Research On Dynamic Offloading Strategy For Parking Vehicle Assisted Edge Computin

Demands of computation resources dramatically increase with the development of popular on-board applications.Data processing is gradually moving from the center to the edge closer to the user,giving rise to mobile edge computing.However,inadequate infrastructure and compute bottlenecks on edge servers hinder the effective use of computeintensive in-vehicle applications.Considering that parked vehicles(PVs)account for a large portion of the global vehicle population,and that parked vehicles in cities have a wide range of available computing resources,these vehicles can be considered lightweight and accessible edge computing nodes,assisting the edge to achieve large-scale computing offload to meet the real-time,secure,low-latency,and responsive computing needs of other intelligent vehicle services.However,on the one hand,as a provider of computing services,the residence time of parking lot vehicles is uncertain,which greatly affects the scheduling of computing tasks.On the other hand,parked vehicles limited by computing power,terminal battery capacity,storage capacity,etc.,are difficult to participate unconditionally and reliably in the execution of computing tasks.Therefore,this paper proposes economic incentive mechanisms based on uniform pricing and differentiated pricing to effectively promote PVs to assist edge service provider(SP)in offloading computation tasks.The main research contents of this thesis are as follows:(1)Based on Stackelberg’s uniform pricing theory,a dynamic computing task offloading method for parked vehicles assisted edge service provider is designed.Considering that the total computing tasks received by SP over time are different,dynamically adjust its computing resource strategy to complete the computing tasks in each time slot.According to the arrival of the computing task,the time is divided into time slots,and the shutdown duration model is integrated to properly allocate the offloading task of PVs to achieve effective execution,and ensure that the computing task received by SP in each time slot can be completed.Resource allocation and uniform pricing strategies were then dynamically developed using the tools of the Stackelberg game to balance the benefits of SP and PVs.In the first stage,SP as the leader decides on uniform-resource pricing.In the second stage,PVs act as followers to decide the computing resources allocated for task execution based on resource pricing.This thesis theoretically proves the existence of a uniform Nash equilibrium.In order to achieve the Nash equilibrium point,the Lagrange dual method is used to solve the constraint optimization problem.The simulation results verify the effectiveness and reliability of the proposed model in a dynamic environment.(2)Based on the Stackelberg differential pricing theory,a dynamic computing task offloading method for parked vehicles assisted edge service provider is designed.Considering the energy consumption cost of each parked vehicle and the impact of parking duration on the system,the simplified pricing strategy has certain limitations,that is,SP,as the leader,determines that a uniform computing resource pricing cannot be applied to all parked vehicles participating in the task calculation in the same time slot,and further proposes a differentiated pricing strategy,which dynamically adjusts the computing resource prices of different parked vehicles.To optimize the utility of each parking vehicle user.The simulation results show that the proposed differential pricing scheme is effective.Compared with the uniform pricing scheme,the differential pricing scheme can obtain better PVs and SP offloading utility,effectively reduce SP operating costs,and meet the incentive demand for PVs.The research results of this thesis are of great significance as they can effectively address the limitations of computational resources and time-sensitive issues in intelligent vehicle applications.By making use of the idle computational resources of parked vehicles and implementing dynamic task offloading computing,the quality and performance of intelligent vehicle applications can be improved.In addition,the research results of this article also provide useful ideas and directions for the development of future connected vehicles and mobile edge computing.By constructing a reasonable resource allocation and pricing scheme,better cooperation between service providers and parked vehicles can be achieved,promoting resource sharing and optimization,and improving the performance and efficiency of the entire edge computing network.