Mode Synchronization Of Pseudo-Random Orthogonal Transform Modulation

With the wide application of wireless communication technology,people are no longer satisfied with the efficiency and bandwidth requirements of data transmission.For the sake of personal privacy,the demand for the stability and security of data transmission is increased.In order to solve the limitation of the stability and security of data transmission on the development of wireless communication technology,a physical layer information encryption technology based on pseudo-random orthogonal transformation has emerged.Although the method based on pseudo-random orthogonal transform can resist interception better,it uses different precoding orthogonal matrices.In order to receive correctly,the receiver needs to synchronize with the transmitter accurately.In recent years,the academic community has conducted extensive research on the synchronization of pseudo-random orthogonal transform.To solve the above problems,combined with the popular deep learning algorithm in recent years,a mode synchronization scheme of pseudo-random orthogonal transform modulation is proposed.The process of pseudo-random orthogonal transform communication is deeply studied,mainly including the construction of orthogonal matrix,pseudo-random orthogonal transform modulation and mode synchronization at the receiver.Firstly,a low complexity implementation scheme of pseudo-random orthogonal transform matrix is proposed.By setting different shift parameters and block lengths of pseudo-random sequences,different pseudo-random sequences can control the generation of different orthogonal matrices.Through simulation experiments,it is found that the constellation of the orthogonal matrix modulated signal generated by pseudo-random sequence control is more scattered and difficult to identify.In addition,considering that in the actual communication,the transmitter and receiver need to frequently modulate and demodulate the signal transmission,which means that there will be a lot of matrix multiplication.In order to solve the delay problem of transmitter and receiver caused by too many matrix multiplication operations,based on the research on the fast algorithm of discrete Fourier transform(DFT),the second-order givens orthogonal matrix is used as the base matrix,and the butterfly operation is used to generate the n-order orthogonal transform matrix.Through theoretical analysis and simulation analysis,compared with the traditional matrix multiplication transformation,the time complexity of butterfly operation is lower,which is more suitable for the needs of efficient modulation and demodulation at the receiver and transmitter.Secondly,a mode synchronization scheme of pseudo-random orthogonal transform modulation based on convolutional neural network is proposed.As a leader in the field of deep learning,convolutional neural network has always been a frequent visitor in the field of classification and recognition,but the traditional convolutional neural network is not ideal for the recognition of communication symbols.Based on the research of convolution neural network,this paper uses one-dimensional convolution kernel instead of traditional two-dimensional convolution kernel,which makes it easier for convolution neural network to find the characteristics of communication signals and classify and recognize them.In addition,in order for the receiver to find the features hidden in the signal faster,this paper adopts the scheme of inserting fixed pilot symbols,and uses the pilot symbols corresponding to the pseudo-random sequence as the pilot symbols of the current pseudo-random orthogonal transform modulation block,so that after the orthogonal transform modulation,the receiver can easily identify the features and realize the mode synchronization with the receiver faster.Finally,combined with the above scheme,the mode synchronization of pseudo-random orthogonal transform modulation is simulated.The transmission data is randomly generated.After 16QAM modulation and inserting a specific pilot,the orthogonal transform modulation controlled by four groups of pseudo-random sequences is adopted,and then through Rayleigh multipath channel.Finally,the pre trained one-dimensional convolutional neural network is used for identification.The experimental results show that the recognition effect can reach 99%by using onedimensional convolutional neural network for classification when snr is 13 dB and recognition based on pseudo-random orthogonal transform modulation inserted with specific pilot symbols.The feasibility and effectiveness of pseudo-random orthogonal transform modulation mode synchronization using deep learning method are verified.