Research On Optimal Algorithm Of Distributed Resource Allocation For Individual User In Ad Hoc Cloud

With the smart mobile devices technology development,mobile applications are becoming more versatile.However,the performance of many mobile devices is affected by the CPU processing speed,storage space and other issues.The mobile device can not bring user a satisfactory application experience.At the same time,there is a problem that the device cannot access the cloud server and have a large transmission delay due to network congestion and a fault condition in many scenarios.A large number of scholars have done a lot of research on the computationally intensive task execution of mobile devices.They have created a new research field called Mobile Ad Hoc Clould Computing.In a mobile Ad Hoc cloud,the local device assists in its own task by utilizing the multi-dimensional resources of nearby available devices without having to access the remote cloud server.However,the related research on mobile Ad Hoc cloud computing is still in its infancy,especially in the problems of the waste caused by uneven resource allocation and the failure of task offloading caused by the randomness of device movement.Therefore,further research on mobile Ad Hoc cloud computing is still needed.Aiming at the resource allocation problem in the above Ad Hoc cloud,a joint optimization allocation model of power resources and computing resources is established from a global perspective.We abstract resource demand devices into buyers,and abstract resource idle devices into sellers in order to encourage resource sharing of mobile devices.A distributed resource allocation optimization algorithm based on buyer/seller games is proposed to obtain the optimal buyer resource requirements and seller prices.The simulation results show that the proposed algorithm greatly reduces the time complexity while the system performance is close to the centralized optimal result.In addition,in the mobile Ad Hoc cloud,the mobility of the device may affect communication connection time,which directly causes the failure of task offloading,especially in an Ad Hoc cloud in which a vehicle is constructed by a self-organizing network.The highspeed movement characteristics of the vehicle make the above problem more conspicuous.For this reason,this thesis constructs the task assignment optimization problem based on maximizing the offloading utility of resource-constrained vehicle.On this basis,Artificial Neural Network(ANN)is used to predict the connection time between vehicles.Then a distributed optimization algorithm for computing resource allocation is proposed to encourage vehicles to participate in cooperation and obtain optimal computing resource allocation results.Finally,a centralized algorithm based on Lagrangian multiplier method is used to obtain the optimal task assignment result.The simulation experiment proves that the method effectively improves the accuracy of time prediction and can make the vehicle obtain higher task offloading utility.