Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute
1
: "shiv rpi¨
Point-to-Point Wireless Communication (I):
Digital Basics. Modulation. Detection. Pulse
Shaping
Shiv Kalyanaraman
Google: 'Shiv RPI¨
shivkuma(ecse.rpi.edu
Based upon slides oI Sorour Falahati, A. Goldsmith,
& textbooks by Bernard Sklar & A. Goldsmith
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
2
: "shiv rpi¨
%he Basics
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
3
: "shiv rpi¨
Big Picture: Detection under AWGN
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
4
: "shiv rpi¨
Additive White Gaussian Noise (AWGN)
%hermal noise is described by a zero-mean Gaussian random process,
n(t) that ADDS on to the signal ÷~ 'additive¨
Probability density function
(gaussian)
|w/Hz|
Power spectral
Density
(flat ÷> ~white¨)
Autocorrelation
Function
(uncorrelated)
Its PSD is Ilat, hence, it is called 90 noise.
Autocorrelation is a spike at 0: uncorrelated at any non-zero lag
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
5
: "shiv rpi¨
ffect of Noise in Signal Space
%he cloud Ialls oII exponentially (gaussian).
Vector viewpoint can be used in signal space, with a random noise vector w
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
6
: "shiv rpi¨
Maximum Likelihood (ML) Detection: Scalar Case
Assuming both symbols equally likely: u
A
is chosen iI
'likelihoods¨
Log-Likelihood ÷>
A simple distance criterion!
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
7
: "shiv rpi¨
AWGN Detection for Binary PAM
) (
1
t o
b
E
b
E
1
s
2
s
) ' (
2
2 5

0
) ' (
1
2 5

2 /
2 /
) ( ) (
0
2 1
2 1

= =

" 2 ! 2 !
e e
s s

2
) 2 (
0

= =

E
" ! !
b
E B
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
8
: "shiv rpi¨
Bigger Picture
General structure oI a communication systems
Formatter
Source
encoder
Channel
encoder
Modulator
Formatter
Source
decoder
Channel
decoder
Demodulator
%ransmitter
Receiver
SOURCE
Info.
%ransmitter
%ransmitted
signal
Received
signal
Receiver
Received
info.
Noise
Channel
Source User
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
9
: "shiv rpi¨
Digital vs Analog Comm: Basics
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
10
: "shiv rpi¨
Digital versus analog
Advantages oI digital communications:
Regenerator receiver
DiIIerent kinds oI digital signal are treated
identically.
Data
Voice
Media
Propagation distance
Original
pulse
Regenerated
pulse
A bit is a bit!
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
11
: "shiv rpi¨
Signal transmission through linear systems
Deterministic signals:
Random signals:
Ideal distortion less transmission:
All the Irequency components oI the signal not only arrive with
an identical time delay, but also are ampliIied or attenuated
equally.
Input
Output
Linear system
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
12
: "shiv rpi¨
Signal transmission (cont`d)
Ideal Iilters:
Realizable Iilters:
RC Iilters Butterworth Iilter
igh-pass
Low-pass
Band-pass
Non-causal!
Duality ÷~ similar problems occur w/
rectangular pulses in time domain.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
13
: "shiv rpi¨
Bandwidth of signal
Baseband versus bandpass:
Bandwidth dilemma:
Bandlimited signals are not realizable!
Realizable signals have inIinite bandwidth!
Baseband
signal
Bandpass
signal
Local oscillator
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
14
: "shiv rpi¨
Bandwidth of signal: Approximations
DiIIerent deIinition oI bandwidth:
a) HalI-power bandwidth
b) Noise equivalent bandwidth
c) Null-to-null bandwidth
d) Fractional power containment bandwidth
e) Bounded power spectral density
I) Absolute bandwidth
(a)
(b)
(c)
(d)
(e)50dB
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
15
: "shiv rpi¨
ncode
%ransmit
Pulse
modulate Sample Quantie
Demodulate/
Detect
Channel
Receive
Low-pass
filter
Decode
Pulse
waveIorms
Bit stream
Format
Format
Digital inIo.
%extual
inIo.
Analog
inIo.
%extual
inIo.
Analog
inIo.
Digital inIo.
source
sink
Formatting and transmission of baseband signal
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
16
: "shiv rpi¨
Sampling of Analog Signals
%ime domain
) ( ) ( ) ( t x t x t x
s
- =
o
) (t x
Frequency domain
) ( ) ( ) ( f X f X f X
s
- =
o
' ) ( ' f X
' ) ( ' f X
o
' ) ( ' f X
s
) (t x
s
) (t x
o
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
17
: "shiv rpi¨
Aliasing effect & Nyquist Rate
LP filter
Nyquist rate
aliasing
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
18
: "shiv rpi¨
Undersampling &Aliasing in
%ime Domain
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
19
: "shiv rpi¨
Note: correct reconstruction does not draw straight lines between samples.
Key: use sinc() pulses for reconstruction/interpolation
Nyquist Sampling & Reconstruction:
%ime Domain
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
20
: "shiv rpi¨
The impulse response of the
reconstruction filter has a classic 'sin(x)/x
shape.
The stimulus fed to this filter is the series
of discrete impulses which are the
samples.
Nyquist Reconstruction: Frequency Domain
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
21
: "shiv rpi¨
Sampling theorem
Sampling theorem: A bandlimited signal with no spectral
components beyond , can be uniquely determined by
values sampled at uniIorm intervals oI
%he sampling rate,
is called Nyquist rate.
In practice need to sample Iaster than this because the receiving
Iilter will not be sharp.
Sampling
process
Analog
signal
Pulse amplitude
modulated (PAM) signal
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
22
: "shiv rpi¨
Quantiation
Amplitude quantizing: Mapping samples oI a continuous amplitude
waveIorm to a Iinite set oI amplitudes.
In
Out
Q
u
a
n
t
i

e
d
v
a
l
u
e
s
Average quantization noise power
Signal peak power
Signal power to average
quantization noise power
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
23
: "shiv rpi¨
ncoding (PCM)
A uniIorm linear quantizer is called Pulse Code Modulation
(PCM).
Pulse code modulation (PCM): Encoding the quantized signals
into a digital word (PCM word or codeword).
Each quantized sample is digitally encoded into an bits
codeword where in the number oI quantization levels and
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
24
: "shiv rpi¨
Quantiation error
"uantizing error: %he diIIerence between the input and output oI a
quantizer
) ( ) ( ` ) ( t x t x t e =

) (t x
) ( ` t x
) ( ) ( `
) (
t x t x
t e

=
AGC
x
) (x q v =
"auntizer
Process oI quantizing noise
) (t x
) ( ` t x
) (t e
Model oI quantizing noise
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
25
: "shiv rpi¨
Non-uniform quantiation
It is done by uniIormly quantizing the 'compressed¨ signal.
At the receiver, an inverse compression characteristic, called 'expansion¨ is
employed to avoid signal distortion.
compressionexpansion companding
) (t v
) (t x
) ( ` t v ) ( ` t x
x
) (x C v =
x`
v`
Compress "auntize
Channel
Expand
%ransmitter Receiver
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
26
: "shiv rpi¨
Baseband transmission
%o transmit inIormation thru physical channels, PCM
sequences (codewords) are transIormed to pulses (waveIorms).
Each waveIorm carries a symbol Irom a set oI size M.
Each transmit symbol represents bits oI the PCM words.
PCM waveIorms (line codes) are used Ior binary symbols (M÷2).
M-ary pulse modulation are used Ior non-binary symbols
(M~2). Eg: M-ary PAM.
For a given data rate, M-ary PAM () requires less bandwidth than
binary PCM.
For a given average pulse power, binary PCM is easier to detect than M-
ary PAM ().

2
log =
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
27
: "shiv rpi¨
PAM example: Binary vs 8-ary
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
28
: "shiv rpi¨
xample of M-ary PAM
B
B
T
01`
B
T
T
B
T
00`
10`
1`
A.
T
0`
T
A.
Assuming real time %x and equal energy 5erTx/,t,bit Ior
binary-PAM and 4-ary PAM:
· 4-ary: T÷T
b
and Binary: T÷T
b
· Energy per symbol in binary-PAM:
4-ary PAM
(rectangular pulse)
Binary PAM
(rectangular pulse)
11`
2 2
10B A =
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
29
: "shiv rpi¨
Other PCM waveforms: xamples
PCM waveIorms category:
Phase encoded
Nultilevel binary
Nonreturn·to·zero (NRZ)
Return·to·zero (RZ)
1 0 1 1 0
0 % 2% 3% 4% 5%
V
-V
V
0
V
0
-V
1 0 1 1 0
0 % 2% 3% 4% 5%
V
-V
V
-V
V
0
-V
NRZ-L
Unipolar-RZ
Bipolar-RZ
Manchester
Miller
Dicode NRZ
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
30
: "shiv rpi¨
PCM waveforms: Selection Criteria
Criteria Ior comparing and selecting PCM waveIorms:
Spectral characteristics (power spectral density and
bandwidth eIIiciency)
Bit synchronization capability
Error detection capability
InterIerence and noise immunity
Implementation cost and complexity
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
31
: "shiv rpi¨
Summary: Baseband Formatting and transmission
InIormation (data- or bit-) rate:
Symbol rate :
Sampling at rate
(sampling time÷%s)
"uantizing each sampled
value to one oI the
levels in quantizer.
Encoding each q. value to
bits
(Data bit duration Tb÷Ts/)
ncode
Pulse
modulate Sample Quantie
Pulse waveIorms
(baseband signals)
Bit stream
(Data bits)
Format
Digital inIo.
%extual
inIo.
Analog
inIo.
source
Mapping every data bits to a
symbol out oI M symbols and transmitting
a baseband waveIorm with duration T
s s
T f / 1 =

2
log =
2
2
log =
|bits/sec| / 1
b b
T R =
ec| |symbols/s / 1 T R =
2R R
b
=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
32
: "shiv rpi¨
Receiver Structure & Matched Filter
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
33
: "shiv rpi¨
Demodulation and detection
Maior sources oI errors:
%hermal noise (AWGN)
disturbs the signal in an additive Iashion (Additive)
has Ilat spectral density Ior all Irequencies oI interest (White)
is modeled by Gaussian random process (Gaussian Noise)
Inter-Symbol InterIerence (ISI)
Due to the Iiltering eIIect oI transmitter, channel and receiver, symbols
are 'smeared¨.
Format
Pulse
modulate
Bandpass
modulate
Format Detect
Demod.
& sample
) (t s
i
) (t g
i
i
2
i
2`
) (t r ) (T z
channel
) (t h
c
) (t n
transmitted symbol
estimated symbol
i , , 1 =
M-ary modulation
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
34
: "shiv rpi¨
Impact of AWGN
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
35
: "shiv rpi¨
Impact of AWGN & Channel Distortion
) 75 . 0 ( 5 . 0 ) ( ) ( T t t t h
c
= o o
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
36
: "shiv rpi¨
Receiver job
Demodulation and sampling:
WaveIorm recovery and preparing the received signal Ior
detection:
Improving the signal power to the noise power (SNR)
using matched Iilter (proiect to signal space)
Reducing ISI using equalizer (remove channel distortion)
Sampling the recovered waveIorm
Detection:
Estimate the transmitted symbol based on the received sample
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
37
: "shiv rpi¨
Receiver structure
Frequency
down-conversion
Receiving
filter
Equalizing
Iilter
%hreshold
comparison
For bandpass signals
Compensation Ior
channel induced ISI
Baseband pulse
(possibly distorted)
Sample
(test statistic)
Baseband pulse
Received waveIorm
Step 1 waveIorm to sample transIormation Step 2 decision making
) (t r
) (T z
i
2`
Demodulate & Sample Detect
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
38
: "shiv rpi¨
Baseband vs Bandpass
Bandpass model oI detection process is equivalent to baseband
model because:
%he received bandpass waveIorm is Iirst transIormed to a
baseband waveIorm.
Equivalence theorem:
PerIorming bandpass linear signal processing Iollowed by
heterodying the signal to the baseband, .
. yields the same results as .
. heterodying the bandpass signal to the baseband ,
Iollowed by a baseband linear signal processing.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
39
: "shiv rpi¨
Steps in designing the receiver
Find optimum solution Ior receiver design with the
Iollowing goals:
1. Maximize SNR: 2,tche/fiter
2. Minimize ISI: equ,izer
Steps in design:
Model the received signal
Find separate solutions Ior each oI the goals.
First, we Iocus on designing a receiver which maximizes the
SNR: 2,tche/fiter
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
40
: "shiv rpi¨
Receiver filter to maximie the SNR
Model the received signal
SimpliIy the model (ideal channel assumption):
Received signal in AWGN
) (t h
c
) (t s
i
) (t n
) (t r
) (t n
) (t r
) (t s
i
Ideal channels
) ( ) ( t t h
c
o =
AWGN
AWGN
) ( ) ( ) ( ) ( t n t h t s t r
c i
+ - =
) ( ) ( ) ( t n t s t r
i
+ =
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
41
: "shiv rpi¨
Matched Filter Receiver
Problem:
Design the receiver Iilter such that the SNR is
maximized at the sampling time when
is transmitted.
Solution:
%he optimum Iilter, is the Matched Iilter, given by
which is the ti2ereverse/and /e,ve/version oI the coniug,te oI the
transmitted signal
) (t h
i t s
i
,..., 1 ), ( =
T 0 t T 0 t
) ( ) ( ) (
*
t T s t h t h
i o5t
= =
) 2 exp( ) ( ) ( ) (
*
fT i f S f H f H
i o5t
x = =
) (t s
i
) ( ) ( t h t h
o5t
=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
42
: "shiv rpi¨
Correlator Receiver
%he matched Iilter output at the sampling time, can be realized
as the correlator output.
Matched Iiltering, i.e. convolution with s
i
*
(T-t simpliIies to
integration w/ s
i
*
(t i.e. correlation or inner product!
> =· =
- =
¦
) ( ), ( ) ( ) (
) ( ) ( ) (
*
0
t s t r / s r
T r T h T z
i
T
o5t
t t t
Recall: correlation operation is the proiection oI the received
signal onto the signal space!
evi/e,: Reiect the noise (N) outside this space as irrelevant:
÷~ maximize S/N
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
43
: "shiv rpi¨
Irrelevance %heorem: Noise Outside Signal Space
Noise PSD is Ilat ('white¨) ÷~ total noise power
inIinite across the spectrum.
We care only about the noise proiected in the Iinite
signal dimensions (eg: the bandwidth oI interest).
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
44
: "shiv rpi¨
Correlation is a maximum when two signals are similar in shape, and are in phase (or
'unshiIted' with respect to each other).
Correlation is a measure oI the si2i,ritv between two signals as a Iunction oI ti2eshift
(',g`t between them
When the two signals are similar in shape and unshiIted with respect to each other, their
product is all positive. %his is like constructiveinterference,
%he bre,/th oI the correlation Iunction - where it has signiIicant value - shows Ior howong
the signals remain similar.
Aside: Correlation ffect
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
45
: "shiv rpi¨
Aside: Autocorrelation
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
46
: "shiv rpi¨
Aside: Cross-Correlation &Radar
Figure: shows how the signal can be located within the noise.
A copy oI the known reIerence signal is correlated with the unknown signal.
%he correlation will be high when the reIerence is similar to the unknown signal.
A large value Ior correlation shows the degree oI conIidence that the reIerence signal is
detected.
%he large value oI the correlation indicates when the reIerence signal occurs.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
47
: "shiv rpi¨
· A copy of a known reference signal is correlated with the unknown signal.
· The correlation will be high if the reference is similar to the unknown signal.
· The unknown signal is correlated with a number of known reference functions.
· A large value for correlation shows the degree of similarity to the reference.
· The largest value for correlation is the most likely match.
· Same principle in communications: reference signals corresponding to
symbols
· The ideal communications channel may have attenuated, phase shifted the
reference signal, and added noise
Source: Bores Signal Processing
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
48
: "shiv rpi¨
Matched Filter: back to cartoon.
Consider the received signal as a vector r. and the transmitted signal vector as s
Matched Iilter 'proiects¨ the r onto signal space spanned by s ('matches¨ it)
iltered signal can now be safely sampled by the receiver at the correct sampling instants,
resulting in a correct interpretation of the binary message
Matched filter is the filter that maximizes the signal-to-noise ratio it can be
shown that it also minimizes the BER: it is a simple projection operation
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
49
: "shiv rpi¨
xample of matched filter (real signals)
T t T t T t 0 T
) ( ) ( ) ( t h t s t v
o5t i
- =
2
A
) (t s
i
) (t h
o5t
T t T t T t 0 T
) ( ) ( ) ( t h t s t v
o5t i
- =
2
A
) (t s
i
) (t h
o5t
T/ T/ T/ T T/
2
2
A

T
A
T
A
T
A
T
A
T
A
T
A
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
50
: "shiv rpi¨
Properties of the Matched Filter
1. %he Fourier transIorm oI a matched Iilter output with the matched signal as input
is, except Ior a time delay Iactor, proportional to the ESD oI the input signal.
2. %he output signal oI a matched Iilter is proportional to a shiIted version oI the
autocorrelation Iunction oI the input signal to which the Iilter is matched.
3. %he output SNR oI a matched Iilter depends only on the ratio oI the signal energy
to the PSD oI the white noise at the Iilter input.
4. %wo matching conditions in the matched-Iiltering operation:
spectral phase matching that gives the desired output peak at time T.
spectral amplitude matching that gives optimum SNR to the peak value.
) 2 exp( ' ) ( ' ) (
2
fT i f S f Z x =
s s s
E R T z T t R t z = = ÷ = ) 0 ( ) ( ) ( ) (
2 /
max
0

E

S
s
T
=

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
51
: "shiv rpi¨
Implementation of matched filter receiver
¦
¦
¦
¦
¦

z
z

1
=
) (t r
) (
1
T z
) (
*
1
t T s
) (
*
t T s

) (T z

Bank of M matched filters
Matched filter output:
Observation
vector
) ( ) ( t T s t r z i
i
- =
-
i ,..., 1 =
) ,..., , ( )) ( ),..., ( ), ( (
2 1 2 1
z z z T z T z T z = =
Note: we are proiecting along the basis directions oI the signal space
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
52
: "shiv rpi¨
Implementation of correlator receiver
/t t s t r z
i
T
i
) ( ) (
0
¦
=
¦
T
0
) ( 1 t s
-
¦
T
0
) (t s
-
¦
¦
¦
¦
¦

z
z

1
=
) (t r
) (
1
T z
) (T z

Bank of M correlators
Correlators output:
Observation
vector
) ,..., , ( )) ( ),..., ( ), ( (
2 1 2 1
z z z T z T z T z = =
i ,..., 1 =
Note: In previous slide we 'Iilter¨ i.e. convolute in the boxes shown.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
53
: "shiv rpi¨
Implementation example of matched filter receivers
¦
¦
¦
¦
¦

2
1
z
z
=
) (t r
) (
1
T z
) (
2
T z

Bank of 2 matched filters
T t
) (
1
t s
T t
) (
2
t s
T
T 0
0
T
A
T
A
T
A
T
A
0
0
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
54
: "shiv rpi¨
Matched FiIter: Frequency domain View
Simple Bandpass ilter:
excludes noise, but misses some signal power
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
55
: "shiv rpi¨
Multi-Bandpass ilter: includes more signal power, but adds more noise also!
Matched ilter: includes more signal power, weighted according to size
=> maximal noise rejection!
Matched FiIter: Frequency Domain View (Contd)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
56
: "shiv rpi¨
Maximal Ratio Combining (MRC) viewpoint
Generalization oI this I-domain picture, Ior combining
multi-tap signal
Weight each branch
SNR:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
57
: "shiv rpi¨
xamples of matched filter output for bandpass
modulation schemes
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
58
: "shiv rpi¨
Signal Space Concepts
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
59
: "shiv rpi¨
Signal space: Overview
What is a signal space?
Vector representations oI 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: %he received signal is transIormed to a received
vectors.
%he signal which has the minimum distance to the received signal is
estimated as the transmitted signal.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
60
: "shiv rpi¨
Schematic example of a signal space
) , ( ) ( ) ( ) (
) , ( ) ( ) ( ) (
) , ( ) ( ) ( ) (
) , ( ) ( ) ( ) (
2 1 2 2 1 1
32 31 3 2 32 1 31 3
22 21 2 2 22 1 21 2
12 11 1 2 12 1 11 1
z z t z t z t z
, , t , t , t s
, , t , t , t s
, , t , t , t s
= ÷ + =
= ÷ + =
= ÷ + =
= ÷ + =

s
s
s
o o
o o
o o
o o
) (
1
t o
) (
2
t o
) , (
12 11 1
, , = s
) , (
22 21 2
, , = s
) , (
32 31 3
, , = s
) , (
2 1
z z =
%ransmitted signal
alternatives
Received signal at
matched Iilter output
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
61
: "shiv rpi¨
Signal space
%o Iorm a signal space, Iirst we need to know the
inner product between two signals (Iunctions):
Inner (scalar) product:
Properties oI inner product:
¦
·
·
>= · /t t v t x t v t x ) ( ) ( ) ( ), (
*
÷ cross-correlation between x(t) and y(t)
> · >= · ) ( ), ( ) ( ), ( t v t x , t v t ,x
> · >= · ) ( ), ( ) ( ), (
*
t v t x , t ,v t x
> · + > >=· + · ) ( ), ( ) ( ), ( ) ( ), ( ) ( t z t v t z t x t z t v t x
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
62
: "shiv rpi¨
Signal space .
%he distance in signal space is measure by calculating the norm.
What is norm?
Norm oI a signal:
Norm between two signals:
We reIer to the norm between two signals as the Euclidean distance between
two signals.
x
E /t t x t x t x t x = = > · =
¦
·
·
2
) ( ) ( ), ( ) (
) ( ) ( t x , t ,x =
) ( ) (
,
t v t x /
v x
=
÷ 'length¨ oI x(t)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
63
: "shiv rpi¨
xample of distances in signal space
) (
1
t o
) (
2
t o
) , (
12 11 1
, , = s
) , (
22 21 2
, , = s
) , (
32 31 3
, , = s
) , (
2 1
z z =
z s
/
,
1
z s
/
,
2
z s
/
,
3
%he Euclidean distance between signals z(t and s(t:
3 , 2 , 1
) ( ) ( ) ( ) (
2
2 2
2
1 1 ,
=
+ = =
i
z , z , t z t s /
i i i z s
i
1
E
3
E
2
E
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
64
: "shiv rpi¨
Orthogonal signal space
N-dimensional orthogonal signal space is characterized by N
linearly independent Iunctions called basis Iunctions.
%he basis Iunctions must satisIy the orthogonality condition
where
II all , the signal space is orthonormal.

Constructing Orthonormal basis Irom non-orthonormal set oI vectors:
Gram-Schmidt procedure
=

i
i
t
1
) (
=
o
ii i i
T
i i i
/t t t t t o o o o o = >= ·
¦
) ( ) ( ) ( ), (
*
0
T t A A 0
i i ,..., 1 , =
|
´
|
= ÷
= ÷
=
i i
i i
ii
0
1
o
1 =
i

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
65
: "shiv rpi¨
xample of an orthonormal bases
Example: /i2ension, orthonormal signal space
Example: /i2ension, orthonornal signal space
1 ) ( ) (
0 ) ( ) ( ) ( ), (
0 ) / 2 sin(
2
) (
0 ) / 2 cos(
2
) (
2 1
2
0
1 2 1
2
1
= =
= >= ·

|

´
|
· A =
· A =
¦
t t
/t t t t t
T t T t
T
t
T t T t
T
t
T
o o
o o o o
x o
x o
T t
) (
1
t o
T
1
0
) (
1
t o
) (
2
t o
0
1 ) (
1
= t o
) (
1
t o
0
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
66
: "shiv rpi¨
Sine/Cosine Bases: Note!
A55roxi2,tev orthonormal!
%hese are the in-phase & quadrature-phase dimensions oI
complex baseband equivalent representations.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
67
: "shiv rpi¨
xample: BPSK
Note: two symbols, but only one dimension in BPSK.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
68
: "shiv rpi¨
Signal space .
Any arbitrary Iinite set oI waveIorms
where each member oI the set is oI duration T, can be expressed
as a linear combination oI N orthogonal waveIorms
where .
where
=

i i
t s
1
) (
=
=

i
i
t
1
) (
=
o A

=
=

i
i ii i
t , t s
1
) ( ) ( o
i ,..., 1 =
A
/t t t s

t t s

,
T
i i
i
i i
i
ii
) ( ) (
1
) ( ), (
1
0
*
¦
>= · = o o T t A A 0
i ,..., 1 =
i ,..., 1 =
) ,..., , (
2 1 i i i i
, , , = s
2
1
ii

i
i i
, E

=
=
Vector representation oI waveIorm
WaveIorm energy
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
69
: "shiv rpi¨
Signal space .

=
=

i
i ii i
t , t s
1
) ( ) ( o
) ,..., , (
2 1 i i i i
, , , = s
¦
¦
¦
¦
¦

i
i
,
,

1
) (
1
t o
) (t

o
1 i
,
i
,
) (t s
i
¦
T
0
) (
1
t o
¦
T
0
) (t

o
¦
¦
¦
¦
¦

i
i
,
,

1
2
s =
) (t s
i
1 i
,
i
,
2
s
Waveform to vector conversion Vector to waveform conversion
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
70
: "shiv rpi¨
xample of projecting signals to an
orthonormal signal space
) , ( ) ( ) ( ) (
) , ( ) ( ) ( ) (
) , ( ) ( ) ( ) (
32 31 3 2 32 1 31 3
22 21 2 2 22 1 21 2
12 11 1 2 12 1 11 1
, , t , t , t s
, , t , t , t s
, , t , t , t s
= ÷ + =
= ÷ + =
= ÷ + =
s
s
s
o o
o o
o o
) (
1
t o
) (
2
t o
) , (
12 11 1
, , = s
) , (
22 21 2
, , = s
) , (
32 31 3
, , = s
%ransmitted signal
alternatives
/t t t s ,
T
i i ii
) ( ) (
0
¦
= o
T t A A 0
i ,..., 1 =
i ,..., 1 =
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
71
: "shiv rpi¨
Matched filter receiver (revisited)
) (t r
1
z
) ( 1 t T
-
o
) ( t T
-
o

z
Bank of N matched filters
Observation
vector
) ( ) ( t T t r z
i i
- = o
i ,..., 1 =
) ,..., , (
2 1
z z z =

=
=

i
i ii i
t , t s
1
) ( ) ( o
A
i ,..., 1 =
¦
¦
¦
¦
¦

z
z
1
=

(note.we2,tchtothe-ass dr0c943s
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
72
: "shiv rpi¨
Correlator receiver (revisited)
) ,..., , (
2 1
z z z =
i ,..., 1 = /t t t r z
i
T
i
) ( ) (
0
o
¦
=
¦
T
0
) (
1
t o
¦
T
0
) (t

o
¦
¦
¦
¦
¦

r
r

1
=
) (t r
1
z

z

Bank of N correlators
Observation
vector

=
=

i
i ii i
t , t s
1
) ( ) ( o i ,..., 1 =
A
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
73
: "shiv rpi¨
xample of matched filter receivers using
basic functions
u2berof2,tche/fiters(orcorre,torsisre/uce/bv compared to
using matched Iilters (correlators) to the transmitted signal!
T t
) (
1
t s
T t
) (
2
t s
T t
) (
1
t o
T
1
0
)
1
z
=
) (t r
matched filter
T t
) (
1
t o
T
1
0
1
z
T
A
T
A
0
0
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
74
: "shiv rpi¨
White noise in Orthonormal Signal Space
AWGN n(t can be expressed as
) (
~
) (
`
) ( t n t n t n + =
Noise projected on the signal space
(colored):impacts the detection process.
Noise outside on the signal space
(irrelevant)
> =· ) ( ), ( t t n n
i i
o
0 ) ( ), (
~
>= · t t n
i
o
) ( ) ( `
1
t n t n

i
i i
=
= o
i ,..., 1 =
i ,..., 1 =
Vector representation of
) ,..., , (
2 1
n n n = n
) ( ` t n
independent zero-mean
Gaussain random variables with
variance
=

i
i
n
1 =
2 / ) var(
0
n
i
=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
75
: "shiv rpi¨
Detection: Maximum Likelihood &
Performance Bounds
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
76
: "shiv rpi¨
Detection of signal in AWGN
Detection problem:
Given the observation vector , perIorm a mapping Irom
to an estimate oI the transmitted symbol, , such that
the average probability oI error in the decision is
minimized.
2
`
i
2
Modulator Decision rule
2
`
i
2

i
s
n

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
77
: "shiv rpi¨
Statistics of the observation Vector
AWGN channel model:
Signal vector is deterministic.
Elements oI noise vector are i.i./,ussi,n
random variables with zero-mean and variance . %he
noise vector pdI is
%he elements oI observed vector are
independent Gaussian random variables. Its pdI is
) ,..., , (
2 1 i i i i
, , , = s
) ,..., , (
2 1
z z z =
) ,..., , (
2 1
n n n = n
n s + =
i
2 /
0

=
0
2
2 /
0
exp
1
) (

5

n
n
n
x

=
0
2
2 /
0
exp
1
) ' (

5
i

i
s
s

x
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
78
: "shiv rpi¨
Detection
Optimum decision rule (2,xi2u2,5osteriori5rob,biitv):
Applying Bayes` rule gives:
. ,..., 1 where
all Ior , ) ' sent Pr( ) ' sent Pr(
iI ` Set

i 2 2
2 2
i
i
=
= >
=

i
5
2 5
5
2 2

i
=
=
all Ior maximum is ,
) (
) ' (
iI ` Set

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
79
: "shiv rpi¨
Detection .
Partition the signal space into decision regions,
such that

Z Z ,...,
1
i

i
2 2
i
5
2 5
5
Z
=
=
`
means %hat
. all Ior maximum is |,
) (
) ' (
ln|
iI region inside lies Vector

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
80
: "shiv rpi¨
Detection (ML rule)
For equal probable symbols, the optimum decision rule
(maximum posteriori probability) is simpliIied to:
or equivalently:
which is known as 2,xi2u2ieihoo/.
i 2 5
2 2

i
=
=
all Ior maximum is ), ' (
iI ` Set

i 2 5
2 2

i
=
=
all Ior maximum is )|, ' ( ln|
iI ` Set

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
81
: "shiv rpi¨
Detection (ML).
Partition the signal space into decision regions,
Restate the maximum likelihood decision rule as Iollows:

Z Z ,...,
1
i

i
2 2
i 2 5
Z
=
=
`
means %hat
all Ior maximum is )|, ' ( ln|
iI region inside lies Vector

i
Z

i
= all Ior minimum is ,
iI region inside lies Vector
s

Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
82
: "shiv rpi¨
Schematic example of ML decision regions
) (
1
t o
) (
2
t o
1
s
3
s
2
s
4
s
1
Z
2
Z
3
Z
4
Z
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
83
: "shiv rpi¨
Probability of symbol error
Erroneous decision: For the transmitted symbol or equivalently signal
vector , an error in decision occurs iI the observation vector does not
Iall inside region .
Probability oI erroneous decision Ior a transmitted symbol
Probability oI correct decision Ior a transmitted symbol
sent) inside lie not does sent)Pr( Pr( ) ` Pr(
i i i i
2 Z 2 2 2 = =
sent) inside lies sent)Pr( Pr( ) ` Pr(
i i i i
2 Z 2 2 2 = =
¦
= =
i
Z
i i i i c
/ 2 5 2 Z 2 !

) ' ( sent) inside lies Pr( ) (
) ( 1 ) (
i c i e
2 ! 2 ! =
i
2
i
s

i
Z
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
84
: "shiv rpi¨
xample for binary PAM
) (
1
t o
b
E
b
E
0
1
s
2
s
) ' (
1
2 5

) ' (
2
2 5

2
) 2 (
0

= =

E
" ! !
b
E B

2 /
2 /
) ( ) (
0
2 1
2 1

= =

" 2 ! 2 !
e e
s s
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
85
: "shiv rpi¨
;0rag0 prob. of symbol error .
Average probability oI symbol error :
For equally probable symbols:
) ' (
1
1
) (
1
1 ) (
1
) (
1
1 1

¦

=
= =
=
= =

i
Z
i

i
i c

i
i e E
i
/ 2 5

2 !

2 !

!

) ` ( Pr ) (
1
i

i
E
2 2 ! = =

=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
86
: "shiv rpi¨
Union bound
nion bound
The probability of a finite union of events is upper bounded
by the sum of the probabilities of the individual events.

=
=
A

i

i i e
! 2 !
1
2
) , ( ) ( s s

=
=
=
A

i

i

i E
!

!
1 1
2
) , (
1
) ( s s
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
87
: "shiv rpi¨
Union bound:
xample of union bound
1
o
1
s
4
s
2
s
3
s
1
Z
4
Z
3
Z
2
Z r
2
o
¦

=
4 3 2
) ' ( ) (
1 1
Z Z Z
e
/ 2 5 2 ! r r
r
1
o
1
s
4
s
2
s
3
s
2
A
r
1
o
1
s
4
s
2
s
3
s
3
A
r
1
o
1
s
4
s
2
s
3
s
4
A
r
2
o
2
o
2
o
¦
=
2
) ' ( ) , (
1 1 2 2
A
/ 2 5 ! r r s s
r
¦
=
3
) ' ( ) , (
1 1 3 2
A
/ 2 5 ! r r s s
r
¦
=
4
) ' ( ) , (
1 1 4 2
A
/ 2 5 ! r r s s
r

=
A
4
2
1 2 1
) , ( ) (

e
! 2 ! s s
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
88
: "shiv rpi¨
Upper bound based on minimum distance

= =
=
¦
·

2 /
2 /
) exp(
1
sent) is when , than closer to is Pr( ) , (
0
0
2
0
2

/
" /u

u

!
i
/
i i i
i
x
s s s s s
i i
/ s s =

A A

=
=
=
2 /
2 /
) 1 ( ) , (
1
) (
0
min
1 1
2

/
" !

!

i

i

i E
s s
i
i
i
/ /
=
=
,
min
min
Minimum distance in the signal space:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
89
: "shiv rpi¨
xample of upper bound on av. Symbol error prob.
based on union bound
) (
1
t o
) (
2
t o
1
s
3
s
2
s
4
s
2 , 1
/
4 , 3
/
3 , 2
/
4 , 1
/
s
E
s
E
s
E
s
E
4 ,..., 1 , = = = i E E
s i i
s
s
s i
E /
i
E /
2
2
min
,
=
=
=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
90
: "shiv rpi¨
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
91
: "shiv rpi¨
b/No figure of merit in digital
communications
SNR or S/N is the average signal power to the average noise
power. SNR should be modiIied in terms oI bit-energy in
digital communication, because:
Signals are transmitted within a symbol duration and hence,
are 030rgv sg3al (z0r4 p40r).
A metric at the -9l0;0l Iacilitates comparison oI diIIerent
DCS transmitting diIIerent number oI bits per symbol.
b
b b
R
W

S
W
ST

E
= =
/
0
b
R
W
: Bit rate
: Bandwidth
Note: S/N ÷ Eb/No x spectral eIIiciency
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
92
: "shiv rpi¨
xample of Symbol error prob. For PAM signals
) (
1
t o
0
1
s 2
s
b
E
b
E
Binary PAM
) (
1
t o 0
2
s
3
s
5
2
b
E
5
6
b
E
5
6
b
E

5
2
b
E

4
s
1
s
4-ary PAM
T t
) (
1
t o
T
1
0
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
93
: "shiv rpi¨
Maximum Likelihood (ML) Detection: '0c94r Case
ps: Vector norm is a natural extension oI 'magnitude¨ or length
Nearest Neighbor Rule:
Proiect the received vector y along the
diIIerence vector direction u
A
- u
B
is a
'sufficientst,tistic¨.
Noise outside these Iinite dimensions is
irreev,nt Ior detection. (rotational
invariance oI detection problem)
By the isotro5ic property oI the Gaussian noise, we expect the error
probability to be the same Ior both the transmit symbols u
A
, u
B
.
Error probability:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
94
: "shiv rpi¨
xtension to M-PAM (Multi-Level Modulation)
Note: h refers to the constellation shape/direction
MPAM:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
95
: "shiv rpi¨
Complex Vector Space Detection
Error probability:
Note: Instead oI v
%
, use v` Ior complex vectors ('tr,ns5ose,n/coniug,te¨)
Ior inner products.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
96
: "shiv rpi¨
Complex
Detection:
Summary
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
97
: "shiv rpi¨
Detection rror ÷> BR
II the bit error is i.i.d (discrete memoryless channel) over the sequence oI bits,
then you can model it as a binary symmetric channel (BSC)
BER is modeled as a uniIorm probability f
As BER (fincreases, the eIIects become increasingly intolerable
f tends to increase rapidly with lower SNR: 'w,terf,¨ curve ("-Iunction)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
98
: "shiv rpi¨
SNR vs BR: AWGN vs Rayleigh
Observe the 'waterIall¨ like characteristic (essentially plotting the "(x) Iunction)!
%elephone lines: SNR ÷ 37dB, but low b/w (3.7kHz)
Wireless: Low SNR ÷ 5-10dB, higher bandwidth (upto 10 Mhz, MAN, and 20Mhz LAN)
Optical Iiber comm: High SNR, high bandwidth ! But cant process w/ complicated codes,
signal processing etc
Need
diversity
techniques
to deal with
Rayleigh
(even 1-tap,
Ilat-Iading)!
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
99
: "shiv rpi¨
Better performance through diversity
Diversity the receiver is provided with multiple copies
of the transmitted signal. The multiple signal copies
should experience uncorrelated fading in the channel.
In this case the probability that all signal copies fade
simultaneously is reduced dramatically with respect to
the probability that a single copy experiences a fade.
As a rough rule:
0
1
e

!

is proportional to
BER BER Average SNR Average SNR
Diversity of
:th order
Diversity of
:th order
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
100
: "shiv rpi¨
SNR
BER
Flat fading channel,
Rayleigh fading,
= 1
AWGN
channel
(no fading)
( )
e
! =
0
( ) =
BER vs. SNR (diversity effect)
= 2 = 4 = 3
We will explore this story later. slide set part II
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
101
: "shiv rpi¨
Modulation %echniques
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
102
: "shiv rpi¨
What is Modulation?
Encoding inIormation in a manner suitable Ior
transmission.
%ranslate baseband source signal to bandpass signal
Bandpass signal: 'modulated signal¨
How?
Vary amplitude, phase or Irequency oI a carrier
Demodulation: extract baseband message Irom carrier
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
103
: "shiv rpi¨
Digital vs Analog Modulation
Cheaper, Iaster, more power eIIicient
Higher data rates, power error correction, impairment
resistance:
Using coding, modulation, diversity
Equalization, multicarrier techniques Ior ISI
mitigation
More eIIicient multiple access strategies, better
security: CDMA, encryption etc
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
104
: "shiv rpi¨
Goals of Modulation %echniques
W High Bit Rate
W High SpectraI Efficiency
W High Power Efficiency
W Low-Cost/Low-Power ImpIementation
W Robustness to Impairments
(2in5owerto,chieve,t,rgetBER)
(2,xB5s/Hz)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
105
: "shiv rpi¨
Modulation: representation
W Any moduIated signaI can be represented as
s(t) = A(t) cos
c
t + (t)(
s(t)
= A(t) cos (t) cos
c
t - A(t) sin (t) sin
c
t
ampIitude
in-phase
quadrature
phase or frequency
W Linear versus nonIinear moduIation ÷ impact on spectral efficiency
W Constant enveIope versus non-constant envelope
÷ hardware impIications with impact on power efficiency
Linear: Amplitude or phase
Non-linear: Irequency: spectral broadening
(÷~ reliability: i.e. target BER at lower SNRs)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
106
: "shiv rpi¨
Complex Vector Spaces: Constellations
Each signal is encoded (modulated) as a vector in a signal space
MPSK
Circular
Square
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
107
: "shiv rpi¨
Linear Modulation %echniques
s(t) = % a
n
g (t-nT)(cos
c
t - % b
n
g (t-nT)( sin
c
t
I(t), in-phase
Q(t), quadrature
LINEAR MODULATIONS
CONVENTIONAL
4-PSK
(QPSK)
OFFSET
4-PSK
(OQPSK)
DIFFERENTIAL
4-PSK
(DQPSK, x/4-DQPSK)
M-ARY QUADRATURE
AMPLITUDE MOD.
(M-QAM)
M-ARY PHASE
SHIFT KEYING
(M-PSK)
M = 4 M = 4 M=4
(4-QAM = 4-PSK)
Square
ConsteIIations
CircuIar
ConsteIIations
n n
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
108
: "shiv rpi¨
M-PSK and M-QAM
M-PSK (CircuIar ConsteIIations)
16-PSK
an
b
n
4-PSK
M-QAM (Square ConsteIIations)
16-QAM
4-PSK
a
n
b
n
Tradeoffs
- Higher-order moduIations (M Iarge) are 24re spectrally
efficient but less p4er efficient (i.e. BER higher).
- M-QAM is more spectraIIy efficient than M-PSK but
aIso more sensitive to system nonIinearities.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
109
: "shiv rpi¨
Bandwidth vs. Power fficiency
Source: Rappaport book. chap 6
MPSK:
MQAM:
MFSK:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
110
: "shiv rpi¨
MPAM & Symbol Mapping
Note: the average energy
per-bit is constant
Gray coding used Ior
mapping bits to symbols
Why? Most likely error is to
conIuse with neighboring
symbol.
Make sure that the
neighboring symbol has only
1-bit diIIerence (hamming
distance ÷ 1)
) (
1
t o
0
1
s 2
s
b
E
b
E
Binary PAM
) (
1
t o 0
2
s
3
s
5
2
b
E
5
6
b
E
5
6
b
E

5
2
b
E

4
s
1
s
4-ary PAM
) (
1
t o 0
2
s
3
s
5
2
b
E
5
6
b
E
5
6
b
E

5
2
b
E

4
s
1
s
4-ary PAM
Gray coding
00
0 0
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
111
: "shiv rpi¨
MPAM: Details
Decision Regions
Unequal energies/symbol:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
112
: "shiv rpi¨
MPSK:
Constellation points:
Equal energy in all signals:
Gray coding
00
0

0
00
0

0
M-PSK (CircuIar ConsteIIations)
16-PSK
an
b
n
4-PSK
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
113
: "shiv rpi¨
MPSK: Decision Regions & Demod`ln
4PSK
8PSK
Coherent Demodulator for BPSK.
s
i
(tn(t
cos(f
c
t
g(T
b
t X
Z

.r0
Z

.r^ 0

2÷0
2÷0or
Z
1
Z
2
Z
3
Z
4
Z
1
Z
2
Z
3
Z
4
Z
5
Z
6
Z
7
Z
8
4PSK: 1 bit/complex dimension or 2 bits/symbol
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
114
: "shiv rpi¨
MQAM:
Unequal symbol energies:
M"AM with square constellations oI size
2
is
equivalent to MPAM modulation with
constellations oI size on each oI the in-phase
and quadrature signal components
For square constellations it takes approximately
6 dB more power to send an additional 1
bit/dimension or 2 bits/symbol while
maintaining the same minimum distance
between constellation points
Hard to Iind a Gray code mapping where all
adiacent symbols diIIer by a single bit
M-QAM (Square ConsteIIations)
16-QAM
4-PSK
a
n
b
n
6QAM: Decision Regions
Z
1
Z
2
Z
3
Z
4
Z
5
Z
6
Z
7
Z
8
Z
9
Z
10
Z
11
Z
12
Z
13
Z
14
Z
15
Z
16
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
115
: "shiv rpi¨
Non-Coherent Modulation: DPSK
InIormation in MPSK, M"AM carried in signal phase.
Requires coherent demodulation: i.e. phase oI the transmitted signal carrier …
0
must
be matched to the phase oI the receiver carrier …
More cost, susceptible to carrier phase driIt.
Harder to obtain in Iading channels
Differential modulation: do not require phase reIerence.
More general: modulation w/ memory: depends upon prior symbols transmitted.
Use prev symbol as the a phase reIerence Ior current symbol
InIo bits encoded as the diIIerential phase between current & previous symbol
Less sensitive to carrier phase driIt (I-domain) ; more sensitive to doppler
eIIects: decorrelation oI signal phase in time-domain
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
116
: "shiv rpi¨
Differential Modulation (Contd)
DPSK: DiIIerential BPSK
A 0 bit is encoded by no change in phase, whereas a 1 bit is encoded as a
phase change oI .
II symbol over time |(1)T
s
T
s
) has phase 0(1) ÷ e
i0i
, 0
i
÷ 0,
then to encode a 0 bit over |Ts( 1)Ts), the symbol would have
phase: 0() ÷ e
i0i
and.
. to encode a 1 bit the symbol would have
phase 0() ÷ e
i(0i
.
DQPSK: gray coding:
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
117
: "shiv rpi¨
Quadrature Offset
Phase transitions oI 180
o
can cause ,rge,25itu/etr,nsitions(through
zero5oint.
Abrupt phase transitions and large amplitude variations can be distorted
by nonlinear ampliIiers and Iilters
Avoided by oIIsetting the quadrature branch pulse g(t) by halI a symbol
period
Usually abbreviated as O-MPSK, where the O indicates the oIIset
"PSK modulation with quadrature oIIset is reIerred to as O-"PSK
O-"PSK has the same spectral properties as "PSK Ior linear
ampliIication,..
. but has higher spectral eIIiciency under nonlinear ampliIication,
since the maximum phase transition oI the signal is 90 degrees
Another technique to mitigate the amplitude Iluctuations oI a 180 degree
phase shiIt used in the IS-54 standard Ior digital cellular is /4-"PSK
Maximum phase transition oI 135 degrees, versus 90 degrees Ior oIIset
"PSK and 180 degrees Ior "PSK
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
118
: "shiv rpi¨
Offset QPSK waveforms
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
119
: "shiv rpi¨
Frequency Shift Keying (FSK)
W Continuous Phase FSK (CPFSK)
- digitaI data encoded in the frequency shift
- typicaIIy impIemented with frequency
moduIator to maintain continuous phase
- nonIinear moduIation but constant-enveIope
W Minimum Shift Keying (MSK)
- minimum bandwidth, sideIobes Iarge
- can be impIemented using I-Q receiver
W Gaussian Minimum Shift Keying (GMSK)
- reduces sideIobes of MSK using a premoduIation fiIter
- used by RAM MobiIe Data, CDPD, and HIPERLAN
s(t) = A cos
c
t + 2 xk
f
¦ d(t) dt(
·
t
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
120
: "shiv rpi¨
Minimum Shift Keying (MSK) spectra
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
121
: "shiv rpi¨
Spectral Characteristics
QPSK/DQPSK
GMSK
(MSK)
1.0
1.0 1.5 2.0 2.5 0.5 0
-120
-100
-80
-60
-40
-20
0
10
0.25
ß
3-dß
T
b
= 0.1ô
NormaIized Frequency (f-f
c
)T
b
P
o
w
e
r

S
p
e
c
t
r
a
I

D
e
n
s
i
t
y

(
d
B
)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
122
: "shiv rpi¨
Bit rror Probability (BR): AWGN
P
b
BPSK, QPSK
DBPSK
DQPSK
0
10
-6
2
5
2
5
2
5
2
5
2
5
10
-1
10
-3
10
-4
10
-5
10
-2
2 4 6 8 10 12 14
For P
b
= 10
-3
BPSK 6.5 dB
QPSK 6.5 dB
DBPSK ~8 dB
DQPSK ~9 dB

b
, SNR/bit, dB
W QPSK is more spectraIIy efficient than BPSK
with the same performance.
W M-PSK, for M>4, is more spectraIIy efficient
but requires more SNR per bit.
W There is ~3 dB power penaIty for differentiaI
detection.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
123
: "shiv rpi¨
Bit rror Probability (BR): Fading Channel
DBPSK
AWGN
BPSK
0
10
-5
2
5
2
5
2
5
2
5
2
5
1
10
-2
10
-3
10
-4
10
-1
5 10 15 20 25 30 35
P
b

b
, SNR/bit, dB
W P
b
is inverseIy proportion to the average SNR per bit.
W Transmission in a fading environment requires about
18 dB more power for P
b
= 10
-3
.
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
124
: "shiv rpi¨
Bit rror Probability (BR): Doppler ffects
0
0 60 50 40 30 20 10
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
DQPSK
RayIeigh Fading
No Fading
f
D
T=0.003
0.002
0.001
0
QPSK
P
b

b
, SNR/bit, dB
W DoppIer causes an irreducibIe error fIoor when differentiaI
detection is used ÷ decorreIation of reference signaI.
The impIication is that DoppIer is n4t an issue f4r high-speed
ireless data.
data rate T P
b
fIoor
10 kbps 10
-4
s 3x10
-4
100 kbps 10
-5
s 3x10
-6
1 Mbps 10
-6
s 3x10
-8
W The irreducibIe P
b
depends on the data rate and the DoppIer.
For f
D
= 80 Hz,
[M. D. Yacoub, F4undati4ns 4f M4-ile Radi4 Engineering ,
CRC Press, 1993]
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
125
: "shiv rpi¨
Bit rror Probability (BR): Delay Spread
+
x
+
+
+
+
+
x
x
x
x
x
10
-2
10
-4
10
-3
10
-2
10
-1
10
-1
10
0
BPSK
QPSK
OQPSK
MSK
ModuIation
Coherent Detection
I
r
r
e
d
u
c
i
b
I
e

P
b
t
T
=
rms deIay spread
symboI period
W ISI causes an irreducibIe error fIoor.
W The rms deIay spread imposes a Iimit on the
maximum bit rate in a muItipath environment.
For exampIe, for QPSK,
t Maximum Bit Rate
MobiIe (ruraI) 25 usec 8 kbps
MobiIe (city) 2.5 usec 80 kbps
MicroceIIs 500 nsec 400 kbps
Large BuiIding 100 nsec 2 Mbps
[J. C.-I. Chuang, "The Effects of Time DeIay Spread on PortabIe Radio
Communications ChanneIs with DigitaI ModuIation,"
IEEE JSAC, June 1987]
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
126
: "shiv rpi¨
Summary of Modulation Issues
W Tradeoffs
- Iinear versus nonIinear moduIation
- constant enveIope versus non-constant
enveIope
- coherent versus differentiaI detection
- power efficiency versus spectraI efficiency
W Limitations
- fIat fading
- doppIer
- deIay spread
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
127
: "shiv rpi¨
Pulse Shaping
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
128
: "shiv rpi¨
Recall: Impact of AWGN only
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
129
: "shiv rpi¨
Impact of AWGN & Channel Distortion
) 75 . 0 ( 5 . 0 ) ( ) ( T t t t h
c
= o o
Multi-tap, ISI channel
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
130
: "shiv rpi¨
ISI ffects: Band-limited Filtering of Channel
ISI due to Iiltering eIIect oI the communications
channel (e.g. wireless channels)
Channels behave like band-limited Iilters
) (
) ( ) (
f i
c c
c
e f H f H
7
=
Non-constant amplitude
Amplitude distortion
Non-linear phase
Phase distortion
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
131
: "shiv rpi¨
Inter-Symbol Interference (ISI)
ISI in the detection process due to the Iiltering eIIects
oI the system
Overall equivalent system transIer Iunction
creates echoes and hence time dispersion
causes ISI at sampling time
) ( ) ( ) ( ) ( f H f H f H f H
r c t
=
i
i
i
s n s z

=
+ + = ¬
ISI eIIect
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
132
: "shiv rpi¨
Inter-symbol interference (ISI): Model
Baseband system model
Equivalent model
%x filter
Channel
) (t n
) (t r
Rx. filter
Detector

z
T t =
=

x` =

x
1
x
2
x
3
x
T
T
) (
) (
f H
t h
t
t
) (
) (
f H
t h
r
r
) (
) (
f H
t h
c
c
quivalent system
) ( ` t n
) (t z
Detector

z
T t =
=

x` =

x
1
x
2
x
3
x
T
T
) (
) (
f H
t h
Iiltered noise
) ( ) ( ) ( ) ( f H f H f H f H
r c t
=
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
133
: "shiv rpi¨
Nyquist bandwidth constraint
Nyquist bandwidth constraint (onequiv,entsvste2):
%he theoretical minimum required system bandwidth to
detect R
s
|symbols/s| without ISI is R
s
/ |Hz|.
Equivalently, a system with bandwidth W÷/T÷R
s
/ |Hz|
can support a maximum transmission rate oI W÷/T÷Rs
|symbols/s| without ISI.
Bandwidth eIIiciency, R/W |bits/s/Hz| :
An important measure in DCs representing data
throughput per hertz oI bandwidth.
Showing how eIIiciently the bandwidth resources are used
by signaling techniques.
Hz| |symbol/s/ 2
2 2
1
> ÷ A =
W
R
W
R
T
s s
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
134
: "shiv rpi¨
quiv System: Ideal Nyquist pulse (filter)
T 2
1
T 2
1
T
) ( f H
f t
) / sinc( ) ( T t t h =
1
0
T T 2 T T 2
0
T
W
2
1
=
Ideal Nyquist Iilter Ideal Nyquist pulse
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
135
: "shiv rpi¨
Nyquist pulses (filters)
Nyquist pulses (Iilters):
Pulses (Iilters) which result in no ISI at the sampling time.
Nyquist Iilter:
Its transIer Iunction in Irequency domain is obtained by
convolving a rectangular Iunction with any real even-
symmetric Irequency Iunction
Nyquist pulse:
Its shape can be represented by a sinc(t/%) Iunction multiply
by another time Iunction.
Example oI Nyquist Iilters: Raised-Cosine Iilter
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
136
: "shiv rpi¨
Pulse shaping to reduce ISI
Goals and trade-oII in pulse-shaping
Reduce ISI
EIIicient bandwidth utilization
Robustness to timing error (small side lobes)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
137
: "shiv rpi¨
Raised Cosine Filter: Nyquist Pulse Approximation
2
) 1 ( Baseband
sSB
s
R
r W + =
' ) ( ' ' ) ( ' f H f H
RC
=
0 = r
5 . 0 = r
1 = r
1 = r
5 . 0 = r
0 = r
) ( ) ( t h t h
RC
=
T 2
1
T 4
3
T
1
T 4
3
T 2
1
T
1
1
0.5
0
1
0.5
0 T T 2 T 3 T T 2
T 3
s
R r W ) 1 ( Passband
DSB
+ =
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
138
: "shiv rpi¨
Raised Cosine Filter
Raised-Cosine Filter
A Nyquist pulse (No ISI at the sampling time)

|

´
|
>
· ·
¦
¦
¦

+
·
=
W f
W f W W
W W
W W f
W W f
f H
' ' Ior 0
' ' 2 Ior
2 ' '
4
cos
2 ' ' Ior 1
) (
0
0
0
2
0
x
xcess bandwidth:
0
W W
Roll-oII Iactor
0
0
W
W W
r

=
1 0 A A r
2
0
0
0 0
| ) ( 4 | 1
| ) ( 2 cos|
)) 2 (sinc( 2 ) (
t W W
t W W
t W W t h

=
x
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
139
: "shiv rpi¨
Pulse Shaping and qualiation Principles
Squ,reRootR,ise/Cosine(SRRC Iilter and Equalizer
) ( ) ( ) ( ) ( ) (
RC
f H f H f H f H f H
e r c t
=
No ISI at the sampling time
) ( ) ( ) ( ) (
) ( ) ( ) (
SRRC RC
RC
f H f H f H f H
f H f H f H
t r
r t
= = =
=
%aking care oI ISI
caused by tr. Iilter
) (
1
) (
f H
f H
c
e
=
%aking care oI ISI
caused by channel
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
140
: "shiv rpi¨
Pulse Shaping & Orthogonal Bases
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
141
: "shiv rpi¨
Virtue of pulse shaping
PSD oI a BPSK signal
Source: Rappaport book. chap 6
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
142
: "shiv rpi¨
xample of pulse shaping
Square-root Raised-Cosine (SRRC) pulse shaping
t/T
A25. |V|
Baseband tr. WaveIorm
Data symbol
First pulse
Second pulse
%hird pulse
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
143
: "shiv rpi¨
xample of pulse shaping .
Raised Cosine pulse at the output oI matched Iilter
t/T
A25. |V|
Baseband received waveIorm at
the matched Iilter output
(zero ISI)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
144
: "shiv rpi¨
ye pattern
Eye pattern:Display on an oscilloscope which sweeps the system
response to a baseband signal at the rate /T (T symbol duration)
time scale
a
m
p
l
i
t
u
d
e

s
c
a
l
e
Noise margin
Sensitivity to
timing error
Distortion
due to ISI
%iming iitter
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
145
: "shiv rpi¨
xample of eye pattern:
Binary-PAM. SRRC pulse
PerIect channel (no noise and no ISI)
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
146
: "shiv rpi¨
xample of eye pattern:
Binary-PAM. SRRC pulse .
AWGN (Eb/0÷0 dB) and no ISI
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
147
: "shiv rpi¨
xample of eye pattern:
Binary-PAM. SRRC pulse .
AWGN (Eb/0÷0 dB) and no ISI
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
148
: "shiv rpi¨
Summary
Digital Basics
Modulation & Detection, PerIormance, Bounds
Modulation Schemes, Constellations
Pulse Shaping
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
149
: "shiv rpi¨
xtra Slides
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
150
: "shiv rpi¨
Bandpass Modulation: I. Q Representation
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
151
: "shiv rpi¨
Analog: Frequency Modulation (FM) vs
Amplitude Modulation (AM)
FM: all inIormation in the phase or Irequency oI carrier
Non-linear or rapid improvement in reception quality beyond a minimum
received signal threshold: 'c,5ture`effect.
Better noise immunity & resistance to Iading
%radeoII bandwidth (modulation index) Ior improved SNR: 6dB gain Ior 2x
bandwidth
Constant envelope signal: eIIicient (70°) class C power amps ok.
AM: linear dependence on quality & power oI rcvd signal
Spectrally eIIicient but susceptible to noise & Iading
Fading improvement using in-band pilot tones & adapt receiver gain to
compensate
Non-constant envelope: Power ineIIicient (30-40°) Class A or AB power amps
needed: ½ the talk time as FM!
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
152
: "shiv rpi¨
xample Analog: Amplitude Modulation

%0,8.8

#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

!.9:70090.943:3/07

#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

3/42574.088  3 9 9.850.9$43949083..0 98.:88.90/.9438.3480  %072.0 98!$81.3.9.3 7.3343 074.34808/08.4770.9 :3.4770.7-0/-. .//9.9 03.0/90 3480 :94. .074 20.:88.//9.090.

 0389 1.75 ..943 .2.7.90/ $.943 :3.97.07!490.3   8.( !407850.943 :3.4770.3 :94.9 90 #03880.:88.3.3.7.3899:90 !74-.4770.:2.-9/03891:3.

07!490.84110543039.05439.75 .7.85.3/423480.0....$5.2.947..947 #03880.0 9.3  '0.3.0 %0. .3-0:80/383.3.7.:88.3899:90 $.110.9414803$3.:2.3   8.4:/1.7.

80 044/8 88:23-4982-4806:.790743 $..7.0: 8.07!490.7.3.3   8.75 .7 .:2.2:2044/  090.2.3.3899:90 8250/89.943$..48031 4 044/ #03880.0..3..

07!490.7! 5   2 5   2 8  8 -  9 !0 2  !0 2  " !  !   " #03880.090.3.3899:90 8 8  .9431473.

  .

7.75 .      $.7.:2.2..3   8.3.

3899:90 $4:7.382990/ #0.943889028 480 314 $ &# $4:7.7.2.4/07 4/:.330 &807 %7. 83.9:7041.07!490.:2.9:70 0307..07 .3.7.3  8.0.0 #0.0/ 314 83..0.0 /0.4/07 .9907 #03880. %7.897:.3829907 472.0..4/07 024/:.330 03.9907 $4:7.0/ %7.422:3.4/07  .07!.0 03.0.947 $.75 .3829907 #0.07 472.3.330 /0.947 #0.

3   8..07!490.4422.8.75 .3899:90 $.:2.9.2.8 #03880.83..7.7.3.3.

 '4.-9 $.0 11070393/841/9.4 /.422:3..94770.:2.3..943/89.0841/9.70970.0. 0/..9.3   8.07!490.3.7.75 .90/ /039..90/ 5:80 !745.3899:90 -98.078:8.3.0 .39. #03880..2..83.07 73. 5:80 #00307.3..9438 #00307.9.7..

3.90/ 06:.77.$3.3/039.7.2.83.788902 09072389./894794308897.3899:90 $.3828843 901706:03.7889028 35:9 30.84..  :95:9 #03880.75 .:2..9903:.09 .8 #.70.8 /0.3.07!490.425430398419083.3/4283.920/0.34943.3828843974:30.2510/47.97..3   8. -:9.7.

..$3.88 :.:8.97.19078 4 5.7574-0284.:7. .3828843 .88  5.982.3/ 5.439 / /0.88 43 .

75 .3   8..7.3.2.3:.75:8083920/42.3 #0.9..3899:90 $.:2.7. 70.07!490.-019078 #19078:99074791907 #03880.3.

3/5.7.7. 4.3//9/022.3/5.88 83..3/..947 ..7034970..-083.3899:90 $.3/290/83. .8. .031390-.88 .078:8-.3//9 #03880.2..07!490.80-.3..-0 #0.75 .3/ 83.80-.:2.3.8.3//94183..3   8. .48.

.3//9 / 0 1 7.:2.3899:90 $.3//9 : 94 3:-.3.3//9 .5407.3//9 48006:.. ./0389 -84:90-.3//9 .97.3 0  /   8..3.9438 1107039/01394341-. / #03880.439.2.55742.75 .7.1 5407-..32039-.3//94183.. .3//9 4:3/0/5407850.7.943.039-.07!490.

3/83..382884341-.9 $. 9.250 ":.993.3829 98970.3/97.9 3.472. 314 9.88 1907 0.014728 %7. 84:7.4/0 !:80 24/:.80-.4 314 4 5.330 83 %09:.4 314 472.314 %09:.4/0 .314 024/:.90.2 472.390 3.0 314 3.90 !:80 .

9 #0.2..3899:90 $.0 #03880.3.7.3   8.:2.7.75 .3.0. 090.07!490.

:2.4$3.07!490.3899:90 $.3   8./42..3 8 9  0 9 L  9 9  8 1  0 1  1   1 0 9  0 1  8 9  8 1  #03880.7.7.75 .3 706:03.2.$.253413.3.3.8 %20/42.

9 6:89#.07!490.83 #03880.7..7.3.830110.90 !1907 6:897.2.3   8.3..:2.90 .3899:90 $..75 .

3.:2.3899:90 $.833 %2042.75 .3.253 .&3/078.3 #03880..3   8.7.7.07!490.2.

43897:.6:89$.3 490.43897:.970.43897:.4770.9308-090038.943/408349/7.897.943.2508  0:8083.943 %2042.253 #0. 5:80814770.

:2.7.7.3.75 .3899:90   8.2.07!490..390754.3 #03880.943 $.3.

3 %025:80708543804190 70..43897:.943706:03. 83  .6:89#0.42..43897:.9431907.88.8.

07!490.50  %0892::810/9498190789080708 41/8.75 .2.3.3899:90 $.:2..3. 8.7.7090 8.709025:808.3   8.7.2508  #03880..

9:314723907.3899:90 ..2537..3.97.425430398-043/ .253904702 -.259:/0 24/:..:088.7.934850.3 1907349-08.89079.253 574. $.:809070.0..398-0.75  $.90/ ! 83.2501..$.088 !:80.3.2.0/6:897. .0300/948.:2..250/. $.07!490.4 83.90  357.9.3   8.90  8.841 %08.253904702 3.3-0:36:0/0907230/- .75 #03880.3/290/83..7.

3   8.07.540794.5407 $3.:08 $3.39.07!490.75 #03880.0 6:.94334805407 $.259:/0 .":.7.06:.259:/08 :9 3 .7.5538.50.3.07.139080941.39.39..943 259:/06:.390/ .2.250841...3899:90 ..0147294.94334805407 ":.393 .:2..4393:4:8.3.

4/0/394..7.3-98 ..4/3 ! :3147230.2508/9.4/047/ .2.943 ! !:80.3.0/!:804/04/:.47/ !47/47.9430.3.8 394.39.4/047/070 3903:2-07416:.943 ! 3.07!490.75 .3   8.3.:2.4/3906:.390/8.76:..39078.3/ #03880.7.08.03.4/024/:.6:./9.3899:90 $.390/83.

3.75 .3907 !74.39.2.7.":.7. 6:.088416:.94307747 ":.07!490.3899:90 $..3933480 ".3933480 9 9 9 09 9  09   9  9 #03880.3   8.:2.39307747 %0/110703.0-090039035:9.:3907 0 9  9  9 6   4/0416:.3.3/4:95:941.

3843.3843 8 02540/94.42570880/ 83.425.3829907  .7.:2.3/ %7. .425708843.0.7.43 :314726:.  99070.39..07 .39390 ..0/ 05./8947943  .33.0.75 #0.3  8.3/3    9  9 9 9  4257088 #03880.0780.07 .3.425708843 05..4/83.2.3.90789.07!490.3899:90  "..330 $.:390 5.943 98/430-:314726:..7.

7097.90  ..80-.05:805407 -3.82-4 1742.3/97.3.3 -3.943.7!80..03/.01472.9.07!490.4/08 ..782-48   .3899:90 $.8094180   4   ..7! 47.3 .99.80794/090.382982-47057080398-984190!47/8 !.07..382931472..3.7!  706:708088-.3828843 %497..3814720/945:808 .75:8024/:.3   8.7! 47.2.7!  #03880..782-48   ..03..:2.75 ..94397:58.3308 ! 806:03..7.08 .014728 30..70:80/147-3.7708.3//99.014728 .70:80/147343 -3.7.4/047/8 .7...97.

8 .7 #03880.3.07!490.3.:2.75 .7.3899:90 $.2503.!0.7..3   8.2.7.

.920%.3/06:.0307507%/.7%%  W 30750782-43-3.7%%.7! 88:2370.2.3   8..3/3.3:.7.3..7.25041 .9.7 !     3.75 .9.7! W  .75:80   .3:.75:80   % %  % %    % %  #03880.3899:90   $.7 !.07!490.-9 147 -3.7! 70.3/ .:2.7! 70.3.9.

9047 !659.2508 !.014728..95.014728..907!..

.6.

95.96 !% %.

6..

08907 ' 07 ' '     #  &354.7 # .2.3.96 % ' ' #.3   8.33/5.7 # ' ' ' 54.3.4/0# ' ' %%%%% %%%%% #03880.9 ' .75 .3.07!490.:50611 3.7.7.3899:90 $.:2..

943.73.0. 7907.!.:2..014728 $50.943..3//9011..3.3   8. 983..7.42509 #03880.03..3/ -.97...943.3/348022:39 2502039./0389.7.2.93!.97.07!490.3/.-9 7747/090..425.3/800.-9 390710703.489.3899:90 $.7.014728$00.9437907.75 .147.5.743.3.90789.8 5407850.5.

314 %09:.3828843 9.3/83.80-.3/97.90 1 8   .7.80-.390 3.253.014728 -.$:22.-98 !:80.250 ":.3/472.2 .8 $.97.4/0 !:80 24/:. 84:7..4 314 472.993.90 $.0 314 3.9 98970.9.

6 ..9.943%-%8.253920%8 3. %8 8.:094   4   -98 .4/30..-9/:7.

3907 31472.8. 47-9 7.3930.250/ .9.0836:. 2  4   82-44:94182-48. .943% ":.943 /.80-.9.5530.-9894.3/97..07/..014729/:7.-.  .382993 .90 #.:0944304190 0.3/..

%.-98.

80.90 #   .( $2-4 7.

% 82-48.

07!490..3.2.( #03880.3. 2#  $.3  8.7.75 .:2.80.7.3899:90 #.

0.0/907 #03880.9..2.07$97:.7.07!490.75 .:2.3.3899:90 $.#0.9:70 .3.3   8.7.

0841077478 %072.:2.943.3//090.01.3   8.943     39 % 024/ 8.90 .024/:.3/70.330 .2.0.9 2 !:80   9 24/:..9 2 090.382990/82-4 0892.9850.3.0 .3.330.70 820.88 8 9 24/:.3480  /89:7-89083.4784:7.90/82-4 .90 97.7.:88.37.7.843 //9.08413907089 90 824/00/-.75 .//9.3.088 .81.3480  3907 $2-4390710703.07 82-48 .3/5.3899:90 .:88..724/:.70/  $.1706:03.250 7 9 472.94197.3.07!490.3/42574.943 472. 9  .0 $ :09490190730110.9 #03880./0389147..97.3829907 .

3.2.25.7.7.:2.3.75 .07!490.941 #03880.3   8...3899:90 $.

3.2.3   8.25.941 .7...3899:90 $.:2.7.75 .3308947943 .07!490.3. 9  0 9 0 9 % #03880.

3.0/1907 5740.0/83.943 2574.39083.0.7.75 .070/.3/5705.4.01472 090.739070.85.0.3   8.074024/:.0 #0/:..3899:90 $.0.07!490.382990/82-4-.0.3/8.2.909097.943 892.3$:8306:.5407949034805407 $#  :832.9.2539070.:2.3.07.250 #03880..07 7024.7...253 .0147270.#0.943.4.147 /090.0/8.80/439070.99483..330/8947943 $.

90 $.250 908989.8 #0.9:70 $905 .3.#0.250 $905 /0.75 .7.8432..80-.3 1907 425038.9 % 79 706:03.01472 .07897:.425.0/.3/5.3 1907 6:.943147 .381472..3 090..0.2.:2.3303/:.0.25097.0/$ %7084/ .943 024/:.01472948.43.989.8883.3   8.07!490.3. #03880..3/5:80 5488-/894790/ .7.3/5:80 $.0.07843 47-.7843 2 #0. /43 .80-.3899:90 $.

9490-.3.:2.3/5.9490-.943574.8824/041/090.20708:98.03994-.783.01472 6:.3/.80-. -..783.7.8 09074/390-..014728178997.88 .3/5.0/-.3/ 24/0-0..8.7.3/5.2.8830..03.80-.3/.88.3814720/94..0.75 .574.80-.3.0904702 !0714723-..80-.-.0883 #03880.3   8.574.3/  0/8908.3/30.80-.8883..07!490.08831440/- 09074/39083.3/5.088806:.3899:90 $.3/5.80-.:80 %070.3/  1440/-.

:843/0833..7.:2.$90583/08339070.07.0.07 3/4592:284:94314770.7.3.9084:94381470.07/083990 14434.0.3   8.9.9. 3/805.75 #03880.7.2.8 789 014.0/1907  320$06:.0/83.2.70.3.07!490.20890 $#2.3899:90 ..20$#2.0.8  .07 $90583/083 4/09070.0.0/1907 $.41904.

7..2090$# 4/09070.3899:90 $.0.3 /0.88:25943  #0.0.3.330.#0. 9 39 $2519024/0 /0. 9 39  79 7 9  8 9 .07!490. 8 9 .2..071907942.3   8.. 9  0 9 8 9 39  79 7 9  8 9 39 #03880.75 .3.3308 ...0.7.0/83.0/83.:2.

7.0719078:. 8 9  9  459 9 % 9 % 9 #03880.382990/ $4:943 %04592:21907 890.43:.990$#8 9 2.0.03- .:2.!74-02 0839070.7.9.75 .90 4190 97.890920 70.3   8.0.07!490.3.9.0/1907 .0/907#0.382990/83.9908.20/.3//0.0780/.3.07  9  459 9  8 % 9  1   459 1  $ 1 05  61% .3899:90 $..9.253920 03     8 9 897.0/.07843 4190.2.

9.4770.0/19073  0 .947#0.0/ .0/19074:95:9.43.9.3-070..07 %02.4770.0.9908.890.943.9474:95:9 .4:94398 % : 82510894 3907.253920 .

4770.20$.943419070.0/ 83.943473307574/:..85.4770.0.9434507.#00....8 :  0 .43949083.9903480  4:98/09885.39 2.0 0/0.9  %  459 % % 7 %  7 : 8 : /:  7 9 8 9  #0..9438905740.0.87700.

3.3   8.:2. $.07!490..3.3899:90 .7.2.75 #03880.7.

90/3901390 83.0 480!$81..75 .7../2038438 090-.9 90 949.34805407 31390.3//9413907089  #03880.748890850.7700..3899:90 $..0%04702480 :98/0$3.$5.-4:99034805740..7.3.97:2 0.:2.07!490.3.3   8.7043.2.3.

8/04770,943110.9 

4770,9438,2,2:2039483,8,7082,738,50 ,3/,7035,80 47 :38190/ 970850.9940,.4907 4770,9438,20,8:70419082,79 -090039483,8,8,1:3.94341920819 , : -09003902 03909483,8,7082,738,50,3/:38190/970850.9940,.4907 907 574/:.98,5489;0 %880.43897:.9;0390710703.0  %0-70,/9 4190.4770,9431:3.943
0709,8831.,39;,:0
848147443 9083,8702,382,7
#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3 
8;75 

8/0:94.4770,943

#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

8/07488 
4770,943 #,/,7 

:7084849083,.,3-04.,90/93903480 .4541903437010703.083,8.4770,90/990:334383,  %0.4770,943-003907010703.0882,79490:334383,  ,70;,:0147.4770,94384890/070041.431/03.09,9907010703.083,8 /090.90/  %0,70;,:04190.4770,943 3/.,90803907010703.083,4..:78 
#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

W .4541,3437010703.083, 8.4770,90/990:334383,  W %0.4770,943-01907010703.0882,79490:334383,  W %0:334383,8.4770,90/9,3:2-07413437010703.01:3.9438  W ,70;,:0147.4770,94384890/07004182,79 94907010703.0  W %0,7089;,:0147.4770,943890248902,9.  W $,20573.503.422:3.,94387010703.083,8.47708543/394 82-48 W %0/0,.422:3.,9438.,3302,,;0,9903:,90/ 5,808190/90 7010703.083, ,3/,//0/3480
#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75 $4:7.04708$3,!74.0883

.08 9 9070/83.92.4770..943 $..108.07.4770.0/83.9434507.0/907-.8.98.990.99.3.9.0..0...9434190-3.72088.. 94 34807.79443 438/079070.085.82505740.250/-9070.3-0 8439.0 .88 ...9.7.382990/83.94.3/9097.0/190789019079.334-08.0.9.07!490.947.0.253389.3899:90   8.398 708:933.330/-8 2.94 9.842320890#98.3 #03880.0/1907 5740.85.3.75 .9477 .2089083..2.:2.7.9.939075709..98 907 439483..

.83.0/1907 70.8 8 9  %  459 9  %  9  8 9   459 9 % 9 % 9 % % 9 8 9  %  459 9  %  9  8 9   459 9 %.250412.9.

 %  % 9  % %.

 % 9 %.

 % %.

:2. % 9   #03880.07!490.2.3   8.7.3.75 .7.3899:90 $.3..

0/  9  #8 9 %   %  #8  8 %04:95:9$#41.94.38147241.920/0.:94.9.41.94419083.947 574547943.4770.059147.943419035:983.9.0/1907/0503/84343907.9.0/19078574547943.94.078434190 .9.9490$ 419035:983.0/83.  1  $ 1  05  61%  %04:95:983.9.90190782.9.1.0/907  %04:70797.8190/.   % 8   .835:9 8 0..9431:3.990190735:9 $ 2.2.2.0/19074:95:99902.!745079084190.2.0307 9490!$4190903480..



%42,9.3.43/94383902,9.0/
190734507,943 850.97,5,802,9.39,9;0890/0870/4:95:950,,9920% 850.97,,259:/02,9.39,9;084592:2$#949050,;,:0 $;:2,7,,3,7,2,3  
8;75

#03880,07!490.3.3899:90

2502039,943412,9.0/190770.0;07 
,3412,9.0/19078

8 % 9 

%

79
8 % 9 

   

,9.0/19074:95:9  -807;,943 ;0.947 

% 

 7 9

8  % 9 

        
$;:2,7,,3,7,2,3  
8;75 

  %   %   %
#03880,07!490.3.3899:90 

4900,705740.93,4390-,88/70.9438419083,85,.0

2502039,94341.4770,94770.0;07 
,341.4770,9478
89
% 

%

79
8  

 

9
% 

4770,94784:95:9  -807;,943 ;0.947  

% 

  %  %   %
% 

     

 7 9 8 9 /9

#03880,07!490.3.3899:90 

   

4903570;4:8 8/00 1907  0 .43;4:90390-408843 $;:2,7,,3,7,2,3  

8;75

7.3899:90 $..078 8 9  % .0/19078  % % 9  % % 79    8 9  % % % 9  %  %  #03880.2502039.9.0.9430.3412.250412.:2.2.7.07!490.9.3   8.3.75 .0/190770.3.

/42.2..3899:90 $.7.:2.5407 #03880.3/5..9.:/083480 -:928808842083.3.3   8.88907 0.3'0 $250.7.75 .0/907706:03.07!490.3.

7.47/39480 2.3   8.5407 090/..3.943 #03880.//824703480.88907 3..3899:90 $.3.42.2.9..3/5.:2.:/08247083.2.0/907 3.5407 -:9.84 .07!490.:/08247083.75 .7.9.0/907706:03.3480700..3'0 439/ :9 .

2..9442-33 # .9:70 147.:2. $# #03880.75 .#.07!490.2.583.3899:90 $.7.35. 090.3.3   8.42-33 2:9 9.05439 0307.3.-7.7..94341981 /42...3.

0208 #03880.:2.2508412.75 .3899:90 $.07!490.3.9.7.0/19074:95:9147-.88 24/:.7..9438..3.3/5.2.3   8.

2.3899:90 $.$5.$3.07!490.043..7.3   8.3.75 ..:2.3.7.0598 #03880.

3/89.85.83.479443.08 -0900383.0949070.890232:2/89.. 98.9478..0/ .80307.8 0892.70039070890/3:..3/:.89097.0 .0/83.0.3/89.894..:2.83..07983.08-0900383.20..9435:754808%070.3894.0.3.07.7.94384183.3/.078..3.70.0 '0.:./0.7.0 /40300/.0 98.897.0 ... #03880.8 .3894.2.$3.0.0.9083.90/..75 ./0.85.83.3814720/94.43.3.3   8.947705708039.0.3899:90 $.382990/83.9478  %083.0.85.8 47/090.07!490.98.3 /203843.0/83.20.3.85.3.

9.  9 . .  9 .9  9    9    9     $. 8   .3.0  9 8  .08 8 9  . . .  9     8   . .  9 8   . #0.  9 .85.07!490.9073. .75 8 9  . .3.25041. %7.7.2...  9 8   .0.:2.  9 8  .83..$.7. 8 9  .0.9. .382990/83.0/83. .02.9.3 2.3899:90   8.0/19074:95:9 #03880.

7 574/:. 9  9  .75 #03880.3899:90 .7.0 %41472.9 -090039483.07!490.7.85.7488 .9438  3307 8.   9  9    9 .   9  9  9  9  9   9  9   9 9  $.$3..83.2..9  . 9  .3/ 9  !74507908413307574/:..3.8 1:3.9 C   9  9  C  9  9 /9 .3.:2..943-09003 9 ..4770.3  8.85.0 17890300/943490 3307574/:.

.0383.8 #03880.:.0 %0/89. 9  9 9   039 41 9 C C 9  /9    .07!490..3.890:.3.75 ..93903472 .:2.85.85.3/89.3.8.7..3899:90 $.$3.3   8..0 -09003 9483.83.7./0.  9 472-090039483.8 /    9  9 07010794903472-090039483.3. 9  .8:70-.0820.983472 47241.2.

2.3.3899:90 .0-0900383.08383.7./0.3.8 9 .07!490. .0  9 8  .:2.    $.3.75 .  / 8    9      / 8   8   . .25041/89.3/8 9  / 8   8 9  9      #03880.3   8.3.85. %0:.   ..3/89.7. / 8  8   . ...

7.-.75 .9478 7.83.85.3   8.94382:898..  9083.85.3.3.:2.43/943 %    9   9    9   9 /9   0  A9 A%      070 0   F    F   1.479443.85.80941.0847943472.73/0503/0391:3.08.79443.93 7943472.7.881:3.9070/-   9 =  30.0/-.881:3...2.0...0/:70 #03880.9438  %0-.98190479443.83.7..07!490.2/9574.9 ..2 $.3899:90 $.881742343 47943472.0  /203843.    43897:.9438.

.25041.347943472. 47943472.0  9   .83.808 .250 /203843.85..48 69 .-.

% % A9 % A9 %  9   9  83 69 .

3899:90 .:2.85.2.07!490. 47943473.0  9  %  9   % 9   9 $.7.3.75 #03880..7. % % %  9    9   9    9   9 /9   9    9   .3.83.250 /203843..3  8.

$30.

7.3.90 47943472.7.4830.70903 5.80-.75 ..80/203843841 .3.2.07!490.:2. %080..3   8.808490 55742.80 6:.039705708039./7.9:70 5.3899:90 $.9438 #03880.4250-.3/06:.

.3.7.:2.250!$ 4909482-48 -:943430/2038433!$ #03880.07!490.7.3899:90 $.3   8.3.75 .2..

7.$3.943% ..947705708039...    .42-3.85.  9     A %   .3..0  3.75 .7...3-00570880/ .01472 #03880.7-97..014728   9 = 8  0700.:2.3   8.7139080941.30.014728  070 9 =      A   8 9  070    .3.2..014720307 $.94341479443.  .202-074190809841/:7. '0.07!490..   8 9   9  8 9   9 /9       A9 A%     8   . .8.7.3899:90      .94341.

.75 ..43.3.7. .94794. '0.  9 8 9 ..0 8 9     .  . .  .  82 82 .0..0147294.07843 . . #03880.01472.85.2.07!490.  .:2.  9 8   .3899:90 $.07843  9 %  9 .3.$3. 8 9  9 % .3   8.947.7.43.

83.9073. .07!490.2. 8 9  .  9 8   .  9 8   . .3 47943472. . 8   .. %7.7. .         $.3 A9 A%  8.:2.3.  8 9   9 /9  .3899:90 .85.894.  9 8   . .08 8 9  .  9 8  .3. .. .0  9 8  . .  9 .  9 ..7.250415740.9383.9.382990/83.75 #03880.  9 % 8 9  .

.3.9490-.7.2.75 .9.0/19078   % 9  79        -807.890/ 349002.9.07 70.947  % 9   8 9  .9.3   8.943 .88/70.07!490..:2.3.  9              7 9   % 9 #03880.3412.0.3899:90 A     $.9438 .0/190770..7.0.

..3.07 70.9478  9 %  79  9 %  7  7   -807.4770.2.75 .890/ .3899:90 A     $.7.943 .  9              7 9   9 /9 #03880.0.3.0.341.7.3   8.94770.07!490.4770.:2.947 8 9  %    .

.9438 8 9  % 8 9 % 9  %  9  % % 9 % 9 2.3.0/- .9478 870/:.3.0/19078 47.7.9.2.250412.0/19078 .:2. #03880.4770.1:3.0/1907  9 79  %  % 9  )    :2-07412.9.7.382990/83.70/94 :832.8.9478 949097.3   8..4770.75 .9.9.0/190770.0.425.078:83 -.07!490.3899:90 $.

3 .94341 39 3   3 9   9  =  3 9   9  3  3 3 3 3/0503/039074 20.$5.0 3 9 ..9034803 7943472.37.088 = 3 9 4804:98/0439083.947705708039.-089 ..3/42..0 .3-00570880/.85..4470/ 25.3.943574.9890/090.90/439083..0 .7..0 7700..8 3 9 3 9 4805740.85...39 39     3   9     '0.7.:88.7 3    .$3.

2.7.3899:90 $.         = 3  #03880.07!490.7.3.3.75 ..:2.3   8.

943.3.2:2044/  !071472.2.75 .3899:90 $.07!490.7..7.:2.04:3/8 #03880.3   8.3.090.3.

03904-807.2.5531742   2 2 94.9 90.947 8  0.07!490.9.7.382990/82-4  8:.947 5071472.0574-.-94107747390/0...943.7.2.07.090.0..3.3.8438 2320/ 3 2 4/:.9434183.30892.75 .90419097.943 574-02 .:2.3   8.8437:0 2 #03880.3899:90 $.3 090.

3. ..841904-807.3 7.947 .:88.7.0.3.-089074 20.3/. 3  3 3 020398413480...989.943'0.947 8   ..7.  .0.0 %0  ..33024/0   8 3 $3.947 3 .70  /.3/42.$9. 8/09072389.

7.:88.37.947        3/0503/039.0.0/.9475/18  6 %00020398414-807.0.3899:90 53 3    .3.07!490.-08 985/18 5   8   #03880..  3480.3/42.

 05 3  .70  6  .

75 .2.:2..3.7.3   8. 05  8   $.7.

:2.3.   5  #03880..-9  $092  2 1 !7 2 8039   K !7 2 8039   147.2:2147.75 .090.7.54890747574-.2:2.3   8.08 $092  2 1 5   2 5 82.07!490.   070    553.8437:0 2.7.2.943 592:2/0.3.08 7:0.3899:90 $.

38 2  2 #03880.7.:2.3   8.3.84370438  8:..07!490.0394 /0.9    '0.9470838/07043  1 5   2 3 5 ( 82.9.3.090.2:2147..85.75 .2.799439083.7.3899:90 $.943 !.920.   5  %.

3..8343.:2.2:2147.-9 882510/94 $092  2 1 5   2 82.3   8.2:2147.   4706:.7.943 7:0 4706:.039 $092  2 1 3 5   2 ( 82.82.07!490.2.7..-082-48 904592:2/0.2:2044/ #03880.3899:90 $.2:254890747574-.574-.   .3.090.75 .8437:0 2.

090.84370438   #089.799439083.3.75 .2.07!490.3.81448 '0.:2.2:2044//0.90902.0394 /0.920.8437:0.9470838/07043  1 3 5   2 ( 82.7.   #03880.7.3899:90 $.2:2147..3   8.38 2  2  '0.   %.943  !..85.9470838/07043  1  8  8232:2147.

3.02.3   8.25041/0.07!490.0.75 .7.9.:2..2.84370438  9  8 8  8 8   9  #03880.7.3.3899:90 $.$.

9/0. 8 ..2.3.-941.75 ..382990/82-4 !7 2  2  !7 2 8039 !7 0838/0  2 8039 !.382990/82-4 !7 2  2  !7 2 8039 !7 /408349038/0  2 8039 !74-.38/07043  !74-.947 .03983.-9410774304:8/0.382990/82-44706:.97.7.0.7.-94182-407747 774304:8/0.97.843 479097.3899:90 5   2 /  $.3   8.!74-.843147.:781904-807..0.07!490.3. 2  !7 0838/0  2 8039  !0 2   !.843147.947/408349 2  1.4770.943.3077473/0.8434.:2. 2 #03880..

250147-3..3899:90 8 8  .7! 5   2 5   2 8  8 -  9 !0 2  !0 2  " !  !   " #03880.07!490.3.

  .

7..:2.7.3.      $.3   8.75 .2.

07.0574-.2.75 .3899:90 5   2 / $.0 574- 4182-407747  .3   8.7.7..07.-94182-407747  !     !7 2  2  4706:.-0 82-48  !     !0 2             !.3. 2       #03880.3.07!490.:2..574-.

3   8.:46.07!490.:6.&343-4:3/ 565/651 '796/.35.5.7.:  !0 2 A     ! 8  8   !  A    ! 8  8        #03880.56565.:2.75 ./3.3.796/.6.::779/6511 /.3899:90 $.3.511.2./3.7..

.:2.7..7.25041:343-4:3/ !0 2   7   8 57 7  2 /7       8 &343-4:3/   8  8  !0 2 A   ! 8  8  8  7 8 7 8   8 7 8   8  8 8 8 8  57 7  2 /7  8  8 ! 8  8   57 7  2 /7 ! 8 8   #03880.07!490.3  8.75  .3.3899:90 ! 8  8  57 7  2 /7  $.2.3.

38  038  88039 C  /   : /  .4807948  9.3.&5507-4:3/-.0 ! 8  8   !7 8.80/43232:2/89.

 05  /: "  6  .

 /   8  8   !  A     / 23 .

 ! 8  8  A   "  .

.75 .3   8.       32:2/89.3.7.3899:90 $.3.2.:2.7.3.07!490.039083.0 / 23  23 /    #03880.85..

07!490..2.3.25041:5507-4:3/43..80/43:343-4:3/ 8    8      9 /     8  / 23   8 8 8 8 8 /  /  8 /  /  8  9 8 8 #03880..7.75 . $2-407747574-  -.7.3.3   8.:2.3899:90 $.

7.3.75 .3   8.2.7.07!490..3899:90 $.:2.#03880.3.

-.

9438 $#47$. ..422:3.41:704120793/9.

784341/1107039 $97.:80 $3.990-9 0.0  ..425.0 1.3/03.70030783.908.8.07..82-4/:7.943 -0..382990/93.382993/11070393:2-0741-9850782-4 $%$      ..890..422:3.083...07. 0745407 2097.943.7097.9.54079490.03480 5407 $#84:/-024/10/39072841-9 03073 /9.

3//9 490$.   ##03880.90 .3.3899:90 # 97.07!490.

-.

.75 .7.4850. $.3.011.3   8.7.97.03.2.:2.

3.25041$2-407747574- 47!83.2.3899:90 % 9  8.75 $..:2.7.7.07!490.7! 8 8 -  9 - 8     .7! 8 8  8    9    9  % #03880..3  .3.8 3.

03989.382982-48: : 7747574-.9:7.70890-47#:0 90849745.090384341 2.943574-02 58'0.0.3   8.2.0/.041/090.0.39:/0 47039 #03880.947/70. 57450794190.  4804:98/090801390/20384388 7700.943  749.80 0.:88.39 147/090.20147-499097...99007747 574-..0.07!490.2:2044/  090.947 .3.99070.33480 0050.3.943: : 8.7.7.-9 !740.7.989.947 . 8:11.3.4390 /110703. 3.0.75 .943.-994-0908.3.943'0.94734728.3899:90 $..:2.

43890.9438.04/:.943 4907010789490.50.90384394 ! :9 0.

3.3   8..:2.75 .7./70.7.3.2.07!490.3899:90 $.943 ! #03880.

-9 4903890.98 #03880.3   8.2.385480.9478 97.0090.7.943 7747574-.90  1473307574/:.4250.% :80.43:.0.75 .7.3.3899:90 $.3/.3. 147.4250'0.947$5.07!490.:2.../41.

3.75 .07!490.7.3899:90 $..7.:2.3   8.4250 090.7 #03880.3.2.943 $:22.

.3   8.090.98-0.75 .3.7..07!490.-3.324/09.7.5/9407$# .3.8..709020247088. .70.70.:2.0790806:03.943 #03880.0 " 1:3.:7.4203.7822097.330 $ #824/00/.9437747# 190-9077478  / /8.2.807.-0 1 903/8943.808 900110.70.3899:90 $..-91 8# 1 3..:31472574-.9071.041-98  9034:.8.8339407.330 4.

8#. 088039.9 1.9 #.7.0 0.0 00/ /.5499390"  1:3.943  %005430308$#/ -:94-.9071.0789 90.36:08 94/0.5  1..$#. 0.03 9.090 ..90789.8#./3  -807.

088.3//9 :594   .3/  59..39574.   700884$#  / 07-.1-07.422$# -..3//9:9.

.425.,90/.4/08  83,574.088309.
#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

099075071472,3.0974:/;0789 
;0789 9070.0;078574;/0/92:950.4508 419097,382990/83, %02:95083,.4508 84:/050703.0:3.4770,90/1,/3 390.,330  398.,8090574-,-99,9, 83,.45081,/0 82:9,304:8870/:.0//7,2,9.,970850.994 90574-,-99,9,830 .45050703.08,1,/0 8,74:7:0

!0 8574547943,94 
# 

  

;078941 947/07 

;07,0$#

#03880,07!490.3.3899:90

$;:2,7,,3,7,2,3  
8;75

#;8 $# /;07890110.9 
# 

!0 

.,330 341,/3 

,91,/3.,330 #,01,/3   

$# 
     

 

005470988947,907
#03880,07!490.3.3899:90 

8/08095,79
$;:2,7,,3,7,2,3 
8;75 

3    8.3.4/:.7.943%0.3899:90 $.:2.7.75 .36:08 #03880.07!490.2..3.

80471706:03.7.2.33078:9.984/:.80-.943 3.3/5. ..90-.7.8883.9-.80-..8883.07!490.3    8.7707 #03880. 24/:.94-.3828843 %7.943097.38.3899:90 $.:2.259:/0 5...3/2088..90/83.083.41.3/5.-0147 97.3.2.9433.4/331472.3.01742.7707 024/:.3/84:7.75 .7. 4 '..

.44/:.3899:90 $.3    8.943 0.:2.943 2:9.0789 6:.7.3.9..770790. #03880.943 470011.3.943 /..83.908 540707747.7594309.088897.039 07/.7.7.3...:79 03.0 &83.07!490.0392:950.75 .4770.8907 24705407011.36:08147$ 29.2.9008 -09907 80.4/3 24/:..943 25.9.72039 70889.507 1.

90 W$50.03.97.36:08 W9#.8414/:.11.4.943%0. 2.58.

9.709# W4 489.0.03.. 23540794. W!40711.0.

2..7.3.:2.75 .3899:90 $.07!490.4 !4072502039.7.3.943 W#4-:8930889425.720398 #03880.3    8.

.75 .7 24/:.078:834330.9438925.2.078:8343 .9  9 ( .90/83.7 .0450 . 30.03.9:70 W30.850.3    8.4389.3.97.3-07057080390/.03.80 6:.48.7.7025.80471706:03./7.97. #03880.-9 0 9.9 3 5..943 25.011.943705708039.3.4/:.9  9 83 9 83..3903.48.7/..71706:03.259:/0 89 5.7.  9 .80 43 30.0450 .:2.709#.7259:/0475.943 W324/:.8 8 9  9 .9 43850.39 03.07!490.-74./033 W4389.9435407011.48 9 .3899:90 70..9407$#8 $..

.70 .:2.7..3899:90 $..803.0.3.947$5.07!490.0 #03880..3.83.7.9473.2.9438 !$ 7.83.7 $6:.:.8.90/ .0843890.85..75 .3    8.4/0/ 24/:.4250'0.

:.9438    " !$   '%   !$ "!$ #03880.3.07!490.70 43890.48.3  9 3% (.9 3 3  9 3 5.9438  #"&#%&# !%&     "  #!$ $%  !$ 7.3899:90 #% $%  !$  !$ "!$ "!$ 6.36:08 8 9  % ./7.943%0.7 43890.80 " 9 6:.9:70 # &% $ $6:.74/:.9  % -3  9 3% ( 83.30.

:2.75 . "!$ $.7.2..7.3.3    8.

 011.039 -:90885407011.:.7.97.9438 .3 %7.3    8.3899:90 $.97.84247080389.702470850.2.70 .9438 -3  "  !$ .0948890234330.039  0 #07   "82470850.75 . !$.7..07!490.:2.3/ "  !$ 7.011.3  " $6:.9438 -3  !$  !$ .3.0399./04118 07 47/0724/:.3.7043890.3 !$-:9 .743890.7908 #03880.

8 !40711.:2.3.7.3. !$ " $ #03880.2..5 $.07!490.3//9.03.5479-44 .75 ..3    8.55.0#.3899:90 $4:7..7.

3   8.990 30-47382-4.2.7! 8 - 8  9 - 8     .553 49090..4389.553-989482-48 489007747894 .! $2-4.39 7.3..00307 507 -98.08:709.4/3:80/147 2..0 .223 /89.75 .0 #03880.07!490.3.7.:2.3.3899:90 3.843  -9/110703.7! 8 8    8    9   $.7.4/3 8      .07.431:80930-473 82-4  .7! 8 8  8    9    7..

!09.8 &306:.030708.

3899:90 $..7.75 .3.82-4 #03880.843#0438   8.7.2.07!490.:2.3.3 0.

:2.743890.2.3. !$ 7..9438 !$ -3  !$  !$ .07!490.7.:..94354398 6:.4/3  #03880.03073.8   7.7.3.3899:90    $.3 43890.83.3   8.75 .

!$0.843#0438 024/ 3             !$ !$-9.

.4250/20384347-98.

7..3.2.947147!$ #03880.9 407039024/:.3   8.75 .82-4 !$ 2 8 9  3 9   %- 9 7 7^ 2 2 47 .48 1.3.:2.7.07!490.3899:90 $.

.94384180430.43890.82-4030708 "986:.. -9.94384180 8 06:.9438 " -3  "  !$ .9439 .7043890.70.03994!24/:.43890. " $6:.41903 5./7.943899.43890..9:7083.//943.3 &306:.55742.3.90 /2470540794803/.3/6:.80 .425430398 4786:.08.70.

/20384347-98.

0 -09003.82-40 2.830-9 #03880./.94354398 .4/02.:2.75  .39. .20232:2/89.2.3.7/9413/.7..43890.07!490.3.33908...3  8.843#0438 $.553070.7.3.3899:90                 "0.03982-48/1107-.7.

/3.24/:.7707… 470.807010703..80/719 .489 8:8..943!$ 31472..943.80  #06:708.5.382990/83..059-094.943 0 5.0.943/4349706:705.80419097..0 4700307.3308 1107039.9.770/383.7707… 2:89 -02.24/:.77075.9433!$ ".43 4070394/:.407039/024/:.07.7/07944-9.0/94905.8041 9070..331.

5.5.2.:77039 570.5.77075.094.4770.4:882-4 08880389.803920 /42.382990/  &80570.:7703982-4 314-9803.07!490.80/719 1 /42.75 .9434183.890.:2.80-09003.3.3.98/0.3 #03880.20247/0503/8:543574782-4897.7...807010703.0147.4/0/.890/1107039.7.82-4.3   8.3 247080389.3899:90 $.094/45507 0110.

4/0/.80Û  0Û .80.80Û   0Û Û    9039403.2.-9803. 5.75 .4/:.80Û  0 Û   "!$7.8..3.3   8.1107039.4/0..3899:90 $.8.85.7.3041 182-44. -94.3035.80 070.7.4/0/-34.4/3 #03880.07920   %8 %8 .07!490..4/0.3/ 9403..943 439/ !$1107039.3.0 5.-99082-44:/...0 5.07%8    %8 9082-44:/.!$  -9803.:2.

7/147/9.94396:.251.70...97..82-4 5074/ &8:.36:09429.:3/0734330.9:70-7.20850.9438.--70.:80.389438 974: 0745439 -7:595.259:/0.8 !$ 07090 3/.3-0/894790/ -34330.574507908.9:70 11809 !.943  83.3/.5:80 9 -.":.2:25.011./7.80819:80/390$ 89.3/.38943841 4 ./7./7.9090.70.7.8 /07008 3490790.9:7041180987010770/94.259:/097..8 "!$ "!$.8908.7.0902.8"!$14730.8097.4/0/-41180993906:.0:.943 -:9.807850..251.03.90/.7.90890411809 "!$24/:..389438.7 ..8097.3.251078. /0700 5..3/19078 ..943841.3.8097.78.97.9:.1.38943419083.259:/01:..

8097.:2.3899:90 $.078:8 /07008147411809 "!$.7.3.3/ /07008147"!$ #03880.3   8.07!490.3894341/07008 . "!$ .2:25..2.7.75 .3.

7.7.3.11809"!$..:2.3.3899:90 $.3   8.75 ..2.07!490.014728 #03880.

4389.9 61 9 C / : /:( 34330.3.3899:90 $.3/!# #03880.:2.57024/:..80$ !$  /9.819 95.3-025020390/:83 "70. ! .:88.706:03.39 03.2.70 .07 W..9431907 :80/-#4-0.0450 W32:2$1903 $ 232:2-.75 .$1903 $ W 4393:4:8!.947942.088/04-0841$:83.4393:4:85.03.7.3. 24/:.0.3.724/:.07!490./.3   8.332:2$1903 $  70/:.943-:9..39.4/0/3901706:03.25020390/91706:03.9.7.9.3//9 8/04-08.80 8 9 .48.

07!490. #03880.:2..97.3.7.3899:90 $.3   8.3.7.2.32:2$1903 $ 850.75 .

90789.97..7.8  !407$50.$50..97.0389 / "!$.

75 472.3."!$ $     /%.07!490.0/706:03.3899:90  .       $        $.7.3  8. %#03880.3..7.:2. 1 1.2.

$#.943 .0399.3!$ 9908.97.           97747!74-.9147/1107039.011. /090.011.3.039 -:9706:7082470$#507-9 W%0708=/5407503.  !$ !$ "!$ !                  "!$ !$ / "!$ / !$=/ "!$=/ W"!$82470850.0 W !$ 147 82470850.-9 #  47!.97.205071472.

.7.3   8.7.75 .3.3899:90 $.2.-9 / #03880.:2.3.07!490.

97747!74-.330           !$ !-                   !$ W!.1.-4:9 /24705407147!.   .7432039706:708.83.38288433.07.-9 # ./3.07805745479439490./303.$#.0$#507-9  W%7..

2.7.3   8.75 .3.7.-9 / #03880..3.07!490.3899:90 $.:2.

..97747!74-.4:.-0!.083.9438:80/ /0.   ./4330073  #!7088 ( .3770/:.388:0147 8500/ 70088/.3/9045507  471   /.  "!$ "!$   #.9.943417010703.-9 # 45507110./0503/84390/. 1%    4.-007747144703/1107039./3             %025.7.90.98 W45507.7.4:3/.9./3   W%0770/:.$#.. /090.90 %  8  8  8 !-1447        -58  -58 -58  !.0.94389.9.9438414-0#.4770.9455078349.:808.

2.7.3   8.75 .7.3..3.:2.-9 / #03880.07!490.3899:90 $.

943 !$ "!$ "!$ 4/:.3770/:..294390 2.2:95.07!490.943  $ 770/:.74.-0077471447   407039090.8570..90 4-0 7:7.2:29#.:808.4/:.-9 # 0.9  380.8570.330899.2:2-97.3899:90 $.9438.3.250 147"!$ : .$570.90 3.903.9841%200./4 422:3.:2../43!479.08  380.$570.2.70:/3  380.-0#.  -58 .  -58 . -58    :.97747!74-./ :  82-45074/ %  #03880.7. -58 4-0 ./ W$.3 %0110./254808.3   8. 380.943  $ :30(          728/0.7.75 .-0!-       W%0728/0.7432039  470.3.

4389.011.3899:90 $.75 .078:834330.91.03.7.2.7414/:.3903.3.03./090.7.:2.0450 .39 03.9438 1./3 /45507 /0.8570.0450.4389../04118 30.943 5407011.3   8. W 29.7.078:8/1107039.078:8343 .$:22.407039.943 ./ #03880.97..078:8850.94388:08 W %7.07!490.3.724/:.

7.3899:90 $.3.75 .7.3.53 #03880.3   8..!:80$.2.:2.07!490.

7.#0.25.3899:90 $.:2..75 .3.94143 #03880..2.3   8.7.07!490.3..

25.75 .3.7.2.3..7.3308947943 .330 #03880.5 $.. 9  0 9 0 9 % :9 9.07!490.3899:90 $.941 .3   8.:2..

.:2.39.75..80 !.3899:90 $.00-.80/8947943 #03880.330 0  70088.3/ 290/907341.75 .7.$110.07!490.94190.4389.3   8.3.98.7.2. 1  ...3308-0.259:/0 259:/0/8947943 7 .. 1 0 43 .330 $/:094190730110.9438 .3.3308 .422:3.3/ 290/19078 . 1 43 30.

9    8 #03880.:2.253920 $0110.943574.3.70. 8 $.3899:90 3  .0398890297.75  .0920/8507843 .98 419088902 .7..2.3  8.07. 1 7 1 .943  1  9 1 .3907 $2-4390710703.3.3/03.088/:09490190730110.98.9080..408.:808$.06:.7.381071:3..07!490.0 $ $390/090.

947  = 39 19070/3480  1  9 1 .3   8. 9 79 39 # 1907 7 9 7 1  9  % 090.80-.3.75 .330 . 1 7 1 #03880.947  =  9 1 % .3899:90 $.03988902 9 %  9  %  9  1 090.7.03924/0  =   % 6:... 1 6:.3.2..7.07!490.0 $ 4/0  =   % .3907 82-4390710703.3/8890224/0 %1907 9 9 .:2.

.39 6:89-.03988902  %0904709.43897.6:89-.3//9.39 4306:.43897.3//9..232:2706:70/88902-.9#8 82-48.3//994 /090.

8(94:9$8#8.

 (  6:.039 .3//9.889029-..

%#8.

2.2:297.38288437.9041. ( ..38:55479.

%#8 82-48.

8(94:9$  #8  A %  #8 K 82-4.

8.

07!490.(  #03880. #.03.3.3//9011.3899:90 .

 -98.

8.

 974:5:950707941-.2.03990-.8:70387057080393/.390.9.:2.70:80/ -83.3920.3//97084:7.08.3   8.3.36:08 $.7.75 .7.3//9 $434011..( 325479.

9 .6:891907  1 % /0.6:.$8902/0.6:895:80 1907 /0.6:895:80  9  83.

%   %  %  % 1 % % % % 9  #03880.7.7..2.07!490.:2.3   8.3899:90 $.3.3.75 .

30/- .6:895:808 19078 6:895:808 19078  !:808 19078 . 9.9439.3-07057080390/-.0.50..9908.83.708:9334$.71:3.94331706:03.253920 6:891907 9897.9.381071:3.4.70.3:.3./42.384-9.370.1706:03.43.1:3.943 6:895:80 988.03 822097.

250416:8919078#..9432:95 -.943 .:2.3   8.3.2.349079201:3.80/ 48301907 #03880.3.07!490.% 1:3.7.3899:90 $.7.75 .

2.7.3..8/04-08 #03880.8.3.0$ 11./0 41135:80 8.3899:90 $.75 .07!490.3/97.3//9:9.:2.53 #0/:.0$ 4.039-.539470/:.!:808.3   8.943 #4-:8930889492307747 82.7.

7.80/48309076:89!:8055742..07!490.75 .943   1   # 1   7  7    9  # 9   7   7  7   7    % % %   % %  % % % % % % % #8 .#.3.3.3899:90 !.3/ 8$  7  #03880.3/ $  7 #8 $.88-.3   8.80-.:2.2.7.

3.2.947 7  A 7 A      8.07!490.3//9   #03880.7.3..  9     9 ( .80/4830907 #.7.088-..9908.3 .:2.80/ 4830907 6:895:80 4$.#.48     147 1    147   1   147 1   .3899:90 #4 4111.75 $.253920  6  1     1  .486   9 (  9   83.

2.3/6:.7041$ .508 4$.7.!:80$.943!73.70 #449#.3. 1 %....7041$ .:80/-.80/4830 $## 1907...3   8.:80/-97 1907 %.07!490.9908.253920  # 1   9 1  .7.07  # 1   9 1  7 1  7 1   9 1   # 1   $## 1 0 1   .75 .53.:2. 1  7 1  0 1 $6:.3.3/6:.3.330 #03880.3..3899:90 $..

75 .808 #03880.3899:90 $..!:80$.53  79443..3.3.3   8.2.07!490.:2.7.7.

3   8.!$83.5479-44 .7.55.'79:0415:808..2..0#.75 .5 !$41.53 #03880.:2.7.3. $.3899:90 $4:7.07!490.3.

80-.80/ 4830 $## 5:808...53 .01472 %7/5:80 9.53 25 '( $6:.70 7449#.250415:808.3/97 .

43/5:80 ..7.3.7.3.:2.07!490.9.3899:90 .75 #03880.% 7895:80 $0.82-4 $.2.3   8.

9.53 #.9.9904:95:9412.0/19074:95:9 074$ 9.250415:808.80-.0.0/.80/48305:80.3/70.9 902.01472.0/1907 25 '( ...

:2.2.07!490.75 .3.% #03880.3899:90 $.7.7..3   8.3.

80-.05..90.43.3/83.9907385.48.-.450.99073 05.9907.800589088902 7085438094.348.

7.75 .0 4802.259:/08.07!490.73 $0389.7.0 $..% % 82-4/:7.3899:90 9208..943  8947943 /:094$ .3..3   8.3.:2.994 92307747 %239907 #03880.2.

.3/34$ #03880..:2.99073 3.75 .25041005.330 343480..07!490.7.7 ! $##5:80 !0710.3899:90 $.3.7.9.3   8.3.2.

.25041005.7 ! $##5:80  -.99073 3.

:2.7.2.  / .3899:90 $..7.3   8.3.07!490.75 .3/34$ #03880.3.

.25041005.7 ! $##5:80  -.99073 3.

:2.7.3.3899:90 $.7..  / .3/34$ #03880.3.2.3   8.07!490.75 .

.943 !071472.75 .7.9438 !:80$.7.$:22.0 4:3/8 4/:.943$..3.3.3.0208 43890.53 #03880.943 090.3899:90 $.2.7 9.3   8.:2.8.07!490.8 4/:.

:2.7.3.2.97..75 .3899:90 $.3.3   8.$/08 #03880.7.07!490.

7.07!490..2..943 "#05708039.3899:90 $.3.3.3   8.7.3/5.943 #03880.75 .:2.884/:.

394 .258 300/0/ 909./3 .97084/ .7/0503/03.425038.5/2574.8 #03880.011.0436:.3//9 4389.59:70 0110.9433905.31472.3//9 24/:.3903.41.0/83.232:2 70.75 .:2..3903.0/$#/.4/:.0.07.3.2.943  .0941.0.07!490..9 5407417. $50.7707 43 30.943  ./.039    .3899:90 $.3147 -.3/54994308 .02039370.97.3.02039:833 -..9-043/.039-:98:8.920.4389.045083.4706:03.2584 30./3 %7.3.7477..5970.7.9  09907348022:39 70889./83.3   8.9433/0 1472574.0450!4073011./32574.3.059-0943480 1..90 43 .80471706:03./0411-.8 259:/04/:.88475407.011.059436:.7.039   .885407.

3.2.3.07!490..75 .4259:/04/:.7.:2.7.943 #03880.3899:90 $.3   8..2503.