Non-data-aided chromatic dispersion estimation for Nyquist spectrally shaped fiber transmission systems

We investigate a non-data-aided chromatic dispersion estimation based on a polynomial fitting algorithm for root-raised-cosine (RRC) Nyquist spectrally shaped fiber transmission systems. We compare the performances of the estimation method on RRC spectrally shaped systems with regular non-return-to-zero systems. At a same symbol rate, a signal with a narrower spectrum has a lower estimation accuracy using a same estimation time window. The estimation accuracy can be improved by increasing the time window. With an OSNR at a bit error rate at a FEC-limit or 3.8×10^-3, 112 Gbit/s Nyquist PDM-QPSK with RRC 0.1 spectral shaping requires 8192 symbols to achieve a measured standard deviation below 200 ps/nm, and 224 Gbit/s Nyquist PDM-16QAM with the same spectral shape requires 4096 symbols to achieve a measured standard deviation below the same level.