Docstoc

Discrete-Time Convolution - Department of Electrical _amp; Computer

Document Sample
Discrete-Time Convolution - Department of Electrical _amp; Computer Powered By Docstoc
					 EE313 Linear Systems and Signals          Fall 2010



 Discrete-Time Convolution
             Prof. Brian L. Evans
Dept. of Electrical and Computer Engineering
      The University of Texas at Austin

     Initial conversion of content to PowerPoint
             by Dr. Wade C. Schwartzkopf
     Discrete-time Convolution
• Output y[n] for input x[n]

• Any signal can be decomposed
  into sum of discrete impulses

• Apply linearity properties of
  homogeneity then additivity

• Apply shift-invariance

• Apply change of variables
                                  7-2
        Discrete-time Convolution
• Filtering viewpoint               x[n]                y[n]



   Hold impulse response h[n] in place and change variables
   Flip and slide input signal x[n] about impulse response
• Example of finite impulse response (FIR) filter
   Impulse response has finite extent (non-zero duration)
      h[n]   Averaging filter
             impulse response    y[n] = h[0] x[n] + h[1] x[n-1]
                                      = ( x[n] + x[n-1] ) / 2
                          n
  0      1   2     3                                           7-3
   Convolution in Both Domains
• Continuous-time convolution of x(t) and h(t)

   For each value of t, we compute a different (possibly)
     infinite integral.
   Discrete-time definition is the continuous-time definition
     with integral replaced by summation


• Linear time-invariant (LTI) system
   Output signal in time domain is convolution of impulse
     response and input signal
   Impulse response uniquely characterizes the LTI system
                                                           7-4
           Convolution Demos
• Johns Hopkins University Demonstrations
  http://www.jhu.edu/~signals
  Convolution applet to animate convolution of simple
     signals and hand-sketched signals
  Convolve two rectangular pulses of same width gives a
     triangle (see handout E)
• Some conclusions from the animations
  Convolution of two causal signals gives a causal result
  Non-zero duration (called extent) of convolution is sum of
    extents of two signals being convolved minus one
                                                          7-5
        Fundamental Theorem
• The Fundamental Theorem of Linear Systems
  If one inputs a complex sinusoid into an LTI system, then
     the output will be a complex sinusoid of the same
     frequency that has been scaled by the frequency
     response of the LTI system at that frequency
  Scaling may attenuate the signal and shift it in phase
  Example in continuous time: see handout G
  Example in discrete time. Let x[n] = e j W n,



    H(W) is the discrete-time Fourier transform of h[n] and
    is also called the frequency response                 7-6
          Frequency Response
• For continuous-time LTI system



• For discrete-time LTI system



• Note: Identity for cosine input assumes a real-
  valued impulse response
                                                7-7
   Example Frequency Response
• System response to complex exponential e j W n
  for all possible frequencies W in rad/sample
            |H(W)|                   Ð H(W)
            passband
stopband               stopband


                           W                      W
      -Ws -Wp     Wp Ws



• Passes low frequencies, a.k.a. lowpass filter
                                                  7-8
Differentiator/Difference Operation
• Continuous                  • Discrete
  f(t)            y(t)          f[n]       y[n]




   We can remove scaling
   by 1/Ts without changing
   lowpass response
                                                  7-9
                     First-Order FIR Filters
  x[n]

                                      y[n] = ½ x[n] + ½ x[n-1]
                                                                     y[n] = ½ x[n] - ½ x[n-1]




                               n                                n
    Signal with a spike                  Output of
                                       averaging filter
signal = [ 1 1 1 1 1 10 1 1 1 1 1 ];                                                               n
figure(1); stem(signal);                                                       Output of
                           averagingFilter = [ 0.5 0.5 ];
                           average = conv(averagingFilter, signal);
                                                                          difference filter
                           figure(2); stem(average);
                                                        differenceFilter = [ 0.5 -0.5 ];
                                                        difference = conv(differenceFilter, signal);
                                                                                         7 - 10
                                                        figure(3); stem(difference);
     Mandrill Demo (DSP First)
• First-order difference FIR filter
                                        h[n]   First-order difference
                                                 impulse response
   Highpass filter (sharpens
     input signal)                                            n
   Impulse response is {1, -1}
• Five-tap discrete-time (scaled) averaging FIR
  filter with input x[n] and output y[n]

   Lowpass filter (smooth/blur input signal)
   Impulse response is {1, 1, 1, 1, 1}
                                                             7 - 11
     Mandrill Demo (DSP First)
• DSP First, Ch. 6, Freq. Response of FIR Filters,
   http://www.ece.gatech.edu/research/DSP/DSPFirstCD/
• From lowpass filter to highpass filter
   original image ® blurred image ® sharpened/blurred image
• From highpass to lowpass filter
   original image ® sharpened image ® blurred/sharpened image
• Frequencies that are zeroed out can never be
  recovered (e.g. DC is zeroed out by highpass filter)
• Order of two LTI systems in cascade can be
  switched under the assumption that computations
  are performed in exact precision              7 - 12
     Mandrill Demo (DSP First)
• Input image is 256 x 256 matrix
   Each pixel represented by eight-bit number in [0, 255]
   0 is black and 255 is white for monitor display
• Each filter applied along row then column
   Averaging filter adds five numbers to create output pixel
   Difference filter subtracts two numbers to create output pixel
• Full output precision is 16 bits per pixel
   Demonstration uses double-precision floating-point data and
    arithmetic (53 bits of mantissa + sign; 11 bits for exponent)
   No output precision was harmed in making of this demo J
                                                            7 - 13
  Linear Time-Invariant System
• Any linear time-invariant system (LTI)
  system, continuous-time or discrete-time,
  can be uniquely characterized by its
   Impulse response: response of system to an impulse
   Frequency response: response of system to a two-
     sided complex exponential input signal for all
     possible frequencies
   Transfer function: Laplace transform (or z-transform)
     of impulse response
• Given one of the three, we can find other
  two provided that they exist                             7 - 14

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:7/24/2013
language:English
pages:14