Multi-path routing method based on reinforcement learning and transfer learning

Abstract

The invention provides a multi-path routing method based on reinforcement learning and transfer learning, which is used for solving the technical problem of poor equivalent path load balance in a network environment with less flow data in the prior art, and comprises the following implementation steps of: constructing a real network Z and an experimental network G consistent with a Z topological structure; establishing a two-dimensional array H; constructing a multi-path routing model based on reinforcement learning; initializing a flow matrix DM and an equivalent path flow proportion matrix PM; performing iterative training on the multi-path routing model based on reinforcement learning in the experimental network G; migrating global neural network weight parameters in a routing decisionmodel obtained by training in an experimental network G to a real network Z based on a transfer learning method; and carrying out adaptive training on the global neural network initialized in the realnetwork Z to obtain a multi-path routing result conforming to the real network environment characteristics. The method can be used for data center networks and other scenes.

Description

technical field [0001] The invention belongs to the technical field of computer networks and relates to a multi-path routing method based on reinforcement learning and migration learning, which can be used in fields such as data center networks. Background technique [0002] For the network, the routing decision stipulates how the data flow reaches another node from a specified node in the network. The routing decision can schedule the traffic, so it determines the load balance of different transmission paths in the network, and measures the network load balance status. The value of is the difference in bandwidth utilization of all equivalent paths between two communication pairs in the network. The smaller the difference, the better the load balance of the equivalent path. Routing decision-making can be divided into traditional routing decision-making methods and routing decision-making methods based on reinforcement learning. The traditional routing decision-making algorit...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the shortcomings of the above-mentioned prior art, and propose a multi-path routing method based on reinforcement learning and transfer learning, which is used to solve the equivalent path load in the network environment with less traffic data existing in the prior art Poorly balanced technical issues

Images