Task Offloading And Mobility Management In Internet Of Vehicles Task Offloading And Mobility

As a carrier of mobile information,a networked vehicle can perform data interaction through V2I(Vehicle-to-Infrastructure)or V2V(Vehicle-to-Vehicle).With the development of vehicular network technology,the demand for high-speed computing and fast communication in automotive application is becoming more and more prominent,and the latency is very sensitive.However,the current vehicular network systems cannot meet the increasing latency requirements of vehicular applications.The vehicular network based on mobile edge computing(MEC)is envisioned as a potential solution to meet the application needs.MEC sinks cloud services to the edge of wireless networks and provides computing services close to users,thereby making up for the delay fluctuations caused by remote cloud computing and improving the vehicle's Qo S(Quality of Service).However,the limited computing resources of MEC and the problem of network changes when vehicles move at high speed will affect the performance of the network as the number of vehicle users increases.Therefore,in the vehicular network scenario based on MEC,this thesis studies the problem of resource allocation during task offloading and the problem of mobility management when the vehicle is moving at high speed.The main contributions are as follows:1.The trade-off between energy consumption and latency when task offloading in the vehicular network is studied.The weighting factor is redefined with the energy residual rate to perform task offloading by real-time energy-aware scheme.A MEC-based offloading framework is built,where the task can be offloaded to the MEC server or computed locally.First,the bi-lever optimization algorithm is used to decouple the original problem into a low-level problem of resource allocation and an upper-level problem of task offloading.Considering the mobility of vehicle users and the availability of cloud resources,a deep Q-learning network(DQN)based offloading scheme enables users to make near optimal offloading decisions.The simulation results show that the proposed mechanism reduces the total cost of the system under the constraint of delay,and the proposed DQN scheme makes the obtained offloading strategy more effective.2.MEC and Content Distribution Network(CDN)are integrated to form a logic framework for organizing cloud resources,including central,edge and vehicle cloud resources.This framework allows vehicles to choose their cloud service flexibly.The high-speed movements of the vehicles are considered,the joint optimization problem of cell handover and virtual machine(VM)migration is formulated to optimize the total latency of the user.First,the dynamic channel allocation algorithm with overhead selection avoids the ping-pong effect and reduces the handover time of vehicles between cells.Then,the cooperative game algorithm based on roadside unit(RSU)is used to perform VM migration and a learning-based price control mechanism is developed to process vehicular computation resources efficiently.The simulation results show that the proposed algorithm can improve resource utilization compared to the existing algorithms.