Large-scale network dynamic adaptive path planning method based on deep learning

Abstract

The invention discloses a large-scale network dynamic adaptive path planning method based on deep learning. The method comprises the following steps: firstly, establishing a situation information management system; collecting and preprocessing node situation information, updating by using a specific multicast mechanism and obtaining global situation information, and then constructing a situation information model suitable for a large-scale network; finally, extracting environment features based on deep learning, and introducing a decision switch mechanism to carry out sparse processing on the problem scale; designing a network layer subsystem with an extensible routing mechanism based on the decision switch mechanism, and proposing a dynamic adaptive path planning method; the invention provides an emerging situation information modeling method and a situation updating maintenance mechanism suitable for the large-scale network, and introduces a large-scale network dynamic adaptive path planning method based on deep learning, so that the problem complexity of the large-scale network and the overall system overhead are reduced, and the problem that the adaptive capacity of the high-dynamic environment is insufficient is solved.

Description

technical field [0001] The present invention relates to the technical field of large-scale wireless network communication, and mainly relates to a large-scale network dynamic adaptive path planning method based on deep learning. Background technique [0002] In the existing technology, people use technology such as machine vision to enable a single communication node to have dynamic adaptability, but its defect is that the total amount of information obtained by machine vision-based technology per unit time is limited, so under this condition The nodes have insufficient adaptability to the situation of high environmental change rate. In a highly dynamic environment, on the one hand, if a priori decision-making mechanism is used, that is, the source node does not make decisions in the subsequent sequence, regardless of the routing scheme designed in advance, it will fail due to the highly dynamic environment; on the other hand If a real-time decision-making mechanism is used...

AI Technical Summary

Problems solved by technology

However, in the actual environment, large-scale networks are very prone to "multi-hop" phenomenon, so information will flow in the network, bringing a certain lag, resulting in a certain delay in the situation information in the network; and each node's Feel the field of view not always coincide with the intended environment area

In addition, the visual field of view of the entire network system is positively correlated with the scale. The larger the scale, the greater the overall load of the system, which will bring high computational overhead to the path planning algorithm.

Images