Recently, the Laboratory of Intelligent Fiber Ecosystem (LIFE, Laboratory of Intelligent Fiber Ecosystem), Department of Electronic Engineering, Shanghai Jiao Tong University, proposed a chaotic secure communication scheme based on coherent detection and neural network synchronization, using coherent detection combined with DSP algorithm to compensate for chaotic encrypted signal channel transmission damage , the chaos synchronization is realized by the neural network, and the experimental demonstration of the transmission of the 30 Gb/s phase chaotic encrypted signal in the 340 km optical fiber is realized. This scheme not only greatly simplifies the difficulty of chaotic synchronization, but also effectively prolongs the transmission distance of chaotic optical communication. The relevant results are titled "Coherent chaotic optical communication of 30 Gb/s over 340-km fiber transmission via deep learning", and will be published in the international high-level optical journal "Optics Letters" in June 2022. Doctoral student Yang Zhao is the first author, and Professor Lilin Yi is the corresponding author.

Research Background

my country's backbone network has entered the all-optical network 2.0 era, and the speed and distance of the existing optical fiber communication system have been greatly improved. However, the all-optical network is also a transparent optical network, and the optical fiber transmission process is basically in an undefended state. Therefore, ensuring the physical layer security of the optical fiber communication system is one of the key issues to be solved urgently. As a physical layer encryption scheme, chaotic optical communication has the advantages of noise-like, wide spectrum, continuous non-periodicity, and high compatibility with existing optical communication systems. It has received continuous research attention in recent years.

However, chaotic optical communication still has the following challenges: (1) The hardware-synchronized chaotic optical communication system requires high parameter matching performance of physical devices. If one of the devices is damaged, all matching physical devices need to be replaced to ensure the consistency of device parameters. (2) Point-to-multipoint chaotic optical communication system is difficult to realize. Multiple sets of paired physical devices are required, which makes chaotic optical communication very expensive and difficult to maintain. (3) The transmission impairment makes it difficult to synchronize the chaotic optical communication system. The traditional chaotic synchronization process is completed in the optical domain, and the compensation for the transmission signal damage needs to be completed in the optical domain before synchronization. However, high-order dispersion compensation and nonlinear compensation in the optical domain are extremely difficult, which limits the high-speed and long-distance transmission of encrypted signals.

Research Path

Figure 1   Chaotic optical communication system based on coherent detection and neural network modeling

To solve the above problems, we first convert the chaotic analog signal to the digital domain through coherent detection, and then use digital signal processing (DSP) technology to compensate channel transmission damage, so as to avoid the traditional channel damage compensation scheme in the optical domain. The limitation of the transmission distance of chaotic optical communication. The use of neural network chaos modeling in the digital domain can realize chaos synchronization decryption, get rid of the dependence of traditional optical domain chaos synchronization on hardware parameter matching at the transceiver end, and reduce the difficulty of chaos synchronization, which is beneficial to point-to-multipoint chaotic optical communication applications. The combination of DSP technology based on coherent detection and neural network synchronization in the digital domain can not only achieve a breakthrough in the transmission index of chaotic optical communication, but also be fully compatible with existing optical communication systems, greatly enhancing the practicability of chaotic optical communication.

In the previous work, we studied the feasibility simulation verification of the combination of coherent detection and chaotic optical communication, and the experimental demonstration of 20 km optical fiber transmission in which the neural network models the chaotic optical communication system. In this paper, we combined coherent detection and neural network for the first time to realize channel damage compensation and chaos synchronization in the digital domain, and experimentally demonstrated the transmission index of 30 Gb/s QPSK chaotic encrypted signal in 340 km optical fiber.

Figure 1 is a chaotic optical communication experimental system based on coherent detection and neural network modeling. First, the neural network modeling of the chaotic model is completed in a back-to-back environment. The modeling principle is to use encrypted signals (chaotic carrier and QPSK signals) as The input of the neural network, the expected chaotic synchronous carrier is used as the output of the neural network, and the functional relationship between the encrypted signal and the chaotic carrier is learned through the neural network. Then keep the trained neural network unchanged, and carry out the 340 km optical fiber transmission experiment. After coherent detection at the receiving end, the chaotic encrypted signal after 340 km optical fiber transmission is compensated for transmission damage through the DSP algorithm, and the phase noise is estimated and compensated through dispersion compensation, constant modulus algorithm (CMA) channel equalization and extended Kalman filter (EKF) After that, the chaotic encrypted signal can basically recover to the back-to-back level. Finally, the restored encrypted signal is input into the trained neural network for chaos synchronization and QPSK signal decoding.

Research Results

Figure 2   Chaotic synchronization and constellation diagram of 30 Gb/s QPSK signal after original, encrypted and decrypted

The 30 Gb/s single-polarization QPSK signal is encrypted by phase chaos with a bandwidth of 15 GHz, and the chaotic signal transmitted through a 340 km optical fiber is shown in Figure 2. The chaotic signal received in the experiment after transmission and the chaotic signal synchronized by the neural network are shown in (a) and (b) respectively, and the normalized correlation coefficient is 0.9116, indicating that the two have a strong correlation. From the chaotic encrypted signal Subtract the chaotic signal predicted by the neural network to recover the QPSK signal. Figure (c) shows the QPSK signal constellation without chaotic encryption after 340 km optical fiber transmission. The chaotic encrypted signal is shown in figure (d). The constellation diagram shows the phase Figure (e) shows the demodulation constellation diagram of the 30 Gb/s QPSK signal after optical fiber transmission, and the bit error rate of the demodulated QPSK signal is 8.9×10-4. At present, the transmission distance is mainly limited by the ability of the blind phase recovery algorithm. In the future, it is planned to study a more powerful and efficient blind phase recovery algorithm to further extend the transmission distance of chaotic encrypted signals.

Full text link of the paper: https://doi.org/10.1364/OL.453696

The LIFE research group is committed to the key technology research of chaotic optical communication systems, involving chaotic encrypted transmission of high-speed signals, chaotic long-distance transmission damage suppression, chaotic synchronization based on deep learning, etc., in order to build a high-speed long-distance safe and practical chaotic optical communication system , to provide reliable security for the high-speed transmission of optical fiber communication.

√  In 2018, the experimental demonstration of 30 Gb/s chaotic optical signal transmission in 100 km optical fiber based on hardware synchronization was realized, breaking the previous chaotic transmission record (10 Gb/s chaotic optical signal 100km optical fiber transmission), the paper was published In Optics Letters, full-text link: https://doi.org/10.1364/OL.43.001323.

√  In 2019, proposed to use neural network to model chaotic systems to realize chaotic synchronization, and realized the experimental demonstration of chaotic encrypted transmission and decryption of 32 Gb/s 16QAM signals. The paper was published in Optics Letters, and the full link is https:// doi.org/10.1364/OL.44.005776.

√  In 2020, combined chaotic optical communication and coherent detection for the first time, compensated for fiber link damage through DSP algorithm, and completed 1000 km fiber transmission simulation of 10 Gb/s DPSK chaotic encrypted signal, the paper was published in Journal of Lightwave Technology , full-text link: https://ieeexplore.ieee.org/document/9091335.

√  In 2021, on the basis of the chaotic synchronization realized by the neural network, the influence of ADC quantization bit number and sampling rate, chaotic complexity and signal rate on the neural network synchronization model was further studied. The paper was published in Optics Letters, the full text link: https://doi.org/10.1364/OL.414966.