Research On IQ Mismatch Compensation Algorithm In Coherent Optical OFDM Systems

Coherent optical orthogonal frequency division multiplexing (CO-OFDM) has recentlyattracted much interest in the optical fiber communication systems because of its robustnessagainst intersymbol interference (ISI) caused by chromatic dispersion (CD) and polarizationmode dispersion (PMD), as well as its high spectral efficiency. Unfortunately, CO-OFDM issensitive to in-phase and quadrature (IQ) mismatch in the transmitter (Tx) and receiver (Rx)front-ends. IQ mismatch will cause inter-carrier interference (ICI) and thus degrade theperformance of CO-OFDM systems. In this paper，the IQ mismatch compensation methodsare studied and simulated systematically. The details are described as follows:(1) A novel training symbol structure is proposed to make the two estimation processesabout Tx IQ mismatch factor and channel distortion independent. The conventional estimationmethods are the sequential process by using pilot signals or training symbols for Tx IQmismatch. The imperfectly first process estimation value results in noise enhancement in thesecond process estimation stage and consequently degrades the second process estimationperformance. The proposed structure enabling Tx IQ mismatch and channel distortionestimation separately, which overcomes the noise enhancement in sequential estimationprocess. The advantages of the proposed structure are displayed via channel mean squareerror (MSE) simulation and IQ mismatch factor MSE simulation, and the improvement of biterror rate (BER) is shown in simulations simultaneously.(2) A method is presented to estimate channel and Rx IQ mismatch separately for Rx IQmismatch. Conventional compensation methods, such as using Gram-Schmidtorthogonalization procedure (GSOP) or two same training symbols degrade channelestimation performance due to the correlation between channel estimation procedure andcompensation procedure. The method uses the characteristic of signal model for Rx IQmismatch and the relationship between two Rx IQ mismatch factors and the dedicated pilotstructure for estimating IQ mismatch factors and channel. Based on the compensation methodfor Tx IQ mismatch, the compensation model for compensating Rx IQ mismatch and channeldistortion together is derived. Then the estimation results are proved more efficiently by usingthe derived compensation method in simulations.(3) A joint scheme to compensate for Tx IQ mismatch, Rx IQ mismatch and channeldistortion simultaneously is proposed. Conventional method compensation for Tx IQmismatch, channel distortion and Rx IQ mismatch requires three continuous processes. Theestimation and compensation noise accumulation continues until the third procedure iscompleted. The influence parameters are defined including IQ mismatch information andchannel distortion which makes them estimated together via analyzing the signal model. Atthe same time, the compensation model is derived to fit for Tx and Rx IQ mismatch. When this model is simplified, it also can compensate Tx or Rx IQ mismatch only. Numerical resultsindicate that the proposed method can improve BER performance in the presence of TX andRx IQ mismatch.