Decryption Of All-optical Chaotic Communication By Neural Network

With the continuous development of information technology,people pay more and more attention to the security of information,and there are two mainstream encryption methods: software encryption and hardware encryption.Software encryption has been widely used because its encryption method is simple and does not damage the performance of the transmitted signal.However,due to the emergence of quantum computers and the continuous improvement of computer performance,software encryption algorithms will gradually be cracked by brute force solutions in a short time.So now people are paying more and more attention to hardware-level encryption.Chaotic encryption,as a physical layer encryption method,can realize high-speed and long-distance secure transmission of signals,and is compatible with existing optical communication systems,and has gained persistent research interest at home and abroad.However,because chaotic decryption requires the harsh conditions that the hardware parameters in transmitter and receiver are completely matched,there are certain limitations in practical applications.This paper studies this problem,the goal is to simplify the receiver as much as possible,and furtherly advance chaotic optical communication to applications.This paper proposes a decryption solution for all-optical chaotic communication based on neural network synchronization.The main method is to first obtain chaotic signals and encrypted signals at the transmitter,and use them as the input and output of the neural network to train the parameters.The network can simulate the entire chaotic synchronization process to replace the hardware decryption device.The advantage of this decryption method is that on the basis of ensuring the security of chaotic encryption,the complexity of the receiving end is greatly reduced,and the parameters can be flexibly adjusted,and the encryption method can be quickly adjusted after being attacked.Based on the best neural network prediction results obtained in the simulation,this paper also conducted back-to-back and 20 km single-mode fiber transmission neural network decryption experiments,and analyzed the decryption bit error rate of signals with different transmission rates.In order to simplify the complexity of the transmitting end,this paper proposes a chaotic encryption scheme for external modulation of signals,and verifies the effect of decryption using neural networks in experiments.Finally,the neural network is successfully decrypted after 5 Gbit/s signals are transmitted over 20 km single-mode fiber.