The symbiotic combination of polarization division multiplexing (PDM), multilevel modulation formats, coherent optical detection, and digital signal processing (DSP) has resulted in the coherent optical transmission system, which provides higher channel capacity than traditional optical transmission system due to its high spectral efficiency. Coherent optical communications are attracting strong attention and are considered for next generation optical backbone networks. This thesis describes the composition of coherent optical transmission system and researches on the DSP algorithms of channel equalization and frequency offset (FO) estimation in digital coherent optical receivers.The main contributions of this thesis are as follows:(1) A time-domain equalization method for compensation of chromatic dispersion (CD) using Wiener filter in digital coherent optical receivers is proposed. The Wiener filter can compensate CD, filter out the noise from the corrupted signal, and provide an optimal estimate of the signal of interest in the sense of minimum mean square error. The tap weights of the Wiener filter applied to a multirate system are determined by the channel impulse response and the signal-to-noise ratio. Numerical Results of 16-ary quadrature amplitude modulation (16-QAM) coherent optical multirate transmission systems show that the proposed equalizer outperforms the existing ones in term of bit error ratio (BER) performance.(2) The algorithm of FO estimation based on fast Fourier transform of the sample signal’s argument for quadrature phase shift keying (QPSK) coherent optical system has been investigated, in which the FO estimation range is only [-Rs/12, Rs/8] (Rs is the symbol rate). In this thesis, an improved algorithm is proposed with the sign of FO being determined by means of an improved judging method. The estimation range of the proposed algorithm can be up to [-Rs/8, Rs/8], with the same accuracy as that of the previous algorithms being obtained, yet at the same time requiring very little additional computational burden.(3) This thesis presents a novel FO estimation algorithm based on the differential phase of all consecutive symbols for square 16-QAM coherent optical systems. Different from the existing algorithm using only consecutive Class I symbols (CCIS), all of the constellation points (i.e., both Class I and Class II symbols) are used here to estimate FO. By considering the compatibility of symbols in different Class,8th power operation is implemented to remove the modulated data phase, before which rotating by π/8 is needed for Class Ⅱ symbols. The proposed algorithm leads to 0.7 dB signal-to-noise ratio per symbol improvement at bit-error rate of 10-3 due to its high estimation accuracy compared with the conventional CCIS scheme. |