Вы находитесь на странице: 1из 19

Tutorial # 6

Learning Outcomes
2

 At the end of this tutorial session, students should get


familiarized or recall:
 Design of Band-Limited Channels : Selected Exercises
 The Nyquist Criterion
 Controlled ISI : EX!

 - Optimum Receiver for Channels with ISI and AWGN


 ML detection
 Selected Exercises
Band-Limited Channels

 Bandwidth constraints on
 Communication system design or
 Type of the information bearing signal

 ISI arises when the channel is dispersive


 Time limited  bandwidth unlimited  bandpass channel
(Frequency limited)  time unlimited  ISI
Signal design for band-
4
limited channels

 Nyquist criteria
 Pulse amplitudes can be detected correctly despite pulse
spreading or overlapping, if there is no ISI at the decision-
making instants  the design of pulse shaping filter)
Nyquist criteria
5

 The condition for No ISI is given by

1 k 0 
m
x( t  kT )  x( kT )  xk  
k 0
 X(f  )  B( f )  T
0 m   T

 When T = 1/2w
T f W  sin( πt T )
X(f)   x(t)   sinc( πt T)
0 otherwise πt T

 When T > 1/2w; Raised Cosine


 1- β
T 0 f 
2T

T   πT 1 β  1- β 1 β
X rc (f)    1  cos  ( f  )  :  f 
 2  β 2 T  2T 2T
0 otherwise

Raised Cosine
6

 Plots of this function with β as a parameter are shown


below, both in the frequency and time domains

Pulses having a raised


cosine spectrum

Excess bandwidth
Matlab Demo
7

% Script for plotting the time domain and frequency domain representation
% of raised cosine filters for various values of alpha
clear all
fs = 10;
% defining the sinc filter
sincNum = sin(pi*[-fs:1/fs:fs]); % numerator of the sinc function
sincDen = (pi*[-fs:1/fs:fs]); % denominator of the sinc function
sincDenZero = find(abs(sincDen) < 10^-10);
sincOp = sincNum./sincDen;
sincOp(sincDenZero) = 1; % sin(pix/(pix) =1 for x =0

alpha = 0;
cosNum = cos(alpha*pi*[-fs:1/fs:fs]);
cosDen = (1-(2*alpha*[-fs:1/fs:fs]).^2);
cosDenZero = find(abs(cosDen)<10^-10);
cosOp = cosNum./cosDen;
cosOp(cosDenZero) = pi/4;
gt_alpha0 = sincOp.*cosOp;
GF_alpha0 = fft(gt_alpha0,1024);
Matlab Demo
8

alpha = 0.5;
cosNum = cos(alpha*pi*[-fs:1/fs:fs]);
cosDen = (1-(2*alpha*[-fs:1/fs:fs]).^2);
cosDenZero = find(abs(cosDen)<10^-10);
cosOp = cosNum./cosDen;
cosOp(cosDenZero) = pi/4;
gt_alpha5 = sincOp.*cosOp;
GF_alpha5 = fft(gt_alpha5,1024);

alpha = 1;
cosNum = cos(alpha*pi*[-fs:1/fs:fs]);
cosDen = (1-(2*alpha*[-fs:1/fs:fs]).^2);
cosDenZero = find(abs(cosDen)<10^-10);
cosOp = cosNum./cosDen;
cosOp(cosDenZero) = pi/4;
gt_alpha1 = sincOp.*cosOp;
GF_alpha1 = fft(gt_alpha1,1024);
Matlab Demo
9
close all
figure
plot([-fs:1/fs:fs],[gt_alpha0],'b','LineWidth',2)
hold on
plot([-fs:1/fs:fs],[gt_alpha5],'m','LineWidth',2)
plot([-fs:1/fs:fs],[gt_alpha1],'c','LineWidth',2)
legend('alpha=0','alpha=0.5','alpha=1');
grid on
xlabel('time, t')
ylabel('amplitude, g(t)')
title('Time domain representation of raised cosine pulse shaping filters')

figure
plot([-512:511]/1024*fs, abs(fftshift(GF_alpha0)),'b','LineWidth',2);
hold on
plot([-512:511]/1024*fs, abs(fftshift(GF_alpha5)),'m','LineWidth',2);

plot([-512:511]/1024*fs, abs(fftshift(GF_alpha1)),'c','LineWidth',2);
legend('alpha=0','alpha=0.5','alpha=1');
axis([-2 2 0 14])
grid on
xlabel('frequency, f')
ylabel('amplitude,|G(f)|')
title('Frequency domain waveform of raised cosine pulse shaping filters')
Result
10
Result
11
Examples
12

 Example 1: The elements of seq. { ak } are independent


sequences taking a value of ±1with equal probability. If
the data sequence is used to modulate a basic rectangular
pulse as

 𝑠 𝑡 = 𝑘=−∞ 𝑎𝑘 𝑔(𝑡 − 𝑘𝑇)

 Find:
 The power spectral density of 𝑠(𝑡);
Cont..
13

 For g(t) and knowing the randomness of


seq. ak

𝐺 𝑓 2 𝑗2𝜋𝑛𝑓𝑇
 𝑆𝑠 𝑓 = 𝑅
𝑛 𝑛 𝑒
𝑇

= 𝐴2 𝑇 𝑠𝑖𝑛𝑐 2 (𝑇𝑓)
 Controlled ISI: Assume we want to have a null in the spectrum
at 𝑓 = 1/2𝑇 . This is done by a precoding of the form
𝑏𝑛 = 𝑎𝑛 + 𝛼𝑎𝑛−1 . Find the α that provides the desired null.
EX…. Proakis!!!
Controlled ISI Trade-off
14

 Binary data transmission over a physical baseband


channel can be accomplished at a rate close to the
Nyquist rate, using realizable filters with gradual
cutoff characteristics.
 However, these desirable characteristics are achieved
at a price :
A large SNR is required to yield the same average
probability of symbol error in the presence of noise.
Optimum Receivers with
15
ISI & AWGN
 For Received signal
r(t)   I n h(t  nT)  z(t)
n

 The probability of error for zero ISI & AWGN channel


with zero mean & No/2 variance

Pe  Q ( SNR )
Optimum Receivers with
16
ISI & AWGN
 With Controlled ISI, the matched filter output sampled
at 𝑡 = 𝐾𝑇 =/2𝑊

y k  ak   ak  1  z ' k ;
 The probability of error for ML optimum detection
d2
Pe  Q ( )
v 2
Optimum Receivers with
17
Zero ISI & AWGN
 Example 2: Considering Example above Determine The
error probability of the optimal detector for an
additive white Gaussian noise channel with noise
power spectral density equal to N0/2.
Optimum Receivers with
18
Controlled ISI & AWGN
 Example :3 In a binary PAM system, the input to the detector is
𝑦𝑚 = 𝑎𝑚 + 𝑛𝑚 + 𝑖𝑚

 where 𝑎𝑚 = ±1 is the desired signal, nm is a zero-mean Gaussian


random variable with variance 𝜎 2 , and 𝑖𝑚 represents the ISI due to
channel distortion. The ISI term is a random variable that takes the
values − 1/2 , 0, and 1/2 with probabilities 1/4 , 1/2 , and 1/4 ,
respectively.

 Determine the average probability of error as a function of 𝜎 2 and


compare the result with the above example.
Ex:
19

 Proakis; 5Th Ed. 9.9 , 9.12, 9.14, 9.13 ; page 676-679

Вам также может понравиться