Research On IQ Mismatch Estimation Algorithm For Polarization Division Multiplexed Coherent Optical OFDM Systems

Polarization division multiplexed coherent optical orthogonal frequency division multiplexing(PDM CO-OFDM)technology gathers not only the advantages of OFDM technology with high spectral efficiency,high chromatic dispersion(CD)tolerance,and strong capability against polarization mode dispersion(PMD),but also the advantages of coherent detection technology with high receive sensitivity and long transmission distance,as well as the superiority of polarization division multiplexed technology double transmission rate and channel capacity.However,in the process of in-phase and quadrature phase(IQ)modulation and coherent detection,the amplitude and phase imbalance in the transmitter(Tx)and receiver(Rx)will be introduced due to the nonideal characteristics of the device,which destroy the orthogonality between sub-carriers.In the process of data transmission,the interference between the two polarization states will be introduced due to the influence of CD,PMD and noise in the fiber.Both factors will degrade the system performance seriously.In this thesis,the channel and IQ imbalance estimation and compensation methods are studied.The main contents include:(1)A new training symbol structure is proposed to estimate the channel and Tx IQ imbalance.The compared method gives the mathematical model and regards the the channel and IQ imbalance as a whole factor.Then the estimation and compensation algorithms are proposed based on the proposed training symbols.However,only half of the subcarriers are used,which add the estimated error and degrade the system performance.In our thesis,according to the relationship between the IQ imbalance factors,two estimation methods about channel and IQ imbalance are given.An method achieves the channel and IQ imbalance estimation independently,and the error accumulation is eliminated.The other reduces the overhead of training symbols and improves the spectrum efficiency.Simulation results show that under the 100Gb/s system after 480 km standard single mode fiber transmission,when the phase and amplitude mismatch are 15 o and 3dB respectively,the proposed two methods can realize the effective transmission under the BER 1e-3.But the compared method can not realize any transmission.Furthermore,the complexity of the two proposed methods reduce nearly 64% after comparison and analysis.(2)A new training symbol structure is proposed to estimate the channel and Rx IQ imbalance.The compared method adopts the Gram-Schmidt orthogonalization procedure(GSOP)algorithm combining with the traditional training symbols to realize the IQ imbalance and channel estimation and compensation.However,the IQ imbalance can not be eliminated completely by using GSOP algorithm,then the value of channel estimation will be influenced by residual IQ imbalance which degrades the system performance.In our thesis,the mathematical model is deduced and a new training symbol structure is given.Part of residual IQ imbalance influence will be eliminated,the estimation error of the channel will be reduced,and the system performance will be improving,when adopts the proposed training symbols combining with GSOP algorithm.Otherwise,a frequency domain estimation method is given,which can achieve the estimation of channel and IQ imbalance independently,the accumulation of estimation error eliminated,and the system performance improved.Simulation results show that under the 100Gb/s system after 480 km standard single mode fiber transmission,when the phase and amplitude mismatch are 15 o and 3dB respectively,and the bit error rate is 1e-3,5° and 15° phase tolerance are improved respectively and 4d B and 6dB optical signal to noise ratio are deduced respectively for the two proposed methods.