Research On Optimization Methods Of Mobile Computing Offloading And Scheduling Under Delay Constraints

The popularity of the Internet of Things(Io T)and the vigorous development of fifth-generation(5G)related technologies have promoted the development of smart cars,which in turn has contributed to the rise of vehicle multimedia applications such as augmented reality and location services.Although vehicle-to-infrastructure communication(V2I)can relieve the heavy calculation and analysis pressure caused by multimedia applications,due to the limited mobility and computing resources of vehicles,the functions and management mechanisms of vehicle equipment are inefficient and rely solely on vehicles.The Internet(Io V)still faces many challenges.Thus,there is an urgent need for a solution that can balance computing-intensive tasks between agents(such as a fixed-location cloud computing center and nearby vehicles)based on the remaining resources,and the emergence of the mobile cloud computing is quite qualified for this role.Although mobile cloud computing can flexibly use opportunistic vehicle-to-vehicle communication(V2V)technology,but without a good offloading strategy,the vehicle terminal will not be able to complete the task in the duration of the link’s sustainable connection and will not be able to provide a satisfactory quality of experience(Qo E)for the vehicle terminals.Therefore,adaptive distributed algorithms should be used to deal with the dynamic network topology of the vehicle cloud computing scene.In this paper,the problem of computing offloading is studied from two aspects,including the cooperation between vehicles and the formation of peer-to-peer connection with base stations.(1)First,this paper proposes an online offloading method based on delay and fairness awareness,in which four issues are considered,including the randomness of task arrival,the dynamic topology caused by the vehicle terminal’s movement,two associated tolerable delays and the fairness requirement of Qo E between terminals shoud be ensured.As a result,an online algorithm designed to balance the overall quality of experience and fairness with a utility function parameterized by fairness deviation under the constraints of delay and revenue for each vehicle terminal.By introducing a virtual queue with delay awareness and revenue constraints,and using Lyapunov optimization technology to transform the objective function into a drift-plus-penalty minimum boundary problem,the offloading decision can be made distributed based on real-time network and queue status information.(2)Secondly,this paper studies the price-aware computing offloading method based on peer-to-peer connections.The core idea is to jointly optimize the roadside unit(RSU)association decision on the long-term scale and the computational offloading decision on the short-time scale.In order to minimize the terminal’s resource purchase cost under the constraints of tolerable delay,energy consumption,power and server computing capacity.This paper models it as a Lyapunov boundary minimization problem,and solves the problem in three stages.First,the objective function and queue stability are combined to construct a Lyapunov function to convert to solving the minimum boundary;secondly,given the transmission power of the V2I link,the terminal-RSU correlation problem is constructed as a many-to-one matching.Finally,on the basis of solving the terminal-RSU correlation problem,the Lagrangian multiplier method is used to solve the allocation of transmission power and computing resources.(3)Finally,this paper has carried out experimental verification on the proposed scheme.Aiming at the online offloading scheme based on latency and fairness perception,this paper analyzes the influence of different parameters on system utility,fairness index and task split ratio in online environment,and compare with two benchmark algorithms in terms of system utility and average queue backlog.Aiming at the price-aware computing offloading scheme based on peer-to-peer connections,this paper verifies that the dual time-scale control algorithm can better realize the collaborative computing between the terminal and the RSU without the prior knowledge of the system,and it has achieved good performance in reducing system cost.