Sie sind auf Seite 1von 3

ECE-S390 - Transforms II Summer 2015

Homework 2
Reading (Week 2 Lecture)
From BB Learn

Review posted Matlab codes


Review notes on convolution with Matlab
Review notes on manual computation of DFT and IDFT
Read notes on correlation you need to use these for the instructor problems

From Oppenheim and Willsky

Chapter 7: 7.0, 7.1, (skim 7.1.2) 7.2 , 7.3

Homework Problems
Problems form Oppenheim

Instructor Problems: use linear convolution/linear correlation for these problems not the
DFT see posted notes on how to do correlation in Matlab
Problem 1: Cross correlation: Do the problem by hand you should verify using Matlab
Given the sequences:
x(n) = {1 7 1 0 0 0 5 5 5 0 0 0 4 5 4 0 0 0 -1 1 -1}

y(n) = {0 0 1 1 1 0 0 }

The arrow or highlight defines the 0 of the signal


a) Find rxy(l) , which value(s) of index l describes where y(n) best fits the signal x(n).
b) Compute xy(l) , the normalized cross-correlation for values of l = { -2, -1, 0, 1, 2} so
you only need to compute 5 values not the entire sequence.
c) Find the energy of each signal and verify that | rxy(l)| (ExEy)1/2
d) Find ryx(l) discuss how it is different than rxy(l).
Problem 2: Autocorrelation of noise
The Matlab code snippet will create random values with a normal distribution having a
mean of 1 and a standard deviation of 2 (see help randn, note the line rng() makes the
sequence the same each time you run it for comparisons but it is still random)
rng('default');
r = 1 + 2.*randn(100,1);
create a signal with N = 1000 elements and plot it, also use mean() and std() to compute
the actual values of the mean and standard deviation. These values get better as N
increases.
Perform an autocorrelation of the signal for the cases with mean of {0, 1, 10}
o Discuss what happens in terms of autocorrelation value rxx(0) Hint think energy.

ECE-S390 - Transforms II Summer 2015


Homework 2

Problem 3: Auto correlation to find periodicity (Matlab required)


load file measured_data.p from BB Learn into a directory. A .p file is a file which
can be executed but will not show the Matlab code, note it is impossible to see the code (
at the Matlab command window prompt type measured_data (followed by enter)
you will see a graph of the measured data (single plus noise) and variables in the
workspace, T is the time between samples, N is the total number samples, n is the index
array for plotting and meas is the signal plus noise.
The actual signal consists of a sine wave with a non standard (not 1 or 2 Hz)
frequency, using autocorrelation determine the frequency of the signal.
Hint find the peaks (one way is with the data tip tool) and do something like averaging
the period. You can also write code to solve this.
You must turn in your published code including the autocorrelation plot, properly
annotated with zero in the center and your analysis to determine the frequency.
Problem 4:
You are given a pure sine wave of 180 Hz. Determine what frequency would be seen if
sampling is:
a)
b)
c)
d)

500 Hz
300 Hz
200 Hz
What sampling frequency would be needed so that the aliased signal would be 45 Hz?

Sketch the Fourier transform of the signal and show the sampling frequency and folding
frequency.
Problem 5:
Given the waveform x(t) = sin(2f + ) with f = 100 Hz and = /4.

Describe the signal that would be reconstructed if we sample at exactly 200 Hz.

You must include the following sheet as the first page of your homework submission; fill in the
status of all problems

ECE-S390 - Transforms II Summer 2015


Homework 2

Name_____________________________________________________________________

Due Wednesday 9 July 2014 at lecture

1) Problems form Phillips and Parr and instructor problems


Indicate the status of each problem as:
C completed; P partially completed; N not attempted

Problem
Oppenheim
7.6
Oppenheim
7.10
Oppenheim
7.21
Oppenheim
7.22

Status

Instructor 1
Instructor 2

Instructor 3
Instructor 4
Instructor 5

Overall grade: __________________________________-

Grade

Das könnte Ihnen auch gefallen