Research On Clustering Based Localization Algorithm And Situation Analysis In IoV

With the rapid development of Internet of Vehicle(IoV),the research on vehicle's localization technology has been paid much attention by researchers at home and abroad.Localization technology not only affects the safety of the vehicle,but also plays a decisive role in the development of the IoV.Currently,most of localization schemes adopt the Global Positioning System(GPS)to achieve localization.The schemes can usually achieve excellent results.However,when the vehicle is in the GPS signal blind area,they are unable to effectively complete the real-time localization of vehicle.Therefore,the localization problem of vehicles in the GPS signal blind area needs urgently to be addressed.The situation of vehicle group's behavior can describe the motion rule of the vehicle group in IoV,which has the guiding significance for road planning and traffic control.Although the research on situation of vehicle group's behavior is in the initial stage,it has become an important research direction in IoV.This paper proposes the clustering and localization algorithm for real-time localization of vehicles,which adopts the label propagation.Firstly,an overlapping clustering algorithm based on label propagation is presented through using the relationship of vehicles with the same purpose to reach the identical place in the time slot.Under the condition of the fewer vehicle equipment hardware,the algorithm adopts a distributed strategy to finish the overlapping cluster structure detection in IoV.Then,utilizing the results of overlapping cluster,the localization algorithm and the localization correction method are proposed.The introduced localization algorithm mainly solves the problem of obtaining vehicle's real-time position in the GPS signal blind area.And the localization correction method can rectify the location information when the GPS device is available,providing more accurate data for the future positioning.Experiment results show that the presented localization algorithm has the high localizing precision.A situation analysis scheme based on the vehicle behavior clustering method is presented to acquire the vehicle group's location information.The correlation of different vehicles can be calculated by analyzing vehicle's behavior in the same time slot.And a vehicle's behavior based clustering algorithm is proposed,the algorithm can divide vehicles in IoV into different groups with same and similar behavior.For the analysis of vehicle group's behavior in different time slots,the paper comes up with a vehicle group situation analysis scheme utilizing deep learning method.This scheme can learn the change rule of vehicle group's behavior in different time slots and use this rule to predict vehicle group situation in the future time slot.Consequently,the vehicle group's location information can be obtained.The simulation results show that the prediction accuracy of the proposed scheme can reach above 85%.