Free Space Optical Reliable Routing Algorithm based on Deep Reinforcement Learning

Free space optical network is vulnerable to various factors such as atmospheric conditions and death node. It leads to the interruption of transmission service and reduces the overall performance of network. Faced with complex free space optical network environment, a reliable routing algorithm based on deep Q network is proposed. The algorithm can adjust routes according to the atmospheric environment and node state changes. Results simulated show that the proposed algorithm can provides a high reliability to FSO network.