Research On Resource And Service Optimization In Vehicular Edge Computing

In recent years,the global number of vehicels has aleadly reached one billion and compute-intesive applications represented by autonomous driving are emerged.This gives rise to great challenges for provisioning communication and computing resources in Internet of Vehicels(IoV).To cope with the issues,researchers propose edge computing technologies to the IoV,leading to a new networking paradigam,named by vehicular edge compting.Nowadays,studies on resource and service optimization in vehicular edge computing have become a hot topic in communities of IoV and intelligent transportation system.This dissertation focuses on optimizing usage and scheduling of limited network resources from both sides of user demand and resource suppliers,for ensuring quality-of-service and improving network capacity.Besides,towards secure and reliable service environement,a distributed reaputation syseme is utilized to promptly and accurately evaluate behaviors of vehicles.Meanwhile,security threats for edge computing infrastrustures such as roadside units(RSU) are explored.The main contributions of the dissertation are shown as follows1.Ties among vehicular neighbors to be served simulatenously in 5G environment are investigated.To dynamically manage these vehicular groups,software-defined networking is considered and a software-defined network architecture is elaboratively designed.In the architecture,vehicle network problem among vehicular neighbors is solved by using the universal plugand-play(UPnP) protocol and a case study of vehicular cloud computing is also provided.The proposed architecture is demonstrated with great advantages in extending resource capacity,ensuring reliability of resource provision and improving resource usage2.By proposing the cooperation between parked vehicles and edge servers,collaborative edge computing is achieved and an optimal rescource scheduling scheme is futher presented for maximizing user satisfaction.Main entities with the network functionalities are desciried.At the same time,an interative protocol among them is introduced for guaranteeing security and reliability.Serviceability of different parked vehicles for collaborative edge computing is evalutated to selecte stable task performers Then Stackelberg game approach is used to formulate and solve the optimization problem about resource scheduling among parked vehicles and edge servers.Simulation results demonstate that the amount of users in the proposed scheme is about twice as many as that in the exsiting shcme only considering edge servers for task execution,owing to the extended resource capacity.3.Towards secure and efficient vehicular edge computing,a distributed reputation management system is builded.Overall system model with detailed funictionalities is introduced.Then multi-weighted subject logic is proposed to update reaputation values of vehicels with higher accuracy.Furthermore,a well-considered algorithm is studied to allocate resources for vehicles based on the reputation values and task requirements in computation offloading.Simulation results demonstrate that compared with the general subject logice method,our proposed method exhibits great advantages in timely misbehavior detection,significantly improving recognition rate of misbehaving vehicels.4.A defense scheme is presented for vehicular edge computing after discussing potential security threats for RSU.First,severe damages caused by compromised RSU are summarized,meanwhile,the features regarding to such an adversary for RSU is discussed.Then a new adversary model,called by RSU hotspot attack,is introduced based on the above features.To prevenet the attack,a proactive defense scheme based on dummy traffic delivery is investigated.Particularly,Stackelberge game approach is also to build the incetive mechanism required by the defense scheme.Simulation results demonstrate that our scheme is effective and efficient to protect networks against RSU hotspot attack by reducing the probability of locating a target RSU.