Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

lect13 by ajizai

VIEWS: 4 PAGES: 17

									                      Today’s Schedule
10111
00000
 1111
0000
1010
 1100
0100

        • Reading: Lathi 11.2-11.4 (also 7.2),
1000
0000
00001
10100
11100   • Mini-Lecture 1:
00100
0100       – Power spectral densities
 0011
 0111
 1101   • Mini-Lecture 2:
01010
10010
10111
           – Random processes and linear systems
        • Activity: First Project Group Meeting
 1101
 1111
1010
00011
10000   • Project Discussion and Questions
10110
01010
         Lecture 13         Dickerson EE422        1
10111
00000
            Autocorrelation of Random
 1111
0000
1010
                     Process
 1100

        • Correlation or second moment of a real
0100
1000
0000
00001
10100
          random process with itself at two times:
                        
11100                _________________     
00100
0100    Rx  t1 , t2   x t x t    x1 x2 f x  x1 , x2  dx1dx2
 0011
                            1   2   
 0111
 1101
01010
10010
10111
 1101
 1111
1010
00011
10000
10110
01010
        Lecture 13                  Dickerson EE422                   2
10111
00000
 1111                Wide Sense Stationary
0000
1010
 1100
0100

        • Usually, t1=t and t2=t+t so that t1- t2 =t
1000
0000
00001
10100
11100   • A random process is Wide-Sense
                         ________
          Stationary if: x  t   constant
00100
0100
 0011

                                  Rx  t1 , t2   Rx t 
 0111
 1101
01010
10010
10111
        • Properties              _________
 1101
 1111
                        Rx  0   x 2  t   second moment
                        Rx t   Rx  t 
1010
00011
10000

                        Rx  0   Rx t 
10110
01010
        Lecture 13                Dickerson EE422              3
10111
00000
 1111                Power Spectral Density
0000
1010
 1100
0100                                     ________________
1000
        • Definition                       X      2
                                                       
                          S x    lim 
0000
00001
                                             T
                                                       
10100                                T      T       
11100                                                 
00100
0100
 0011
        • Relationship to Time Autocorrelation
                                        
                             Rx t   S x  
 0111
 1101
01010
10010
10111   • Power of a Random Process
 1101
                                              
                            1
           Px  Rx  0        sx   d  2 sx  f  df
 1111
1010

                           2 
00011
10000
10110                                          0
01010
        Lecture 13             Dickerson EE422                 4
10111
00000
 1111                     Properties of PSD
0000
1010
 1100
0100

        •
1000
0000        Px(f) is always real
00001
10100
11100   •   Px(f) > 0
00100
0100
 0011
        •   When x(t) is real, Px(-f)= Px(f)
                
                    Px  f  df  P  total normalized power
 0111
 1101
01010
        •   
                                                              ____
                                       
10010
10111   •   If x(t) is WSS,  Px  f  df              P  x  Rx  0 
                                                                2

 1101
 1111
1010    •                 
            Px  0    Rx t  dt
00011                     
10000
10110
01010
        Lecture 13                    Dickerson EE422                        5
10111
00000
 1111           General Expression PSD
0000
1010
 1100
0100

        • General expression for the PSD of a
1000
0000
00001
          Digital Signal:            F f  
10100                                      2

                          Ps  f            R  k  e j kT
11100
                                                                  s
00100
0100                                                 Ts   k 
 0011
 0111
 1101
            – F(f) is the Fourier Transform of the Pulse
01010
10010
              Shape f(t)
            – Ts is the sampling interval
10111
 1101
 1111
1010
00011
            – R(k) is the autocorrelation of the data:
                                 I
                      R  k     an ak  n  Pi
10000
10110
01010                           i 1
        Lecture 13               Dickerson EE422                      6
10111
00000
 1111                Autocorrelation Term
0000
1010
 1100
0100
1000
                                 I
                      R  k     an ak  n  Pi
0000
00001
10100
11100                           i 1
00100
0100
 0011
 0111   • an and ak+n are the levels of the nth and
 1101
01010
10010
          (n+k)th symbol positions
10111
 1101   • Pi is the probability of having the ith anan+k
 1111
1010
00011
          product
10000
10110
01010
        Lecture 13                   Dickerson EE422   7
10111
00000
 1111            Digital PSD Discussion
0000
1010
 1100
0100
1000

                                  F f 
0000                                        2
00001                                               
10100
11100
00100
                     Ps  f  
                                    Ts
                                                
                                                k 
                                                        R  k  e j kTs
0100
 0011
 0111
 1101
01010   • PSD only depends on the
10010
10111
 1101
             – Pulse shape (f(t))
 1111
1010         – Statistical properties of the data
00011
10000
10110
01010
        Lecture 13                Dickerson EE422                          8
10111
00000
 1111           Example: Unipolar NRZ
0000
1010
 1100
0100

        •
1000
0000          Possible levels are +A and 0
00001
10100
11100   •     Square pulses of width Tb
00100
0100
 0011
        •     Find the PSD:
 0111
 1101
01010
        1.    Find the spectrum of pulse:
10010
10111                         t               sin  fTb
 1101
                 f  t       F  f   Tb
 1111
1010                          Tb                 fTb
00011
10000
10110
01010
        Lecture 13                Dickerson EE422           9
10111
00000
 1111                 Evaluate Autocorrelation
0000
1010
 1100
0100
1000
0000       • For k = 0: there are 2 possibilities, an=A or an=0:
00001                     2
          R  0     an an i Pi  A2 P 1  02 P  0   A2 / 2 (if P 1  0.5)
10100
11100
00100                    i 1
0100
 0011
 0111      • For k >0: there are 4 possibilities, an=0 orA and
 1101
01010        an+k=0 or A:
10010
                  4
           k     an ank i Pi  A2 P  an  1 and ank  1  0  A  P  an  0 and ank  1
10111
 1101 R
 1111            i 1

                  A  0  P  an  1 and an  k  1  02  P  an  0 and an  k  0   A2 / 4
1010
00011
10000
10110
01010
            Lecture 13                       Dickerson EE422                                  10
10111
00000
 1111   Putting it all together
0000
1010
 1100
0100
                                                  2
                  A Tb  sin  f Tb                      
                                                               j 2 kf Tb 
1000                     2
         P f                                       1   e
0000
                                                                        
                   4   f Tb 
00001
10100
11100
                                                       k              
00100                                                 2
                       A Tb  sin  f Tb 
                              2
                                                           1          
                                                          1    f  
0100
 0011                                  
 0111
 1101
                        4   f Tb                        Tb         
01010
10010
10111   • Simplify using:                                                            
                                                                                1
             – Poisson Sum Formula k e                                                 f n T 
 1101                                                            j 2 kf Tb
 1111                                                                                           b
1010                                                                         Tb   k 
00011
10000
             – sin  f Tb
10110                              0 at f  n / Tb for n  0
01010                 f Tb
        Lecture 13                         Dickerson EE422                                       11
10111
00000
 1111                 Activity
0000
1010
 1100
0100

        • Find the PSD for BiPolar NRZ Signaling
1000
0000
00001
10100
11100   1. Find F.T. of pulse
00100
0100
 0011
        2. Find Autocorrelation of signal
 0111
 1101
01010
10010
10111
 1101
 1111
1010
00011
10000
10110
01010
        Lecture 13     Dickerson EE422        12
10111
00000
 1111                                  Solution
0000
1010
 1100
0100
1000
        1. Find the spectrum of pulse:
                                  t               sin  fTb
0000

                     f  t       F  f   Tb
00001
10100
11100
                                  Tb                 fTb
00100
0100
 0011
 0111   2. Autocorrelation
 1101
01010
        • For k = 0: an=A or an= -A:
                                 2
                       R  0     an an i Pi  A2 P 1    A  P  0   A2
10010                                                               2
10111
 1101                           i 1
 1111
1010
        • For k >0: an=-A or A and an+k=-A or A:
                                                                                        
                       4
          R  k     an an  k i Pi  A2  A   A     A  A    A 
00011
                                                                                                 0
                                                                                     2       1
10000                                                                                        4
10110                 i 1
01010
        Lecture 13                         Dickerson EE422                                            13
10111
00000
 1111                White Noise Process
0000
1010
 1100
0100
1000
0000    • A random process is said to be a white noise
00001
10100     process if the PSD is constant over all
11100
00100     frequencies: P  f   N0               N
                                          R t   0  t 
0100                      x                      x
 0011
                                    2                   2
 0111
 1101
01010
10010                   R(t)                     P(f)
10111
 1101
 1111
1010
00011                  N0/2                                 N0/2
10000
10110
01010                          t                             f
        Lecture 13             Dickerson EE422                   14
10111
00000
 1111                  Linear Systems
0000
1010
 1100
0100    • Recall:
                 y t   h t   x t   Y  f   H  f  X  f 
1000
0000
00001
10100
11100   • This is still valid if x and y are random processes,
00100
0100      x might be signal plus noise or just noise
 0011
 0111
 1101
        • What is the autocorrelation and PSD for y(t)
01010
10010
          when x(t) is known?
10111          x(t)                                y(t)
 1101
 1111         X(f)
                        Linear Network             Y(f)
1010                          h(t)
00011         Rx(t)                                Ry(t)
10000                        H(f)
10110                                              Py(f)
01010
              Px(f)
        Lecture 13               Dickerson EE422                        15
10111
00000
 1111                   Key Theorem
0000
1010
 1100
0100

        • IF a WSS random process x(t) is applied to
1000
0000
00001
10100
11100
          a LTI system with impulse response h(t),
00100
0100
          the output autocorrelation is:
                         
             Ry t     h  1 h  2  Rx t  2  1  d 1d 2
 0011
 0111
 1101                    

                       h  t   h t   Rx t 
01010
10010
10111
 1101
 1111   • And the output PSD is:
1010

                     Py  f   H  f  Px  f 
00011                                       2
10000
10110
01010
        Lecture 13              Dickerson EE422                     16
10111
00000
 1111                  Next Time
0000
1010
 1100
0100

        • Reading: Lathi 11.2-11.4 (also 7.2),
1000
0000
00001
10100
11100   • Mini-Lecture 1:
00100
0100       – Equivalent Noise Bandwidth
 0011
 0111
 1101   • Activity
01010
10010
10111   • Mini-Lecture 2:
 1101
 1111      – Quiz 2
1010
00011
10000   • Mini-Lecture 3:
10110
01010
         Lecture 13         Dickerson EE422      17

								
To top