Deep learning-based mobility prediction method of nodes in vehicle-mounted Ad Hoc network

A vehicle-mounted self-organization and deep learning technology, applied in the field of communication, to achieve the effect of accurate prediction

Active Publication Date: 2020-01-03
军事科学院系统工程研究院网络信息研究所
View PDF6 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to the characteristics of rapid mobility, frequent topology changes, brief connectivity, and large net...

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Deep learning-based mobility prediction method of nodes in vehicle-mounted Ad Hoc network
  • Deep learning-based mobility prediction method of nodes in vehicle-mounted Ad Hoc network
  • Deep learning-based mobility prediction method of nodes in vehicle-mounted Ad Hoc network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0025] The flow chart of realizing the node mobility prediction in the vehicle ad hoc network based on deep learning proposed by the present invention is shown in the attached figure 1 Shown, the present invention comprises the following steps:

[0026] Create mobility models for different types of vehicles; combine different types of vehicle movement models to establish a sample database of historical vehicle journey data and a database of traffic law constraints, and agree on the characteristics of the sample journey data; then use the cyclic neural network to extract the deep-level characteristics of the mobility of the vehicle samples, Establish a mobility prediction model; then use the gradient descent backpropagation algorithm to train the model parameters; finally use the real-time data information of the vehicle's current movement to predict the mobility.

Embodiment 2

[0028] A recurrent neural network is used to extract the deep-level characteristics of the mobility of vehicle samples, and a mobility prediction model is established. The problem of gradient disappearance or gradient explosion is prone to occur in the process of solving the simple RNN structure, so that the long-term sequence dependence problem cannot be solved. The present invention mainly adopts a special RNN structure long short-term memory network (Long Short Term Memory, LSTM) that can deal with long-term dependence problems. The structure diagram of the repeating module in its LSTM chain structure is as follows figure 2 shown. Each module corresponds to input x at different times and output h at different times. When training the network, x is the extracted road network information, traffic law constraint information, vehicle driving habit information, vehicle owner information, historical urban road condition information, average traffic speed, vehicle density infor...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides a deep learning-based mobility prediction method of network nodes in a vehicle-mounted Ad Hoc network. The method effectively utilizes traffic regulations to realize mobility prediction of multiple future time points of the vehicle nodes on mobility constraints of the nodes, history travel data of the vehicle nodes and personalized information of vehicles and drivers. The method includes: combining different types of vehicle motion models to establish a vehicle history travel data sample library and a traffic regulation constraint database, and simultaneously agreeing onsample travel data features; then utilizing a recurrent neural network to extract vehicle sample mobility deep-layer features, and establishing a mobility prediction model; then using a gradient-descent back-propagation algorithm for training of model parameters; and finally, utilizing real-time data information of current movement of vehicles to predict mobility. The invention relates to vehiclemovement model data analysis and neural-network model construction and parameter training realization methods. The prediction method utilizes non-linear prediction capability of deep learning, maps the vehicle running data features to vehicle movement, and realizes mobility prediction of the nodes in the vehicle-mounted Ad Hoc network.

Description

technical field [0001] The invention relates to a network node mobility prediction method based on deep learning in a vehicle-mounted ad hoc network, which belongs to the field of communication technology, and in particular to a vehicle node mobility prediction method in a vehicle-mounted ad hoc network based on a deep learning method. Background technique [0002] In recent years, vehicle ad hoc networks have been paid close attention by research institutions and researchers all over the world. Vehicle Ad Hoc Network (VANET) is a new type of mobile ad hoc network, which consists of communication between vehicles on the road (vehicle-to-vehicle, V2V) and communication between vehicles and infrastructure (vehicle- to-infrastructure (V2I), a self-organizing, easy-to-deploy, low-cost, and open-structure wireless communication network is built on the road. The mobility prediction of vehicle ad hoc network is of great significance to the realization of safe and intelligent trave...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G08G1/01G08G1/015H04W4/40H04W84/18G06Q10/04G06F16/29G06N3/04G06N3/08
CPCG08G1/0125G08G1/0129G08G1/015H04W4/40H04W84/18G06Q10/04G06F16/29G06N3/084G06N3/045G06N3/044
Inventor 贾亦真吴胜董飞鸿胡向晖
Owner 军事科学院系统工程研究院网络信息研究所
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap