt rect rect2 rb t
Tb
2
f A2
2
A2 n
S vv f sinc sinc f n rb 2
16 rb 2 rb 16 n 2
f
P f
1
sinc
2 rb b
2 r
E an , E an
A
2
2 A2
2
2 A2
a ma ,n 0
2
Ra n
2
2
m 2 A , n0
a 4
52
Power spectrum of
Unipolar, binary RZ signal
2
f A2
2
A2 n
S vv f sinc sinc f n rb
16 rb 2 rb 16 n 2
Unipolar Binary RZ
0.07
0.06
• For rb=2
0.05
0.04
0.03
0.02
0.01
0
-8 -6 -4 -2 0 2 4 6 8
freq (f)
2
A2 f A2 t
S vv f sinc sincn f n rb pt rect rectrb t
2
4 rb rb 4 n Tb
2
2
f A2
S vv f f
A f
sinc P f
1
sinc
4 rb b
r 4 rb rb
Unipolar Binary NRZ
0.25
A2
E an , E an
A 2
0.2
2 2
• For rb=2
2 A2
a ma ,n 0
0.15 2
Ra n
2
2
0.1
m 2 A , n0
a 4
0.05
0
-8 -6 -4 -2 0 2 4 6 8
freq (f)
54
Power spectrum of
Polar, binary RZ signal (+/- A/2)
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
t
pt rect rect2 rb t
2
A2 f
S vv f sinc
Tb
16 rb 2 rb 2
f
P f
1
• For rb=2 sinc
0.035
Polar Binary RZ 2 rb 2 rb
0.03
E an 0, E an A
2 2
0.025 4
0.02
0.015 2 m 2 A2 , n 0
Ra n 4
a a
0.01
ma2 0, n0
0.005
0
-8 -6 -4 -2 0 2 4 6 8
freq (f)
55
Power spectrum of
Polar, binary NRZ signal (+/- A/2)
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
f
2
t
pt rect rectrb t
2
S vv f
A
sinc
4 rb rb Tb
f
P f
• For rb=2 1
sinc
0.14
Polar Binary NRZ rb rb
0.12
0.1
2
E an 0, E an A
2
0.08
2 m 2 A2 , n 0
Ra n 4
0.06 a a
0
-8 -6 -4 -2 0 2 4 6 8
freq (f)
56
Spectral Attributes of PCM
If Bandwidth W=1/T,
then WT=1
57
Baseband Binary Receiver
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 11.2-1
yt a k pt kT h t n in t h t
k
y t k ak n t k
58
Regeneration of a unipolar signal
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
(a) signal plus noise (b) S/H output (c) comparator output: Figure 11.2-2
59
Unipolar NRZ Binary Error
Probability
• Hypothesis Testing using a voltage threshold
– Hypothesis 0
• The conditional probability distribution expected if a 0 was sent
pY yk | H 0 pY ak n t k | ak 0 pY n t k
– Hypothesis 1
pY yk | H 0 p N yk
• The conditional probability distribution expected if a 1 was sent
pY yk | H1 pY ak n t k | ak A pY A n t k
pY yk | H1 p N yk - A
60
Decision Threshold and
Error Probabilities
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
V
Pe1 PY V p Y y | H1 dy
Pe 0 PY V p Y y | H 0 dy
V
pY Vopt | H 0 pY Vopt | H1
2
62
Threshold regions for conditional PDFs
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 11.2-4
P H 0 P H1
1 A
Vopt
2 2
63
For AWGN
• The pdf is Gaussian
y2
p Y y | H 0 p N y
1
exp
2
2 2
2
for
2
Q x
1
exp d
2 x 2
V A
Pe 0 PY V p N y dy Q Q
V 2
AV A
V
Pe1 PY V p N y A dy Q Q
2
for
P H 0 P H 1
1 A
Vopt
2 2
A V A
Pe1 Q Q Pe 2 64
2
Modification for
Polar NRZ Signals (+/- A/2)
• Hypothesis Testing using a voltage threshold
– Hypothesis 0
• The conditional probability distribution expected if a 0 was
sent
A A
pY yk | H 0 pY ak n t k | ak pY n t k
2 2
A
pY yk | H 0 p N yk
2
– Hypothesis 1
• The conditional probability distribution expected if a 1 was
sent
A A
pY yk | H1 pY ak n t k | ak pY n t k
2 2
A
pY yk | H1 p N yk -
2
A A
Vopt 0 65
2 2
Modification for
Polar NRZ Signals (+/- A/2)
• Determining the error probability
A V
A Q A
Pe 0 PY V p N y dy Q 2
2 2
V
A V
A Q A
V
Pe1 PY V p N y dy Q 2
2 2
• Notice that the error is the same as Unipolar NRZ
– The distance between the expected signal values is the
same
66
Modification for
Bipolar NRZ Signals (+/- A)
• Hypothesis Testing using a voltage threshold
– Hypothesis 0
• The conditional probability distribution expected if a 0 was
sent
pY yk | H 0 pY ak n t k | ak A pY A n t k
pY yk | H 0 p N yk A
– Hypothesis 1
• The conditional probability distribution expected if a 1 was
sent
pY yk | H1 pY ak n t k | ak A pY A n t k
pY yk | H1 p N yk - A
Vopt A A 0 67
Modification for Bipolar NRZ Signals
• Determining the error probability
AV A
Pe 0 PY V p N y A dy Q Q
V
AV A
V
Pe1 PY V p N y A dy Q Q
68
Relationship to signal power
• Defining the average received signal power
– Unipolar NRZ
1
S R A2 , 0, A
2
T2
A A 1
1 S R E lim xc t dt
2
– Polar NRZ S R A2 , 2 , 2
T T 2
T
4
– Bipolar NRZ S R A2 , A, A
• In terms of SNR
1 S
2 2 N for Unipolar
A A 2
R
2 4 N R S
for Polar
N R
A 2 S
2
A
for Bipolar
N R N R 69
Probability of error
• The probability of detecting a transmitted symbol
correctly is dependent upon the received signal-to-
PH PH
noise ratio …. assuming 1
0 1
2
– Unipolar NRZ (orthogonal)
A 1 S A
2
A2 1 S
Pe Q Q
2 2 N R 2 4 NR 2 N R
Eb
2
A A2 1 S
– Unipolar
2 4 N R 2 N R N0
2 Eb
2
A A2 S
– Polar
2 4 N R N R N0
2 Eb
2
– Bipolar A A2 S
N R N R N0
73
Relationship to Bit Error Probability
A Eb
– Unipolar Perror Q Q
2 N0
A 2 Eb
– Polar Perror Q Q
2 N0
A 2 Eb
– Bipolar Perror Q Q
N0
74
Bit Error Rate Plot
Classical Bit Error Rates
0.5
Orthogonal
0.45 Antipodal
0.4
0.35 EbNo=(0:10000)'/1000;
0.3
Bit Error Rate
% Q(x)=0.5*erfc(x/sqrt(2))
0.25
Ortho=0.5*erfc(sqrt(EbNo)/sqrt(2));
0.2 Antipodal=0.5*erfc(sqrt(2*EbNo)/sqrt(2));
0.15
semilogx(EbNo,[Ortho Antipodal])
0.1 ylabel('Bit Error Rate')
xlabel('Eb/No')
0.05 title('Classical Bit Error Rates')
legend('Orthogonal','Antipodal')
0
-3 -2 -1 0 1
10 10 10 10 10
Eb/No
75
BER Performance, Classical Curves
log-log plot
0
Classical Bit Error Rates
10
Orthogonal
-1 Antipodal
10
-2
10
Bit Error Rate
-3
10
-4
10
-5
10
-6
10
-7
10
-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Eb/No
76
Antipodal and Orthogonal Signals
• Antipodal
– Distance is twice “signal voltage”
– Only works for one-dimensional signals
T
1 for i j
zij si t s j t dt
1
d 2 Eb
E 0 1 for i j
• Orthogonal
– Orthogonal symbol set
– Works for 2 to N dimensional signals
T
1 for i j
z ij s i t s j t dt
1
d 2 Eb
E 0 0 for i j
77
M-ary Signals
sumb
K N
i
b sum b ik b kj
j
k k
1 for i j
z ij k 1 k 1
K 0 for i j
78
Example 11.2-1
• Unipolar computer network with
Rb 106 bps N 0 4 10 20 W / Hz 194dB / Hz
– Desired BER is one bit per hour
Pe 1 3 10 10
3600 Rb
• Solve for the signal energy
A A
Perror Q 3 10 10 From p. 790 6.2
2
2
1 S
2
A A2
R for Unipolar
2 4 NR 2 NR
2 R
2 6.2 N R 2 6.2 N 0 b S R
2
2
S R 2 38.44 4 10 20 0.5 106 W 1.54 10 12W 79
Exercise 11.2-1 (1)
• Unipolar system with equally likely digits and
SNR = 50
• Calculate the error probabilities when the
threshold is set to V=0.4 x A
1 S
2
A
R for Unipolar
2 2 NR
A
50 10
1
2
2
A V
V
Pe1 PY V N
p y A dy Q
V
Pe 0 PY V p N y dy Q
V 80
Exercise 11.2-1
• Calculate the error probabilities when the
threshold is set to V=0.4 x A PH PH
1
0 1
V 2
Pe 0 PY V Q Q0.4 10
Pe 0 Q4.0 3.5 10 5 V=0.5 x A
A V
Pe1 PY V Q Q0.6 10 Pe 0 Pe1 Q0.5 10 Q5.0
Pe1 Q6.0 1.5 10 9
Pe 0 Pe1 Perror 3.5 10 7
f X x
1
exp
x X 2
, for x
2
2
2
Gaussian PDF and pdf
x v X 2 1
FX x
exp dv
1
2
2
2 0.9
v
0.8
0.6
0.5
0.4
0.3
0.2
0.1
0
-8 -6 -4 -2 0 2 4 6 8
ECE 6640 82
Gaussian Distribution
The Gaussian Probability Distribution Function is
x
v X 2
FX x
exp dv
1
2
2
2
v
ECE 6640 83
Q Function
Another defined function that is related to the Gaussian (and used) is the Q-function.:
u2
Qx
1
exp
2
du Q Function Table p. 858
2
ux
x X
FX x 1 Q
84
Using MATLAB
Another way to find values for the Gaussian
The error function
exp u du
x
erf x
2 2
u 0
1 x
Q x 1 erf
2 2
1 x X 1 1 xX
FX x 1 1 erf erf
2
2 2 2 2
From MATLAB:
ERF Error function.
Y = ERF(X) is the error function for each element of X. X must be
real. The error function is defined as:
ECE 6640 85
Using MATLAB (2)
The complementary error function
erfcx 1 erf x
x
Q x
1
erfc
2 2
From MATLAB:
ERFC Complementary error function.
Y = ERFC(X) is the complementary error function for each element
of X. X must be real. The complementary error function is
defined as:
ECE 6640 86
Qfn and Qfninv
• These function are now in the Misc_Matlab zip file on the
web site
function [Qout]=Qfn(x)
% Qfn(x) = 0.5 * erfc(x/sqrt(2));
Qout = 0.5 * erfc(x/sqrt(2));
function [x]=Qfninv(Pe)
% For Qfn(x) = 0.5 * erfc(x/sqrt(2));
% The inverse can be found as
x=sqrt(2)*erfcinv(2*Pe);
87
Properties of Matched
• See ECE3800 Notes
– Review from Chapter 9
• Wikipedia
– http://en.wikipedia.org/wiki/Matched_filter
– “The matched filter is the optimal linear filter for maximizing the
signal to noise ratio (SNR) in the presence of additive stochastic
noise.”
88
Defining a Filter for Pulses
y t ak pt t d k T
k
r 2 B , for B r 2 B
89
Defining a Filter for Pulses
• Possible solutions
1 t 0 pt p t sincr t
pt
0 t T ,2T ,
P f 0 B f p t P f 0 for f
p 0 P f df
1
• Therefore we select
1 f 1
P f P f rect r
r r T
2 f f
P f cos rect
4 4 2
91
Convolving
• Raised Cosine Convolution with Bandlimited Spectrum
1 f
P f P f rect
r r
1 r
r f
2
1 2 r
P f cos 2
r r
f f
r 4 2 2 2
r
0 f
2
r r
where B , with 0 and B BT
2 2
ABC
r
2
1
r
T
GNU FDL:Oli Filth, Raised Cosine Filter Response , en.wikipedia.org, 3 November 2005, Oli Filth
ABC
r
2
1
r
T
GNU FDL:Oli Filth, Raised Cosine Filter, Impulse Response, en.wikipedia.org, 3 November 2005, Oli Filth 93
Nyquist Filter
n
cos
cos2 t M n
pt sincr t pn sinc , for k M n k M
1 4 t
2
M
2
n
1 2
M
% function hnyq=nyquistfilt(alpha,M)
2 r
% or
r
0 0 1 % function hnyq=nyquistfilt(alpha,fsymbol,fsample,k)
2 %
cos r t
% alpha roll-off
pt sincr t % fsample rate
1 2 r t
2
% fsymbol rate
% M = fsample/fsymbol (an integer value)
% k is 1/2 the number of symbols in the filter
n fs
t r % The filter length is euqal to 2*ceil(k*M)+1
fs M %
% A discrete time cosine taperd Nyquist filter
n
r t % Based on frederic harris, Multirate Signal Processing for
Communications
M
% Prentice-Hall, PTR, 2004. p. 89 94
MATLAB Raised Cosine Filters (1)
• Rcosine (obsolete)
– [NUM, DEN] = RCOSINE(Fd, Fs, ‘fir’, R)
– FIR raised cosine filter to filter a digital signal with the digital
transfer sampling frequency Fd. The filter sampling frequency is
Fs. Fs/Fd must be a positive integer. R specifies the rolloff factor
which is a real number in the range [0, 1].
• rcosfir
– B = RCOSFIR(R, N_T, RATE, T)
– Raised cosine FIR filter. T is the input signal sampling period, in
seconds. RATE is the oversampling rate for the filter (or the
number of output samples per input sample). The rolloff factor, R,
determines the width of the transition band. N_T is a scalar or a
vector of length 2. If N_T is specified as a scalar, then the filter
length is 2 * N_T + 1 input samples.
95
MATLAB Raised Cosine Filters (2)
• firrcos
– B=firrcos(N,Fc,DF,Fs)
– Returns an order N low pass linear phase FIR filter with a raised
cosine transition band. The filter has cutoff frequency Fc,
sampling frequency Fs and transition bandwidth DF (all in Hz).
– The order of the filter, N, must be even.
– Fc +/- DF/2 must be in the range [0,Fs/2]
– The coefficients of B are normalized so that the nominal passband
gain is always equal to one.
– B=firrcos(N,Fc,R,Fs,'rolloff') interprets the third argument, R, as
the rolloff factor instead of as a transition bandwidth.
– R must be in the range [0,1]
96
Textbook Waveform Energy
• Waveform Energy
T
Ei si t dt
2
• Matched Filter
t
z t r t ht r ht d
ht u t s * T t
t
z t s s * T t d
0
T T T
z T s s T T d s s d s d
* * 2
0 0 0
98
Symbols and Matched Filters
y (t ) si (t ) h opt (t )
si (t ) h opt (t ) A2
A A
T T
T t T t 0 T 2T t
y (t ) si (t ) h opt (t )
si (t ) h opt (t ) A2
A A
T T
0 0 0
T T
Eb si t dt Eb si t s j * t dt
2
0 0
Ed Eb 2 Eb Eb Eb 1
• Antipodal
– Distance is twice “signal voltage”
– Only works for one-dimensional signals
T
1 for i j
zij si t s j t dt
1
d 2 Eb
E 0 1 for i j
• Orthogonal
– Orthogonal symbol set
– Works for 2 to N dimensional signals
T
1 for i j
z ij s i t s j t dt
1
d 2 Eb
E 0 0 for i j
102
Relationship to Bit Error Probability
– Orthogonal Eb
Perror Q
N0
– Antipodal 2 Eb
Perror Q
N0
103
Bit Error Rate Plot-Linear BER
Classical Bit Error Rates
0.5
Orthogonal
0.45 Antipodal
0.4
0.35 EbNo=(0:10000)'/1000;
0.3
Bit Error Rate
% Q(x)=0.5*erfc(x/sqrt(2))
0.25
Ortho=0.5*erfc(sqrt(EbNo)/sqrt(2));
0.2 Antipodal=0.5*erfc(sqrt(2*EbNo)/sqrt(2));
0.15
semilogx(EbNo,[Ortho Antipodal])
0.1 ylabel('Bit Error Rate')
xlabel('Eb/No')
0.05 title('Classical Bit Error Rates')
legend('Orthogonal','Antipodal')
0
-3 -2 -1 0 1
10 10 10 10 10
Eb/No
Eb
Q
-2
10 Perror
N0
Bit Error Rate
-3
10
-4
10
2 Eb
-5
10
Perror Q
10
-6
N0
-7
10
-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Eb/No
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Signal Processing Functions
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 4
Prentice Hall PTR, Second Edition, 2001.
Bandpass Demodulation and Detection
t 2 f 0 t t
t
2 f 0 t p m 2 t 2 f m 3 d
ECE 6640 6
Phasor Representation
ECE 6640 7
Example: Bandpass Phasor Analysis
of Double Sideband (DSB)
• Given a tone message …
mt A m cos2 f m t
Ac Am
s t cos2 f c f m t cos2 f c f m t
2
• A positive frequency phasor can be defined and drawn
– First define the complex signal as (cos exp)
Ac Am
s pos f t
C
exp j 2 f c f m t exp j 2 f c f m t
4
ECE 6640 8
Phasor Analysis DSB (2)
Ac Am
s pos f t
C
exp j 2 f c f m t exp j 2 f c f m t
4
4
A
m
A
c
f fc 4
A
m
m
A
f c
m
ECE 6640 9
Phasor Analysis AM
ECE 6640 10
Phasor Analysis AM (2)
4
A
fm
fm
c
2
A
4
c
A
fc
ECE 6640 11
Narrowband FM & PM Spectrum
s C t A expj 2 f c t j t
• Forming the Quadrature Representation and transforming
the series expanded rig functions
s C t A expj 2 f c t j t A expj 2 f c t expj t
1
s C t A expj 2 f c t 1 j t j t
2
2!
• Maintaining the 1st order terms …
ECE 6640 12
Narrowband FM & PM Spectrum (2)
SC f A f f c f j f
SC f A f f c j f f c
SC f A f f c f f c
2 2 2
ECE 6640 13
PM and FM Basis
M 3
PM f p M 2 f FM t j f
f
ECE 6640 14
PM Phasor v1
s c t A expj 2 f 0 t j expj p m 2 t
2 t Ac
m
p
fo
ECE 6640 15
PM Phasor v2
s C t A expj 2 f c t 1 j t
• For a cos wave message input
j p j p
s C t A expj 2 f c t 1 expj 2 f m t exp j 2 f m t
2 2
Ac
fo
ECE 6640 16
FM Phasor
s C t A expj 2 f c t 1 j t
• For a cos wave message input
t
s C t A expj 2 f c t 1 j f cos2 f m t d
s C t A expj 2 f c t 1 f exp j 2 f m t f exp j 2 f m t
2 2
ECE 6640 17
Why discuss phasors?
ECE 6640 18
General Notes from ABC
ECE 6640 19
Binary modulated waveforms
a) ASK
b) FSK
c) PSK
d) DSB with
baseband pulse
shaping
• Auto-correlation
E s0 t s0 t 0
Es 0 t s1 t 0
2
E s1 t s1 t cos2 f c
Ac
2 T
• Average Power
POOK P0 Rs0 s0 P1 Rs1s1
2 E
1 1 Ac
2
0 2
Ac
cos0
Ac
POOK 0 T
2 2 2 T 4 21
Amplitude Shift Keying (2)
• Auto-correlation
R s0s0 0 2 E
Ac
Ac
2
T
R s1s1 cos2 f c
2 T
• Symbol Power Spectral Density
2
S OOK c T 2 sinc 2 f c f T sinc 2 f c f T
A
8
• Bandpass Bandwidth
– Nominally: BT=1/T, first null at Bnull=2/T
22
ASK Power Spectrum
• From ABC Chapter 11
S vv f a rb P f ma rb Pn rb f n rb
2 2 2 2
n
• RF Analysis
1
Gc f S vv f f c S vv f f c
4
23
ASK Power Spectrum (2)
2
A2 f A2
S vv f sinc f 1
4 rb rb 4 rb
Tb
1
Gc f S vv f f c S vv f f c
4
Figure 14.1-2 24
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
ASK MATLAB Simulation
Symbol Sequenct in Time
0
Symbol Sequence Circular Auto-correlation
0.5
-50
Magnitude (dB)
Amplitude
-0.5 -100
-1
-150
1 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 5 6
Time -5
Frequency 8
x 10 x 10
OOK Demodulation Eye Diagram
2.5 Symbol Sequence Circular Auto-correlation
0
2 -10
-20
1.5
Magnitude (dB) -30
Amplitude
1 -40
-50
0.5
-60
0 -70
-80
-0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 25
Time -6
3 3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4
x 10 Frequency 7
x 10
ASK Transmission Capability
• Comparing the ratio of the bit rate to the required
signal bandwidth
rb
TP
BT
26
M-ary ASK
• Use multiple amplitude levels to represent more than one
bit per symbol
• MASK
– M-1 one states and the off state
– All positive amplitudes (no phase reversals)
M 1
ma E an
2
a E an
2
2 2
ma
M 2 1
12
S vv f
2
A M 1
2
f A2 M 12
sinc
2
f
12 rb b
r 4
Gc f S vv f f c S vv f f c 27
M-ary ASK Transmission Capability
• Comparing the ratio of the bit rate to the required
signal bandwidth
– For m-ary, the bit rate is
bit rate rs log 2 M
– Note that for m-ary ASK, the OOK system has the
smallest spectral efficiency
28
Binary QAM
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
ma E an 0
a 2 E an 2 A 2 29
Quadrature AM (QAM)
• An M-ary Signal – 4 complex symbols
• Quadrature
p 0 t 1 s 0 t 1 A c cos2 f c t
p1 t i s1 t i A c cos2 f c t
p 2 t 1 s 2 t 1 A c cos2 f c t
p 3 t i s 3 t i A c cos2 f c t
• Average Power
0 1 Ac
2
A c cos0
2
E QAM
T 2 2 30
QAM
• Symbol Cross Correlation
C0, 0 t 1 C0, 2 t 1
T T
C0,1 t i C0,3 t i
T T
Ts rs rs
1 f
P f sinc
r 2
rs b 2 1 f
rs rs 2 S vv f Ac sinc
rs rs
Note that the symbol rate
is one-half the bit rate. Gc f S vv f f c S vv f f c
2
2
Ac 4 2 f
S vv f sinc
rb rb
S vv f a r P f ma r Pn r f n r
2 2 2 2
32
n
QAM Transmission Capability
• Comparing the ratio of the symbol rate to the
required signal bandwidth
rs log 2 M
TP
BT
33
Phase Modulation Methods
• Phase shift keying (PSK) is digital PM
x t A cos2 f t p t k T
c c c k D s
k
This is QAM,
rotated by pi/4
for 4-PSK
M=4 M=8
4-PSK 8-PSK
35
M-PSK
• An M-ary Signal – M complex symbols
• Quadrature (2 possible representations)
s k t A c cos 2 f c t
2 k 1 , for k 0 to M 1
M
2 k 1 2 k 1
p k t I k , Q k cos , sin , for k 0 to M 1
M M
T 2
• Average Power, Amplitude to Energy
0 1 2
2 E
Ac cos0
2 Ac Ac
PQAM
T 2 2 T
36
Binary PSK
• Signal Symbols
s 0 t A c cos2 f c t 0 A c cos2 f c t
s1 t A c cos2 f c t 1 A c cos2 f c t
• Autocorrelation
E s k t s k t
*
1
A c cos2 f c
2
T 2
• Cross Correlation (the definition of antipodal)
1
E s 0 t s1 t A c cos2 f c
* 2
T 2
Rs0 s1 Rs0 s0
37
Binary PSK
• Signal Symbols
s 0 t A c cos2 f c t 0 A c cos2 f c t
s1 t A c cos2 f c t 1 A c cos2 f c t
• Autocorrelation
E s k t s k t
*
1
A c cos2 f c
2
T 2
• Cross Correlation (the definition of antipodal)
1
E s 0 t s1 t A c cos2 f c
* 2
T 2
Rs0 s1 Rs0 s0
38
BPSK Power Spectrum
• From Chapter 11
S vv f a rb P f ma rb Pn rb f n rb
2 2 2 2
n
0.8
-40
0.6
0.4 -60
0.2
Magnitude (dB)
Amplitude
-80
0
-0.2 -100
-0.4
-120
-0.6
-0.8
-140
-1
1 -20
0.5 -40
Magnitude (dB)
Amplitude
0 -60
-0.5 -80
-1 -100
-1.5 -120
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 40
Time -6 Frequency 7
x 10 x 10
Other Forms of PSK
• Differential PSK
– The symbols are the “encoding” of two adjacent bits
– Encoding the bit changes not the bit values
– Typically an exclusive-Or or Exclusive NOR
• QPSK
– Already shown as QAM
• Offset QPSK
– Offset the I and Q bits of QAM by one half the symbol period
– Phase changes at BPSK bit rate, bit absolute phase change is now
always pi/2 (orthogonal)
41
Differential Encoded PSK (DPSK)
• The binary data stream is differentially encoded
– The logical combination of the previous bit sent and the next bit to
be sent. An Exclusive NOR gate can be used.
– Provides an arbitrary start … only phase change by pi is required
to decode the message, not the absolute bit values!
Sample Index 0 1 2 3 4 5 6 7 8 9 10
Information m(k) 1 1 0 1 0 1 1 0 0 1
Figure 14.1-6
43
Digital Frequency Modulation
44
Frequency Shift Keying
• Binary FSK
s0 t Ac cos2 f c f d t
s1 t Ac cos2 f c f d t
45
M-FSK
• An M-ary Signal – M complex symbols
s t A cos2 f t 2 f k t ,
k c start for k 0 step to M 1
• Autocorrelation
E sk t sk t
*
1
Ac cos2 f start f step k
2
T 2
46
BFSK
• Signal Symbols
s0 t Ac cos2 f c f d t
s1 t Ac cos2 f c f d t
• Autocorrelation
1
E sk t sk t Ac cos2 f c f d
* 2
T 2
• Cross Correlation
E s0 t s1 t Ac E cos2 f c f d t cos2 f c f d t
* 2
T
1
E s0 t s1 t Ac E cos2 2 f d t 2 f c f d
* 2
T 2
rb
sk t Ac cos2 f c ak f d t ak 1 fd
2
sk t Ac cos2 f c t cos2 ak f d t Ac sin 2 f c t sin 2 ak f d t
bbk t I k , Qk cos rb t , 1 ak sin rb t
k
48
BFSK Quadrature Representation (2)
bbk t I k , Qk cos rb t , 1 ak sin rb t
k
4 2 2
2
1 r r
Qk
2
sinc f b rb sinc f b rb
4 rb 2 2 2
f
cos
4
rb
Qk 2 2
2
2
rb 2 f
1
rb
2
cos f
1 r r 4 rb
Glp f Gi f Gq f f b f b 2
4 2 2 rb 2 f 2
1
rb 49
Power spectrum of BFSK
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.1-8
2
f
cos
1 r rb 4 rb
Glp f Gi f Gq f f b f 2
4 2 2 rb 2 f 2
1
rb
2 2 f d rb 2
rb
fd
2
50
BFSK MATLAB Simulation 0
Magnitude (dB)
The change in frequency -100
is too small
-150
BFSK Demodulation Eye Diagram 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
1.5 Frequency 8
x 10
0
1
-20
0.5
-40
Amplitude
Magnitude (dB)
0
-60
-0.5
-80
-1
-100
-1.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-120
Time -6
x 10 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 51
Frequency 7
x 10
Spectrum of M-FSK
52
Special Versions of FSK
• Continuous Phase FSK (CPFSK)
t
xc t Ac cos 2 f c t 2 f d x d
0
a0 t , 0t T
a T a t T , T t 2 T
0 1
t
0 x d
k 1
a j T ak t k T , k T t k 1 T
j 0
0 x d
k 1
a j T ak t k T , k T t k 1 T
j 0
• The phase is continuous at the transitions between
bit.
– This is most easily accomplished if the phase is π or a
multiple of π at the start and end of each bit period.
54
Binary CPFSK
55
MSK Baseband
bbk t I k , Qk
xi t cos k ak ck pt k T xq t sin k ak ck pt k T
k k
m , for k even
rb
ck t k T k
2 n , for k odd
2
56
Illustration of MSK.
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
58
Sklar Representations
• Matched Filter
t
z t r t ht r ht d
ht u t s * T t
t
z t s s * T t d
0
T T T
z T s s T T d s s d s d
* * 2
0 0 0
ECE 6640 64
Optimum binary detection
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
65
Conditional PDFs
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.2-2
66
Bandpass binary receiver
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.2-1
hBPF t h1 t h0 t
• BPSK
hBPF t h1 t h0 t 2 h1 t 2 K s1 T t cos2 f c T t
• BFSK
hBPF t h1 t h0 t K s1 T t K s0 T t
hBPF t cos2 f c f d T t cos2 f c f d T t
hBPF t 2 sin 2 f c T t sin 2 2 f d T t 68
Correlation receiver for
OOK or BPSK
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.2-4
T T T T
E s1 t s0 t dt E s1 t dt 2 E s1 t s0 t dt E s0 t dt
2 2 2
0 0 0 0
T
E s1 t s0 t dt E1 2 E10 E0
2 Eb E1 E0 2
0
2
z1 z0 2 Eb 2 E10 Eb E10
2 max 2 N 0 N0
70
Optimal Parallel Matched Filter Receiver
Error Analysis
T
E s1 t s0 t dt
Eb
E10 Eb
E1 E0 0
2
z1 z0 Eb
• OOK E10 0
2 max N 0
2
z1 z0 2 Eb
• PSK E10 1 Eb
2 max N0
2
• FSK E10 0 z1 z0
E
b
2 max N 0
71
Generalized Probability of Error
• Using the optimal BPF filter and sampling for
each symbol, the relationship will be based on:
Eb E10 Eb 1
2
z1 z0
2 max N0 N0
0
2 T
E10 c cos2 2 f c t cos2 2 f d t dt
A
2 0
T
E 1
E10 b expi 2 2 f d t exp i 2 2 f d t dt
T 2 0
Eb 1 expi 2 2 f d T exp i 2 2 f d T
Eb
T 2 i 2 2 f d i 2 2 f d
Eb sin 2 2 f d T f
Eb Eb sinc4 f d T Eb sinc 4 d
T 2 2 f d rb
k
• There are multiple “orthogonal” tone separations. 2 f d f step
2T
• The correlation coefficient can go negative! The minimum occurs at
73
approximately sinc(1.22) = -0.166
MATLAB Coherent Receivers
ECE 6640 74
Noncoherent Binary Systems
75
Noncoherent OOK receiver
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.3-2
76
Conditional PDFs for
noncoherent OOK
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.3-3
Pe 0 Vopt Pe1 Vopt
E
Pe 0 exp b
2 N0
1 1 1 E
Pe Pe 0 Pe1 Pe 0 exp b
2 2 2 2 N0
Ac
Vopt
2
77
Noncoherent detection of binary FSK
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.3-5
78
Noncoherent FSK
• Qualitative comments
– Using envelope detectors on each symbol output, the Rician error
distribution effects the z detection statistic.
1 E
Pe exp b
2 2 N0
79
Binary error probability curves
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
(a) coherent BPSK (b) DPSK (c) coherent OOK or FSK (d) noncoherent
FSK (e) noncoherent OOK: Figure 14.3-4
0
BER Simulation for BPSK and BFSK
10
-1
10
-2
10
BER -3
10
-4
10
BPSK simulation
-5
10 BPSK (theoretical)
BFSK simulation
BFSK (theoretical)
-6
10
0 2 4 6 8 10 12 14 16
E b/No (dB)
80
Binary error probability curves
(a) coherent BPSK (b) DPSK (c) coherent OOK or FSK
(d) noncoherent FSK (e) noncoherent OOK
Figure 14.3-4
81
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Detection for M-ary Systems
• Determine the detection statistic for all symbols
• Select the maximum statistic
• Decode the binary values from the selected symbol
• Notes:
– M-ASK and M-PSK symbols may no longer be orthogonal
– M-FSK symbols may be orthogonal, but the bandwidth W must
increase to contain the symbols.
82
Quadrature-carrier receiver with
correlation detectors
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.4-1
• Applicable for:
– M-QAM
– M-PSK
83
Carrier synchronization
for quad-carrier receiver
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.4-2
84
Coherent M-ary PSK receiver
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.4-3
15
10
Imag
0
• MPSK_Demo.m -5
-10
-20
-20 -15 -10 -5 0 5 10 15 20
Real
85
Decision thresholds for M-ary PSK
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Figure 14.4-4
15
10
Imag
0
-5
-10
-15
-20
-20 -15 -10 -5 0 5 10 15 20
Real
86
PSK signal constellations
(a) M=4 (b) M=8
Figure 14.5-1
87
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
MPSK Eb/N0 Examples
PreDecode, Es /N0 (dB)=1 PreDecode, Es /N0 (dB)=9
8 10
8
6
6
4
4
2
2
Imag
Imag
0 0
-2
-2
-4
-4
-6
-6
-8
-8 -10
-8 -6 -4 -2 0 2 4 6 8 -10 -8 -6 -4 -2 0 2 4 6 8 10
PreDecode, Es /N0 (dB)=19 Symbol Error Rate, M=8 Bit Error Rate, M=8
Real 0 Real 0
20 10 10
15
-1 -1
10 10
10
5 -2 -2
10 10
Imag
SER
BER
0
-3 -3
10 10
-5
-10
-4 -4
10 10
-15
-20
-20 -15 -10 -5 0 5 10 15 20 -20 -10 0 10 20 -20 -10 0 10 20
88
Real Es/No (dB) Eb/No (dB)
Simulated Performance MPSK
• MPSK_Ber and MPSK_PP_Plot
0
MPSK Symbol Error Rate 0
MPSK Bit Error Rate
10 10
-1 -1
10 10
-2 -2
10 10
-3 -3
10 10
SER
BER
-4 -4
10 M=2 Sim 10 M=2 Sim
M=2 Bound M=2 Bound
-5 M=4 Sim -5 M=4 Sim
10 M=4 Bound 10 M=4 Bound
M=8 Sim M=8 Sim
-6 M=8 Bound -6 M=8 Bound
10 10
M=16 Sim M=16 Sim
M=16 Bound M=16 Bound
-7 -7
10 10
-5 0 5 10 15 20 25 -5 0 5 10 15 20
Es /N0 (dB) Eb/N0 (dB)
89
Simulated Performance MFSK
• MFSK_Ber and MFSK_PP_Plot
0
MFSK Symbol Error Rate 0
MFSK Bit Error Rate
10 10
-1 -1
10 10
-2 -2
10 10
-3 -3
10 10
SER
BER
-4 -4
10 M=2 Sim 10 M=2 Sim
M=2 Bound M=2 Bound
-5 M=4 Sim -5 M=4 Sim
10 M=4 Bound 10 M=4 Bound
M=8 Sim M=8 Sim
-6 M=8 Bound -6 M=8 Bound
10 10
M=16 Sim M=16 Sim
M=16 Bound M=16 Bound
-7 -7
10 10
0 2 4 6 8 10 12 14 16 -5 0 5 10 15
Es /N0 (dB) Eb/N0 (dB)
90
Comparing MPSK and MFSK
• MPSK
– More Eb/N0 required for higher M for symbol error rate
– 2- and 4-PSK have the same BER
• Otherwise higher BER for higher M
• MFSK
– More Eb/N0 required for higher M for symbol error rate,
BUT it does not increase as fast as MPSK
– Less Eb/N0 required for higher M for BER!
91
M-ary QAM system
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
92
Performance comparisons of M-ary
modulation systems
Pbe 104
93
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
MATLAB Coherent Receivers
ECE 6640 94
Notes on BER
ECE 6640 95
QAM BER Computation
ECE 6640 96
QAM BER Curves
0
BER Composite Plot
10
4 QMA
-1 16 QAM
10
64 QAM
256 QAM
-2
10
Bit Error Rate
-3
10
-4
10
-5
10
-6
10
-7
10
0 5 10 15 20 25 30
Eb/No (dB)
ECE 6640 97
QAM BER Curves Detail/Differences
0
BER Composite Plot
10
4 QMA
16 QAM
64 QAM
256 QAM
Bit Error Rate
-1
10
-2
10
-1 0 1 2 3 4 5 6 7 8 9 10
Eb/No (dB)
ECE 6640 98
MPSK Nyquist Filter BER
SER vs SNR
Sklar 1
10
MPSK Simulation: Theory vs. Simulation
Theory 0
Plot
10
-1
10
-2
10
Symbol Error Rate
-3
10 T4
S4
T8
-4 S8
10
T16
S16
T32
-5
10 S32
T64
S64
-6 T128
10
S128
T256
S256
-7
10
0 5 10 15 20 25 30 35 40 45 50 55
SNR (dB)
ECE 6640 99
MPSK Nyquist Filter BER
BER vs Eb/No
Sklar MPSK Simulation: Theory vs. Simulation
Theory
0
10
Plot -1
10
-2
10
-3
T4
Bit Error Rate
10
S4
T8
-4 S8
10
T16
S16
T32
-5
10 S32
T64
S64
-6 T128
10
S128
T256
S256
-7
10
-5 0 5 10 15 20 25 30 35 40 45
EbNo (dB)
-1
10
-2
Symbol Error Rate
10
-3
10
T4
-4
10 S4
T16
-5 S16
10
T64
-6 S64
10 T256
S256
-7
10
0 5 10 15 20 25 30 35
ECE 6640 SNR (dB) 101
QAM Nyquist Filter BER
BER vs Eb/No
Sklar 0
10
QAM Simulation: Theory vs. Simulation
Theory
Plot -1
10
-2
10
Bit Error Rate
-3
10
-4
10 T4
S4
-5 T16
10 S16
T64
-6 S64
10
T256
S256
-7
10
-5 0 5 10 15 20 25
ECE 6640 EbNo (dB) 102
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
What is a Link Budget
• An analysis of the entire communications path
– signal, noise, interference, ISI contributions, etc.
– Include gains and losses
• Link Budget
– An estimate of the input to output system performance
– Will the message get communicated?
– What trade-offs can be made and what effect will they have?
ECE 6640 4
The Channel
ECE 6640 5
Error-Performance Degradation
• Established in Chapter 3
– Loss of SNR
– Intersymbol interference
ECE 6640 7
Figure 5.1
ECE 6640 8
Gains and Losses to be Discussed
• Antenna Efficiency
• Pointing
• Atmospheric Noise
• Space Loss
• Receiver
ECE 6640 9
Range Equations
• The power density in a sphere
from a “point source” antenna
(surface area of a sphere)
Pt
pr
4 r 2
• Receiving power collected by
an antenna (using the
effective area of the receiving
antenna so that p(d) can be • Effective Area
collected) total power extracted
A er
Pt A er incident power flux density
Pr pr A er
4 r 2
ECE 6640 10
Antenna Efficiency and Gain
• Antenna Gain
maximum power intensity
G
average power intensity over 4 steradians
Aer
Pr Pt Gt
4 r 2
ECE 6640 12
Antenna Gain in terms of Area
2 EIRP EIRP
Pr iso EIRP
42 r 2 4 r 2 Ls
• Where Ls is called the “free-space” or “path” loss
– Note: It is defined based on an isotropic antenna with G=1!
L s 4 r
2
4 r f
c
2
ECE 6640 15
The Friis Transmission Equation
ECE 6640 16
Path Loss Considerations
ECE 6640 17
Path Loss Considerations (2)
4 4
ECE 6640 18
Radio Receiver Consideration
G prede mod dB G RF dB G1stMixer dB G IF1 dB G 2 ndMixer dB G IF 2 dB
G prede mod dB G RF dB L1stMixer dB G IF1 dB L 2 ndMixer dB G IF 2 dB
ECE 6640 21
Noise Figure
xt y t
Caution,
PSin Noise Figure is
F
SNRin
N in often referred to in
SNRout G PSin dB instead as a
G N in N amp
linear term
PSin
SNR in N in
F
SNR out G1 G 2 PSin
G 2 G1 N in N amp1 N amp 2
N amp 2
F
SNRin G2 G1 N in N amp1 N amp 2
N in N amp1
G1
SNRout G1 G2 N in N in
N amp1 1 N amp 2 1 F 1
ECE 6640
F 1 1 F1 1 F2 1 F1 2 23
N in G1 N in G1 G1
Basic Receiver
Bandpass Bandpass Lowpass
Amplifier
Filter Filter Filter
x c t x Pr eD t x M t
Demod
GPr eD dB GRF BPF dB GLNA dB G1stMixer dB GBPF dB G2 ndMixer dB GLPF dB
F1stMixer 1 F 1 FBPF 1
FPr eD FRF BPF 1stMixer
GRF BPF G Amp GRF BPF G Amp GRF BPF G Amp G1stMixer
F2 ndMixer 1 FLPF 1
ECE 6640 GRF BPF G Amp G1stMixer GBPF GRF BPF G Amp G1stMixer GBPF G2 ndMixer 24
Thermal Noise Power
228.6 dBW / K Hz
T0 290K IEEE ref
ECE 6640
N 0 204 dBW / Hz 174 dBm / Hz 25
Receiver Operating Characteristics
• FM receiver
• 200 kHz BW
• 12-bit ADC with
10-bit
performance
• Multiple signal
environment
• SOI detection
threshold
ECE 6640 28
Putting It All Together
ECE 6640 29
Table 5.2 Earth Terminal to Satellite
Link Budget
ECE 6640 30
Table 5.3 Link Budget Example
ECE 6640 31
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Signal Processing Functions
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 4
Prentice Hall PTR, Second Edition, 2001.
Encoding and Decoding
Data
Encoder Codeword
Block
Data
Decoder Codeword
Block
ECE 6640 5
Waveform Coding Structured Sequences
• Waveform Coding:
– Transforming waveforms in to “better” waveform representations
– Make signals antipodal or orthogonal
• Structured Sequences:
– Transforming waveforms in to “better” waveform representations
that contain redundant bits
– Use redundancy for error detection and correction
– Bit sets become longer bit sets (redundant bits) with better
“properties”
– The required bit rate for transmission increases.
ECE 6640 6
Antipodal and Orthogonal Signals
• Antipodal
– Distance is twice “signal voltage”
– Only works for one-dimensional signals
T
1 for i j
z ij s i t s j t dt
d 2 Eb 1
E 0 1 for i j
• Orthogonal
– Orthogonal symbol set
– Works for 2 to N dimensional signals
T
1 for i j
z ij s i t s j t dt
1
d 2 Eb
E 0 0 for i j
ECE 6640 7
M-ary Signals Waveform Coding
sumb
K N
i
b sum b ik b kj
j
k k
1 for i j
z ij k 1 k 1
K 0 for i j
ECE 6640 8
Hadamard Matrix Orthogonal Codes
H D 1 H D 1
HD
H D 1 H D 1
• Start with the data set for one bit and generate a Hadamard
code for the data set
0 0 0
1 H 1 0 1
0 0 0 0 0 0
0 1 H1 0
H1 1 0 1
H2
1 0 H1 H1 0 0 1 1
0 1 1 0
1 1
1 for i j
z ij
ECE 6640 0 for i j 9
Hadamard Matrix Orthogonal Codes (2)
0 0 0 0 0 0 0 0 0 0 0
0 1 0
0
1 0 1 0 1 0 1
0 1 0 0 0 1 1 0 0 1 1
0 1 1 H H 2 0 1 1 0 0 1 1 0
Data Set Codeword
H3 2
1 0 0 H 2 H 2 0 0 0 0 1 1 1 1
0
1 0 1 1 0 1 1 0 1 0
1 1 0 0 0 1 1 1 1 0 0
1 1 1 0 1 1 0 1 0 0 1
0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 1 0
1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
See Hadamard in 0
0
0 1 0
0 1 1
0
0 1 1 0 0 1 1 0 0 1 1 0 0 1 1
0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0
1 0 0
MATLAB
0 1 0 1
0
0
0
1
0
0
0
1
1
1
1
0
1
1
1
0
0
0
0
1
0
0
0
1
1
1
1
0
1
1
1
0
0
1 1 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0
H 3 0 1
H4=hadamard(16) 0
1
1 1 1
0 0 0
H
H4 3
H3
H 3 0
1
0
1
0
0
0
1
0
0
0
0
0
1
0
0
1
1
1
1
1
0
1
1
1
0
1
0
1 1
1 0 0 1 0 1 0 1 0 1 0 1 1 0 1 0 1 0 1 0
H4xcor=H4'*H4=H4*H4'
1 0 1 0
0
0
0 1 1 0 0 1 1 1 1 0 0 1 1 0 0
1 0 1 1 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1
0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0
ans = 16*eye(16) 1
1
1 0 0
1 0 1
0
1 0 1 1 0 1 0 1 0 1 0 0 1 0 1
1 1 1 0 0 0 1 1 1 1 0 0 1 1 0 0 0 0 1 1
1 0 0 10
1 1 1
ECE 6640 1 1 0 1 0 0 1 1 0 0 1 0 1 1
Hadamard Matrix Orthogonal Codes (2)
Es M 2D
PE M M 1 Q
N0 Es D E b
ECE 6640 11
Symbol error to Bit Error
PB k 2 k 1 PB M
M
k M 2k 2
PE k 2 1 PE M M 1
• Substituting bit error probability for symbol error
probability bounds previously stated
Es
PE M M 1 Q
N0
k Eb M E S
PB k 2 k 1
Q
PB M Q
N0 2 N 0
ECE 6640 12
Biorthogonal Codes
1 1 0 1 1 0 0
1
0 1 1
1
0 0 1 1 0 0 1
1
1
1 0 0
1
1
1 1 1 0 0 0 0
0 1 0 0 1 0 1
1 1 1 1 0 0 1
1 0 1
1 1 1 0 1 1 0 0 0 0 1 1
1 1 0 0 1 0 1 1 0
1 1 1
ECE 6640 14
B3 Example
H2=hadamard(4)
B3=[H2;-H2]
B3*B3'
ans =
4 0 0 0 -4 0 0 0
0 4 0 0 0 -4 0 0
0 0 4 0 0 0 -4 0
0 0 0 4 0 0 0 -4
-4 0 0 0 4 0 0 0
0 -4 0 0 0 4 0 0
0 0 -4 0 0 0 4 0
0 0 0 -4 0 0 0 4
ECE 6640 15
Biorthogonal Codes (3)
Es 2 Es M 2 k 1
PE M M 2 Q
Q
N0 N0 Es k E b
PB M PE M , for M 8
2
ECE 6640 16
Biorthogonal Codes (4)
M E S M 2k
Hadamard
PB M Q
2 N 0 Es k E b
ECE 6640 17
Waveform Coding System Example
ECE 6640 18
Waveform Coding System Example (2)
ECE 6640 20
Waveform Coding System Example (4)
Stated Improvement Tb k T
2 k
C
ECE 6640 21
Types of Error Control
• Types of codes
– Block Codes or Linear Block Codes (Chap. 6)
– Convolutional Codes (Chap. 7)
– Turbo Codes (Chap. 8)
ECE 6640 23
Channel Models
ECE 6640 24
Code Rate Redundancy
ECE 6640 25
Binary Bit Error Probability
j
n! j
P j , n p 1 p n j
j!n j !
E x n p
E x 2 n p 1 p
ECE 6640 26
Code Example Triple Redundancy
0!3 0 !
3!
P1,3 0.001 0.999 3 0.001 0.999 0.002994
1 2 1 2
1!3 1!
3!
P2,3 0.001 0.999 3 0.001 0.999 0.000002997
2 1 2 1
2!3 2 !
3!
P3,3 0.001 0.001 10 9
3 3
3!3 3!
Code Example Triple Redundancy
ECE 6640 29
(4,3) Single-Parity-Code
ECE 6640 31
Matlab Results
-3
10
-4
10
-5
10
-6
ECE 6640 10 32
2 4 6 8 10 12 14 16
Channel Bits
Rectangular Code
ECE 6640 33
Rectangular Code Example
j t 1 j
n t 1
PM p 1 p n t 1
ECE 6640
t 1 34
Error-Correction Code Tradeoffs
ECE 6640 35
Coding Gain
Eb Eb
CodeGaindB
dB
dB
No uncoded N o coded
ECE 6640 36
Data Rate vs. Bandwidth
Eb PS W PS 1
N 0 N 0 W R N 0 R
ECE 6640 37
Linear Block Codes
• Encoding methodology
– A giant table lookup
– Generators that can span the subspace to create the desired linear
independent sets can be defined
ECE 6640 38
Generator Matrix
k
V vk 1 vk 2 vkn
ECE 6640 39
Table 6.1 Example
% Simple single bit parity
n = 6; k = 3; % Set codeword length and message length.
msg = [0 0 0; 0 0 1; 0 1 0; 0 1 1; ...
1 0 0; 1 0 1; 1 1 0; 1 1 1]; % Message is a binary matrix.
gen = [ 1 1 0 1 0 0; ...
0 1 1 0 1 0; ...
1 0 1 0 0 1];
0 0 0 0 0 0
1 0 1 0 0 1
0 1 1 0 1 0
1 1 0 0 1 1
1 1 0 1 0 0
0 1 1 1 0 1
1 0 1 1 1 0
0 0 0 1 1 1
ECE 6640 40
Systematic Linear Block Codes
G P I k
ECE 6640 42
Parity Check Matrix
H I nk PT
I
U H T m P | I k H T m P | I k n k 0 mr ,n k
P
ECE 6640 43
Parity Check Matrix
[parmat,genmat] = hammgen(3)
rem(genmat * parmat', 2)
[parmat,genmat] = cyclgen(7,cyclpoly(7,4))
rem(genmat * parmat', 2)
ECE 6640 44
Syndrome Testing
ECE 6640 46
Standard Array Format (n,k)
ECE 6640 48
Decoding with the Standard Array
ECE 6640 49
The Syndrome of a Coset
ECE 6640 50
Error Correct
ECE 6640 51
Error Correction Example
• see Table61_example
– each step of the process is shown in Matlab for the (6,3) code of
the text
ECE 6640 52
Decoder Implementation
ECE 6640 53
Hamming Weight and Distance
ECE 6640 55
Hamming Distance Capability
d 1
t min
2
– for (6,3) dmin=3: t = 1
j t 1 j
ECE 6640 56
Hamming Distance Capability
• Error-detection capability
e d min 1
– a block code with a minimum distance of dmin guarantees that all
error patterns of dmin-1 or fewer error can be detected.
– for (6,3) dmin=3: e = 2 bits
j 1
– for (6,3) A(0)=1, A(3)=4, A(4)=3, all else A(i)=0
Pnd 4 p 3 1 p 3 p 4 1 p
4 3
4 p p 3 1 p
3
ECE 6640 58
– for p=0.001: Pnd ~ 3.9 x 10-6
Simultaneous Error
Correction and Detection
• It is possible to trade capability from the maximum
guaranteed (t) for the ability to simultaneously detect a
class of errors.
• A code can be used for simultaneous correction of alpha
errors and detection of beta errors for beta>=alpha
provided
d min 1
ECE 6640 59
Visualization of (6,3) Codeword
Spaces
ECE 6640 60
More to Come
• see http://www.mathworks.com/help/comm/error-
detection-and-correction.html
ECE 6640 61
(8,2) Block Code
ECE 6640 62
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Signal Processing Functions
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 4
Prentice Hall PTR, Second Edition, 2001.
Waveform Coding Structured Sequences
• Structures Sequences:
– Transforming waveforms in to “better” waveform representations
that contain redundant bits
– Use redundancy for error detection and correction
• Block Codes are memoryless
• Convolution Codes have memory!
ECE 6640 5
Convolutional Encodings
ECE 6640 6
Convolutional Encoder Diagram
• Each message, mi, may be
a k-tuple. (or k could be a
bit)
• K messages are in the
encoder
• For each message input,
an n-tuple is generated
• The code rate is k/n
• We will usually be
working with k=1 and
n=2 or 3
ECE 6640 7
Proakis Convolution Encoder
ECE 6640 9
Connection Representation
• k=1, n=3
• Generator Polynomials
– G1 = 1 + X + X2
– G2 = 1 + X2
• To end a message, K-1 “zero” messages are transmitted.
This allows the encoder to be flushed.
– effective code rate is different than k/n … the actual rate would be
(2+k*m_length)/n*m_length
– a zero tailed encoder ….
ECE 6640 10
Impulse Response of the Encoder
• allow a single “1” to transition through the K stages
– 100 -> 11
– 010 -> 10
– 001 -> 11
– 000 -> 00
• If the input message where 1 0 1
– 1 11 10 11
– 0 00 00 00
– 1 11 10 11
– Bsum 11 10 00 10 11
– Bsum is the transmitted n-tuple sequence …. if a 2 zero tail follows
– The sequence/summation involves superpoition or linear addition.
• The impulse response of one k-tuple sums with the impulse responses of
successive k-tuples!
ECE 6640 11
Convolutional Encoding the Message
ECE 6640 12
Proakis (3,1), rate 1/3, K=3 Pictorial
• Generator Polynomials
– G1 = 1
– G2 = 1 + X2
– G3= 1 + X + X2
Solid Lines
are 0 inputs
Dashed Lines
are 1 inputs
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 16
Edition, 2001. ISBN: 0-07-232111-3.
Tree Diagram
ECE 6640 17
Proakis (3,1) K=3
Polynomial and Tree
g1 1 0 0
g 2 1 0 1
g 3 1 1 1
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 18
Edition, 2001. ISBN: 0-07-232111-3.
Trellis Diagram
ECE 6640 19
Trellis Diagram
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 21
Edition, 2001. ISBN: 0-07-232111-3.
A more complicated example follows
ECE 6640 22
Proakis (3,2) K=2
Pictorial
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 24
Edition, 2001. ISBN: 0-07-232111-3.
Proakis (3,2) K=2
State Diagram
Solid Lines
are 0 inputs
Dashed Lines
are 1 inputs
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 25
Edition, 2001. ISBN: 0-07-232111-3.
Proakis (3,2) K=2
Trellis Diagram
ECE 6640 JohnG. Proakis, “Digital Communications, 4th ed.,” McGraw Hill, Fourth 26
Edition, 2001. ISBN: 0-07-232111-3.
Encoding
ECE 6640 27
Decoding Convolutional Codes
ECE 6640 28
ECE 5820 MAP and ML
PY y | X x PX x
P X x | Y y
PY y
ECE 6640 29
ECE 5820 MAP and ML
ECE 6640 30
ECE 5820 Markov Process
– where Zi is the ith branch of the received sequence Z, zji is the jth
code symbol of Zi and similarly for U and u.
ECE 6640 32
ML Computed Using Logs
ECE 6640 33
Channel Models:
Hard vs. Soft Decisions
• Our previous symbol determinations selected a detected symbol with
no other considerations … a hard decision.
• The decision had computed metrics that were used to make the
determination that were then discarded.
• What if the relative certainty of decision were maintained along with
the decision.
– if one decision influenced another decision, hard decisions keep certainty
from being used.
– maintaining a soft decision may allow overall higher decision accuracy
when an interactions exists.
ECE 6640 34
ML in Binary Symmetric Channels
P Z | U m p dm 1 p
L dm
1 p
log P Z | U m dm log L log1 p
p
– The constant is identical for all possible U and can be pre-computed
– The log of the probability ratios is also a constant
log P Z | U m dm A B
ECE 6640 37
Viterbi Decoder Trellis
encoder trellis
ECE 6640 38
Viterbi Example
• m: 1 1 0 1 1
• U: 11 01 01 00 01
• Z: 11 01 01 10 01
merging
paths
ECE 6640 39
Viterbi Example
• m: 1 1 0 1 1
• U: 11 01 01 00 01
• Z: 11 01 01 10 01
ECE 6640 40
Add Compare Select
Viterbi Decoding Implementation
• Section 7.3.5.1, p. 406
Possible Connections
ECE 6640 41
Add Compare Select
Viterbi Decoding Implementation
• State Metric Update based on new Branch Metric Values
– Hard coding uses bit difference measure
– Soft coding uses rms distances between actual and expected branch
values
– The minimum path value is maintained after comparing incoming
paths.
– Paths are eliminated that are not maintained.
• When all remaining paths use the same branch, update the output
sequence
ECE 6640 42
Properties of Convolutional Codes
• Distance Properties
– If an all zero sequence is input and there is a bit error, how and
how long will it take to return to an all zeros path?
– Find the “minimum free distance”
• The number of code bit errors required before returning
• Note that this is not time steps and not states moved through
• This determines the error correction capability
d 1
t f
2
ECE 6640 44
Computing Distance, Number of Branches,
and Branch Transition caused by a One
• Split the state diagram to start at 00.. and end at 0..
– Show state transitions with the following notations
– D: code bit errors for a path
– L: one factor for every branch
– N: one factor for every branch taken due to a “1” input
• Define the state equations using the state diagram
– Determine the result with the smallest power of D and interpret
– See Figure 7.18 and page p. 412
ECE 6640 45
Performance Bounds
• Upper Bound of bit error probability
dTD, N
PB
dN N 1,D 2 p1 p
D5 N
– for Figure 7.18 and Eq. 7.15 on p. 412 TD, N
1 2 D N
dTD, N D5 D5 N
2 D
dN 1 2 D N 1 2 D N 2
1 2 D N 2 N D D 5
1 2 D N 2
D5
1 2 D N 2
PB
D 5
2 p 1 p
5
ECE 6640
1 2 D N 2 N 1, D 2 p1 p 1 4 p 1 p
2
46
Performance Bounds
• For
EC E k E
r b b
N0 N0 n N0
5 Eb
PB Q exp 5 E b 1
2 N 2
N0 0 Eb
1 2 exp
2 N
0
ECE 6640 47
Coding Gain Bounds
ECE 6640 48
Soft Decision Viterbi
ECE 6640 49
Sequential Decoding
• Complexity
– Viterbi grows exponentially with constraint length
– Sequential is independent of the constraint length
• Can have buffer memory problems at low SNR (many trials)
ECE 6640 50
Feedback Decoding
ECE 6640 51
References
• http://home.netcom.com/~chip.f/viterbi/tutorial.html
• http://www.eccpage.com/
ECE 6640 52
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
Reed-Solomon Codes
ECE 6640 5
R-S Error Probability
p 1 p 2 1 j
1
j
m
PE m
2 1 j t 1 j
ECE 6640 6
Burst Errors
ECE 6640 8
R-S Encoding/Decoding
ECE 6640 9
Reed-Solomon Summary
ECE 6640 10
Interleaving
• Convolutional codes are suitable for memoryless channels
with random error events.
Digital Communications I: Modulation and Coding Course, Period 3 – 2006, Sorour Falahati, Lecture 13
ECE 6640 11
Interleaving …
• Interleaving is done by spreading the coded symbols in
time (interleaving) before transmission.
• The reverse in done at the receiver by deinterleaving the
received sequence.
• “Interleaving” makes bursty errors look like random.
Hence, Conv. codes can be used.
• Types of interleaving:
– Block interleaving
– Convolutional or cross interleaving
Digital Communications I: Modulation and Coding Course, Period 3 – 2006, Sorour Falahati, Lecture 13
ECE 6640 12
Interleaving …
• Consider a code with t=1 and 3 coded bits.
• A burst error of length 3 can not be corrected.
A1 A2 A3 B1 B2 B3 C1 C2 C3
2 errors
Interleaver Deinterleaver
A1 B1 C1 A2 B2 C2 A3 B3 C3 A1 A2 A3 B1 B2 B3 C1 C2 C3
1 errors 1 errors 1 errors
ECE 6640 13
Digital Communications I: Modulation and Coding Course, Period 3 – 2006, Sorour Falahati, Lecture 13
Convolutional Interleaving
ECE 6640 14
Forney Reference
• Forney, G., Jr., "Burst-Correcting
Codes for the Classic Bursty
Channel," Communication
Technology, IEEE Transactions on ,
vol.19, no.5, pp.772,781, October
1971.
ECE 6640 15
Convolutional Example
Channel
Output Outer Inner
Deinterleaver Demodulate
data decoder decoder
ECE 6640 17
Digital Communications I: Modulation and Coding Course, Period 3 – 2006, Sorour Falahati, Lecture 13
Practical example: Compact Disc
“Without error correcting codes, digital audio
would not be technically feasible.”
ECE 6640 19
Compact disc – cont’d
• CIRC encoder and decoder:
Encoder
C2 D* C1 D
interleave encode interleave encode interleave
C2 D* C1 D
deinterleave decode deinterleave decode deinterleave
Decoder
ECE 6640 20
Digital Communications I: Modulation and Coding Course, Period 3 – 2006, Sorour Falahati, Lecture 13
CD Encoder Process
16-bit Left Audio
16-bit Right Audio
(24 byte frame)
RS code 8-bit symbols
RS(255, 251)
24 Used Symbols
227 Unused Symbols
Equ. RS(28, 24)
RS(255, 251)
28 Used Symbols
223 Unused Symbols
Equ. RS(32, 28)
ECE 6640 22
Advanced Topic: Turbo Codes
ECE 6640 23
Turbo Code MATLAB
ECE 6640 24
Turbo Code Performance
ECE 6640 25
MATLAB Simulations
0
LTE Turbo-Coding LTE Turbo-Coding
0
10 10
N = 2048, 1 iterations N = 2048, 2 iterations
-2 -2
10 10
-4 -4
10 10
BER
BER
-6 -6
10 10
-8 -8
10 10
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Eb/N0 (dB) Eb/N0 (dB)
ECE 6640 26
MATLAB Simulations
LTE Turbo-Coding 0
LTE Turbo-Coding
0
10 10
N = 2048, 3 iterations N = 2048, 4 iterations
-2 -2
10 10
-4 -4
10 10
BER
BER
-6 -6
10 10
-8 -8
10 10
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Eb/N0 (dB) Eb/N0 (dB)
ECE 6640 27
References
• http://home.netcom.com/~chip.f/viterbi/tutorial.html
• http://www.eccpage.com/
• http://www.csee.wvu.edu/~mvalenti/turbo.html
• http://www.eg.bucknell.edu/~kozick/elec47601/notes.html
• Digital Communications I: Modulation and Coding Course, Period 3 – 2006,
Sorour Falahati, Lecture 13
ECE 6640 28
ECE 6640
Digital Communications
ECE 6640 2
Sklar’s Communications System
Notes and figures are based on or taken from materials in the course textbook:
ECE 6640 Bernard Sklar, Digital Communications, Fundamentals and Applications, 3
Prentice Hall PTR, Second Edition, 2001.
System Level Tradeoffs
ECE 6640 4
Error Probability Plane
ECE 6640 5
BER vs Eb/No Curves
ECE 6640 6
Nyquist Minimum Bandwidth
ECE 6640 7
Example 9.1: Digital Schemes
• Orthogonal Signaling
– expect improvement in BER as k or M increases
• Non-orthogonal signaling
– expect a decrease in BER as k or M increases
ECE 6640 8
Example 9.1
• Expected trade-offs
• M-FSK
– as M increases, the required transmission bandwidth increases for
minimum frequency spacing.
– to maintain a constant bit rate, the symbol transmission rate
decreases with increasing M
• M-PSK
– while there is degradation as M increases, the symbol transmission
rate may be decreased as M increases
– M-PSK systems plot equal-bandwidth curves, as the bit
transmission rate increases.
ECE 6640 9
Shannon-Hartley Capacity Theorem
ECE 6640 10
Shannon-Hartley Capacity Theorem
C S
log 2 1
W N
1
W S
log 2 1
C N
ECE 6640 11
S-H Equivalent Equations
• For
C
Eb S 1 Rb Eb Rb
2 1
W
N0 N 1 W N0 W
• Letting C = Rb
Eb W W
C C
E C
2 1 b
W
2 1
N0 W N0 C
C Eb C
log 2 1
ECE 6640 W N0 W 12
Shannon Capacity Theorem
E 1
1 b log 2 1 x 1
Eb
N0
log 2 1 x x
1
N0 x
1 lim
x 0
Eb
N0
1
log 2 1 x x b log 2 e
E
N0
Eb 1
0.693 1.6dB
N 0 log 2 e
ECE 6640 13
Shannon Limit
Eb 1
0.693 1.59dB
N 0 log 2 e
• As C/W 0 or W/C∞
• In practice, it is not possible to
reach the bound.
• Provides an improvement bound
for encoding and decoding.
• For example: raw BPSK requires
approximately 9.6 dB Eb/No to
achieve a BER of 10-5 which
suggests that up to an 11.2 dB
improvement is possible.
– Turbu Codes can achieve ~ 10 dB.
ECE 6640 14
Entropy
H p log 2 p 1 p log 2 1 p
ECE 6640 15
Entropy for a Binary System
• The entropy is based on the
probability, p, of an event.
• This can also be looked at as
the randomness of successive
events or how correlated
individual events are.
• Note that maximum entropy is
achieved when the probability
is 50%
– A sample provides no information
about a succeeding sample.
ECE 6640 16
Example 9.2 English Language
ECE 6640 18
Equivocation Example
H X | Y 0.081
ECE 6640 19
Effective Transmission Rate
ECE 6640 20
Pb vs Eb/No Curves
ECE 6640 21
Deriving an Effective Eb/No
Eb E N 0.1
b 0 1.176 0.7dB
N 0 eff H eff 0.085
ECE 6640 22
Bandwidth-Efficieny Plane
ECE 6640 23
Figure 9.6:
Bandwidth-Efficiency Plane
• Factors of note:
– MPSK and QAM nominally
maintain the same bandwidth
will increasing the bits per
symbol and required Eb/No
– MFSK uses an increasing
bandwidth as the bits per
symbol increases while the
Eb/No is decreasing
– BPSK and QPSK have the
same Eb/No but different bits
per symbol
ECE 6640 24
Bit and Symbol Rate Considerations
• For MPSK
R k Rs log 2 M Rs log 2 M Rs
log 2 M
R
1
WIF Rs WIF Rs
Ts
– R/W increases with M
• For MFSK
R k Rs log 2 M Rs R log 2 M Rs log 2 M
M
WIF M Rs WIF M Rs M
Ts
– R/W decreases with M
ECE 6640 25
Bandwidth versus Power
ECE 6640 26
Digital Comm. System Engineering
ECE 6640 27
System Example #1:
Bandwidth Limited
• W = 4000 Hz, Pr/No=53 dB-Hz, R=9600 bps, PB=1e-5
log 2 M
Es Eb
N0 N0
PE M 2 Q 2 s sin
E
N0 M
PE M
PB
ECE 6640 log 2 M 28
System Example #1:
Bandwidth Limited
• W = 4000 Hz, Pr/No=53 dB-Hz, R=9600 bps, PB=1e-5
log 2 M
Pr E E Es Eb
b R s Rs
W 4000 Hz
N0 N0 N0 N0 N0
Pr/No 53 dB‐Hz
PE M
R 9600 bps
PB
PE M 2 Q 2 s sin
Pb 1.00E‐05 BER E
log 2 M
Pr/No 199526.23 Hz N0 M
Eb/No 20.78
Eb/No 13.18 dB
ECE 6640 29
System Example #2:
Power Limited
• W = 45 kHz, Pr/No=48 dB-Hz, R=9600 bps, PB=1e-5
log 2 M
Es Eb
N0 N0
M 1 1 E
PE M exp s
2 2 N0
2 k 1
PB PE M k
2 1
ECE 6640 30
System Example #2:
Power Limited
• W = 45 kHz, Pr/No=48 dB-Hz, R=9600 bps, PB=1e-5
Pr E E
log 2 M
b R s Rs Es Eb
W 45000 Hz N0 N0 N0 N0 N0
Pr/No 48 dB‐Hz
R 9600 bps
Pb 1.00E‐05 BER
M 1 1 Es 2 k 1
Pr/No 63095.73 Hz
PE M exp PB PE M k
2 2 N0 2 1
Eb/No 6.57
Eb/No 8.18 dB
ECE 6640 31
Coded System Example
ECE 6640 32
System Example #3:
Encode-Decode
• W = 4000 Hz, Pr/No=53 dB-Hz, R=9600 bps, PB=1e-9
• Starting with the previous 8-PSK system, we need
additional coding gain
R log 2 M Rs
n
Pr E E E Rc
b R c Rc s Rs k
N0 N0 N0 N0
E k
log 2 M b log 2 M
E s Ec
N0 N0 N0 n
PE M 2 Q 2 s sin
E
N0 M
1 n n j
PB j Pc 1 Pc
n j
PE M n j t 1 j
PC
ECE 6640 log 2 M 33
Solution is Steps
• Step 1: Compute the Es/No
E k
log 2 M b log 2 M
E s Ec
Pr E E E
b R c Rc s Rs
N0 N0 N0 N0 N0 N0 N0 n
n j t 1 j
ECE 6640 34
Excel Computations
• Alternate Approach
– the coding gain formula can be used.
E
G in dB b in dB Eb in dB
N 0 uncoded N 0 coded
ECE 6640 36
QPSK and Offset QPSK
ECE 6640 37
QPSK versus Offset QPSK
• OQPSK makes 90
degree phase transitions
• 180 degrees phase
changes may result in
significant amplitude
variation
ECE 6640 38
Minimum Shift Keying (MSK)
d
st cos 2 f 0 k t xk , k T t k 1 T
4 T
k
xk mod xk 1 d k 1 d k ,2
2
ECE 6640 39
MSK Quadrature Representation
k
xk mod xk 1 d k 1 d k ,2
2
ECE 6640 41
Modulation and Coding for
Bandlimited Channels
• Research Areas (as of 2001 copyright):
– Optimum signal constellation boundaries (choosing a closely
packed signal subset from any regular array or lattice of candidate
points)
– Higher density lattice structures (adding improvement to the signal
subset choice by starting with the densest possible lattice for the
space)
– Trellis-coded modulation (combined modulation and coding
techniques for obtaining coding gain for bandlimited channels).
• Ungerboeck Partitioning
ECE 6640 42
Evolution of Telephone
Modem Standards (1)
• Telephone modems have dealt with the limited power and bandwidth
problem for a considerable time.
• Progress was made at different times for both leased-lines and dial-line
services.
ECE 6640 43
Evolution of Telephone
Modem Standards (2)
• Home modem standards
– Mostly replaced by telephony DSL or cable TV access
ECE 6640 44
Signal Constellation Boundaries
• Various QAM constellations that
have been investigated.
– optimal packing of points with
maximum separation
– reduce maximum amplitude
– optimize PE(M)
ECE 6640 45
Trellis-Coded Modulation (TCM)
ECE 6640 47
ECE 6640
Digital Communications
• Communication Objects
– see MatlabModDemod directory for trial code
– AM Modulation
• comm.PAMModulator
• comm.RectangularQAMModulator
– PM Modulation
• comm.BPSKModulator
• comm.PSKModulator
• comm.DBPSKModulator
• comm.DQPSKModulator
• comm.OQPSKModulator
– FM Modulation
• comm.FSKModulator
ECE 6640 2
MATLAB Simulations
• Communication Objects.
– see MatlabModDemod directory for trial code
– CPM Modulation
• comm.CPFSKModulator
• comm.CPMModulator
• comm.GMSKModulator
• comm.MSKModulator
ECE 6640 3
MATLAB Simulations
• Communication Objects.
– see ViterbiComm directory for demos
– TCM Modulation
• comm.PSKTCMModulator
• comm.RectangularQAMTCMModulator
• comm.GeneralQAMTCMModulator
– Convolutional Coding
• comm.ConvolutionalEncoder
• comm.ViterbiDecoder (Hard and Soft)
ECE 6640 4