| In the next generation of optical network, Coherent optical orthogonal frequencydivision multiplexing (CO-OFDM) technology is an important research direction. Itcombined OFDM technology and coherent optical communication system perfectly, whichutilized high sensitivity of the advantages belong to the coherent optical communicationtechnology. At the same time, It applied the unique orthogonal channel segmentationtechnology of OFDM system cleverly, So CO-OFDM system effectively reduce theinfluence of the received signal which by the fiber channel dispersion, polarization modedispersion and so on, and then, achieve the dynamic allocation of resources and bandwidth.In CO-OFDM systems, channel estimation is one of the most important signalprocessing methods, it is the precondition of recover data accurately. Because in most cases,the condition of fiber channel is better than wireless channel, the channel estimation methodbased on CO-OFDM systems mainly relied on LS and LMMSE estimation, and did not takethe fiber channel of special circumstances into consideration, such as the fiber in desert orsea, channel transfer function changed with the time quickly. To view the limitations of thethese two algorithms which existing today. In this paper, two new channel estimationalgorithm is proposed, which based on the study of the CO-OFDM basis knowledgerespectively. And used Matlab and Optisystem to structure simulation platform to validatethe new algorithms of CO-OFDM channel estimation.Aiming at based channel estimation algorithm in CO-OFDM system, LS algorithmignores the noise of system, the system’s BER is larger when the SNR is nothigh,while,LMMSE algorithm is too complex,at the condition of optical fiber channel, theefficiency of algorithm is not satisfactory.We present a new channel estimation method basedon channel Impulse Response processed with Window algorithm (IRW). This method relieson LS estimation, and then through the time domain window processed the channel impulseresponse to filter the noise which outside the window, and to ensure the length of channelimpulse response in OFDM guard interval. On the side of the algorithm complexity,IRWmethod is not more enormous than LS algorithm,On the other side of system’s BERperformance,this method lower than LS algorithm nearly an order of magnitude. Thesimulation results show that:in the256subcarrier number CO-OFDM systems, and with thesame terms of system’s BER, The requirements of SNR in IRW algorithm lower than LSalgorithm nearly2.5dB, while only0.7dB higher than LMMSE algorithm. At the same time,the number of complex multiplying number of IRW algorithm is increased only8timeshigher than LS, but reduced nearly32times lower than LMMSE.In view of special environment,such as the desert or sea in that case,channel transferfunction change quickly with time,shows that—fast-varying fiber channel.We present a newchannel estimation method based on Analysis of Symbols Mapping(ASM) that applies tothis fast-varying fiber channel. And this method thought the analysis the constellation diagram of received OFDM symbol, and then amend channel’s transfer function of channelconstantly.That is to say that this method can fast tracking the change of fast-varying fiberchannel,and at he same time,this way didn’t need a large number of pilot for channelestimation,Thought simulation,we can show that:in40Gb/s fast-varying fiber channelCO-OFDM system,On the one hand, while the Signal-to-Noise Ratio(SNR) is10dB andsystem’s Bit Error Ratio(BER) is103,the pilot numbers of ASM algorithm reduced halfthan conventional Zero Forcing Estimation(ZFE) algorithm,On the other hand,while the pilotinterval is10(every ten OFDM symbols inset one pilot) and system’s BER is103,Compared with conventional ZFE algorithm,ASM algorithm has3.5dB SNRimprovement. |
PDF Full Text Request