Virtual optical network mapping method based on multi-agent deep reinforcement learning

Abstract

The invention provides a virtual optical network mapping method based on multi-agent deep reinforcement learning. The method comprises the following steps: determining whether other virtual network services leave in an environment module or not; extracting the node and link information of a current physical network of the virtual network request service; by a node intelligent agent module, obtaining probability distribution that each node of a physical network providing mapping for a current virtual network request service by utilizing a node strategy network, and selecting a proper node as anode to which the current virtual network request service is to be mapped; by a link agent module, calculating frequency slot probability selection distribution of the shortest path between the source node and the destination node through a link strategy network; by a judgment module, obtaining the accumulated rewards of the instant reward calculation multi-step mapping from the environment module; and by the evaluation module, calculating evaluation values of the node action and the link action, and updating parameters of the intelligent agent module according to the evaluation values. According to the invention, the resource utilization rate of nodes and links can be effectively improved, the blocking rate of the network is reduced, and the mapping success rate of the virtual network isimproved.

Description

technical field [0001] The present invention relates to the technical field of communication and reinforcement learning, in particular to a virtual optical network mapping method based on multi-agent deep reinforcement learning. When the elastic optical network is used as the underlying physical network communication facility, multi-agent deep reinforcement learning training Learning historical data enables virtual network services to be independently and efficiently mapped to the underlying physical network. Background technique [0002] The Internet of Things is booming in various fields, and the era of Internet of Everything is approaching. However, with the development of 5G, virtual reality, and autonomous driving, and the rapid increase of IoT devices, in the face of the explosive growth of network data centers, the traditional Internet has deficiencies in data security, scalability, network management and control capabilities, and service quality assurance. It is bri...

AI Technical Summary

Problems solved by technology

[0012] Aiming at the technical problems of low network utilization and high blocking rate in the traditional elastic optical network virtual network mapping method, the present invention proposes a virtual optical network mapping method based on multi-agent deep reinforcement learning, adopting a multi-agent reinforcement learning framework, through The node agent and the link agent interact and learn in the underlying network environment, obtain the reward value for each mapping, and evaluate the node actions performed by the node agent and the link actions performed by the link agent each time, so as to optimize The next action selection, complete the collaborative cooperation, and finally make the virtual request achieve the best mapping

Images