Research On Computation Offloading Oriented Mobile Ad Hoc Networks

Mobile ad hoc networks(MANETs)play an important role in communication,monitoring and even computing.Nodes of mobile ad hoc networks can collect information from the environment and perform corresponding computing tasks based on gathered information.However,due to the limited battery capacity of nodes in the network,it is difficult to maintain long-term communication and computing functions.Therefore,combination of computation offloading and energy harvesting technology can effectively balance the computing and energy resources in the network.Due to the selfishness of the nodes in the network,task offloading does not usually proceed spontaneously.Hence an incentive mechanism is needed to promote cooperation between nodes.Because of the uncertainty caused by the random arrival of tasks,it is necessary to apply an online method to studying the problem of computation offloading.Therefore,a Vickrey-Clarke-Groves(VCG)based reverse auction and stochastic network optimization technology are applied to handling the computation offloading problem in MANET with energy harvesting.The main work of this paper is as follows:Firstly,in order to motivate nodes in the network to offload tasks,the computation offloading problem is studied by a VCG based reverse auction mechanism in this paper,with the goal of maximizing social welfare.Because the proposed VCG based reverse auction mechanism is truthful and has the has the property of individual rationality,buyer nodes will bid honestly,and the utility obtained by each seller node is not less than the utility obtained by not participating in the auction.The modeled social welfare maximization problem is an integer linear programming problem,which is NP-hard,so a simulated annealing algorithm is used to solve the problemSecondly,considering that the number of arriving tasks in the network varies with time.which makes it difficult to optimize offloading decision and computation stage duration,the stochastic network optimization technology is utilized to study the time-average utility maximization.In this paper,the time-average utility maximization problem is transformed into the minimization problem of the infimum bound of drift-minus-profit for each time slot by stochastic network optimization.Then further transformation,a mixed integer linear programming problem(MINLP)is obtained,which is solved by the branch and bound algorithm.An online task offloading algorithm is developed to optimize the offloading decision and computation phase durationFinally,extensive experiments are conducted on the computation offloading models based on the VCG based reverse auction mechanism and stochastic network optimization.First,the VCG based reverse auction mechanism offloading scheme is simulated.As simulation results show,the social welfare and buyer nodes' utility with energy harvesting is higher than that without energy harvesting,and the social welfare and buyers' utility is mainly affected by the number of seller nodes and tasks.Then,the computation offloading method based on stochastic network optimization is simulated.As the simulation results show,the method proposed in this paper can achieve higher network utility,and can effectively balance the computing and energy resources.At the same time,we also analyze the impact of the control parameter V and the task arrival rate on the system in this paper.