OFDM-OQAM involves special modulation and demodulation methods, owing to which offers several advantages, including flexible time-frequency lattice, very low out-of-band spectral leakage, and excellent spectral efficiency. Presently, it is widely applied in fields such as wireless communication and optic-fiber communication. In the past, several frequency-domain channel estimation methods have been proposed for CO-OFDM-OQAM. However, these methods are associated with various drawbacks. For instance, more guard intervals need to be inserted between the training sequence and payload to reduce the influence of inherent imaginary part interference (IMI), which can lead to excessive spectrum resources. Thus, this work proposes and studies an improved frequency-domain channel estimation method for PDM CO-OFDM-OQAM systems. The proposed method remarkably enhances the spectrum efficiency and does not severely degrade the power peak-to-average ratio (PAPR) of the signal.

A fully-loaded training sequence was designed using real-valued pilots based on the demodulation principle of a polarization multiplexing system and the symmetry rules of an inherent imaginary interference. This method was combined with the interpolation method to ensure that number of frequency-domain symbols to be occupied by the training sequence on each polarization state was reduced to four. The training sequence and the payload at the transmitter's end form a data frame, which assist in generating a baseband transmission signal. The training sequences of x-polarized and y-polarized states occupied four frequency-domain symbols and the real-valued pilots were placed in the 2nd and 3rd frequency-domain symbols. All the odd-numbered subcarriers of the x-polarized state were zero, while the even-numbered subcarriers were not zero. In contrast, the odd-numbered subcarriers of the y-polarized state were not zero, while all the even-numbered subcarriers were zero. For convenience, we denoted the transmission training sequences in the x- and y- polarized directions as [z,px1,px2,z] and [z,py1,py2,z], respectively. These transmission sequences satisfy the given symmetry conditions. The channel frequency response coefficients at the receiver's end for a portion of the subcarriers were estimated using the analytical formula. Alternatively, the channel response coefficients for other even and odd subcarrier positions were calculated using the interpolation method. Finally, intra-symbol frequency-domain average (ISFA) was used to lessen the influence of noise and interference and to further improve the channel estimation accuracy. In addition, the PAPR values of PDM CO-OFDM-OQAM were calculated using different channel estimation methods.

In this paper, a numerical simulator for a polarization multiplexing CO-OFDM-OQAM system is developed. By using the simulator, the spectral efficiency, peak-to-average power ratio, and channel estimation performance of the method have been verified under three scenes of back-to-back (BtB), 100-km, and 200-km fiber transmissions. Our findings reveal that E-IAM-C method has the greatest impact on the PAPR performance of the system, while the proposed method has the least impact. Furthermore, the investigation of channel estimation performance of the proposed method in the BtB case reveals that the BER of the proposed method after 10 cycles of ISFA shows better results than that of the interference approximation method using real-valued pilots (IAM-R) in the OSNR range under investigation (10 dB-30 dB). This work also focuses on the channel estimation performance of the proposed method in nonlinear fiber channels. Overall, our results demonstrate that the proposed method provides an efficient channel estimation capability in both linear and nonlinear fiber channels.

This work proposes and studies an improved frequency-domain channel estimation method for PDM CO-OFDM-OQAM systems. The main advantage of this method is that it only requires four frequency-domain symbols for the training sequence of each polarization branch. Compared with IAM-R and the E-IAM-C method that uses complex pilots, number of symbols occupied by the training sequence of the proposed method is reduced by 33.3% and 50%, respectively. Thus, the proposed method substantially improves spectral efficiency. In addition, in the proposed method, the real-valued pilot is a random sequence, which does not worsen the PAPR of the signal. The channel estimation capability of the proposed method is verified quantitively in an actual fiber channel by considering the effects of fiber dispersion, polarization mode dispersion, and nonlinearity. Moreover, the results show that the proposed method achieves better BER performance even for nonlinear fiber channels. The findings of this work provide useful insights into the research and development of short-distance fiber communication systems based on OFDM-OQAM.