Research on Pulse Position Modulation and Demodulation System of Space Laser Communication Based on DBN-SVM

The pulse position modulation (PPM) demodulation system for space laser communication based on the cascade of deep belief network (DBN) and support vector machine (SVM) is studied in this paper, and a simulation system of space laser communication is established, and then a demodulator based on DBN-SVM is proposed. The lognormal atmospheric channel is simulated in MATLAB to simulate atmospheric turbulence channel, and the bit error rate of pulse position modulation signal affected by atmospheric turbulence is analyzed. The system performance is evaluated by comparing the average Signal to noise ratio and bit error ratio of different algorithms. In the DBNSVM based demodulator, the received signal is normalized before it is sent to the deep belief network. The processed signal data set is used to train the deep belief network parameters layer by layer to realize the feature extraction and recognition of the pulse position modulation signal. Then, the feature matrix is classified by support vector machine to obtain the demodulation result. The simulation results show that the demodulator based on DBN-SVM can effectively alleviate the influence of channel fading. Numerical simulation shows that DBN-SVM detection algorithm is superior to the least mean square error method and approximate to the maximum likelihood detection algorithm in atmospheric turbulence, which verifies the good performance of DBN-SVM detection algorithm.