Font Size: a A A

Research On Driving Behavior Based Routing Algorithms In Vanets

Posted on:2019-03-08Degree:MasterType:Thesis
Country:ChinaCandidate:Y J LiFull Text:PDF
GTID:2322330542956393Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
In recent years,the rapid increase of car ownership has led to deterioration of traffic in urban areas and rural areas,causing traffic jams and traffic accidents.The ITS can largely solve such problems.As a key to support data transmissions in such system,routing algorithms are especially important.Different to the routing table mechanism used by traditional MANETs,in VANETs the nodes have the characteristics of high-speed mobility,which leads to frequent changes of topology.It is easy to cause the routing table to fail,and the routing efficiency is low.Therefore,existing routing algorithm of MANETs cannot meet requirements of VANET.Position-based routing algorithms eliminate the process of maintaining routing tables,where communication can be done with position information,solve the problem of routing resources waste.At present,most position-based routing algorithms use independent characteristic such as distance,speed and direction to improve the routing efficiency.All these characteristic are based on expert experience to set the weight of each feature value in routing.However,in position-based routing algorithms,the selection of relay nodes is affected by many features and is not limited to the above mentioned.When the number of characteristic increases,it is difficult to manually set its weight value.Therefore,this paper proposes a method to extract node characteristic through the simulation to generate the dataset of node characteristic.The characteristic values include the distance,speed,acceleration,and direction vector.After calculating the relative distance,the characteristic of neighbor node combine the characteristic of the current node and destination node to generate a training set ultimately used for machine learning.Finally,machine learning is used to fit multiple characteristic into a new routing metric as the driving behavior.The new routing metric is applied to improve the original position-based routing algorithm.We adopt the new route metrics to improve the classic Greedy Perimeter Stateless Routing algorithm.Through experimental comparison,the original GPSR algorithm has been improved in terms of the average delay and packet loss rate,proving the possibility of multi-characteristic improvements to a position-based routing algorithm.
Keywords/Search Tags:VANET, GPSR, Multiple characteristic, Machine learning
PDF Full Text Request
Related items