Research On Channel Estimation Technique For Asymmetrically Clipped Optical Orthogonal Frequency Division Multiplexing System Of Space Optical Communication

Channel estimation technique is used for estimating the channel state information(CSI), which is close to the signal recovery, synchronization, equalization at receiver andadaptive optical power technique at transmitter, etc. Channel estimation has significanteffect on improving the performance of the system. Due to the particular Hermitianstructure of asymmetrically clipped optical orthogonal frequency division multiplexing(ACO-OFDM) symbols and optical power is precious for ACO-OFDM systems of spaceoptical communication, cause that the traditional channel estimation technique oforthogonal frequency division multiplexing (OFDM) systems is not applicable. As thespace optical communication technology is attracting more and more attention, theresearch on the channel estimation technology of space optical communication systemhave important significance on its reliable communication and application.Firstly, situation on channel estimation technology of ACO-OFDM communicationsystem is analyzed, its advantages and disadvantages and existing problems are presentedin this thesis. Due to the noise-sensitive problem of the existing channel estimationmethod for space optical communication system, a new channel estimation methodwhich is based on the period superimposed training sequences and the linear minimummean square error (LMMSE) algorithm estimation is proposed. In this method, trainingsequences is cyclical added at variable power ratio to the information sequence at thetransmitter prior to transmission, statistical average method is employed to separatetraining and information sequences at the receiver. Then, LMMSE algorithm is employedto accomplish channel estimation. Compared with the traditional channel estimationmethods, period superimposed training sequence has significantly improved transmissionrate of system. The proposed algorithm in this thesis also overcomes the problem ofnoise-sensitive and significantly improves the estimation performance under thecondition of lower signal-to-noise ratio (SNR). This method can meet the requirements ofACO-OFDM systems which have lower SNR at the receiver.Secondly, due to the scattering channel property and high power consumption oftraditional channel estimation methods for non-line of sight optical communicationsystem, a channel estimation method which is based on the compressed sensing (CS)theory and partial superimposed training sequences is proposed for ACO-OFDM systemsin this thesis.(1) A training sequence is added at variable power ratio to the OFDMsymbols partially at the transmitter.(2) The local “observation matrix” is structured by the training sequence symbols at the receiver.(3) Channel state information is estimatedusing sparse signal reconstruction algorithm of CS theory. Compared with periodsuperimposed training sequence, partial superimposed training sequence decreases theinterference induced by training sequence symbols to data symbols and improves thereliability of communication systems. In addition, compared with traditional least squares(LS) and LMMSE estimate algorithm, the proposed CS signal reconstruction algorithmbased on channel estimation can both improve the estimation accuracy and the speed,which is meaningful in the development and adaptation of ACO-OFDM systems.At last, the author summarizes the whole thesis and proposed the future directions ofthe research.