Unbiased Channel Estimation Based on the Discrete Fresnel Transform for CO-OFDM Systems

In this letter, the deviation of the channel estimator based on intra-symbol frequency-domain averaging (ISFA) for coherent optical orthogonal frequency-division multiplexing (CO-OFDM) is investigated. The deviation-induced estimation error is derived analytically as a function of pulse broadening caused by chromatic dispersion and averaged noise power, and thereby the optimum averaging window size can be determined. To avoid the deviation, we propose an unbiased channel estimation algorithm based on the discrete Fresnel transform (DFnT) for CO-OFDM systems, utilizing the convolution-preservation property of DFnT for intra-symbol averaging. It is shown that the DFnT-based channel estimator converges to the actual channel response under estimation, and achieves better performance than the ISFA estimator, especially in highly dispersive channels. Finally, numerical results are provided to confirm the analysis and its advantages.