Page 161 - Fiber Optic Communications Fund
P. 161

142                                                               Fiber Optic Communications


                                                                       ∗
                                 ∗
            same as that of having a a =−1. Therefore, the ensemble average of a a is zero when k ≠ l. When k = l,
                                l k                                  l k
                                                  ∗        2
                                              < a a >=< |a | >= 1.                            (4.11)
                                                 l k      k
            The terms in Eq. (4.9) can be divided into two groups; terms with k = l and terms with k ≠ l,
                                             [                                        ]
                                                                   L
                                                L            ∑    ∑
                                                              L
                                               ∑
                                                                           ∗
                          2
                                                        2
                               2
                   (f)= A |̃p(f)| lim  1         < |a | > +          < a a > e i2f(l−k)T b  .  (4.12)
                   m      0                           k                  l k
                                L→∞ (2L + 1)T
                                            b  k=−L          l=−L k=−L,k≠l
            Using Eqs. (4.10) and (4.11) in Eq. (4.12), we find
                                                                    2    2
                                                         2L + 1   A |̃p(f)|
                                                                    0
                                                  2
                                            2
                                     (f)= A |̃p(f)| lim       =         .                   (4.13)
                                     m      0      L→∞ (2L + 1)T
                                                               b     T b
            4.4.1   Polar Signals
            Consider a polar signal with RZ pulses. The pulse shape function p(t) is
                                                        (    )
                                                           t
                                                p(t)= rect     ,                              (4.14)
                                                          xT b
                                               ̃ p(f)= xT sinc (xT f),                        (4.15)
                                                      b
                                                             b
            where sinc (y)= sin (y)∕(y). Using Eq. (4.15) in Eq. (4.13), we find
                                                     2 2
                                             RZ
                                                             2
                                             (f)= A x T sinc (xT f).                        (4.16)
                                             m       0   b      b
            When x = 1, this corresponds to a NRZ pulse and the PSD is
                                              NRZ      2     2
                                               (f)= A T sinc (T f).                         (4.17)
                                              m        0 b     b
            Fig. 4.3 shows the PSD of a polar signal with NRZ and RZ pulses. As can be seen, the signal bandwidth of RZ
            pulses with 50% duty cycle is twice that of NRZ. For a polar signal with NRZ, the effective signal bandwidth
            (up to the first null) is B Hz. This is twice the theoretical bandwidth required to transmit B pulses per second
            [1]. Therefore, the polar signal is not the most bandwidth-efficient modulation format.

            4.4.2   Unipolar Signals
            The PSD of a unipolar signal is given by (see Example 4.5)

                                                    [                  ]
                                              2    2         ∞
                                            A |̃p(f)|     1  ∑       l
                                              0
                                      (f)=          1 +       (f −  ) .                   (4.18)
                                      m
                                               4T        T           T
                                                 b        b l=−∞      b
            For RZ pulses, ̃p(f) is given by Eq. (4.15). Substituting Eq. (4.15) in Eq. (4.18), we find
                                                        [        ∞          ]
                                          2 2
                                         A x T b              1  ∑       l
                                          0
                                  RZ
                                                  2
                                  (f)=       sinc (xT f) 1 +      (f −  ) .               (4.19)
                                  m                   b
                                           4                 T           T
                                                              b l=−∞      b
   156   157   158   159   160   161   162   163   164   165   166