Bit-Error Ratio Performance Improvement Using Iterative Decoding for Polybinary-Shaped Super-Nyquist Wavelength Division Multiplexed Signals

Abstract

In super-Nyquist wavelength division multiplexed (WDM) systems with frequency spacing smaller than the signal baudrate, the maximum-likelihood (ML) decoder in the receiver is usually introduced to compensate for intersymbol interference due to tight spectral filtering, such as polybinary shaping. After the ML decoder, symbol errors tend to propagate, causing excess continuous errors. Considering that forward error correction (FEC) is commonly introduced, the excess continuous errors degrade bit-error ratio (BER) performance after FEC, so-called post-FEC BER. In order to suppress the performance degradation, we introduce iterative decoding between the first ML decoder for polybinary shaping and the second FEC decoder in the receiver. First, we calculate BER characteristics of polybinary-shaped super-Nyquist WDM quadrature phase-shift keying (QPSK) signals. The results show that iterative decoding is effective for improving post-FEC BER performance. A lager pre-FEC BER threshold for post-FEC BER <; 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-5</sup> is obtained in super-Nyquist WDM case than in the Nyquist WDM case, although a higher signal-to-noise ratio (SNR) is required. Next, we measure the BER characteristics of three-channel duobinary-shaped super-Nyquist WDM 12.5-Gbaud dual-polarization QPSK signals. The iterative decoding reduces the optical SNR penalty by 0.8 dB. A larger pre-FEC BER threshold of 3.1 × 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-2</sup> is obtained in the duobinary-shaped super-Nyquist WDM case, compared with the threshold of 2.2 × 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-2</sup> in the Nyquist WDM case.