Method and system for optimizing Volterra equalizer structure based on deep reinforcement learning

Abstract

The invention provides a method and a system for optimizing a Volterra equalizer structure based on deep reinforcement learning. The method comprises the following steps: initializing memory length states of an agent, an experience playback pool and a Volterra equalizer; the method comprises the following steps of: randomly generating actions on an Agent, updating a memory length state of the Volterra equalizer until an end state, calculating a reward value according to the complexity of the Volterra equalizer and an error rate after signal equalization, taking a transfer process as experience, and storing the transfer process into an experience playback pool; sampling experience from the experience playback pool, and performing training and soft updating on the Agent; and determining the memory length of each order of the Volterra equalizer according to the convergence value. The method for automatically searching the optimal structures of different types of Volterra equalizers under the condition that computing resources are given is achieved, and compared with traditional greedy search, the method has the advantages that the equalization effect can be further improved, and the complexity of the equalizers is greatly reduced.

Description

technical field [0001] The present invention relates to the technical field of optical communication, in particular to a method and system for optimizing a Volterra equalizer structure based on deep reinforcement learning. Background technique [0002] Volterra nonlinear equalizers are widely used in optical fiber communication systems to alleviate linear and nonlinear damage to signals during transmission. In the long-distance optical fiber communication system, the nonlinear damage mainly comes from the fiber nonlinearity, while in the short-distance optical fiber communication system, the nonlinear damage mainly comes from the transceiver devices, such as the nonlinear response of the modulator and the square-law detection of the photodetector. Wait. Equalization effect and implementation complexity are important indicators for evaluating equalizers. In order to realize real-time hardware, the high-performance and low-complexity Volterra nonlinear equalizer structure has...

AI Technical Summary

Problems solved by technology

[0004] However, the existing methods still need to determine the memory length of each order, pruning threshold or correlation distance of the Volterra nonlinear equalizer through manual experience or greedy search, which is inefficient and cannot give full play to the best performance of the Volterra nonlinear equalizer. Performance, it is also difficult to achieve a compromise between balance effect and complexity

Images