Research On PPM Signal Demodulation Technology Of Space Optical Communication Based On Machine Learning

Space laser communication is a communication mode that uses laser as carrier to realize wireless transmission of information.It has gradually become a research hotspot of communication because of its high transmission rate,strong anti-interference ability and good confidentiality.The modulation and demodulation of laser signal is an important link in the laser communication link.Due to the influence of atmospheric turbulence,there will be light intensity fluctuation and beam drift in the transmission process of laser in the atmospheric channel,which makes the traditional demodulation algorithm more difficult to demodulate and can’t meet the requirements of low bit error rate in complex atmospheric environment.In recent years,the rapid development of machine learning algorithm can extract the characteristics of signal from a deeper level,and has good robustness and fault tolerance.Therefore,according to the requirements of pulse position modulation and demodulation（PPM）system for laser communication under atmospheric turbulence,this paper designs a demodulation algorithm based on machine learning to reduce the symbol error rate and improve the performance of the communication system.The main contents of this paper include:（1）The basic theory of atmospheric turbulence channel is analyzed and discussed.Firstly,the atmospheric attenuation effect and atmospheric turbulence effect are described,and then the turbulence channel model is analyzed,including lognormal distribution turbulence channel and Gamma-Gamma distribution turbulence channel.Finally,the two turbulence channels are simulated and modeled by MATLAB,which lays a foundation for the establishment of PPM modulation and demodulation simulation system of space laser communication.（2）The basic principle of PPM modulation is analyzed theoretically,and the demodulation algorithm of PPM modulation is studied under the condition of simulating atmospheric turbulence.Firstly,the traditional maximum likelihood estimation algorithm and fixed threshold algorithm are analyzed.Then,four machine learning algorithm models are analyzed theoretically,including deep belief network（DBN）,support vector machine（SVM）,k-nearest neighbor（k NN）and adaptive boosting（Ada Boost）.The demodulation algorithm of DBN-SVM and k NN adaptive enhanced demodulation algorithm are designed.Finally,according to the characteristics of PPM signal,the parameters of DBN algorithm and the number of k NN adaptive enhancement classifiers are improved and verified by simulation.（3）The simulation system of space laser communication under atmospheric turbulence is established and the experiment is built.In the actual communication system,the two machine learning demodulation algorithm models proposed in this paper are verified and analyzed,and compared with the maximum likelihood estimation demodulation algorithm and the fixed threshold demodulation algorithm.Experimental results show that k NN adaptive enhanced demodulation algorithm and DBN-SVM demodulation algorithm can significantly improve the demodulation performance of communication system.Among them,the performance of k NN adaptive enhanced demodulation algorithm is the best,followed by DBN-SVM algorithm.For 4-PPM signal,when the signal-to-noise ratio is 14d B,the symbol error rate of k NN adaptive enhancement demodulation is lower than 10^-7,and the symbol error rate of DBN-SVM demodulation is lower than 10^-6.In addition,when comparing k NN adaptive enhanced demodulation with DBM-SVM demodulation with traditional demodulation algorithms,it is found that when the symbol error rate is 10^-6,the demodulation performance of the former is about 1d B better,and the latter is almost the same.