Research On Task Offloading Method Based On Mobile Edge Computing Strategy In IoV

Internet of Vehicles(Io V)is becoming a domestic and international research hotspot in the era of 5G Internet,and challenges are encountered in the rapid development of Io V.First of all,the scale of vehicles is rising exponentially,which generates a huge amount of data information,resulting in insufficient resources for the vehicles themselves to meet the computation demand in time.Secondly,vehicles have fast driving speed on the road,and high mobility has an adverse impact on link connection stability,and frequent network changes reduce the service experience of vehicle users.Cloud computing(CC)can alleviate the pressure on vehicle resources to a certain extent,but there are still bottlenecks.Mobile Edge Computing(MEC),an extension of cloud computing,not only has powerful computing capabilities,but is also closer to the edge of the mobile network,enabling services to be provided near the end devices.Computation offloading is an important technique to improve user experience in MEC.In this thesis,we propose two offloading schemes to reduce the delay and energy consumption in the computation offloading process for the task offloading problem in vehicular edge networks.In response to the great challenges posed by the high mobility of connected vehicles for communication and computation,this thesis first proposes a vehicle network model SNM,in which vehicles are divided into three sub-networks according to the turning direction at the next intersection.Then,the sub-networks with offloading demands are constructed into a bipartite graph,and the optimal matching of the graph is solved using the KM algorithm(Kuhn-Munkres algorithm,KM),i.e.,the offloading scheme with the minimum delay in processing the computational task is obtained.Finally,through simulation experiments,the algorithm proposed in this thesis can effectively reduce the packet loss rate and average delay in the offloading process,which verifies the effectiveness of the offloading scheme.As the popularity of smart vehicles generates a large amount of computational demand,offloading tasks to edge servers for processing is an effective solution.However,when the task volume is too large,the edge server runs out of computing resources and will not be able to meet the demand of all vehicles.Therefore,this thesis proposes a task offloading scheme combining Multi-Objective Grey Wolf Optimization(MOGWO)with Gravity Reference Point Technique(GRPT)in a scalable,multi-user environment relying on a three-tier cloud-edge-end architecture.GRPT)task offloading scheme.By generalizing the time delay and energy consumption into two objective functions,an optimal search method is designed to "capture" the optimal solution by imitating the hierarchical relationship and hunting activities of the grey wolf population,and to achieve the joint optimization of energy consumption and computational time delay.Experiments show that the proposed scheme has better performance compared with other existing algorithms.