Research On Communication And Caching Technologies In Vehicular Edge Information System

The emerging new driving applications(such as surveillance video,online data stream,and HD map)and repaid development of mobile communication networks bring about not only chances but also challenges for vehicular communications.Although Io V(Internet of Vehicle)technologies have been developed rapidly recent years,it is still under-powered to support those new applications with ultra-high requirements in network performance and computing efficiency.As a new computing architecture,edge computing is widely recognized in recent years due to its advantages in resource management,response time,and network burden alleviating.Consequently,edge computing empowered Io V is a new field,which can take the advantages of communications,caching,and computing in a cooperation way,accordingly forming a new Vehicular Edge Information System(VEIS).To strengthen VEIS,a novel communication architecture needs to be developed firstly,in order to meet the demands of large-volume data transmissions.Secondly,caching algorithms in VEIS,which is to develop the proactive edge caching scheme in edge nodes,is also needed to be improved in terms of service scalability.As the major focus of the thesis,the research on these two parts plays a basic role in VEIS,especially for the function implementation of computing functions.It is well known that those new vehicular applications need to be supported by stable link quality,fast response time,and high resource utilization.Typically,researches focus more on the dynamics in vehicular environments and take them as negative factors.In the thesis,the intrinsic attributes of vehicles,including stability,mobility and non-stationary characteristic are investigated in detail.The negative and positive impact of these factors are then discussed and employed to develop VEIS.My research focuses on the stability of vehicular communication,network efficiency of vehicular caching services in VEIS.Specifically,the first work is developing an inter-vehicle backbone communication scheme based on the stochastic analysis on practical transportation datasets;the second work is introducing the network energy efficiency as an important performance metric and then designing mobility-aware vehicular caching algorithm by transferring the role of vehicles from service consumers to service providers;the third work is designing an online caching algorithm for the non-stationary characteristics of vehicular requests and achieve a trade-off between caching hit ratio and cache service interruption ratio.The major contributions are shown as follows.(1)Developing a multi-hop inter-vehicle communication architecture by exploiting the existence of stability of vehicles.By stochastic analysis on two real-world traffic flow datasets from US highway roads,we can verify the existence of stability of vehicles.A stable vehicle decision model is then proposed based on the knife diffraction model and relative mobility model.A backbone inter-vehicle transmission link thus could be developed by connected the stable vehicles in a linear topology,which can provide stable multi-hop forwarding services for other general vehicles.Considering the arbitrary traffic arrival process,we design a theoretical analysis model for the multi-hop transmission link based on G/G/1 queuing model.The theoretical analysis framework can also be used in other similar multi-hop networks.(2)Designing an online vehicular caching algorithm to maximize network energy efficiency based on mobility of vehicles.In this part,an analysis model is firstly developed to evaluate the connectivity among mobile users,vehicles,and base station by developing a 2-D Markov process.We then formulate the network energy maximization problem into a fractional optimization problem,in order to make otpimal cache decisions in vehicular caching.Finally,by nonlinear programming technology and Lyapunov optimization theory,we derive an efficient solution with low computing complexity.Finally,an online vehicular caching algorithm is formulated.(3)Proposing an online edge caching algorithm to maximize utility with trade-off between caching hit ratio and cache service interruption.We firstly analyze the impact of stationary and non-stationary requests of vehicular users on the caching performance among difference algorithms.A slot-based online cache algorithm is designed based on classical LRU and TTL algorithms,in order to achieve better caching hit ratio in vehicular environments.Besides,we develop an analysis model based on the diffusion approximation theory to obtain interruption rate of edge caching.Finally,an utility maximization model is developed to make cache decisions.