ML/MAP Estimation of Frequency Offset and Linear Phase Noise in Coherent Optical Communications

We propose a joint maximum likelihood (ML)/ maximum a posteriori probability (MAP) frequency offset (FO) and linear phase noise (LPN) estimation scheme for coherent optical communications. By designing the pilot symbols as unmodulated signal with known amplitude, the joint estimation can be simplified to the estimation of frequency and phase of a complex single sinusoid. We then utilize an ML Mth power approach to iteratively estimate the phase on those unknown transmitted symbols. The Cramér-Rao lower bounds are derived and the optimum estimation performance of the proposed algorithms are verified through simulations.