Routing optimization method and system based on graph neural network and deep reinforcement learning

Abstract

The invention discloses a routing optimization method and system based on a graph neural network and deep reinforcement learning, and belongs to the field of network routing optimization. The method comprises the following steps: measuring a current network state s, and selecting k shortest paths from a source node to a target node as an action set a according to a traffic demand distributed by a current network state request; inputting the action set a into the graph neural network, aggregating and iteratively updating link features , and obtaining a network state s and an estimated Q value of the action set a through a Q function; and performing deep reinforcement learning according to the estimated Q value to obtain a routing strategy in the current network state, and feeding back the routing strategy to the network topology to execute a corresponding routing action. The invention provides a network routing optimization system structure based on the graph neural network and deep reinforcement learning, and aims to utilize the graph neural network to learn relationships among graph elements in topology and rules forming the graph elements and utilize a deep reinforcement learning algorithm to make decisions so as to optimize network routing.

Description

technical field [0001] The invention belongs to the field of network routing optimization, and more specifically relates to a routing optimization method and system based on graph neural network and deep reinforcement learning. Background technique [0002] In the network field, finding the optimal routing configuration from a given traffic matrix is ​​a basic problem, and it is also a non-deterministic polynomial problem (NP problem for short). A variety of existing solutions based on Deep Reinforcement Learning (DRL) usually preprocess the data from the network state and present it in the form of a fixed-size matrix, which is then processed by a traditional neural network (such as a fully connected neural network, convolutional neural network) to solve the routing optimization problem. Research deep reinforcement learning as a key technology for network routing optimization, with the goal of building a self-driving software-defined network (Software Defined Network, SDN)....

AI Technical Summary

Problems solved by technology

[0004] In view of the defects of the prior art, the purpose of the present invention is to provide a routing optimization method and system based on graph neural network and deep reinforcement learning, aiming to solve the traffic distribution problem of dynamic and unknown network topology that is difficult to solve by existing deep reinforcement learning

Images