Vehicular Crowd Sensing Coverage Methods And Applications For Smart Transportation

Smart transportation depends on the sensing and utilization of urban traffic information,such as the real-time road traffic conditions.With the rapid development of wireless communication technology and the popularity of smart mobile devices,urban vehicular crowd sensing,as a new emerging sensing paradigm that leverages drivers’ smart devices to obtain sensing data from the urban environment,can complete sensing tasks that are difficult for traditional wireless sensor networks to handle.Urban vehicular crowd sensing is widely used in many applications of smart transportation due to the reasons of large number of sensing nodes,fast moving speed,large sensing data scale and low deployment cost.The foundation of the development of smart transportation systems lies in the accurate and timely acquisition of large-scale urban traffic environment data.However,the current vehicular crowd sensing coverage still has the status of incomplete coverage and coarse granularity spatially and temporally.Solving this problem faces challenges such as complex urban environments,variable sensing tasks,many types of sensing vehicles,and complex user behaviors.To this end,we first investigate the joint trajectory scheduling and incentive mechanisms for the urban vehicular crowd sensing systems,aiming at increasing the spatial sensing coverage with lowest social cost.Furthermore,considering the complexity of the urban environment and the difficulty of obtaining sensing demands,in order to achieve spatio-temporal fine-grained sensing coverage,we further propose a hybrid sensing approach that utilizes not only for-hire vehicles(FHVs),but also dedicated sensing vehicles(DSVs)that are exclusively dedicated for sensing.We utilize DSVs to make up the intrinsic spatio-temporal distribution bias of FHVs,which enables fine-grained spatio-temporal sensing coverage.Next,depending on the improvement of fine-grained sensing coverage performance,we study how to achieve optimal control of the electric vehicle charging system.This thesis finds that under the fine-grained sensing coverage,ie.,the urban area is divided into smaller regions and the road condition data is updated faster,the platform can increase the long-term time average profit.On the one hand,the platform can more accurately calculate each vehicle’s arrival time to the charging station.Through dynamic control methods such as dynamic pricing,the platform can balance the spatiotemporal charging demand over the whole city.On the other hand,the platform can estimate the time and energy consumption of an electric taxi for serving an order more accurately,and thus improve the reliability of the charging decisions and the accuracy of dispatch decisions effectively.This thesis mainly focuses on the urban vehicular crowd sensing coverage methods and applications.The main results of this thesis are as follows:Firstly,we capture the interactive effects between the trajectory scheduling and incentive mechanisms,and propose a joint trajectory scheduling and incentive algorithm which improves the spatial sensing coverage of the vehicular crowd sensing systems.The social cost yielded by the proposed algorithm is close-to-optimal,and we derive the approximation ratio and prove the property of truthfulness,individual rationality and computational efficiency.Theoretical analysis and experimental results show that the algorithm designed in this thesis can effectively improve the spatial sensing coverage while reducing the overall social costs.It serves as a theoretical basis for the collection of large amounts of high-quality sensing data required by the vehicular crowd sensing systems.Secondly,in order to achieve fine-grained spatio-temporal sensing coverage,this thesis proposes to use a hybrid approach,where the platform leverages not only for-hire vehicles but also DSVs to bridge FHVs’ coverage gaps,in order to improve the granularity of sensing coverage spatially and temporally.In terms of research methods,we integrate stochastic dynamic programming with distributionally robust optimization.We design a repositioning policy for DSVs which utilizes only partial statistical knowledge of sensing demands,and achieves cost minimization in a robust manner.Furthermore,we carefully design a linear decision rule that significantly improves the computational efficiency of our solution approach.Theoretical analysis and experimental results show that under our hybrid approach,exploiting the first and second moments information of the sensing demand distribution achieves not only lower repositioning cost but also better spatio-temporal sensing coverage.Thirdly,on the platform side of the electric vehicle charging system,this thesis controls the lengths of the demand queues at multiple charging stations under the fine-grained spatiotemporal sensing coverage.We construct independent sets of charging stations to decouple the complex correlations of vehicles’ charging demands over different charging stations.This thesis develops a price control policy that maximizes the platform’s profit while ensuring bounded queue lengths at charging stations.Furthermore,we give the lower bound of profit and the upper bound of the queue length.Under the fine-grained spatio-temporal sensing coverage of urban road segments,the platform can accurately calculate the vehicle’s arrival time at the charging station.Through price and demand control methods,the length of the charging demand queue at each station in the city is balanced spatially and temporally,which improves the platform’s long-term time average profit.Lastly,on the user side of the electric vehicle charging system,this thesis investigates the problem of joint order dispatch and charging in an urban electric self-driving taxi system under the fine-grained spatio-temporal sensing coverage.Given the charging price of each charging station,the platform decides the charging time and interval based on the real-time road conditions,and allocates the orders based on the remaining energy of the electric taxis.Under the fine-grained sensing coverage,the platform can accurately estimate the service time of each order as well as the time travelling to the nearest charging station.By integrating the expected future profits into the objective function of the current decision,this thesis derives the longterm decision to maximize the long-term profit of the system.Results show that fine-grained spatio-temporal sensing coverage can improve the accuracy of system order matching decisions and the reliability of charging decisions.Our multi-dimensional joint decision-making method is computationally efficient and can be extended to a city-scale scenario.