Research On Offloading Mechanisms Of Vehicular Fog Computing Networks With Uncertain Vehicle Mobility

The development of 5G technology and the demand for intelligent travel make the Internet of vehicles applications more carried on vehicles.Subject to the long transmission link,the traditional cloud computing network is not enough to meet the needs of low latency and high stability for vehicular networking applications.By moving the service framework to the edge of the network,fog computing achieves lower computing delay and becomes the best choice to solve the problem of vehicle networking.In recent years,how to use the computing resources of idle vehicles on the road to assist in the processing of computing tasks in Internet of vehicles applications has become the focus of many scientific researchers.The rapid change of network topology caused by vehicle movement may seriously affect the reliability of computing task offloading.In the field of vehicle networking,determining the movement behavior of vehicles has become a prerequisite for further decision-making.In most of the existing studies,the description of vehicle movement behavior ignores its variability and uncertainty,which makes it difficult for the algorithm to effectively deal with the problems caused by the sudden change of vehicle movement behavior and affects its utility in the actual environment.Therefore,thesis focuses on the problem of reliable and efficient computing task offloading in vehicle fog environment.In the environment where the vehicle movement behavior is uncertain,the exact duration of the communication link cannot be obtained.On this basis,computing task offloading must bear the risk of communication link interruption and wasting of limited computing resources.In this regard,Thesis proposes a task offloading algorithm based on decision-making error risk aversion(ORA)to minimize the probability of communication link interruption while making full use of idle vehicle computing resources.The algorithm analyzes the different risks of computing task offloading decision,and calculates the probability of different situations according to the perceived vehicle state information.In the thesis,the NSGA-II algorithm is used to solve the multiobjective optimization problem to maximize the system revenue and the utilization of service vehicle computing resources.Finally,the performance of ORA algorithm is verified based on simulation experiments.Furthermore,Thesis considers the influence of more complex stream computing task structure on task offloading problem in this environment,uses directed acyclic graph to describe the dependencies between subtasks,and proposes an optimization problem aiming at the average makespan of computation task offloading decision.In this regard,thesis introduces a computing task re-offloading mechanism,which shortens the average makespan of computing tasks while ensuring reliability.In addition,Thesis proposes an optimization of re-offloading mechanism based on path truncation(ORPT)algorithm.By shortening the length of the longest execution path of subtasks on the same execution node,ORPT algorithm can reliably reduce the average makespan of computing tasks in high-load vehicle fog computing environment.