Page 517 - Fiber Optic Communications Fund
P. 517

498                                                               Fiber Optic Communications


                                  Received
                                                              y I      y I,l
                                optical signal  90°      PDs      ADC
                                             Hybrid      and  y Q      y Q,l  DSP
                                                       front end  ADC
                                        LO

            Figure 11.1 Block diagram of a coherent IQ receiver. LO = local oscillator, PDs = photodiodes, ADC =
            analog-to-digital converter, DSP = digital signal processing.

                       √
            where K = R P P   ∕2, n(t) represents noise due to ASE and shot noise, and  and  are the delays
                          r LO                                                 I     Q
                          ∘
            introduced by 90 hybrids and other parts of the coherent receiver. The constant K has no impact on the
            performance. So, from now on, we set it to unity. An ADC discretizes the analog signal at a sampling rate
            R samp  ≥ B , where B = 1∕T is the symbol rate. Typically, two samples per symbol are required. The samples
                             s
                                   s
                     s
            are combined into a complex number. The outputs of the ADC are written as
                             y = Re{s exp [−i(2f (t −  )+Δ )] + n },                    (11.6)
                              I,l     I,l        IF l  I,l    l    l
                             y Q,l  = Im{s Q,l  exp [−i(2f (t −  )+Δ )] + n },  l = 1, 2, … ,  (11.7)
                                                       Q,l
                                                 IF l
                                                                    l
                                                               l
            where s and s  are the samples of s(t −  ) and s(t −  ), respectively, at t = lT  , T  = 1∕R  . n is
                   I,l   Q,l                   I         Q                   samp  samp    samp  l
            the sample of the noise at t = lT  . DSP performs the complex addition to obtain the received signal as
                                      samp
                                                 y = y + iy .                                 (11.8)
                                                  l   I,l  Q,l
            In general,  could be different from  . Therefore, s and s  may be different, and the real and imaginary
                      I                    Q           I,l   Q,l
            parts of ̃y may not correspond to the same symbol, which could lead to symbol errors. However, this is a
                    l
            systematic error and can be corrected easily. Using the DSP, the delays experienced by I- and Q-channels can
            be removed. After correcting for  and  ,wehave
                                        I    Q
                                   y = x exp [−i(2f t +Δ )] + n ,  l = 1, 2, … ,          (11.9)
                                    l   l         IF l   l    l
            where x = s ≡ s(lT samp ).
                   l
                      l
            11.3   Laser Phase Noise
            The output of a single-frequency laser is not strictly monochromatic but rather has frequency deviations that
            change randomly. The output field of a fiber-optic transmitter may be written as
                                        q (t)= A s(t) exp {−i[2f t − (t)]},              (11.10)
                                         T      T             c
            where s(t) is the data, f is the laser mean frequency, and (t) is the laser phase noise. The instantaneous
                               c
            frequency deviation can be written as (see Eq. (2.165))
                                                        1 d
                                                  f =−      .                                (11.11)
                                                   i
                                                       2 dt
            The instantaneous frequency deviation is a zero-mean Gaussian noise process with standard deviation  .
                                                                                                 f
            Integrating Eq. (11.11), it follows that
                                                             t
                                            (t)= (t )− 2  ∫  f ()d                 (11.12)
                                                    0
                                                              i
                                                            t 0
   512   513   514   515   516   517   518   519   520   521   522