Study On Wavefront Sensorless Algorithm In Coherent Free Space Optical Communication System

Free space optical communication(FSOC)is a wireless communication technology that loads signals onto light waves and transmits information in free space.Its spectrum does not require certification and it has a strong anti-electromagnetic interference capacity.In addition,it also has a series of other advantages.It can be applied to a variety of scenarios such as space,ground and water.It has a considerable room for development in the future.The FSOC technology in which intensity modulation is used at the transmitting end and directly detected at the receiving end is called incoherent optical communication.This kind of communication method is relatively mature in technology and has the advantages of simple structure and low cost.In addition,there is another type of FSOC technology-coherent optical communication.Compared with incoherent optical communication,coherent FSOC technology has the characteristics of higher receiving sensitivity and more flexible modulation mode.It is an effective means to realize high-speed and long-distance information transmission in the future.When the coherent FSOC system transmits information in the atmospheric environment,the communication performance will be affected due to the interference of atmospheric turbulence.Relevant studies have shown that adaptive optics(AO)technology can effectively reduce the impact of atmospheric turbulence on the transmission performance of the FSOC system.Based on the coherent FSOC technology,this paper adopts the AO technology that does not require the wavefront sensor which is called the wavefront sensorless technology to compensate for wavefront distortion in the coherent FSOC system.At first,the effect of the Stochastic Parallelism Gradient Descent(SPGD)algorithm on the mixing efficiency and bit error rate of the homodyne detection coherent FSOC system with binary phase shift keying modulation is evaluated.In order to reduce the number of iterations of SPGD algorithm for aberration correction,this paper combines Convolutional Neural Networks(CNN)and SPGD algorithm to propose the CNN-SPGD algorithm.The main contents of this paper are as follows:(1)Considering that the traditional AO system will be limited by the wavefront sensor technology to a certain extent in its application,this paper uses the wavefront sensorless algorithm-SPGD algorithm to compensate for wavefront distortion in the coherent FSOC system.Based on the approximate relationship between strehl ratio and mixing efficiency in the FSOC system with homodyne detection,the influence of SPGD algorithm on the mixing efficiency and bit error rate of the homodyne detection FSOC system with binary phase shift keying modulation is evaluated.Using MATLAB software to carry out a simulation experiment and analyze the performance changes of the system.The simulation results show that when the number of iterations of the SPGD algorithm reaches about 560,the mixing efficiency of the system converges to about0.94.Assuming that the quantum efficiency?=1 and the number of photons per bit at the receiving endN_P(28)5,the system BER can eventually drops to 10^-6 which shows that the performance of FSOC system with homodyne detection has been improved.By setting differentN_P values for simulation experiments,it can be concluded that when theN_P?5,the BER of the distortion-compensated system can drop to10^-6 and below.(2)Aiming at the shortcomings of the SPGD algorithm with a large number of iterations and slow aberration correction speed,the CNN-SPGD algorithm is proposed to improve the speed of aberration correction.First,the CNN model is used to classify the input light spots,then the optimal voltage is selected as the initial voltage through judging the pre-prepared correction voltage and the wavefront aberration correction is completed by the SPGD algorithm.The effect of this algorithm on the mixing efficiency and bit error rate of the homodyne detection FSOC system with binary phase shift keying modulation is evaluated through simulation experiments and compared with the SPGD algorithm.The simulation results show that when correcting the same wavefront aberration as in Chapter 3,the number of iterations required by the CNN-SPGD algorithm to achieve mixing efficiency of 0.80,0.85 and 0.90 are 4,37 and 141,respectively.The number of iterations required to reduce the bit error rate to10^-6 is152.Taking into account the process of spot classification and voltage selection,compared with the SPGD algorithm,the number of iterations required for the CNN-SPGD algorithm to achieve the same system performance is reduced and the aberration correction speed is improved.The wavefront sensorless algorithms used in this paper can all correct the wavefront aberration in the coherent FSOC system,thereby improving the performance of the system.The research contents of this paper has certain reference value for the design of coherent FSOC system based on wavefront sensorless technology.