A Generation Method of Routing Metrics Based on Multimodal Data

Abstract

The invention belongs to the technical field of wireless networks, especially the technical field of network layer routing search in a vehicular ad-hoc network, and particularly provides a multimodal data-based routing metric generation method. By analyzing multiple types of modal data representing environment, driver, vehicle and traffic information, traffic-data based routing metric (TDR) is established, so that the judgment precision of reliability of an intermediate node is improved; a small amount of labeled data is learned and cooperatively trained by using a semi-supervised multimodal machine learning architecture; the cooperative training process is divided into two iterative processes including positive training and label upgrading; and the routing metric is constructed. A mechanism for generating the routing metric based on a machine learning algorithm is proposed; massive routing constraint points are considered; the influence of related data on routing is analyzed from the data mining perspective; and the routing metric for accurately judging the routing reliability is generated. Massive unlabeled data and a small amount of the labeled data are analyzed by utilizing the semi-supervised multimodal architecture, and a training modeling process is finished with the minimum cost.

Description

technical field [0001] The invention relates to a semi-supervised multimodal machine learning framework, a method for constructing routing metrics through learning and collaborative training on a small amount of labeled data, specifically a method for generating routing metrics based on multimodal data. This analysis includes a variety of modal data representing the environment, drivers, vehicles, and traffic information, and establishes a routing metric TDR (Traffic-Databased Routing Metric) to improve the reliability of intermediate nodes (next hop nodes). Background technique [0002] Since entering the 21st century, the total number of cars in my country has increased from more than 20 million to nearly 190 million in 2016, and will continue to rise at a growth rate of 10%. Along with this, problems such as road congestion and traffic accidents will follow. The Internet of Vehicles can provide services such as vehicle lane change reminders, intersection reminders, accide...

AI Technical Summary

Problems solved by technology

ETX and its derived routing metrics only monitor the reliability of the link. Although the throughput is increased, it cannot provide better transmission reliability because the routing metrics do not consider other routing factors.

Images