A coherent optical qpsk decision method and system based on optical neural network

Abstract

The invention discloses a coherent optical QPSK judgment method based on an optical neural network. First, according to the QPSK judgment problem, an ONN model structure of an optical neural network is designed, model parameters are trained, and a judgment photon circuit is built and set according to the parameters. The judgment photon circuit is A 2×4 optical network with two input ports and four output ports; then, the QPSK received signal and the local oscillator signal are input to the two input ports of the decision photonic circuit at the same time, and are processed by the four output ports In the output of the port, the phase difference between the received signal and the local oscillator signal is judged according to the port with the highest output power, so as to complete the QPSK judgment. This kind of judgment method can adapt to the requirements of all-optical communication network and optical mobile communication system, directly process the received signal and local oscillator signal in the optical band, and realize the judgment or judgment of the optical signal directly based on the phase difference information of the two input signals. branch. The invention also discloses a coherent optical QPSK decision system based on the optical neural network.

Description

technical field [0001] The invention belongs to the field of judgment, and particularly relates to a coherent optical QPSK judgment method and judgment system based on an optical neural network. Background technique [0002] The homodyne detection scheme in the traditional coherent optical communication system uses a 90-degree optical mixer (90-degree hybrid) and a balanced receiver (balanced receiver) to extract the I-channel and Q-channel information of the received signal. The mixer divides, shifts and combines the received signal and the local oscillator signal, and then completes the photoelectric conversion and differential subtraction through the balanced receiver to extract the modulation information of the I channel and the Q channel. The above system can restore the modulation information well under the condition of ideal synchronization, and the balanced receiver is also beneficial to reduce the influence of noise while completing the photoelectric conversion. Re...

AI Technical Summary

Problems solved by technology

The ONN processing object proposed by Yichen Shen et al. is a real number vector, and the weight matrix in the neural network is a real number matrix. For the judgment of the optical signal, when dealing with the optical signal with phase information, the real number matrix cannot complete the optical signal. Universal phase shifting operation, so it needs to be adjusted for the structure of traditional ONN to meet the needs of coherent optical QPSK decision system

Images