Fig. 3. Numerical model of a
long-haul coherent transmission system implementing DSP
The physical properties characterize the S-SMF; attenuation α = 0.2 dB,
dispersion D = 16.75 ps/nm/km, dispersion slope = 0.075 ps/nm2/km, and
nonlinearity coefficient γ = 1.2 km-1W-1. A 16 dB gain Erbium-doped
fiber amplifier (EDFA) with a noise figure of 4 dB is used for optical
amplification to compensate for power loss. On the receiver side, the
incoming in-phase and quadrature-phase signals in the x and y
polarization states are detected by a homodyne coherent polarization and
phase diversity receiver. The receiver comprises two
90o optical hybrids and four pairs of balanced PIN
photodetectors. The local oscillator (LO) laser for coherent detection
operates at a frequency, power, and linewidth of 1550 nm, 10 dB, and 0.1
MHz, respectively. After coherent detection, the received signal
undergoes analog-to-digital conversion by four samples per symbol
down-sampling method at the reference symbol rate. The DSP module
processes it after that.
Digital Signal Processing Module
The DSP module, implemented with MATLAB, comprises segmental blocks for
polarization multiplexing, carrier phase estimation, linear dispersion
compensation, and nonlinearity compensation. The constant modulus phase
algorithm implements an adaptive finite impulse response (FIR) butterfly
equalizer to demultiplex the dual-polarized signal. The
Viterbi-and-Viterbi joint-phase estimation algorithm is adopted to
account for relative phase error between the transmitter laser and the
LO laser, which arises due to their non-zero laser linewidth (Viterbi &
Viterbi, 1983). The DBP-based nonlinearity techniques are implemented by
the SSFM method based on the Weiner-Hammerstein model where the
nonlinear part is computed at the midpoint of each step with two
surrounding virtual linear blocks . We use the step sizes given in Table
1 for the 10-step SSFM calculation. The DSP linear blocks have been
characterized in Table 2.
Table 2: Characterization of DSP
linear blocks