Adaptive Phase Estimation in the Presence of Nonlinear Phase Noise for Coherent Optical Detection

Abstract

For phase-modulated systems, digital carrier phase estimation (CPE) has been generally accepted as the fundamental solution to suppressing the challenging effects of laser phase noise (LPN). The optimal average length FIR filter used in carrier phase estimation algorithms, which is the best trade-off between tracking the time-varying phase offset and reducing effects of the Gaussian noise, has been identified as crucial to estimating phase reference from consecutive multi-symbols and impacts performance of digital coherent receivers. The optimal average length depends on factors such as laser linewidth and optical signal-to-noise ratio (OSNR) as well as nonlinearities of the channel, factors that are very difficult to observe in actual networks because they differ device by device, channel by channel, and subject to change with time, since their statistical knowledge may be unknown especially in reconfigurable optical systems. This paper proposes a simple adaptive phase estimation scheme that uses a phase noise reference to suitably set the properties of the FIR filter (i.e., average length) in phase tracking circuits under different channel conditions to mitigate the combined effects of both intrinsic laser phase noise and nonlinear phase noise in coherent phase-modulated optical systems.