Modulation, Demodulation and

Coding Course

Period 3 - 2005
Sorour Falahati
Lecture 3
 Last time we talked about:
 Transforming the information source to a
form compatible with a digital system
 Sampling
 Aliasing
 Quantization
 Uniform and non-uniform
 Baseband modulation
 Binary pulse modulation
 M-ary pulse modulation
 M-PAM (M-ay Pulse amplitude modulation)

2005-01-26 Lecture 3 2
 Formatting and transmission of baseband
signal
Digital info. Bit stream Pulse waveforms
(Data bits) (baseband signals)
Textual Format
source info.
Pulse
Analog Sample Quantize Encode modulate
info.

Sampling at rate Encoding each q. value to

f s  1 / Ts l  log 2 L bits
(sampling time=Ts) (Data bit duration Tb=Ts/l)

Quantizing each sampled Mapping every m  log 2 M data bits to a

value to one of the symbol out of M symbols and transmitting
L levels in quantizer. a baseband waveform with duration T

 Information (data) rate: Rb  1 / Tb [bits/sec]

 Symbol rate : R  1 / T [symbols/sec]
 For real time transmission: Rb  mR
2005-01-26 Lecture 3 3
 Qunatization example
amplitude
x(t)
111 3.1867

110 2.2762 Quant. levels

101 1.3657

100 0.4552

011 -0.4552 boundaries

010 -1.3657

001 -2.2762 x(nTs): sampled values

xq(nTs): quantized values
000 -3.1867
Ts: sampling time
PCM t
codeword 110 110 111 110 100 010 011 100 100 011 PCM sequence

2005-01-26 Lecture 3 4
 Example of M-ary PAM
0 Ts 2Ts 4-ary PAM 3B
(rectangular pulse)
2.2762 V 1.3657 V
B
‘11’
0 Tb 2Tb 3Tb 4Tb 5Tb 6Tb T T ‘01’
-B ‘00’ T T

1 1 0 1 0 1 ‘10’
-3B

T
A.
0 T 2T 3T 4T 5T 6T Binary PAM
(rectangular pulse)
‘0’ ‘1’
-A.
T

Assuming real time tr. and equal energy per tr. data bit for
binary-PAM and 4-ary PAM:
•4-ary: T=2Tb and Binay: T=Tb
• 2
0 T 2T 3T A  10B 2

2005-01-26 Lecture 3 5
 Today we are going to talk about:
 Receiver structure
 Demodulation (and sampling)
 Detection
 First step for designing the receiver
 Matched filter receiver
 Correlator receiver
 Vector representation of signals (signal
space), an important tool to facilitate
 Signals presentations, receiver structures
 Detection operations

2005-01-26 Lecture 3 6
 Demodulation and detection
mi Pulse g i (t ) Bandpass si (t ) M-ary modulation
Format
modulate modulate i  1,  , M
channel
transmitted symbol hc (t )
estimated symbol n(t )
Demod.
Format Detect
m̂i z (T ) & sample r (t )

 Major sources of errors:

 Thermal noise (AWGN)
 disturbs the signal in an additive fashion (Additive)
 has flat spectral density for all frequencies of interest (White)
 is modeled by Gaussian random process (Gaussian Noise)
 Inter-Symbol Interference (ISI)
 Due to the filtering effect of transmitter, channel and receiver,
symbols are “smeared”.
2005-01-26 Lecture 3 7
 Example: Impact of the channel

2005-01-26 Lecture 3 8
 Example: Channel impact …

hc (t )   (t )  0.5 (t  0.75T )

2005-01-26 Lecture 3 9
 Receiver job
 Demodulation and sampling:
 Waveform recovery and preparing the
received signal for detection:
 Improving the signal power to the noise power
(SNR) using matched filter
 Reducing ISI using equalizer

 Sampling the recovered waveform

 Detection:
 Estimate the transmitted symbol based on
the received sample

2005-01-26 Lecture 3 10
 Receiver structure

Step 1 – waveform to sample transformation Step 2 – decision making

Demodulate & Sample Detect

z (T ) m̂i
r (t ) Threshold
Frequency Receiving Equalizing
comparison
down-conversion filter filter

For bandpass signals Compensation for

channel induced ISI

Received waveform Baseband pulse

(possibly distored) Baseband pulse Sample
(test statistic)

2005-01-26 Lecture 3 11
 Baseband and bandpass
 Bandpass model of detection process is
equivalent to baseband model because:
 The received bandpass waveform is first
transformed to a baseband waveform.
 Equivalence theorem:
 Performing bandpass linear signal processing
followed by heterodying the signal to the
baseband, yields the same results as
heterodying the bandpass signal to the
baseband , followed by a baseband linear signal
processing.

2005-01-26 Lecture 3 12
 Steps in designing the receiver
 Find optimum solution for receiver design with the
following goals:
1. Maximize SNR
2. Minimize ISI
 Steps in design:
 Model the received signal
 Find separate solutions for each of the goals.
 First, we focus on designing a receiver which maximizes
the SNR.

2005-01-26 Lecture 3 13
 Design the receiver filter to maximize
the SNR
 Model the received signal
si (t ) hc (t ) r (t ) r (t )  si (t ) h c (t )  n(t )

n(t )
AWGN

 Simplify the model:

 Received signal in AWGN
Ideal channels si (t ) r (t ) r (t )  si (t )  n(t )
hc (t )   (t )

n(t )
AWGN

2005-01-26 Lecture 3 14
 Matched filter receiver
 Problem:
 Design the receiver filter h(t ) such that the SNR is
maximized at the sampling time when si (t ), i  1,..., M
is transmitted.
 Solution:
 The optimum filter, is the Matched filter, given by
*
h(t )  hopt (t )  si (T  t )
H ( f )  H opt ( f )  S i ( f ) exp( j 2fT )
*

which is the time-reversed and delayed version of the

conjugate of the transmitted signal
si (t ) h(t )  hopt (t )

0 T t 0 T t
2005-01-26 Lecture 3 15
 Example of matched filter
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

T/2 T t T/2 T t 0 T/2 T 3T/2 2T t

A A  A 2T
2
T T

2005-01-26 Lecture 3 16
 Properties of the matched filter
1. The Fourier transform of a matched filter output with the matched signal as
input is, except for a time delay factor, proportional to the ESD of the input
signal.

Z ( f ) | S ( f ) |2 exp( j 2fT )
2. The output signal of a matched filter is proportional to a shifted version of
the autocorrelation function of the input signal to which the filter is
matched.

z (t )  R (t  T )  z (T )  R (0)  E
3. The output SNR of as matched filter depends sonly on thes ratio of the signal
energy to the PSD of the white noise at the filter input.

S Es
max  
4. N matched-filtering
Two matching conditions inthe T N0 / 2 operation:
 spectral phase matching that gives the desired output peak at time T.
 spectral amplitude matching that gives optimum SNR to the peak value.

2005-01-26 Lecture 3 17
 Correlator receiver
 The matched filter output at the
sampling time, can be realized as the
correlator output.

z (T )  hopt (T )  r (T )
T
  r ( )si ( )d  r (t ), s (t ) 
*

2005-01-26 Lecture 3 18
 Implementation of matched filter
receiver
Bank of M matched filters

z1 (T )
 z1 
*
s (T  t )
1
Matched filter output:
r (t )   z z Observation
vector
 
*  z M 
sM (T  t ) zM (T )

zi  r (t )  s i (T  t ) i  1,..., M
z  ( z1 (T ), z2 (T ),..., z M (T ))  ( z1 , z 2 ,..., z M )

2005-01-26 Lecture 3 19
Implementation of correlator receiver
Bank of M correlators
s 1 (t )
T z1 (T )
0
 z1  Correlators output:
r (t )   z
z Observation
vector
 

s M (t )
T  z M 
0 z M (T )

z  ( z1 (T ), z2 (T ),..., z M (T ))  ( z1 , z 2 ,..., z M )
T
zi   r (t )si (t )dt i  1,..., M
0

2005-01-26 Lecture 3 20
 Example of implementation of
matched filter receivers
s1 (t )
Bank of 2 matched filters
A
T

0 T t A z1 (T )
T  z1 
r (t )
0 T   z
z
s2 (t )  
0 T
 z2 
z 2 (T )
0 T t
A A
T T

2005-01-26 Lecture 3 21
 Signal space
 What is a signal space?
 Vector representations of signals in an N-dimensional
orthogonal space
 Why do we need a signal space?
 It is a means to convert signals to vectors and vice
versa.
 It is a means to calculate signals energy and Euclidean
distances between signals.
 Why are we interested in Euclidean distances
between signals?
 For detection purposes: The received signal is
transformed to a received vectors. The signal which has
the minimum distance to the received signal is
estimated as the transmitted signal.

2005-01-26 Lecture 3 22
 Schematic example of a signal space
 2 (t )
s1  (a11 , a12 )

 1 (t )
z  ( z1 , z 2 )

s 3  (a31 , a32 )

s 2  (a21 , a22 )
s1 (t )  a11 1 (t )  a12 2 (t )  s1  (a11 , a12 )
Transmitted signal
alternatives s2 (t )  a21 1 (t )  a22 2 (t )  s 2  (a21 , a22 )
s3 (t )  a31 1 (t )  a32 2 (t )  s 3  (a31 , a32 )
Received signal at
matched filter output z (t )  z1 1 (t )  z 2 2 (t )  z  ( z1 , z 2 )
2005-01-26 Lecture 3 23
 Signal space
 To form a signal space, first we need to
know the inner product between two
signals (functions):
 Inner (scalar) product:

 x(t ), y (t )  
*
x (t ) y (t )dt

= cross-correlation between x(t) and y(t)
 Properties of inner product:
 ax(t ), y (t )  a  x(t ), y (t ) 
 x(t ), ay (t )  a *  x(t ), y (t ) 
 x(t )  y (t ), z (t )  x(t ), z (t )    y (t ), z (t ) 
2005-01-26 Lecture 3 24
 Signal space – cont’d
 The distance in signal space is measure by
calculating the norm.
 What is norm?
 Norm of a signal:


2
x(t )   x(t ), x(t )   x(t ) dt  E x

= “length” of x(t)
ax(t )  a x(t )
 Norm between two signals:

d x , y  x(t )  y (t )
 We refer to the norm between two signals as the
Euclidean distance between two signals.

2005-01-26 Lecture 3 25
 Example of distances in signal space
 2 (t )
s1  (a11 , a12 )

E1 d s1 , z
 1 (t )
E3 z  ( z1 , z 2 )
d s3 , z E2 d s2 , z
s 3  (a31 , a32 )

s 2  (a21 , a22 )
The Euclidean distance between signals z(t) and s(t):
d si , z  si (t )  z (t )  (ai1  z1 ) 2  (ai 2  z 2 ) 2
i  1,2,3

2005-01-26 Lecture 3 26
 Signal space - cont’d
 N-dimensional orthogonal signal space is characterized by N linearly
independent functions called basis functions. The basis
functions must satisfy the orthogonality condition
 (t )
j
N
j 1

where
T
0t T
     i is orthonormal.
  j (t ) dt  K i ji
*
(t ), (t ) (
If all i , thej signal space
t )

0
j , i  1,..., N

1  i  j
 ij  
0  i  j
Orthonormal basis
K i procedure
 Gram-Schmidt 1

2005-01-26 Lecture 3 27
 Example of an orthonormal basis
functions
 Example: 2-dimensional orthonormal signal space
 2

 1 (t )  cos(2t / T ) 0t T  2 (t )
 T

 (t )  2 sin(2t / T ) 0t T
 2 T
 1 (t )
T 0
  1 (t ), 2 (t )   1 (t ) 2 (t )dt  0
0

 1 (t )   2 (t )  1

 Example: 1-dimensional orthonornal signal space

 1 (t )
1  1 (t )  1
T  1 (t )
0
0 T t
2005-01-26 Lecture 3 28
 Signal space – cont’d
 
Any arbitrary finite set of waveforms i i 1
s (t ) M

where each member of the set is of duration T,

can be expressed as a linear combination of N
  N
orthonogal waveforms j j 1 where N  M .
(t )
N
si (t )   aij j (t ) i  1,..., M
j 1 NM
where
T
1 1 j  1,..., N
aij 
Kj
 si (t ), j (t )  
Kj 0
si (t ) *j (t )dt
i  1,..., M
0t T

N
Ei   K j aij
2
s i  (ai1 , ai 2 ,..., aiN )
j 1
Vector representation of waveform Waveform energy
2005-01-26 Lecture 3 29
 Signal space - cont’d
N
si (t )   aij j (t ) s i  (ai1 , ai 2 ,..., aiN )
j 1
Waveform to vector conversion Vector to waveform conversion

 1 (t )  1 (t )
T ai1 ai1
0
 ai1  sm
 ai1 
si (t )     sm   si (t )
 N (t )      N (t )
T aiN  aiN 
0 aiN aiN

2005-01-26 Lecture 3 30
 Example of projecting signals to an
orthonormal signal space
 2 (t )
s1  (a11 , a12 )

 1 (t )

s 3  (a31 , a32 )

s 2  (a21 , a22 )

s1 (t )  a11 1 (t )  a12 2 (t )  s1  (a11 , a12 )

Transmitted signal
alternatives s2 (t )  a21 1 (t )  a22 2 (t )  s 2  (a21 , a22 )
s3 (t )  a31 1 (t )  a32 2 (t )  s 3  (a31 , a32 )
T
aij   si (t ) j (t )dt j  1,..., N i  1,..., M 0t T
2005-01-26 0 Lecture 3 31
 Signal space – cont’d
 To find an orthonormal basis functions for a given
set of signals, Gram-Schmidt procedure can be used.
 Gram-Schmidt procedure:
 Given a signal set , compute an orthonormal basis
1. Define  si (t ) iM1
2. For  j (t ) Njcompute
1
If let
 1 (t )  s1 (t ) / E1  s1 (t ) / s1 (t ) i 1
If , do not assign any basis function.
i  2,..., M d i (t )  si (t )    si (t ), j (t )   j (t )
3. Renumber the basis functions such that basis is
d i (t )  0  i (t )  d i (t ) / d i (t ) j 1

d i (t )  0

 This is only necessary if for any i in step 2.

 Note
 1 (t ), 2 (t ),..., N (t )
that

d i (t )  0
NM
2005-01-26 Lecture 3 32
 Example of Gram-Schmidt procedure
 Find the basis functions and plot the signal space for
the following transmitted signals:
s1 (t ) s2 (t )
A
T 0 T t
A
0 T t T

 Using Gram-Schmidt procedure:

T  1 (t ) s1 (t )  A 1 (t )
1 E1  0 s1 (t ) dt  A
2 2
1 s2 (t )   A 1 (t )
T
 1 (t )  s1 (t ) / E1  s1 (t ) / A s1  ( A) s 2  ( A)
T
2  s2 (t ), 1 (t )   s2 (t ) 1 (t )dt   A 0 T t
0
s2 s1
d 2 (t )  s2 (t )  ( A) 1 (t )  0  1 (t )
-A 0 A
2005-01-26 Lecture 3 33
Implementation of matched filter receiver

Bank of N matched filters

z1
 (T  t )
 Observation
1
 z1  vector
r (t )
  z
z
 
  N (T  t )  z N 
zN
N
si (t )   aij j (t ) i  1,..., M
j 1
z  ( z1 , z 2 ,..., z N ) NM
z j  r (t )  j (T  t ) j  1,..., N

2005-01-26 Lecture 3 34
 Implementation of correlator receiver
Bank of N correlators

 1 (t )
T z1
0
 r1 
r (t )   z
z Observation
vector
 N (t )  
T rN 
0 zN
N
si (t )   aij j (t ) i  1,..., M
j 1
z  ( z1 , z 2 ,..., z N ) NM
T
z j   r (t ) j (t )dt j  1,..., N
0
2005-01-26 Lecture 3 35
 Example of matched filter receivers using
basic functions
s1 (t ) s2 (t )  1 (t )
A 1
T T
0 T t
0 T t A 0 T t
T

1 matched filter
 1 (t )
1
r (t ) T z1
 z1   z z

0 T t

er of matched filters (or correlators) is reduced by 1 compared to using matched filters (correlators) to the transmitted signal.

d number of filters (or correlators)

2005-01-26 Lecture 3 36
White noise in orthonormal signal space
 AWGN n(t) can be expressed as
n(t )  nˆ (t )  n~ (t )
Noise projected on the signal space Noise outside on the signal space
which impacts the detection process.

N
nˆ (t )   n j j (t ) Vector representation of nˆ (t )
j 1
n j  n(t ), j (t )  j  1,..., N n  (n1 , n2 ,..., nN )

 n~ (t ), (t )  0 j  1,..., N n  N
independent zero-mean
j j 1
j Gaussain random variables with
variance var(n j )  N 0 / 2

2005-01-26 Lecture 3 37
Digital Communications I: Modulation
and Coding Course

Period 3 - 2007
Catharina Logothetis
Lecture 4
Last time we talked about:
 Receiver structure
 Impact of AWGN and ISI on the transmitted
signal
 Optimum filter to maximize SNR
 Matched filter receiver and Correlator receiver

Lecture 4 39
Receiver job
 Demodulation and sampling:
 Waveform recovery and preparing the received
signal for detection:
 Improving the signal power to the noise power (SNR)
using matched filter
 Reducing ISI using equalizer
 Sampling the recovered waveform
 Detection:
 Estimate the transmitted symbol based on the
received sample

Lecture 4 40
 Receiver structure

Step 1 – waveform to sample transformation Step 2 – decision making

Demodulate & Sample Detect

z (T ) m̂i
r (t ) Frequency Receiving Equalizing Threshold
down-conversion filter filter comparison

For bandpass signals Compensation for

channel induced ISI

Received waveform Baseband pulse

Baseband pulse Sample
(possibly distored)
(test statistic)

Lecture 4 41
Implementation of matched filter receiver

Bank of M matched filters

z1 (T )
 z1 
*
s (T  t )
1
Matched filter output:
r (t )   z z Observation
  vector

*  z M 
sM (T  t ) zM (T )

zi  r (t )  s i (T  t ) i  1,..., M
z  ( z1 (T ), z2 (T ),..., z M (T ))  ( z1 , z 2 ,..., z M )

Lecture 4 42
Implementation of correlator receiver
Bank of M correlators
s 1 (t )
T z1 (T )
0
 z1  Correlators output:
r (t )   z
z Observation
 
 vector
s M (t )
T  z M 
0 z M (T )

z  ( z1 (T ), z2 (T ),..., z M (T ))  ( z1 , z 2 ,..., z M )
T
zi   r (t )si (t )dt i  1,..., M
0

Lecture 4 43
Today, we are going to talk about:

 Detection:
 Estimate the transmitted symbol based on the
received sample
 Signal space used for detection
 Orthogonal N-dimensional space
 Signal to waveform transformation and vice versa

Lecture 4 44
Signal space
 What is a signal space?
 Vector representations of signals in an N-dimensional
orthogonal space
 Why do we need a signal space?
 It is a means to convert signals to vectors and vice versa.
 It is a means to calculate signals energy and Euclidean
distances between signals.
 Why are we interested in Euclidean distances between
signals?
 For detection purposes: The received signal is transformed to
a received vectors. The signal which has the minimum
distance to the received signal is estimated as the transmitted
signal.

Lecture 4 45
 Schematic example of a signal space
 2 (t )
s1  (a11 , a12 )

 1 (t )
z  ( z1 , z 2 )

s 3  (a31 , a32 )

s 2  (a21 , a22 )
s1 (t )  a11 1 (t )  a12 2 (t )  s1  (a11 , a12 )
Transmitted signal
s2 (t )  a21 1 (t )  a22 2 (t )  s 2  (a21 , a22 )
alternatives
s3 (t )  a31 1 (t )  a32 2 (t )  s 3  (a31 , a32 )
Received signal at z (t )  z1 1 (t )  z 2 2 (t )  z  ( z1 , z 2 )
matched filter output
Lecture 4 46
Signal space
 To form a signal space, first we need to know
the inner product between two signals
(functions):
 Inner (scalar) product:

 x(t ), y (t )  
*
x (t ) y (t )dt

= cross-correlation between x(t) and y(t)

 Properties of inner product:

 ax(t ), y (t )  a  x(t ), y (t ) 
 x(t ), ay (t )  a *  x(t ), y (t ) 
 x(t )  y (t ), z (t )  x(t ), z (t )    y (t ), z (t ) 
Lecture 4 47
Signal space …
 The distance in signal space is measure by calculating
the norm.
 What is norm?
 Norm of a signal:


2
x(t )   x(t ), x(t )   x(t ) dt  E x

= “length” of x(t)
ax(t )  a x(t )
 Norm between two signals:

d x , y  x(t )  y (t )
 We refer to the norm between two signals as the
Euclidean distance between two signals.
Lecture 4 48
Example of distances in signal space
 2 (t )
s1  (a11 , a12 )

E1 d s1 , z
 1 (t )
E3 z  ( z1 , z2 )
d s3 , z E2 d s2 , z
s 3  (a31 , a32 )

s 2  (a21 , a22 )

The Euclidean distance between signals z(t) and s(t):

d si , z  si (t )  z (t )  (ai1  z1 ) 2  (ai 2  z 2 ) 2
i  1,2,3

Lecture 4 49
Orthogonal signal space
 N-dimensional orthogonal signal space is characterized by
N linearly independent functions  j (t ) j 1 called basis
N

functions. The basis functions must satisfy the orthogonality

condition
T
0t T
  i (t ), j (t )   i (t ) (t )dt  K i ji
*
j
0
j , i  1,..., N
where
1  i  j
 ij  
0  i  j

 If all K i  1, the signal space is orthonormal.

Lecture 4 50
 Example of an orthonormal bases
 Example: 2-dimensional orthonormal signal space
 2

 1 (t )  cos(2t / T ) 0t T  2 (t )
 T

 (t )  2 sin(2t / T ) 0t T
 2 T
T
 1 (t )
0
  1 (t ), 2 (t )   1 (t ) 2 (t )dt  0
0

 1 (t )   2 (t )  1
 Example: 1-dimensional orthonornal signal space

 1 (t )
1  1 (t )  1
T  1 (t )
0
0 T t
Lecture 4 51
 Signal space …
 Any arbitrary finite set of waveforms  si (t ) iM1
where each member of the set is of duration T, can be
expressed as a linear combination of N orthonogal
 
waveforms  j (t ) jwhere
N
1
N  M.

N
si (t )   aij j (t ) i  1,..., M
j 1 NM
where
T
1 1 j  1,..., N
aij 
Kj
 si (t ), j (t )  
Kj 0
si (t ) *j (t )dt
i  1,..., M
0t T

N
Ei   K j aij
2
s i  (ai1 , ai 2 ,..., aiN )
j 1
Vector representation of waveform Waveform energy
Lecture 4 52
 Signal space …
N
si (t )   aij j (t ) s i  (ai1 , ai 2 ,..., aiN )
j 1
Waveform to vector conversion Vector to waveform conversion

 1 (t )  1 (t )
T ai1 ai1
0
 ai1  sm
 ai1 
si (t )     sm   si (t )
 N (t )      N (t )
T aiN  aiN 
0 aiN aiN

Lecture 4 53
Example of projecting signals to an
orthonormal signal space
 2 (t )
s1  (a11 , a12 )

 1 (t )

s 3  (a31 , a32 )

s 2  (a21 , a22 )

s1 (t )  a11 1 (t )  a12 2 (t )  s1  (a11 , a12 )

Transmitted signal
alternatives s2 (t )  a21 1 (t )  a22 2 (t )  s 2  (a21 , a22 )
s3 (t )  a31 1 (t )  a32 2 (t )  s 3  (a31 , a32 )
T
aij   si (t ) j (t )dt j  1,..., N i  1,..., M 0t T
0 Lecture 4 54
 Signal space – cont’d
 To find an orthonormal basis functions for a given
set of signals, Gram-Schmidt procedure can be
used.
 Gram-Schmidt procedure:
Given a signal set  si (t ) i ,1 compute an orthonormal basis  j (t ) j 1
M N


1. Define  1 (t )  s1 (t ) / E1  s1 (t ) / s1 (t ) i 1
2. For i  2,..., M compute d i (t )  si (t )    si (t ), j (t )   j (t )
If d i (t )  0 let  i (t )  d i (t ) / d i (t )
j 1

If d i (t )  ,0do not assign any basis function.

1. Renumber the basis functions such that basis is
 1 (t ), 2 (t ),..., N (t )
 This is only necessary if d i (t )  0 for any i in step 2.
 Note that N  M

Lecture 4 55
 Example of Gram-Schmidt procedure
 Find the basis functions and plot the signal space for the
following transmitted signals:
s1 (t ) s2 (t )
A
T 0 T t
A
0 T t T

 Using Gram-Schmidt procedure:

T  1 (t ) s1 (t )  A 1 (t )
1 E1  0 s1 (t ) dt  A
2 2
1 s2 (t )   A 1 (t )
T
 1 (t )  s1 (t ) / E1  s1 (t ) / A s1  ( A) s 2  ( A)
T
2  s2 (t ), 1 (t )   s2 (t ) 1 (t )dt   A 0 T t
0
s2 s1
d 2 (t )  s2 (t )  ( A) 1 (t )  0  1 (t )
-A 0 A
Lecture 4 56
Implementation of matched filter receiver

Bank of N matched filters

z1
 (T  t )

Observation
1
 z1  vector
r (t )
  z
z
 
  N (T  t )  z N 
zN
N
si (t )   aij j (t ) i  1,..., M
j 1
z  ( z1 , z 2 ,..., z N ) NM
z j  r (t )  j (T  t ) j  1,..., N

Lecture 4 57
Implementation of correlator receiver
Bank of N correlators

 1 (t )
T z1
0
 r1 
r (t )   z
z Observation
 N (t )   vector
T rN 
0 zN
N
si (t )   aij j (t ) i  1,..., M
j 1
z  ( z1 , z 2 ,..., z N ) NM
T
z j   r (t ) j (t )dt j  1,..., N
0
Lecture 4 58
 Example of matched filter receivers using
basic functions
s1 (t ) s2 (t )  1 (t )
A 1
T T
0 T t
0 T t A 0 T t
T

1 matched filter
 1 (t )
1
r (t ) T z1
 z1   z z

0 T t

 Number of matched filters (or correlators) is reduced by 1 compared to using

matched filters (correlators) to the transmitted signal.

Lecture 4 59
White noise in orthonormal signal space
 AWGN n(t) can be expressed as
n(t )  nˆ (t )  n~ (t )
Noise projected on the signal space Noise outside on the signal space
which impacts the detection process.
N
nˆ (t )   n j j (t ) nˆ (t )
Vector representation of
j 1
n j  n(t ), j (t )  j  1,..., N n  (n1 , n2 ,..., nN )

 n~ (t ), (t )  0
j
j  1,..., N n  N
independent zero-mean
j j 1
Gaussain random variables with
variance var(n j )  N 0 / 2

Lecture 4 60
S-72.227 Digital Communication Systems

Spread spectrum and

Code Division Multiple Access (CDMA)
communications
 Spread Spectrum Communications - Agenda Today

 Basic principles and block diagrams of spread spectrum communication

systems
 Characterizing concepts
 Types of SS modulation: principles and circuits
– direct sequence (DS)
– frequency hopping (FH)
 Error rates
 Spreading code sequences; generation and properties
– Maximal Length (a linear, cyclic code)
– Gold
– Walsh
 Asynchronous CDMA systems

Timo O. Korhonen, HUT Communication Laboratory 62

 How Tele-operators* Market CDMA

Coverage Capacity Cost

$ $
For Coverage, CDMA saves CDMA’s capacity supports at A carrier who deploys CDMA
wireless carriers from deploying least 400% more revenue-producing instead of GSM will have
the 400% more cell site that subscribers in the same spectrum a lower capital cost
are required by GSM when compared to GSM

Clarity Choice Customer satisfaction

CDMA with PureVoice CDMA offers the choice of simultaneous The Most solid foundation for
provides wireline clarity voice, async and packet data, FAX, and attracting and retaining subscriber
SMS. is based on CDMA

Timo O. Korhonen, HUT Communication Laboratory *From Samsumg’s narrowband CDMA (CDMAOne®) marketing (2001) 63
 Direct Sequence Spread Spectrum (DS-SS)
 This figure shows BPSK-DS transmitter and receiver
(multiplication can be realized by RF-mixers)
spreading

AA2 2
PP av  AA 22PPav
av  2  av
2

DS-CDMA is used in WCDMA, cdma2000 and IS-95 systems

Timo O. Korhonen, HUT Communication Laboratory 64
 Characteristics of Spread Spectrum
 Bandwidth of the transmitted signal W is much greater than the original
message bandwidth (or the signaling rate R)
 Transmission bandwidth is independent of the message. Applied code is
known both to the transmitter and receiver

Narrow band signalWideband signal

 (data)
Interference and noise (transmitted
immunity of SS system isSS signal)
larger, the larger the
processing gain
 Multiple SS systemsLc can / R  Tb in
W co-exist / Tcthe same band (=CDMA). Increased user
independence (decreased interference) for (1) higher processing gain and
higher (2) code orthogonality
 Spreading sequence can be very long -> enables low transmitted PSD-> low
probability of interception (especially in military communications)

Timo O. Korhonen, HUT Communication Laboratory 65

 Characteristics of Spread Spectrum (cont.)
 Processing gain, in general
Lc  W / R  (1/ Tc ) /(1/ Tb )  Tb / Tc , Lc , dB  10 log10 ( Lc )
– Large Lc improves noise immunity, but requires a larger
transmission bandwidth
– Note that DS-spread spectrum is a repetition FEC-coded systems
 Jamming margin
M J  Lc  [ Lsys  ( SNR ) desp ]
– Tells the magnitude of additional interference and noise that can be
injected to the channel without hazarding system operation.
Example:
Lc  30dB,available processing gain
Lsys  2dB, margin for system losses
SNRdesp  10dB, required SNR after despreading (at the RX)
 M j  18dB,additional interference and noise can deteriorate
received SNR by this amount
Timo O. Korhonen, HUT Communication Laboratory 66
 Characteristics of Spread Spectrum (cont.)
 Spectral efficiency Eeff: Describes how compactly TX signal fits into the
transmission band. For instance for BPSK with some pre-filtering:
Eeff  Rb / BT  Rb / BRF Lc  Tb / Tc  Lc / Tb  1/ Tc

BRF , filt  BRF , filt : bandwidth for polar mod.

1/ Tc Lc 
BRF   
k log 2 M Tb log 2 M  M : number of levels
k : number of bits
Rb 1 Tb log 2 M log 2 M 
 Eeff   
BRF Tb Lc Lc M 2 k
 k  log 2 M 

 Energy efficiency (reception sensitivity): The value of  b  Eb / N 0

to obtain a specified error rate (often 10-9). For BPSK the error rate is

1
pe  Q( 2 b ), Q (k ) 
2 k 
exp( 2 / 2)d 

 QPSK-modulation can fit twice the data rate of BPSK in the same
bandwidth. Therefore it is more energy efficient than BPSK.
Timo O. Korhonen, HUT Communication Laboratory 67
 A QPSK-DS Modulator
q
d (t ) s (t )
S/P 2 P sin  ot c2 (t )

i
2 P cos  ot c1 (t )
QPSK-modulator Constellation
diagram
 After serial-parallel conversion (S/P) data modulates the orthogonal carriers
 Modulation2on P cos(  o t ) and
orthogonal carriers sin( ot )by codes c1 and c2
2 Pspreaded
 Spreading codes c1 and c2 may or may not be orthogonal (System performance is
independent of their orthogonality, why?)
 What kind of circuit can make the demodulation (despreading)?

Timo O. Korhonen, HUT Communication Laboratory 68

 DS-CDMA (BPSK) Spectra (Tone Jamming)
 Assume DS - BPSK transmission, with a single tone jamming (jamming
power J [W] ). The received signal is
r (t )  2 Pc1 (t  Td )cos   0t   d (t )   2 J cos   0t   '
 The respective PSD of the received chip-rate signal is
1 1
S r ( f )  PTc sinc 2  f  f 0  Tc   PTc sinc 2  f  f 0  Tc 
2 2
1
 J   ( f  f 0 )   ( f  f 0 ) Spreading of jammer power
2
 At the receiver r(t) is multiplied with the local code c(t) (=despreading)
d (t )  2 Pc1 (t  Td )c(t  Tˆd ) cos   0t   d (t )  data
 2 J c(t  Tˆd ) cos   0t   '
 The received signal and the local code are phase-aligned:
1 1
c1 (t  Tˆd )c(t  Td )  1  S d ( f )  PTb sinc 2  f  f 0  Tb   PTb sinc 2  f  f 0  Tb 
2 2
Data spectra 1 1
after phase modulator  JT c sinc 2

  f  f 0  Tc 
  JTc sinc 2
 f  f 0  Tc 
2           2        
Timo O. Korhonen, HUT Communication Laboratory F  2 J c (t Tˆd )cos 0t  '  69
 Tone Jamming (cont.)
 Despreading spreads the jammer power and despreads the signal power:

Timo O. Korhonen, HUT Communication Laboratory 70

 Tone Jamming (cont.)
 Filtering (at the BW of the phase modulator) after despreading
suppresses the jammer power:

Timo O. Korhonen, HUT Communication Laboratory 71

 Error Rate of BPSK-DS System*
 DS system is a form of coding, therefore number chips, eg code weight
determines, from its own part, error rate (code gain)
 Assuming that the chips are uncorrelated, prob. of code word error for a binary-
block coded BPSK-DS system with code weight w is therefore

 2 Eb 
Pe  Q  Rc wm  , Rc  k / n ( code rate)
 N0 
 This can be expressed in terms of processing gain Lc by denoting the average
signal and noise power by , respectively, yielding
Pav , N av
Eb  PavTb , N 0  N avTc 
 2 PavTb   2 Pav 
Pe  Q  Rc wm   Q  Lc Rc wm 
 Note that the symbol error rate is upper bounded due to repetition code nature of
 N avTc   N av 
the DS by

n m n
   of erroneous
Pesnumber
where t denotes the m
p (1  p ) nbits   1can
, t that ( dbe  1) 
min corrected in the coded
m  t 1  m  2
word

Timo O. Korhonen, HUT Communication Laboratory *For further background, see J.G.Proakis: 72
Digital Communications (IV Ed), Section 13.2
 Example: Error Rate of Uncoded Binary BPSK-DS

 For uncoded DS w=n, thus Rc w  (1/ n)n  1 and

 2 Eb   2 Eb 
Pe  Q  Rc wm   Q  
 N 0   N 0 

 We note that Eb  PavTb  Pav / Rb and J 0  J av / W yielding

Eb Pav / R W /R
 
J 0 J av / W J av / Pav

 2W / R 
 Pe  Q  
J / P
 av av 
 Therefore, we note that increasing system processing gain W/R, error
rate can be improved

Timo O. Korhonen, HUT Communication Laboratory 73

 Code Generation in DS-SS
chip interval maximal length (ML)
spreading code

DS modulator Spreading sequence period

ML code generator
delay elements (D-flip-flops) -> XOR - circuit

- code determined by feedback taps

- code rate determined by clock rate 74
Timo O. Korhonen, HUT Communication Laboratory
 Some Cyclic Block Codes
 (n,1) Repetition codes. High coding gain, but low rate
 (n,k) Hamming codes. Minimum distance always 3. Thus can detect 2
errors and correct one error. n=2m-1, k = n - m, m  3
 Maximum-length codes. For every integer k  3 there exists a
maximum length code (n,k) with n = 2k - 1,dmin = 2k-1. Hamming codes
are dual1 of of maximal codes.
 BCH-codes. For every integer m  3 there exist a code with n = 2m-1,
k  n  mt and d min  2t  1 where t is the error correction capability
 (n,k) Reed-Solomon (RS) codes. Works with k symbols that consist of
m bits that are encoded to yield code words of n symbols. For these
codes n  2  1, number of check symbols n  k  2t and d min  2t  1
m

 Nowadays BCH and RS are very popular due to large dmin, large number
of codes, and easy generation
 For further code references have a look on self-study material!

Timo O. Korhonen, HUT Communication Laboratory

1: Task: find out from net what is meant by dual codes! 75
 Maximal Length Codes

autocorrelation

power spectral density

Timo O. Korhonen, HUT Communication Laboratory 76

 Maximal Length Codes (cont.)
 Have very good autocorrelation but cross correlation not granted
 Are linear,cyclic block codes - generated by feedbacked shift registers
 Number of available codes* depends on the number of shift register
stages: 5 stages->6 codes, 10 stages ->60 codes, 25 stages ->1.3x106 codes
 Code generator design based on tables showing tap feedbacks:

Timo O. Korhonen, HUT Communication Laboratory 77

*For the formula see: Peterson, Ziemer: “Introduction to Spread Spectrum Communication”, p. 121
 Design of Maximal Length Generators
by a Table Entry
 Feedback connections can be written directly from the table:

Timo O. Korhonen, HUT Communication Laboratory 78

 Other Spreading Codes
 Walsh codes: Orthogonal, used in synchronous systems, also in WCDMA
downlink
H Hn 1 
 Generation recursively: H 0  [0] Hn   n 1 
 Hn 1 H n 1  0 0 0 0
0 1 0 1
 All rows and columns of the matrix are orthogonal: H2   
0 0 1 1
 (1)(1)  (1)1  1(1)  1  1  0  
0 1 1 0
 Gold codes: Generated by summing preferred pairs of maximal length
codes. Have a guarantee 3-level crosscorrelation:  t (n) / N ,1/ N ,(t(n)  2) / N 
 For N-length code there exists N + 2 codes in a code family and
1  2( n 1) / 2 ,for n odd
N  2  1 and t (n)  
n
( n  2) / 2
1  2 ,for n even
 Walsh and Gold codes are used especially in multiple access systems
 Gold codes are used in asynchronous communications because their
crosscorrelation is quite good as formulated above
Timo O. Korhonen, HUT Communication Laboratory 79
 Frequency Hopping Transmitter and Receiver
 In FH-SS hopping frequencies are determined by the code and the
message (bits) are usually non-coherently FSK-modulated

BW  Wd BW  Ws

BW  Ws BW  Wd

 This method is applied in BlueTooth®

Timo O. Korhonen, HUT Communication Laboratory 80

 Frequency Hopping Spread Spectrum (FH-SS)
(example: transmission of two symbols/chip)

4-level FSK modulation

Hopped frequency
slot determined by
hopping code

2 L levels
2k slots

Wd  2 L f d (  data modulator BW)

Tc :chip duration
Ws  2 Wd ( total FH spectral width) T : bit duration
k
Tb b

Ts : symbol duration
  
L2
Timo O. Korhonen, HUT Communication Laboratory 81
 Error Rate in Frequency Hopping
 If there are multiple hops/symbol we have a fast-hopping system. If
there is a single hop/symbol (or below), we have a slow-hopping
system.
 For slow-hopping non-coherent FSK-system, binary error rate is
Pe  1 exp   b / 2  ,  b  Eb / N 0
2
and the respective symbol error rate is (hard-decisions)
Pes  1 exp   b Rc / 2  , Rc  k / n  1
2
 A fast-hopping FSK system is a diversity-gain system. Assuming non-
coherent, square-law combining of respective output signals from
matched filters yields the binary error rate (with L hops/symbol)
1
exp   b / 2   i 0 K i   b / 2  ,  b  L c
L 1 i
Pe  2 L 1
2
1 L 1i  2 L  1
Ki   r 0  
i!  r 

(For further details, see J.G.Proakis: Digital Communications (IV Ed), Section 13.3 )
Timo O. Korhonen, HUT Communication Laboratory 82
 DS and FH compared
 FH is applicable in environments where there exist tone jammers that
can be overcame by avoiding hopping on those frequencies
 DS is applicable for multiple access because it allows statistical
multiplexing (resource reallocation) to other users (power control)
 FH applies usually non-coherent modulation due to carrier
synchronization difficulties -> modulation method degrades performance
 Both methods were first used in military communications,
c  10 2...10 7
– FH can be advantageous because the hopping span canLbe very large
(makes eavesdropping difficult)
– DS can be advantageous because spectral density can be much
smaller than background noise density (transmission is unnoticed)
 FH is an avoidance system: does not suffer on near-far effect!
 By using hybrid systems some benefits can be combined: The system
can have a low probability of interception and negligible near-far effect
at the same time. (Differentially coherent modulation is applicable)

Timo O. Korhonen, HUT Communication Laboratory 83

 Multiple access: FDMA, TDMA and CDMA

•FDMA, TDMA and CDMA yield

conceptually the same capacity
• However, in wireless communications
CDMA has improved capacity due to
• statistical multiplexing
• graceful degradation
•Performance can still be improved by
adaptive antennas, multiuser detection,
FEC, and multi-rate encoding
Timo O. Korhonen, HUT Communication Laboratory 84
 Example of DS multiple access waveforms

channel->

polar sig.->

detecting A ... ->

Timo O. Korhonen, HUT Communication Laboratory 85

 FDMA, TDMA and CDMA compared (cont.)
 TDMA and FDMA principle:
– TDMA allocates a time instant for a user
– FDMA allocates a frequency band for a user
– CDMA allocates a code for user
 CDMA-system can be synchronous or asynchronous:
– Synchronous CDMA can not be used in multipath channels that destroy
code orthogonality
– Therefore, in wireless CDMA-systems as in IS-95,cdma2000,
WCDMA and IEEE 802.11 user are asynchronous
 Code classification:
– Orthogonal, as Walsh-codes for orthogonal or
near-orthogonal systems
– Near-orthogonal and non-orthogonal codes:
 Gold-codes, for asynchronous systems

 Maximal length codes for asynchronous systems

Timo O. Korhonen, HUT Communication Laboratory 86

 Capacity of a cellular CDMA system
 Consider uplink (MS->BS)
 Each user transmits
Gaussian noise (SS-signal) whose
deterministic characteristics
are stored in RX and TX
 Reception and transmission
are simple multiplications
 Perfect power control: each
user’s power at the BS the same
 Each user receives multiple copies of power Pr that is other user’s
interference power, therefore each user receives the interference power
I k  (U  1) Pr (1)
where U is the number of equal power users

Timo O. Korhonen, HUT Communication Laboratory 87

 Capacity of a cellular CDMA system (cont.)
 Each user applies a demodulator/decoder characterized by a certain
reception sensitivity Eb/Io (3 - 9 dB depending on channel coding,
channel, modulation method etc.)
 Each user is exposed to the interference power density (assumed to be
produced by other users only) I 0  I k / BT [W/Hz] (2)
where BT is the spreading (and RX) bandwidth
 Received signal energy / bit at the signaling rate R is
Eb  Pr / R [ J ]  [W ][ s ] (3)
 Combining (1)-(3) yields the number of users
I I B
I k  (U  1) Pr  U  1  k  o T 
 1/ R  BT W / R
 (4)
Pr Eb R Eb  1/ I 0  Eb / I 0
 This can still be increased by using voice activity coefficient Gv = 2.67
(only about 37% of speech time effectively used), directional antennas,
for instance for a 3-way antenna GA = 2.5.

Timo O. Korhonen, HUT Communication Laboratory 88

 Capacity of a cellular CDMA system (cont.)
 In cellular system neighboring cells introduce interference that
decreases capacity. It has been found out experimentally that this
reduces the number of users by the factor
1  f  1.6
 Hence asynchronous CDMA system capacity can be approximated by
W / R Gv GA
U
Eb / I o 1  f
yielding with the given values Gv=2.67, GA=2.4, 1+f = 1.6,
4W / R
U
Eb / I o
 Assuming efficient error correction algorithms, dual diversity antennas,
and RAKE receiver, it is possible to obtain Eb/Io=6 dB = 4, and then
WThis is of order of magnitude larger value than
U
Rwith the conventional (GSM;TDMA) systems!
Timo O. Korhonen, HUT Communication Laboratory 89
 Lessons Learned
 You understand what is meant by code gain, jamming margin, and
spectral efficiency and what is their meaning in SS systems
 You understand how spreading and despreading works
 You understand the basic principles of DS and FH systems and know
their error rates by using BPSK and FSK modulations
 You know the bases of code selection for SS system. (What kind of
codes can be applied in SS systems and when they should be applied.)
 You understand how the capacity of asynchronous CDMA system can
be determined

Timo O. Korhonen, HUT Communication Laboratory 90

March. 2007 doc.: 15-07-0624-00-004c

Outline
• One formal PHY tech proposal from Chinese
companies, universities and partners.

• Major tech points are upgraded from the OQPSK

Modulation on IEEE 802.15.4-2006
– Spreading sequence
– SFD design
– Pulse shaping filter
– Synchronization and demodulation performance

Submission Slide 91 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Proposed New Operation Frequency Bands

• Fc= 314.3, 314.8, 315.3, 315.8 (MHz); BW=400kHz

• Fc= 430.3, 430.8, 431.3, 431.8 (MHz); BW=400kHz
• Fc= 433.3, 433.8, 434.3, 434.8 (MHz); BW=400kHz
• Fc=780, 782,786, 788 (MHz) BW=2MHz

Submission Slide 92 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Communication Modulation and Data Rate

Spreading Parameters Data Parameters
Frequency
PHY
Bands
(MHz)
(MHz) Chip Rate Bit Rate Symbol Rate
Modulation
(Mchip/s) (kb/s) (ksymbol/s)

Chirp Sequence
315 0.4 0.2 50 12.5
+ MPSK
Chirp Sequence
430 0.4 0.2 50 12.5
+ MPSK
Chirp Sequence
780 2 1 250 62.5
+ MPSK

Submission Slide 93 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Upgrade from PHY Layer Operation of IEEE802.15.4-2006

• The new PHY tech proposal is similar to the OQPSK

ones used in IEEE802.15.4-2006 at sub 1GHZ on:
• The principal structures of the transmitter and
receiver
• Operation procedure of the transceiver
• The important parameters for RF parts
• Some different techniques are applied in this new
PHY proposal.

Submission Slide 94 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Reference Design of the Wireless Transceiver based on

New PHY tech Proposal
Transeiver
Q Pulse
Filter
Pre-
PA Mapping Bit Streams
I Pulse
processing
Filter
RF
switc MAC
h 90o
PHY
VCO

VGA
I LPF ADC
Digital Signal
Processing
BPF LNA  Acquisition
Bit Streams
LPF ADC  Sync
Q  Demodulation

Submission Slide 95 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Difference 1: Spreading Sequence and Mapping

• The Direct Sequence Spread Spectrum (DSSS) tech is applied
– 16 orthogonal spreading sequences are designed to map 4
information bits. The base sequence is a 16 length chirp
sequence and the other 15 sequences are its cyclic shifts.

Submission Slide 96 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Spreading Sequence

• New Proposal:
– Chirp code is orthogonal among its cyclic shifts
• Perfect auto-correlation property of the Preamble sequence,
• Perfect orthogonal property of the 16 spreading sequences,
• Reduce inter-chip interference in multipath environments.
– Chirp code is robust to frequency offset
• Low cost implementation of transmitter and receiver.

• OQPSK in 15.4-2006 in sub-1GHz :

– 16 sequences are quasi-orthogonal.
– The auto-correlation property of the Preamble sequence is not
very well.
– The code is susceptible to frequency offset.

Submission Slide 97 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Sliding Correlation Values of the Preamble Sequence

16 16

14 14

12 12

Auto-correlation Values
Auto-correlation Values

10 10

8 8

6 6

4 4

2 2

0 0
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15
Sliding Chips Sliding Chips

With Sequence in New Proposal With Sequence in 15.4-2006 Std

Submission Slide 98 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

Difference 2: Pre-processing on Transmission

• Conjugate the first sequence to obtain SFD
– The SFD sequence is the conjugate of the Preamble sequence, that means the phases
of the SFD chips are adverse to the phases of the Preamble chips.
• DC component removal
– All chips of each symbol in the head and load of PPDU should be multiplied by –1 or 1
based on the follow PN code serials.

G ( x)  x 7  x 3  1

r6 r5 r4 r3 r2 r1 r0 PN
code

Submission Slide 99 Liang Li

March. 2007 doc.: 15-07-0624-00-004c

SFD Design

• New Proposal :
– The 16 spreading sequences are cyclic shift of each
other.
– The SFD sequence is the complex conjugate of the
first spreading sequence.

• OQPSK in 15.4-2006 in sub-1GHz :

– The first 8 spreading sequences are cyclic shift of
each other, and the last 8 spreading sequences are
the complex conjugate of the first 8 spreading
sequences.
– The SFD sequences are chosen from the spreading
sequences.
Submission Slide Liang Li
100
March. 2007 doc.: 15-07-0624-00-004c

Difference 3: Pulse Shaping on Transmission and

Spectrum of Transmit Waveform
• The Pulse Shaping Filter on I and Q path is designed as a raised cosine
filter with roll-off factor 0.5:
sin(t / Tc ) cos(rt / Tc )
p (t ) 
t / Tc 1  4r 2 t 2 / Tc2
•The Transmit waveform and Spectrum are :

Submission Slide Liang Li

101
March. 2007 doc.: 15-07-0624-00-004c

Pulse Shaping Filter

New Proposal: Raised cosine filter 15.4-2006: Half-sine filter

The chip duration is 1us. To O-QPSK PHY, the zero-to-zero bandwidth
The zero-to-zero bandwidth is 1.5MHz. is 1.5MHz.
To New PHY proposal, the zero-to-zero
bandwidth is 3MHz.

Submission Slide Liang Li

102
March. 2007 doc.: 15-07-0624-00-004c

Synchronization Performance of
0
10
C-WPAN
15.4-2006 Simulation Conditions:
-1
10 1) AWGN channel
environment
-2
10 2) Chip rate sampling
3) Basic sliding correlation
Sync Error Rate

-3
receiver
10
4) Synchronized on other chips
means an error
-4
10

-5
10

-6
10
3 4 5 6 7 8 9 10
Eb/N0

Submission Slide Liang Li

103
March. 2007 doc.: 15-07-0624-00-004c

System Packet Error Performance

0
10
C-WPAN
15.4-2006 Simulation Conditions:
1) AWGN channel
-1
10 environment
2) Chip rate sampling
3) Basic sliding correlation
-2 receiver
PER

10
4) Ideal synchronization
5) 32 data octets in each packet
-3
10

-4
10
3 4 5 6 7 8 9 10
Eb/n0

Submission Slide Liang Li

104
March. 2007 doc.: 15-07-0624-00-004c

The Independent Simulation Results from I2R

0 AWGN Channel Performance (PER) 0
10 10

-1 -1
10 10

-2
PER

10 -2

PER
10

-3 16FSK(Non Coherent)
10 COBI16 (Ideal) -3
10 DSSS(Non Coherent)
DSSS (Ideal) COBI16(Non Coherent)
DSSS(Coherent)
COBI16 (Sync)
COBI16(Coherent)
-4 DSSS (Sync)
10 -4
0 2 4 6 8 10 12 10
0 2 4 6 8 10 12
E b /No (dB) Eb/N0

• I2R implements the independent Simulation based on New PHY proposal. The left figure shows
its results.
• BUAA obtained the Same results in the right picture. And compare with the ones of I2R on same
page. Simulation condition are
• Packet Length = 20bytes; AWGN channel, Ideal Sync.
• Coherent detection: Decision is based on the real parts of the correlation values
• Noncoherent detection: Decision is based on the norms of the correlation values

Submission Slide Liang Li

105
March. 2007 doc.: 15-07-0624-00-004c

System Performance of Receiver based on New Proposal

in Multipath Channel
• Tapped-Delay-Line Channel Model
– IEEE P802.15 Working Group for WPANs, Multipath Simulation
Models for Sub-GHz PHY Evaluation, 15-04-0585-00-004b, Oct.
2004.
– Power delay profile is exponentially declined.
– Each path is independently Rayleigh fading.
– The average power of the channel response over many packets
is 1, but in each packet the power is varied.
• Short Delay Environments
– Without rake receiver
• Long Delay Environments
– With 3-tap rake receiver
Submission Slide Liang Li
106
March. 2007 doc.: 15-07-0624-00-004c

System Performance of Rake Receiver based on New

Proposal in Multipath Channel
• Test Conditions
– RMS delay spread 0~600ns
– Tx nonlinear amplifier, Rapp’s model, p=3, backoff=1.5dB
– Tx and Rx frequency offset ±80ppm, phase noise -110dBc/Hz
@±1MHz
– Tx and Rx IQ imbalance 2dB, 10o
– 3bit AD sampling, 8bit baseband processing
– Rx will implement time and frequency synchronization and data
detection
– 5000 packets are tested for each SNR, each packet comprises
20 octets
– The packet error rate is counted for 90% coverage

Submission Slide Liang Li

107
March. 2007 doc.: 15-07-0624-00-004c

PER in Short Delay Environments without Rake Receiver

Submission Slide Liang Li

108
March. 2007 doc.: 15-07-0624-00-004c

PER in Long Delay Environments with 3-tap Rake Receiver

Submission Slide Liang Li

109
March. 2007 doc.: 15-07-0624-00-004c

Thank you!

Submission Slide Liang Li

110
1 Dr. Uri Mahlab
 INTRODUCTION
In order to transmit digital information over *
bandpass channels, we have to transfer
the information to a carrier wave of
.appropriate frequency

We will study some of the most commonly *

used digital modulation techniques wherein
the digital information modifies the amplitude
the phase, or the frequency of the carrier in
.discrete steps

2 Dr. Uri Mahlab

 The modulation waveforms for transmitting
:binary information over bandpass channels

ASK

FSK

PSK

DSB

3 Dr. Uri Mahlab

 OPTIMUM RECEIVER FOR BINARY
:DIGITAL MODULATION SCHEMS

The function of a receiver in a binary communication *

system is to distinguish between two transmitted signals
.S1(t) and S2(t) in the presence of noise

The performance of the receiver is usually measured *

in terms of the probability of error and the receiver
is said to be optimum if it yields the minimum
.probability of error

In this section, we will derive the structure of an optimum *

receiver that can be used for demodulating binary
.ASK,PSK,and FSK signals

4 Dr. Uri Mahlab

 Description of binary ASK,PSK, and
: FSK schemes

-Bandpass binary data transmission system

Transmit Local carrier

carrier Noise
Clock pulses (n(t Clock pulses

Input +
Channel Demodulator
Modulator
(Hc(f
ּ+ (receiver)
Binary
data
+
(V(t
{bk} (Z(t
Binary data output

{bk}
5 Dr. Uri Mahlab
:Explanation *
The input of the system is a binary bit sequence {bk} with a *
.bit rate r b and bit duration Tb

The output of the modulator during the Kth bit interval *

.depends on the Kth input bit bk

The modulator output Z(t) during the Kth bit interval is *

a shifted version of one of two basic waveforms S1(t) or S2(t) and
:Z(t) is a random process defined by

.1
s1[t  (k  1)Tb ] if b k  0
Z (t )  
 s2 [t  (k  1)Tb ] if b k  1
for : (k  1)Tb  t  kTb
6 Dr. Uri Mahlab
 The waveforms S1(t) and S2(t) have a duration *
of Tb and have finite energy,that is,S1(t) and S2(t) =0

if t  [0, Tb ] and
Tb

Energy E1   [ s1 (t )]2 dt  
:Term 0
Tb

E2   [ s2 (t )]2 dt  
0

7 Dr. Uri Mahlab

 :The received signal + noise

s1[t  (k  1)Tb t d ]  n(t )


V (t )   or (k  1)Tb t d  t  kTb  t d
s [t  (k  1)T t ]  n(t )
2 b d

8 Dr. Uri Mahlab

Choice of signaling waveforms for various types of digital*
modulation schemes
c
S1(t),S2(t)=0 for t  [0, T ]; f b c
2
.The frequency of the carrier fc is assumed to be a multiple of rb
Type of
S1 (T );0  t  Tb s (t );0  t  T
2 b
modulation

A cos wc t ASK
0
(or A sin wc t ) PSK
 A cos wc t A cos wc t
FSK
(or  A sin wc t ) ( A sin wc t )

A cos{( wc  wd )t} A cos{( wc  wd )t}

[(or A sin{( wc  wd )t}] [or A sin{( wc  wd )t}]

9 Dr. Uri Mahlab

 :Receiver structure

(V0(t
Threshold
Filter
device or A/D
(H(f
converter
output

v(t )  z(t )  n(t ) Sample every

Tb seconds

10 Dr. Uri Mahlab

 :{Probability of Error-{Pe*
The measure of performance used for comparing *
!!!digital modulation schemes is the probability of error

The receiver makes errors in the decoding process *

!!! due to the noise present at its input

The receiver parameters as H(f) and threshold setting are *

!!!chosen to minimize the probability of error

11 Dr. Uri Mahlab

 :The output of the filter at t=kTb can be written as *

V0 (kTb )  s0 (kTb )  n0 (kTb )

12 Dr. Uri Mahlab

 :The signal component in the output at t=kTb
kTb

s0 (kTb ) 

 Z ( )h(kT b   ) d
kTb

  Z (
k 1)Tb
 ) h ( kTb   ) d  ISI terms

 h( ) is the impulse response of the receiver filter*

ISI=0*

kTb

s0 (kTb )   Z ( )h(kT
( k 1)Tb
b  )d

13 Dr. Uri Mahlab

 Substituting Z(t) from equation 1 and making*
change of the variable, the signal component
:will look like that

Tb
  s1 ( )h(Tb  )d  s01 (kTb ) when b k  0
0
s0 (kTb )  Tb
 s ( )h(T   )d  s (kT ) when b  1
 0 2 b 01 b k

14 Dr. Uri Mahlab

 :The noise component n0(kTb) is given by *

kTb

n0 (kTb )  

n ( ) h ( kTb   ) d
.The output noise n0(t) is a stationary zero mean Gaussian random process
:The variance of n0(t) is*

N 0  E{n0 (t )}   Gn ( f ) H ( f ) df
2 2


:The probability density function of n0(t) is*

1  -n 
2
f n0 (n)  exp  ;    n  
2N 0  2N0 
15
The probability that the kth bit is incorrectly decoded*
:is given by

.2
Pe  P{bk  0 and V0 (kTb )  T0
or bk  1 and V0 (kTb )  T0 }
1
 P{V0 (kTb )  T0 | bk  0}
2
1
 P{V0 (kTb )  T0 | bk  1}
2
16 Dr. Uri Mahlab
 :The conditional pdf of V0 given bk = 0 is given by*

.3

1  - (V0  s01 ) 
2
fV0 \bk 0 (V0 )  exp  , -   V0  
2N 0  2N0 
:It is similarly when bk is 1*

1  - (V0  s02 ) 2 
fV0 \bk 1 (V0 )  exp  , -   V0  
2N 0  2N0 

17 Dr. Uri Mahlab

 Combining equation 2 and 3 , we obtain an*
:expression for the probability of error- Pe as


.4 1 1  - (V0  S 01 ) 
2
Pe   exp dV0
2 T0 2N 0  2N0 
T0
1 1  - (V0  S 02 ) 2

  exp dV0
2  2N 0  2N0 

18 Dr. Uri Mahlab

 :Conditional pdf of V0 given bk

f v0 bk  0 ( v0 )
f v0 b k  1( v0 )

:The optimum value of the threshold T0* is*

S S
*
T 0  01 02
19
2
Dr. Uri Mahlab
 Substituting the value of T*0 for T0 in equation 4*
we can rewrite the expression for the probability
:of error as

1  (V0  s01 ) 2


exp  dV0
Pe  
( s02  s01 ) / 2
2N 0  2N0 


1  Z 2

  2
exp 
 2
dZ

( s02  s01 ) / 2 N 0

20 Dr. Uri Mahlab

 
The optimum filter is the filter that maximizes*
the ratio or the square of the ratio 2
(maximizing eliminates the requirement S01<S02)
2

S 02 (Tb )  S 01 (Tb )

N0

21 Dr. Uri Mahlab

 :Transfer Function of the Optimum Filter*
The probability of error is minimized by an *
appropriate choice of h(t) which maximizes 2

[ s02 (Tb )  s01 (Tb )] 2

Where  2

N0
Tb

s02 (Tb )  s01 (Tb )   [ s2 ( )  s1 ( )]h(Tb   )d

0

 2
And N 0   Gn ( f ) H ( f ) df

22 Dr. Uri Mahlab
 If we let P(t) =S2(t)-S1(t), then the numerator of the*
:quantity to be maximized is

S 02 (Tb )  S 01 (Tb )  P0 (Tb )

Tb 
  P( )h(Tb   )d   P( )h(T b   )d
0 

:the Fourier transform of P0 is  

Since P(t)=0 for t<0 and h( )=0 for <0*

P0 ( f )  P( f ) H ( f )

P0 (Tb )   P( f ) H ( f ) exp( j 2fT )df

b

23 Dr. Uri Mahlab

 
:Hence 2 can be written as*

 2

 H ( f ) P( f ) exp( j 2fT )df


b

2   (*)
 H( f )
2
Gn ( f )df


We can maximize  2 by applying Schwarz’s*

:inequality which has the form

2


 X ( f )X
1 2 ( f )df



 X 2
2
( f ) df (**)

2 
X 1 ( f ) df
24 
Dr. Uri Mahlab
 Applying Schwarz’s inequality to Equation(**) with-

X 1 ( f )  H ( f ) Gn ( f )
and
P ( f ) exp( j 2fTb )
X2( f ) 
Gn ( f )
We see that H(f), which maximizes ,is given by-
2

P ( f ) exp( j 2fTb )
*
(***) H( f )  K
Gn ( f )
!!! Where K is an arbitrary constant
25 Dr. Uri Mahlab
 Substituting equation (***) in(*) , we obtain-
:the maximum value of as
2

 2
P( f )
   G ( f ) df
2
max
 n

:And the minimum probability of error is given by-


1  Z 2
   max 
Pe 

 2
exp 
 2
dZ  Q
  2 

max/ 2

26 Dr. Uri Mahlab

 :Matched Filter Receiver*

If the channel noise is white, that is, Gn(f)= /2 ,then the transfer -
:function of the optimum receiver is given by

H ( f )  P ( f ) exp( j 2fTb )
*

From Equation (***) with the arbitrary constant K set equal to /2-
:The impulse response of the optimum filter is


h(t )   ( f ) exp(2jfTb )] exp(2jft )df
*
[ P


27 Dr. Uri Mahlab

 Recognizing the fact that the inverse Fourier *

of P*(f) is P(-t) and that exp(-2 jfTb) represent
:a delay of Tb we obtain h(t) as

h(t )  p(Tb  t )
:Since p(t)=S1(t)-S2(t) , we have*

h(t )  S 2 (Tb  t )  S1 (Tb  t )

The impulse response h(t) is matched to the signal *
:S1(t) and S2(t) and for this reason the filter is called
MATCHED FILTER

28 Dr. Uri Mahlab

 :Impulse response of the Matched Filter *

(S2(t 1
t
0 2 \Tb (a)
(S1(t
0
2 \Tb t
1- (b)

2
(P(t)=S2(t)-S1(t 2 \Tb
0 t
2 (c) Tb
(P(-t
t
Tb- 0 (d)
2
(h(t)=p(Tb-t (h(Tb-t)=p(t
2 \Tb
0 t
29 (e) Tb Dr. Uri Mahlab
 :Correlation Receiver*
The output of the receiver at t=Tb*
Tb

V0 (Tb )   V ( )h(Tb   )d




Where V( ) is the noisy input to the receiver

Substituting
h( )  S and noting *
(Tb   )  S1 (Tb   )
2
: that h( )  0 for   (0, Tb )
we can rewrite the preceding expression as

Tb

V0 (Tb )   V ( )[ S 2 ( )  S1 ( )]d
0 (# #)
Tb Tb

  V ( ) S 2 ( )d   V ( ) S1 ( )d
0 0
30 Dr. Uri Mahlab
Equation(# #) suggested that the optimum receiver can be implemented *
as shown in Figure 1 .This form of the receiver is called
A Correlation Receiver

S1 (t ) integrator
Figure 1
Tb


0
 S1 (t )  n(t )

V (t )   or
S (t )  n(t ) - Threshold
 2
+  device
(A\D)
Sample
every Tb
Tb
seconds

0

S 2 (t ) integrator
31 Dr. Uri Mahlab
 In actual practice, the receiver shown in Figure 1 is actually *
.implemented as shown in Figure 2
In this implementation, the integrator has to be reset at the
- (end of each signaling interval in order to ovoid (I.S.I
!!! Inter symbol interference

:Integrate and dump correlation receiver

White (n(t Closed every Tb seconds

Gaussian Filter
noise + to Threshold
+ limit c device
noise (A/D)
(Signal z(t power R
High gain
Figure 2 S1 (t )  S 2 (t ) amplifier

The bandwidth of the filter preceding the integrator is assumed *

!!! to be wide enough to pass z(t) without distortion
32
 Example: A band pass data transmission scheme
uses a PSK signaling scheme with

S 2 (t )  A cos wc t , 0  t  Tb , wc  10 / Tb
S1 (t )   A cos wc t , 0  t  Tb , Tb  0.2m sec

The carrier amplitude at the receiver input is 1 mvolt and

the psd of the A.W.G.N at input is 10 11 watt/Hz. Assume
that an ideal correlation receiver is used. Calculate the
.average bit error rate of the receiver

33 Dr. Uri Mahlab

 :Solution

34 Dr. Uri Mahlab

 :Solution Continue

=Probability of error = Pe *

35 Dr. Uri Mahlab

 * Binary ASK signaling
schemes:
The binary ASK waveform can be described as

s1[t  (k  1)Tb ] if b k  0

z (t )   (k - 1)Tb  t  kTb
s [t  (k  1)T ] if b  1
 2 b k

Where and
S 2 (t )  A cos  c t s1 (t )  0
We can
represent :Z(t) as
Z (t )  D (t )( A cos  c t )
36 Dr. Uri Mahlab
Where D(t) is a lowpass pulse waveform consisting of
.rectangular pulses

:The model for D(t) is


d (t )   b g[t  (k  1)T ], b
k  
k b k  0 or 1

1 0  t  Tb
g (t )  
0 elswhere

D(t )  d (t  T )
37 Dr. Uri Mahlab
 :The power spectral density is given by
A2
Gz ( f )  [GD ( f  f c )  GD ( f  f c )
4
The autocorrelation function and the power spectral density
:is given by
 1 Tb  
  for   Tb
RDD ( )   4 4Tb
0 for   Tb


1 sin 2 fTb 
GD ( f )    ( f )  2 2 
4  f Tb 
38 Dr. Uri Mahlab
 :The psd of Z(t) is given by
2
A
Gz ( f )  ( ( f  f c )   ( f  f )
16
sin Tb ( f  f c )
2
 2
 Tb ( f  f c ) 2

sin TB ( f  f c )
2
 
 Tb ( f  f c )
2

39 Dr. Uri Mahlab

 If we use a pulse waveform D(t) in which the individual pulses
g(t) have the shape

 a 
 1  cos(2rbt   ) 0  t  Tb
g (t )   2 
0 elsewere


40 Dr. Uri Mahlab

 Coherent ASK
We start with s2 (t )  A cos  c t and s1 (t )  0
The signal components of the receiver output at the
:of a signaling interval are

Tb

s01 (kTb )   s1 (t )[ s2 (t )  s1 (t )]dt  0

0

and
Tb
A2
SO2 (kTb )   s2 (t )[ s2 (t )  s1 (t )]dt  Tb
0
2
41 Dr. Uri Mahlab
 :The optimum threshold setting in the receiver is

s01 (kTb )  s02 (kTb ) A 2

T  *
0  Tb
2 4
:The probability of error P can be computed as
e

A2Tb
 2
max 



1  z 2
  A 2
T 
pe   exp dz  Q b 

2  2   4 
1
 max  
2

42 Dr. Uri Mahlab

:The average signal power at the receiver input is given by
2
A
sav 
4
We can express the probability of error in terms of the
:average signal power
 S avTb 
pe  Q 

  
The probability of error is sometimes expressed in *
: terms of the average signal energy per bitP,  Q Eav  as
e   


Eav  ( sav )Tb

43 Dr. Uri Mahlab
 Noncoherent ASK
:The input to the receiver is *
 A cos  c t  ni (t ) when b k  1
V (t )  
ni (t ) when b k  0

ni (t )  the noise at the receiver input

which is assumed to be zero mean,
Gaussian, and white.
44 Dr. Uri Mahlab
Noncoharent ASK Receiver

At the filter output we have :

Y (t )  Ak cos  c t  n(t ) 
 Ak cos  c t  nc (t ) cos  c t 
ns (t ) sin  c t
where A k  A when the kth
transmitted bit b k  1 and A k  0
when n(t) is the noise at the output
of the bandpass filter
45
 :The pdf is r  r2 
f R|bk 0 (r )  exp  , r  0
N0  2N0 

r  Ar   r 2  A2 
f R|bk 1 (r )  I 0   exp  , r  0
N0  N0   2N0 
N 0  noise power at the output of the
bandpass filter.
2
N 0  BT 
TB
2
1
46
I0 ( X ) 
2  exp( x cos(u ))du
0 Dr. Uri Mahlab
 pdf’s of the envelope of the noise and the signal *
:pulse noise

47 Dr. Uri Mahlab

 :The probability of error is given by
1 1
pe  p (error | b k  0)  p (error | b k  1)
2 2
1 1
 pe 0  pe1
2 2
where

r  r2   A2 
pe 0   exp  dr  exp  
A N0  2N0   8N 0 
2

and
A
 
2
1  (r  A)  2  A 
pe1   exp  dr  Q 
  2N 0  2N0   2 N0 
 
Using the approximation
  x2 
exp 
Q( x)   2 
48 x 2 Dr. Uri Mahlab
 for large x, we can reduce pe1 to
4N0  A2 
pe1  exp
  

2A 2
 8N0 
Hence,
1 4N0   A 2

pe  1  2 
exp 
 8N 

2 2A   0 
1  A 2

 exp 
 8N 
 if A 2
 N 0
2  0 
49 Dr. Uri Mahlab
 BINERY PSK SIGNALING
SCHEMES
:The waveforms are *
s1 (t )   A cos  c t for b k  1
s2 (t )  A cos  c t for b k  0
:The binary PSK waveform Z(t) can be described by *

Z (t )  D(t )( A cos  c t )
.D(t) - random binary waveform *

50 Dr. Uri Mahlab

:The power spectral density of PSK signal is

2
A
GZ ( f )  [GD ( f  f c )  GD ( f  f c )]
4

Where,
sin fTb
2
GD ( f )  2 2
 f Tb
51 Dr. Uri Mahlab
 Coherent PSK
:The signal components of the receiver output are

kTb

s01 (kTb )      2
s1 (t )[ s 2 (t ) s1 (t )]dt A Tb
( k 1)Tb
kTb

s02 (kTb )     2
s 2 (t )[ s 2 (t ) s1 (t )]dt A Tb
( k 1)Tb

52 Dr. Uri Mahlab

:The probability of error is given by
 
Pe  Q max 
 2 
where
T
2 b 4 A2Tb
 max
2
  (2 A cos  c t ) dt 
2

0 
or
 A2T 
pe  Q b 
  
 
The average signal power sav end the
signal energy per bit Eav for the PSK
scheme are
A2
s av 
2
and
 A2 
Eav   Tb
 2 
53 Dr. Uri Mahlab
we can express the probability of error :
 2 savTb 
pe   

  
 2 Eav 
 Q 

  

54 Dr. Uri Mahlab

 DIFFERENTIALLY COHERENT *
:PSK
DPSK modulator
BINERY
 dk 
SEQUENCE
LOGIC o or 1 LEVEL  A cos  C t

NETWORK SHIFT 1 Z(t)

 d k 1
DELAY
A cos  c t
Tb

55 Dr. Uri Mahlab

 DPSK demodulator
n(t )
Filter to Lowpass Threshold
 b̂ 
k
 limit noise filter or device
Z(t) power integrator (A/D)
sample
Delay at kTb
Tb

56 Dr. Uri Mahlab

 Differential encoding & decoding

Input
Seque-
nce
1 1 0 1 0 0 0 1 1
Encoded
sequence 1 1 1 0 0 1 0 1 1 1
Transmit
Phase
0 0 0 pi pi 0 pi 0 0 0
Phase
Compari-son
output
+ + - + - - - + +
Output
Bit
sequence
1 1 0 1 0 0 0 1 1

57 Dr. Uri Mahlab

 * BINARY FSK SIGNALING
SCHEMES
:The waveforms: of FSK signaling
S1 (t )  A cos( c t   d t ) for b k  0
S 2 (t )  A cos( C t   d t ) for b k  1
:Mathematically it can be represented as
 
Z (t )  A cos  c t   d  D(t ' )dt ' 
 
 

 1 for b k  1
D(t )  
58
 1 for b k  0
Dr. Uri Mahlab
 Power spectral density of FSK signals

wd
fd 
2
we
fe 
2

Power spectral density of a binary FSK signal

with
2 f d  rb

59 Dr. Uri Mahlab

 Coherent FSK
:The local carrier signal required is

s2 (t )  s1 (t )  A cos( c t   d t )  A cos( c t   d t )
The input to the A/D converter at sampling time

t  kTb is s01 (kTb ) or s02 (kTb ) where

Tb

s02 (kTb )   s2 (t )[ s2 (t )  s1 (t )]dt

0
Tb

s01 (kTb )   s1 (t )[ s2 (t )  s1 (t )]dt

0
60 Dr. Uri Mahlab
The probability of error for the correlation receiver is
:given by

  max 
Pe  Q 
 2 
where
Tb
2
 2
max   [ s2 (t )  s1 (t )] dt
2

0
when
s2 (t )  A cos( c t   d t ) and
s1 (t )  A cos( c t   d t )
61 Dr. Uri Mahlab
 .Which are usually encountered in practical system

:We now have

2 A Tb 2
 sin 2 d Tb 
 2
max  1  
  2 T
d b 
:When

wcTb  1 , w c  wd
62 Dr. Uri Mahlab
 Noncoherent FSK
Assuming that s1 (t )  A cos( c   d )t has been transmitted during the kth signaling
interval, the pdf of the envelopeR 1 (kTb ) of the bottom filter is :
r1  Ar1   r12  A2 
f R1|s1 (t ) (r1 )  I 0   exp  , r1  0
N0  N0   2n0 
and
r2  r22 
f R2 |s1 (r2 )  exp  , r2  0
N0  2N0 

63 Dr. Uri Mahlab

 Noncoharenr demodulator of binary FSK

Bandpass R2 (kTb )
ENVELOPE
filter DETECTOR
fc  f d
+
THRESHOLD
 DEVICE
(A/D)
bandpass -
ENVELOPE T 0*  0
filter DETECTOR R1 (kTb )
fc  fd
Z(t)+n(t)

1  A2 
Pe  exp  
2  4N0 
64 Dr. Uri Mahlab
Probability of error for binary digital modulation *
:schemes

65 Dr. Uri Mahlab

 M-ARY SIGNALING
SCHEMES
:M-ARY coherent PSK
The M possible signals that would be transmitted
:during each signaling interval of duration Ts are

 k 2 
S k (t )  A cos  c t  , k  0,1,...M  1, 0  t  Ts
 M 

:The digital M-ary PSK waveform can be represented


Z (t )  A  g (t  kTs ) cos( c t   k )
k  
66 Dr. Uri Mahlab
  
Z (t )  A cos  c t  (cos  k ) g (t  kTs )  A sin  c t  (sin  k ) g (t  kTs )
k   k  

:In four-phase PSK (QPSK), the waveform are

S1 (t )  A cos  c t
S 2 (t )   A sin  c t
S3 (t )   A cos  c t
S 4 (t )  A sin  c t
for all 0  t  TS
67 Dr. Uri Mahlab
 Phasor diagram for QPSK

A cos( c t  45 ) and A cos( c t  45 )

0 0

That are derived from a coherent local carrier

A cos  t c
reference

68
 If we assume that S 1 was the transmitted signal
:during the signaling interval (0,Ts),then we have
Ts

S 01 (Ts )   ( A cos  c t ) A cos( c t  )dt 
0
4
A2 
 Ts cos  L0
2 4
Ts
 
S 02 (Ts )   ( A cos  c t ) A cos  c t  dt
0  4
A2 
 Ts cos  L0
2 4
69 Dr. Uri Mahlab
 QPSK receiver scheme

A cos( c t  45 )
TS V01 (kTS )

0


Z(t) TS

 V02 (kTS )
n(t ) 0

A cos( c t  45 )

70 Dr. Uri Mahlab

 :The outputs of the correlators at time t=TS are

V01 (Ts )  S 01 (Ts )  n01 (Ts )

V02 (Ts )  S 02 (Ts )  n02 (Ts )

where n01 (Ts ) & n02 (Ts ) are zero mean Gaussian random variables defined by

Ts

n01 (Ts )   n(t ) A cos( ct  450 )dt

0

TS

n02 (Ts )   n(t ) A cos( ct  450 )dt

0

71 Dr. Uri Mahlab

 Probability of error of
QPSK:
Pec1  P (n01 (Ts )   L0 )
 P (n02 (Ts )  L0 )
 L0   A2T 
 Q   Q s 
 Pec 2
 N   2 
 0   
72 Dr. Uri Mahlab
 Pc - The probability that the transmitted signal is received correctly

Pc  (1  Pec1 )(1  Pec 2 )

Pe for the system is :

 A2T 
Pe  1  Pc  2 Pec1  2Q s 
 2 
 

for M  4
 A2T  
Pe  2Q s
sin 2 
  M 
 

73 Dr. Uri Mahlab

 Phasor diagram for M-ary PSK ; M=8

74 Dr. Uri Mahlab

 The average power requirement of a binary PSK
:scheme are given by

( S av ) M  Z 12  1

 2  
( S av ) b  Z 2   sin 2   
 
M 

If Pe is very small & Z1  Z 2

( S av ) M 1

( S av ) b 2  
 sin  
M 
75 Dr. Uri Mahlab
 * COMPARISION OF POWER-BANDWIDTH
:FOR M-ARY PSK

Pe  10 4

Value ( Bandwidth ) M ( S av ) m
of M ( Bandwidth ) b ( S av ) b

4 0.5 0.34 dB
8 0.333 3.91 dB
16 0.25 8.52 dB
32 0.2 13.52 dB

76 Dr. Uri Mahlab

 * M-ary for four-phase
Differential PSK:
RECEIVER FOR FOUR PHASE DIFFERENTIAL PSK

Integrate
and dump
n(t ) filter V01 (t )
Delay
TS

900
Delay
Z(t) phase
shift TS
Integrate
and dump
filter V02 (t )
77 Dr. Uri Mahlab
 :The probability of error in M-ary differential PSK

 A2T   
Pe  2Q S 2
2 sin  
   2 M 

:The differential PSK waveform is

Z (t )  A g (t  kTS ) cos( c t   k )
k

78 Dr. Uri Mahlab

 :Transmitter for differential PSK*

Serial to Diff Envelope

Binary parallel phase
modulator BPF
Data converter mod. (Z(t
rs  1200
rb  2400 f c  1800 Hz
M 4

3

Clock 4
signal 600 Hz
2400 Hz

79 Dr. Uri Mahlab

 * M-ary Wideband FSK
Schemas:
Let us consider an FSK scheme witch have the
: following properties
 A cos  i t 0  t  Ts
S i (t )  
0 elsewhere

and
 A2TS
 FOR i  j
TS
S i (t ) S j (t )   2
0 0
 FOR i  j

80 Dr. Uri Mahlab

 :Orthogonal Wideband FSK receiver

gausian
noise TS

n(t ) 
0 Y1 (t )
S1 (t )
 TS
MAXIMUM
Z(t)  SELECTOR
0
.
Y2 (t )
S 2 (t ) .
.
.
TS


0 YM (t )
S M (t )
81 Dr. Uri Mahlab
 :The filter outputs are
Ts

Y j (Ts )  S
0
j (t )[n(t )  S1 (t )]dt , j  1,2,...., M

TSS TS

 S
0
j (t ) S1 (t )dt  S
0
j (t ) n(t )dt

 S 0 j (Ts )  n j (Ts )
where
S 0 j (Ts ) - The signal component of the j - th filter output

n j (Ts ) - The noise component

82 Dr. Uri Mahlab

 :N0 is given by
 
N 0  A Ts   2

4
:The probability of correct decoding as
Pc1  p{Y2  Y1 , Y3  Y1 ,..., YM  Y1 | S1 sent}


  2 1  s1 sent and
P{Y y ,..., YM y |
1 Y1  y1 } f Y1 |S1 ( y1 ) dy1
-

:In the preceding step we made use of the identity


P( X  Y )   P( X  y | Y  y ) f

Y ( y )dy
83 Dr. Uri Mahlab
 The joint pdf of Y2 ,Y3 ,…,YM *
:is given by

M
fY 2...YM |S1:Y1  y1 ( y2 ,..., yM )   fYi ( yi )
i 2

84 Dr. Uri Mahlab

where 1  y 2i 
fYi ( yi )  exp   ,    yi  
2N 0  2N0 
and

 y
 1 1 M
y


Pc1     ...   f Yi ( yi ) dyi  f Y1 |S1 ( y1 ) dy1
 
    i  2 


M 1

 y1 
    f Y ( y )dy  f Y1 |S1 ( y1 )dy1
- 
 
where
1  y2 
fY ( y )  exp   ,   y  
2N 0  2N0 
1  ( y1  S 01 ) 2 
fY1 |S1 ( y1 )  exp   ,    y1  
2N 0  2N0 
and
 A2 
N 0   Ts 
 2 2
A2
S 01  Ts
85 2
Dr. Uri Mahlab
Probability of error for M-ary orthogonal *
: signaling scheme

86 Dr. Uri Mahlab

 The probability that the receiver incorrectly *
decoded the incoming signal S1(t) is

Pe1 = 1-Pe1
The probability that the receiver makes *
an error in decoding is

Pe = Pe1

e assume that M 2
, and rb  rs log 2 M  rs ( a positive inteegr )
e can see that increasing values of M lead to smaller power
quirements and also to more complex transmitting
ceiving equipment.

87 Dr. Uri Mahlab

In the limiting case as M  the probability of error P satisfies
e

1 if Sav / rb  0.7


Pe  
0 if S / r  0.7
 av b

The maximum errorless rb at W data can be transmitted

using an M- ary orthogonal FSK signaling scheme

S av S av
rb   log 2 e
0.7 

88
The bandwidth of the signal set  as M 
Dr. Uri Mahlab