Study Of Modeling And Detecting Random Optical Signal In Terrestrial Laser Communications

Terrestrial Laser Communication (TLC) transmits information using laser ascarrier through atmosphere channel. An optical field that traverses the TLC channelis scattered and/or absorbed resulting in power loss. Furthermore, the atmosphericturbulence causes random fluctuations in the index of refraction of the atmosphere,resulting in random fluctuations of the intensity and phase of the optical signal. Thus,the bit-error-rate (BER) increases and the performance of the TLC system isdegraded. Therefore, the effect of atmosphere, especially the effect of atmosphericturbulence, is the primary factor that blocks the popularization of TLC.This thesis investigates the modeling and simulation of laser beam propagatingthrough atmospheric turbulence, and algorithms for detecting random opticalintensity signal. The contributions of this work are summarized as follows:(1) Modeling the whole TLC link, which is composed of a transmitter, atmospherechannel and a receiver; The physical theory and characteristics of atmosphericturbulence are investigated, and three widely used distribution functionscorresponding to different turbulence strength are analyzed.(2) Simulating the atmosphere random phase screen, which obeys Kolmogorovtheory, by Fourier transformation. Simulating the intensity and phase of acollimated Gaussian-beam propagating Kolmogorov turbulence. To simulate theTLC channel, a Gaussian filter approximated by cascading Butterworth filters in essence, is proposed to model the temporal spreading behavior of Gaussian laserpulses propagating through atmospheric turbulence. The spreading of theGaussian pulses is combined with the parameters of atmospheric turbulence bythis model. To simulate the arrival time of a laser pulse propagating through theatmospheric turbulence, a filter bank is constructed by cascading digital unitdelay filters and a fractional delay filter. This filter bank incorporates theturbulence structure parameter, mean refractive index, turbulence outer scale andpath-length. Then a theoretical arrival-time model with femtosecond precision isproposed based on the filter bank.(3) The noise characteristics of p-i-n (PIN) diode with an electronic transimpedanceamplifier, PIN diode with an optical preamp and avalanche photodiode (APD)are analyzed theoretically, then, according to the theoretical analysis, a TLCsystem and an experimental precept are designed. To measure the scintillationand construct the model of generating test sequence for algorithms of detectingrandom intensity signal, experiments of measuring atmosphere channel over1.7km and7.7km are carried out. Based on the data obtained in the experiments,the turbulence channel is analyzed statistically. A model of generating testsequence for algorithms of detecting random intensity signal is constructedaccording to the measured data and simulated data, then, the scheme of thismodel is given.(4) Aiming at two existing problems, which are the high complexity of maximumlikelihood sequence detection algorithms and errors in the channel estimationaided maximum likelihood detection algorithms, corresponding solutions areinvestigated. For TLC systems operating over lognormal fading and highsignal-to-noise channel, a closed-form suboptimal maximum likelihoodsequence detection (MLSD) algorithm is proposed, and its correspondinglow-complexity algorithm is deduced. The BER performance is analyzedthrough the constructed model of generating test sequence for algorithms ofdetecting random intensity signal and Monte Carlo method, then the computation performance is tested on a personal computer (PC) and aTMS320DM642DSP platform. To mitigate the impact of the error between theestimated channel fading coefficient and the perfect fading coefficient on theBER, a priori conditional probability density function averaging the estimationerror is defined. Then, an improved maximum-likelihood (ML)symbol-by-symbol detection is derived for the TLC systems, which implementpilot symbol assisted modulation. To reduce complexity, a closed-formsuboptimal improved ML detection is deduced using distribution approximation.The BER performance is analyzed and compared through the constructed modelof generating test sequence for algorithms of detecting random intensity signaland Monte Carlo method.The simulation and modeling of atmospheric turbulence channel and thedetection of random intensity signal are key technologies for TLC. This thesisinvestigates the primary intensity detection algorithms in the current TLC filed. Itprovides the important theory for designing TLC systems, and the results aremeaningful and can be referred.