Workload Offloading Algorithm In Ad Hoc Cloud Computing

The rapid spread of mobile smart devices has brought great convenience to people's lives,but its performance is still limited by CPU processing power,battery life,storage capacity and other factors.The application of mobile cloud computing technology provides a good idea for improving the task processing efficiency of mobile terminals.However,there are no available cloud infrastructures in many scenarios,or large data transmission delay due to far away servers.Therefore,a more concise Ad Hoc cloud scheme which needs no cloud infrastructures,and is widely studied.In Ad Hoc cloud,mobile devices expand computing capacity by offloading workload to other devices,and without having to access the remote cloud server.Most of the researches focus on workload offloading delay and energy consumption in Ad Hoc cloud computing.However,the uneven distribution of tasks and resources leads to the imbalance between resource waste and delay/energy consumption.Aiming at this problem,and the multi-objective decision problem in Ad Hoc cloud computing,a joint distribution model of workload offloading and power allocation based on the control system delay and energy consumption is established from the system point of view.The algorithm dispatches the idle resources that can be used to distribute and offloading the calculation workload in order to balance the delay and energy consumption on the workload of offloading.Simulation results show that the algorithm can not only effectively coordinate the workload offloading and power allocation,but also improve the performance of balancing system delay and energy consumption.Then,based on the above-mentioned centralized joint optimization model,the client and agent are abstracted into buyer and seller respectively,and a distributed joint optimization algorithm for workload offloading and power allocation is proposed for the incentive agent terminal resource sharing in order to motivate the resource sharing of agent terminal.And interaction between the client and the agent is simulated,and then a quotation strategy based on the Stackelberg Equilibrium(SE)is proposed as a paid workload offloading mechanism.In addition,the distribution of agents in the cloud is unbalanced due to the dynamic arrival of the workload and the randomness of the departure when the workload processing is completed,which causes low acceptance rate of the subsequent requests,a dynamic workload offloading algorithm based on SMDP is proposed in order to solve the above problem.The algorithm obtains the optimal decision for workload assignment control by cooperating with cloud server resources,which aiming at maximizing the cumulative revenue of the system.The simulation results show that the proposed algorithm can improve the performance of the request receiving rate and obtain more payoff in Ad Hoc cloud.