Page 480 - Fiber Optic Communications Fund
P. 480

Nonlinear Effects in Fibers                                                        461


           Taylor series expansion of the exponential function in Eq. (10.261) yields
                                            (             2 2        )
                                                          Z  4
                                                    2
                                   u(T, Z)=  1 + i|k| Z −  |k| +···  k.                 (10.262)
                                                          2!
           Comparing Eqs. (10.260) and (10.262), we see that they match up to the first-order term in .Ifwesolve
           Eq. (10.253) and add the second-order correction in Eq. (10.260), we find that Eqs. (10.260) and (10.262)
           would match up to second order in  (see Example 10.11).
            Now, let us consider a more general problem in which neither  nor  is zero. Let the optical field envelope
                                                              2
           at the fiber input be
                                                         ∞
                                                        ∑
                                     q(T, 0)= a(0)u(T, 0)=  b f(T − nT ),                 (10.263)
                                                             n       s
                                                        n=−∞
           where T is the symbol interval and f(T) is the pulse shape. For systems based on OOK,
                 s
                                                  {
                                                    1for ‘1’,
                                              b =                                         (10.264)
                                               n
                                                    0for ‘0’.
           For systems based on PSK or DPSK,
                                                  {
                                                    1   for ‘1’,
                                             b =                                          (10.265)
                                              n
                                                    −1 for ‘0’.
           Eq. (10.251) can be solved using the Fourier transform technique (see Chapter 2). The solution is

                                                              2
                                     u (T, Z)=  −1 {̃u (, 0) exp [i S(Z)∕2]},        (10.266)
                                                   0
                                      0
           where ̃u (, 0)= [u (T, 0)], u (T, 0)= u(T, 0), and S(Z) is the accumulated dispersion
                            0
                                    0
                 0
                                                      Z
                                              S(Z)=     (x)dx.                          (10.267)
                                                    ∫   2
                                                     0
           Taking the Fourier transform of Eq. (10.252), we obtain
                                     d̃u 1   (Z)  2        2
                                            2
                                                                 ̃
                                    i    +      ̃u (, Z)=−a (Z)F(, Z),              (10.268)
                                                   1
                                      dZ     2
                                      ̃
                                                        2
           where ̃u (, Z)= [u (T, Z)] and F(, Z)= [|u (T, Z)| u (T, Z)]. Eq. (10.268) is a first-order ordinary dif-
                 1
                                                  0
                                                          0
                            1
           ferential equation which can be solved to yield
                                                      L tot
                                                             ̃
                                                          2
                                   ̃ u (, L )IF(L )= i ∫  a (x)F(, x)IF(x)dx,         (10.269)
                                   1
                                        tot
                                              tot
                                                     0
           where the integrating factor is
                                                     [      2  ]
                                           IF(Z)= exp iS(Z) ∕2 ,                        (10.270)
           and L  is the total transmission distance. We assume that the dispersion is fully compensated either in the
               tot
           optical or the electrical domain before the decision device. So, S(L )= 0 and Eq. (10.269) becomes
                                                                tot
                                                   L tot
                                                       2
                                                          ̃
                                      ̃ u (, L )= i  ∫  a (x)F(, x)IF(x)dx.           (10.271)
                                      1
                                           tot
                                                  0
   475   476   477   478   479   480   481   482   483   484   485