Bayesian Filter-based Dynamic Joint Estimation of Carrier Frequency Offset and Linear Phase Noise for CO-OFDM Systems

Coherent optical orthogonal frequency-division multiplexing (CO-OFDM) is a promising technique due to its high spectral efficiency and dispersion tolerance. However, CO-OFDM is sensitive to the carrier frequency offset (CFO) which is caused by the frequency difference between the transmitter laser and local oscillator, and linear phase noise (LPN) which is closely related to the laser linewidth. The existing of CFO and LPN can introduce the inter-carrier interference (ICI) and cause the rotation of constellations, thus degrade the system performance. In this paper, we propose to dynamically and jointly track and compensate the CFO and LPN by utilizing the Gaussian particle filter (GPF) and extended kalman filter (EKF). Firstly, the GPF and EKF can dynamically track the CFO and LPN in a real-time manner, thus can achieve accurate estimation result at even high phase noise variance. Secondly, GPF can successfully prevent the particle impoverishment (PI) problem of the conventional sequential importance resampling (SIR) PF, by recursively updating the mean and variance of the particles based on the weights computed from the posterior density and the designed importance sampling function. Thirdly, the joint estimation of CFO and LPN utilizes merely one OFDM symbol, which ensures the effective data transmission efficiency. Also, EKF can convert the nonlinear problem into linearity which helps to reduce the computation complexity. Simulations are carried out to verify the accuracy, robustness and efficiency of the proposed approach by considering the estimation error variance with the Cramér-Rao lower bound (CRLB), the convergence speed of the GPF and EKF, and the real-time dynamic tracking errors.