Adaptive Phase Estimation in the Presence of Nonlinear Phase Noise for Carrier Phase Recovery of PM-QPSK Signals in Coherent Optical Receivers

Abstract

There are a number of architectural choices to improving feedback and feedforward carrier phase recovery (CPR) algorithms: filter length, filter shape, error signal, etc. The optimal choice depends on the relative strengths of phase noise and additive noise. On the other hand, the optimal average length of CPE used for filtering the phase noise estimate is quite subjective. In principle, a short average length enables fast carrier phase tracking, and a longer average length enables large ASE noise tolerance. Consequently, the optimal average length becomes relatively so short in the presence of fiber nonlinearities when phase noise is large to prevent any symbol with occurring large ASE noise from severely distorting the CPE process. This may affect the stability of CPE performance and subsequently causing performance degradation. For the steady performance of CPE in the presence of large phase noise, this paper proposes adaptive filtering technique to detect and isolate symbols with significantly large ASE noise from the filtering process to improve stability and performance of CPE algorithms. The results showed that the proposed adaptive filtering CPE technique combined with normalization processing in CPE provides about 1.0 dB Q-factor performance improvement over the conventional CPE method in mitigating the combined effects of laser linewidth and fiber nonlinearity for coherent phase-modulated optical signals. The technique automatically identifies and eliminates symbols with significantly large noise from the filtering process to improve stability and estimation accuracy of the carrier phase estimators. The complexity of additional hardware required to implement the scheme is quite simple.