Research On Computation Offloading Strategy In Mobile Cloud Computing Network

Mobile cloud computing(MCC)provides a resourceful cloud computing environment for various mobile devices.Mobile devices can offload their requests to the cloud servers.In this way,it enhances the computing ability,alleviates the resource constraints,and reduces the energy consumption.However,it is worth mentioning that the execution delay always increases according to the condition of wireless network.How to balance the energy consumption,execution delay and other performance has become a hot research in MCC.In this thesis,we focus on studying the computation offloading strategies in MCC networks.Firstly,in the onstructed computation offloading model in the ad hoc mobile cloud,by discussing the prices which the master device offers to the slave devices and the number of execution units which the slave devices provide to the master device according to the price and the inconvenience coefficients,a two-level Stackelberg game is constructed.After proving the existence and uniqueness of Nash equilibrium point,we obtain the optimal solution in this model by utilizing Lagrangian multiplier method.Secondly,we study the single-user computation offloading strategies with/without fog server accessed.When there is a fog server,by introducing queue theory to simulate the execution delay of the mobile device and the fog server,and combining the load-balance of the fog server,we formulate an energy consumption and execution delay(E&D)minimized multi-objective optimization problem.When tthere is no fog server,the mobile device directly offloads its requests to the remote central cloud through the Wide Area Network(WAN),which will cause long transmission delay.Similarly,the E&D problem in this case is constructed by introducing queuing theory to simulate the task execution process of the mobile device and the remote central cloud.Finally,the optimal offloading probabilities under different situations in this model are derived based on the interior point method(IPM).Thirdly,we analyze the multi-users computation offloading strategies in fog computing.Similarly,we utilize the queue theory to model the execution delay of the mobile devices,the fog server,and the remote central cloud.By combining the fog load balance,we construct an E&D&P(Energy consumption & execution Delay &Price cost)multi-objective optimization problem and solve it with IPM.At last,we obtain the optimal offloading probability and transmission power for each mobile device in this model based on IPM.Additionally,we study the socially-aware dynamic computation offloading strategies for fog computing system.Similarly,the queue theory is used to simulate the execution delay of the mobile device,the fog server and the central cloud server.As the mobile device can harvest energy,the energy consumption of the mobile device comes from the harvesting energy and the remaining energy in the previous slot.Based on this,we construct a jointly convex generalized Nash equilibrium problem(GNEP)whose aim is to minimize the average social group execution cost.Then we use the semi-smooth Newton method with Armijo line search to solve the transformed KKT conditions of the system,which is equal to the original GNEP.At last,we obtain the optimal execution strategy of all mobile devices in this model at all time slots.At last,we study the dynamic computation offloading strategies in fog computing based on Lapunov optimization.The mobile devices in the system access the sub-channels to transmit requests in OFDMA manner,we can dynamic adjust the number of selected channels and the corresponding transmission power.The mobile device's battery energy is coupled between time slots.We solve the average execution cost of the system based on Lyapunov optimization.At last,we obtain the optimal execution strategy,the optimal sub-channel assignment and the optimal transmission power of all mobile devices in this model at all time slots.