Resource Allocation And Adaptive Streaming For QoS Guarantee In Vehicular Networks

As a main part of future intelligent transportation system(ITS),vehicular networks should be able to guarantee quality of service(QoS)of various vehicular applications.However,the vehicles'high mobility,the competitions for limited channel resources among applications/services with different priorities,and the possibly sparse deployment of road side infrastructures together lead to the intermittent connection between vehicles and the infrastructures,rapid changes of topologies among vehicles,high dynamic channel quality,and ever-changing number of available channels.In addition to the aforementioned complex network conditions,vehicular networks need different resource allocation strategies to satisfy the QoS requirements from different kinds of applications.Further,different QoS metrics of one application might be mutually contradictory,e.g.,visual quality will be sacrificed to ensure smooth video playback.This thesis aims to explore the QoS guarantee technologies for vehicular networks,design resource allocation and streaming schemes for video streaming and beaconing services which are typical infotainment service and intelligent transportation service,respectively.For the vehicular networks using cognitive channels and with possibly sparse deployed infras-tructures,quality-driven adaptive video streaming is studied,an effective scheme is designed and its performances are analyzed.Mathematic expressions of single-hop or two-hop end-to-end con-nections are given.A utility maximization problem is modeled to obtain optimal request number of video layers,so as to promote visual quality on the premise of guaranteeing smooth video play-back.The utility is expressed as function of the number of stored data in the vehicle,the number for data for future playback without interruption,the vehicle's future trajectory and communication connectivity,the number of receivable data with the required number of video layers.An adaptive video streaming scheme is proposed,and its performances are verified through simulation.For the vehicular networks that the number of available channels is ever-changing because of the occupation of the other services with higher priorities,the resource allocation and video stream-ing among multiple vehicles are investigated,an effective scheme is designed and its performances are analyzed.An auction-based resource allocation scheme is first set up,taking the motion of vehicles and the dynamic changes of channel resources into consideration.The scheme transforms the resource allocation at the RSU side to the auction among the vehicles.A vehicle's bid is a function of the numbers of channels and vehicles,the number of data units in the vehicle's buffer,the vehicle's location and future trajectory,current transmission rate,and the intensity of vehicles'competition,and thus reflecting the vehicles' urgent level to the amount of allocated channel re-source.A utility sum maximization problem is modeled to obtain optimal request number of video layers,aiming to promote visual quality on the premise of guaranteeing smooth video playback.Environment states and their transmission probabilities are defined to rationalize the vehicle's bid.A mean of the total utility value accumulated(MTUA)maximization problem is modeled,which is with the form of Bellman equation.A dynamic programming-based low-complex algorithm is proposed to obtain the optimal bid and requested number of video layers in various vehicle states and environment states.For the beaconing service in vehicular networks,which is a typical intelligent transportation application,adaptive beaconing and resource allocation for collision avoidance and location track-ing accuracy are studied.A scheme is proposed for adjacent RSUs to negotiate their resource usages orderly and fairly,so as to avoid the vehicles in the adjacent RSUs' coverages to be allo-cated the same time-frequency resource.The mathematic expressions of two beacon interval(BI)requirements are defined,which reflect the vehicle's driving safety and location tracking accura-cy requirements,respectively.The BI requirements can also reflect the vehicle's different driving states.A utility function is defined based on the satisfaction degree of vehicle's BI requirements.The resource allocation among the vehicles within a single RSU's coverage is modeled as a utility summation maximization problem.To solve this 0-1 integer optimization problem,a scheme is proposed to transform it into a maximum weighted independent set(MWIS)problem in an undi-rected weighted graph,and a low-complex solving algorithm is developed.The above schemes can avoid the beacon collisions among the vehicles located in either adjacent RSUs' coverages or the same RSU's coverage,and thus guarantee the vehicles' driving safety and location tracking accuracy requirements as possible.