Page 539 - Fiber Optic Communications Fund
P. 539

520                                                               Fiber Optic Communications


            and from Eq. (11.138), we have
                                                 Δz
                                                         2
                                                                                2
                              (t, Δz)= (t, 0)−   |A(t, z)| dz = (t, 0)− Δz |A(t, 0)| ,  (11.140)
                                              ∫
                                                                        eff
                                                0
            where
                                                     exp (Δz)− 1
                                              Δz eff  =         .                           (11.141)
                                                          
            Substituting Eqs. (11.140) and (11.139) in Eq. (11.136), we find
                                                                    2
                                   q (t, Δz)= q (t, 0) exp (−iΔz |q (t, 0)| + Δz),      (11.142)
                                    b         b             eff  b
            where
                                            q (t, 0)= A(t, 0) exp [i(t, 0)].              (11.143)
                                             b
                          l
            With q (t, 0)= q (t, Δz), Eq. (11.142) becomes
                  b
                          b
                                            l                  l     2
                                  q (t, Δz)= q (t, Δz) exp (−iΔz |q (t, Δz)| + Δz).     (11.144)
                                   b        b               eff  b
            Fig. 11.24 illustrates the unsymmetric SSFS. This technique can be summarized as follows:
                                                                       ̂
              (i) Initial field q (t, 0) is known. First, the nonlinear and loss effects (N) are ignored and the output of a
                           b
                                  l
                lossless, linear fiber q (t, Δz) is calculated using the Fourier transformation technique.
                                  b
                                   ̂
             (ii) Next, fiber dispersion (D) is ignored. The NLSE is analytically solved with the initial condition q (t, 0)=
                                                                                            b
                 l
                q (t, Δz) and the field envelope at Δz, q (t, Δz) is calculated using Eq. (11.144).
                 b                              b
            (iii) q (t, 2Δz) is calculated with q (t, Δz) as the initial condition by repeating (i) and (ii). This process is
                 b
                                         b
                repeated until z = L. The step size Δz should be chosen sufficiently small that the absolute value of
                the nonlinear phase shift Δ accumulated over a distance Δz should be much smaller than .From
                Eq. (11.140), it follows that
                                                                        2
                                      |Δ| = |(t, Δz)− (t, 0)| = Δz |A(t, 0)| ≪.   (11.145)
                                                                eff
              A disadvantage of the unsymmetric SSFS is that the step size has to be really small since the error scales
                     2
            since as Δz [24]. The step size can be made significantly larger using the symmetric SSFS, which is described
            as follows. From Eq. (11.125), we have
                                                   Δz
                                                [                   ]
                                   q (t, Δz)= exp    [N (t, z)+ D (t)]dz q (t, 0).          (11.146)
                                    b            ∫     b       b      b
                                                  0
              q b (t, 0)
                 l
                                                                    l
                                l
              = q (t, 0)  Dispersion  q (t, ∆z)  Nonlinear  q b (t, ∆z)  Dispersion  q (t, 2∆z) Nonlinear operation  q b (t, 2∆z)
                 b              b                                   b
                        only             operation           only          exp(∫ ∆z N
                      exp(D ∆z)       exp(∫ ∆z   b (t, z)dz)  exp(D b ∆z)      0    b (t, z)dz)
                                           N
                          b
                                          0
                                                                         l
                                                     q b (t, L)  Nonlinear operation  q (t, L)  Dispersion
                                                                         b
                                                           exp(∫ ∆z   b (t, z)dz)  only
                                                                N
                                                               0                exp(D b ∆z)  q (t, L – ∆z)
                                                                                          b
                          Figure 11.24  Unsymmetric split-step Fourier scheme for backward propagation.
   534   535   536   537   538   539   540   541   542   543   544