Joint intra symbol frequency domain averaging channel estimation for polarization-division-multiplexed CO-OFDM/OQAM system

Abstract

For the advantage of high spectral efficiency and lower out of band power radiation, orthogonal frequency-division multiplexing offset quadrature amplitude modulation (OFDM/OQAM) system have aroused wide interest. For OFDM/OQAM, prototype filter with promising time-frequency localization properties have been employed to suppress the out of band radiation. Time offset between the real and imaginary part of the QAM symbol and the phase offset module guaranteed the real filed orthogonal condition of the sub-carriers for OFDM/OQAM. Most recently, numerical experimental demonstrations for coherent optical OFDM/OQAM (CO-OFDM/OQAM) have been reported. Polarization-division-multiplexed structure could promote the spectral efficiency of CO-OFDM/OQAM further. However, chromatic dispersion (CD) and polarization mode dispersion (PMD) effect brought serious intrinsic imaginary interference (IMI) to PDM CO-OFDM/OQAM. Thus, effective channel estimation methods have to be developed to combat IMI and maintain system performance. Although numerical frequency domain channel estimation method based on the concept of pseudo pilot have been discussed, the correlation of the amplified spontaneous emission (ASE) noise brought by the time offset module decreased the channel estimation accuracy. To our best knowledge, channel estimation method based on weight averaging have not been discussed for PDM CO-OFDM/OQAM system. In this paper, we proposed joint intra symbol frequency domain averaging (JISFDA) channel estimation method based on the maximum-likelihood criterion for the PDM CO-OFDM/OQAM system. Compared to the conventional channel estimation method which only use single subcarrier for channel estimation, the JISFDA method considers the correlation of the ASE noise on each subcarrier and utilizes weight averaging method for the channel estimation matrix. As shown in the theoretical analysis and Montel Carlo simulation results, the interference induced by CD, PMD, and the correlation property of the ASE noise could be suppressed significantly with the using of the JISFDA method.