ECE 6640

Digital Communications

Dr. Bradley J. Bazuin

Assistant Professor
Department of Electrical and Computer Engineering
College of Engineering and Applied Sciences
 Chapter 2
2. Formatting and Baseband Modulation.
1. Baseband Systems.
2. Formatting Textual Data (Character Coding).
3. Messages, Characters, and Symbols.
4. Formatting Analog Information.
5. Sources of Corruption.
6. Pulse Code Modulation.
7. Uniform and Nonuniform Quantization.
8. Baseband Modulation.
9. Correlative Coding.

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

• Insure that the message is compatible with digital

processing
• Transmit formatting is where the source information is
translated into digital symbols
• When data compression is also employed, the process is
called source coding. (see Chap. 13)

ECE 6640 5
 Baseband Signaling

• Generation of the baseband waveform from the digital

symbols provide by formatting or source coding.
• This could take the form of pulse modulation or pulse code
modulation.
• The baseband signal may be sent using a wired connection
or network to a receiver.

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

• 5-bit coding – Baudot: 32 characters, alphabet plus 6

• 7-bit coding – ASCII: American Standard Code for
Information Interchange
– Originally for telegraph therefore extra fields
• 8-bit coding – EBCDIC: Extended Binary Coded Decimal
Interchange Code
– IBM system
• 16-bit coding – Unicode

• 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

The transmission is a burst occurring at an unknown time but

ECE 6640 9
with known bit periods.
 The EIA-232E Electrical
Specifications

Electrical signal on a pair of wires … signal and ground.

ECE 6640 10
 Messages, Characters and Symbols

• Message is encoded into a sequence of bits

– The bit stream may be a basedband signal
– ASCII can generate a continuous bit stream if “idle characters are
“1’s”

• Grouping of k-bits can be formed into symbols

– M-ary systems use symbols sets where M=2^k
– For k=1, the bit rate and symbol rate are the same
– Defined waveforms represent each of the symbols

• Therefore a message based bit streams can be represented

as a string of Octal or Hex characters in sequences!
ECE 6640 11
– See Text Figure 2.5
 A Review of Sampling Theorem
• We use digital signal processing to transmit and receive all forms of
communications.

• Digital communications inherently describes bit values and symbol

values that “conceptually” exist for a defined period of time and the
“instantaneously” switch to another value.
– The transmitted signals can not physically do this!
– Transmitted signals must exist at defined frequencies and within defined
bandwidths … limited bandwidths often start at baseband.

• 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   Xf   X  f 

X s f    X     X f     d


1 
X s f        n  f s   Xf     d
T
  s k  

1 
X s f        n  f s   Xf     d
Ts k   
1 
ECE 6640 X s f     Xf  n  f s  14
Ts k  
 Fourier Domain

• Spectral replication at steps of fs

• Appears as the convolution of the original spectrum by a
comb waveform spaced as fs
– If the Nyquist rate is not maintained, the convolved elements will
overlap and become distorted

• See Figure 2.6 on p. 64

• Note: Signals are not typically band limited; therefore,

there will be some aliasing whenever sampling is
performed

ECE 6640 15
 Sampling Pulses and Filters

• While Nyquist Theory and Impulse Sampling is

mathematically wonderful ….
– Sampling rates above Nyquist are more practical (Fig. 2.7)
• 2.2 fmax for audio example (20 kHz vs. 44.1 ksps CD rate)
– Impulses must be approximated by signals with real duration and
magnitude (Section 2.4.1.2 Natural Sampling and Fig. 2.8)
• Sample by infinite sequence of rects
• Math equivalent of convolving sampling impulses with rect in time
In frequency, convolve infinite replicas with sinc … amp mod
impulses
• When this sampling signal is used (mult. in time, conv. In freq.)
you get Fig. 2.8

1 
X s  f     ck  X  f  n  f s 
ECE 6640 Ts k   16
 Sample and Hold, Zero Order Hold

• Typical ADCs use a “sample and hold” prior to the ADC

• Sampling is typically an integration of the signal for a
fixed sampling period
• Hold is to insure the ADC has a stable signal for a defined
period of time (conversions time)
  
x sp t   pt   x t   x  t   pt     x t   t  n  Ts 
 n   
 t 
pt   rect 
 Ts 

ECE 6640 17
 ZOF Spectral Domain
1  
X sp f   Pf      X f  n  f s 
 Ts k   
1  
X sp f   Ts  sincf  Ts      X f  n  f s 
 Ts k   


X sp f   sincf  Ts    Xf  n  f 
s
k  

• The spectrum is shaped by the sinc function.

– Note that if spectral analysis is being performed, an inverse sinc
weighting should be applied to “correct” the output.
ECE 6640 18
 Filter and Aliasing

• If a signal is “under-sampled” the output will have spectral

content that is not desired.
– “Engineering Nyquist” 2.2 x fmax
– With digital post filtering, sample at 4 x fmax (or 4.4) and then use
a half-band filter decimator (may be less expensive)
• If additional digital filtering will be employed, aliased
regions of the spectrum may be digitally removed.
– This allows the transition bands to overlap and the stopband to be
placed nearer to the passband edge.

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

If additional digital filtering

will be employed, aliased
regions of the spectrum may be
digitally removed.
This allows the transition
bands to overlap and the
stopband to be placed

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

– See Analog Devices – Radio 101

– http://www.analog.com/static/imported-files/tech_articles/480501640radio101.pdf

• Intersymbol Interference (ISI)

ECE 6640 23
 Quantization Noise

• Round-off Error
– +/- one half of the LSB q
2
– Uniform error distribution about the     pe 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

– Uniform error distribution from one q2


quantized value to the next 12
– Error mean  1/2
– Error Variance  q^2/12
ECE 6640 24
 Quantization Levels

• The levels defined for a typical L-level (2^k=L) ADC

Quantized

qL
rmsVp  
values

ECE 6640 25
 Quantized Peak SNR

• For an L level quantized system, letting power be the

square of one half the rms value of a maximum sine wave
2
  
rmsVp   
2 q L
 
rmsVp 
 2 2  2

• The estimated signal to noise ratio is SNR q 

2
2
 qL 
2 2  q 2  L2 12 3 2
SNR q      2  L
q2 8 q 2
12
For 8-bits or 256 levels SNR q  1.5  256 2  49.9 dB
ECE 6640 Nominally 6 dB per bit 26
 Intersymbol Interference

• 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

• Other classes (ECE6950 MRSP Chap. 4)

– A raised cosine window/filter is a form of Nyquist filtering
• As combined transmitting and receiving filters, each uses a square-
root raised cosine filter.
ECE 6640 27
 2.6 Pulse Code Modulation

• Use the digital word/symbol generated for each character

or ADC value
• Note that the more information or accuracy per symbol, a
higher bit rate is required to maintain the symbol rate
– Fs demands a fixed, constant communication rate
– Text may be sent at any rate that is acceptable (non real-time)

• This can also be referred to as Amplitude Shift Keying

– http://en.wikipedia.org/wiki/Amplitude-shift_keying

• Note that ADC values if sent as pulse would be called

PAM or pulse amplitude modulation
ECE 6640 28
– PAM may be discrete or quantized
 2.7 Uniform vs. Nonuniform

• Uniform (linear) quantizing:

– No assumption about amplitude statistics and
correlation properties of the input.
– Robust to small changes in input statistic by not finely
tuned to a specific set of input parameters
– Simply implemented
• Non-uniform quantizing:
– Using the input statistics to tune quantizer parameters
– Larger SNR than uniform quantizing with same number
of levels
– Non-uniform intervals in the dynamic range with same
ECE 6640 quantization noise variance 29
 Uniform vs. Nonuniform (2)

• Application of linear quantizer:

– Signal processing, graphic and display
applications, process control applications

• Application of non-uniform quantizer:

– Commonly used for speech
– u-law in US, A-law in Europe

ECE 6640 30
 Nonuniform Quantization

• When some portions of the voltage range are not often

used, additional emphasis can be given to those that are.

• μ-law algorithm (North America μ=255)

– http://en.wikipedia.org/wiki/Mu-law

• A-law algorithm (Stnadard Value A=87.6)

– http://en.wikipedia.org/wiki/A-law_algorithm

ECE 6640 31
 Baseband Signaling

• Generation of the baseband waveform from the digital

symbols provide by formatting or source coding.
• This could take the form of pulse modulation or pulse code
modulation.
• The baseband signal may be sent using a wired connection
or network to a receiver.

ECE 6640 32
 PCM Transmission

• Pulse code modulation (PCM) is used when a binary data

stream is to be sent

• In PCM the binary sequence is used to define logical signal

levels for transmission.
– A logical level may map to bits (e.g. 0-High, 1-Low)
– A bit value may define whether a level changes or not
Mark : change whenever the bit is a one
Space: change whenever the bit is a zero
– Period half-cycles can take on various structures based on a bit
value or the sequence of bits
• See Figure 2.22 on p. 87
ECE 6640 33
 PCM Common Waveform Types

• Marks (1’s) and Spaces (0’s)

• Non-return-to-zero (NRZ) – Level, Mark, Space

• Return-to-zero (RZ) – unipolar, bipolar, AMI (alternate
mark inversion)
• Manchester – biphase level, biphase mark, biphase space

ECE 6640 34
 PCM Types

NRZ Bipolar RZ Manchester

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

• M-ary modulation is used when symbol data stream is to

be sent
• M-ary waveform include:
– PAM: Pulse-Amplitude Modulation
– PPM: Pulse-Position Modulation
– PDM: Pulse-Duration Modulation or PWM: Pulse-Width
Modulation

– Multiple “level” can be transmitted as one symbol

• Other M-ary waveforms
– QAM: Quadrature-Amplitude Modulation

ECE 6640 38
 Correlative Codes

• Web site: “Complex technology made real • Complex

communications technology made easy” or
“Complex to Real” by Charan Langton
– http://complextoreal.com/
• Tutorial 16 – Partial Response signaling and Quadrature
Partial Response (QPR) modulation
– http://complextoreal.com/wp-content/uploads/2013/01/qpr.pdf

ECE 6640 39