Research On Offloading Method Of Mobile Edge Computing Based On Integrated Air-Ground Internet Of Things

In 5G era,the number of access devices and the amount of task data have exploded.At the same time,tasks of devices are definitely toward the computation-intensive and delay-sensitive.Therefore,one of the major challenges for the 5G communication is to ensure Quality of Services(QoS)of networks and Quality of Experience(QoE)of users.Mobile Edge Computing(MEC)deploys the ability of computation and storage at the edge of networks,which can decrease not only the task execution latency but also the transmission energy consumption and waiting time.Meanwhile,as the supplement to applications of 5G,one of the key goals for Beyond 5G(B5G)is providing the ubiquitous networks accessibility.Thus,it is vitally important to investigate the Air Ground and Sea(AGS)networks convergence and collaboration.Above all,based on the integrated air-ground internet of things networks,this thesis studies the devices collaboration,computation offloading and resource allocation in emergency or particular cases with no ground communication services,by the related knowledge such as the optimization theory and deep reinforcement learning.The main works of this thesis are as follows:(1)By applying the Unmanned Air Vehicles,an integrated air-ground MEC system model is designed,in which the air communication network can provide not only the necessary computing assistance but also certain energy support to the ground communication network.As for the ground communication network,avoiding the QoE decreasing caused by the heavy load,from the user’s point of view,the multi-task charging-offloading scheduling is formulated as an optimization problem aiming to minimize the total system task completion latency,by jointly optimizing the task offloading decisions,connection scheduling,charging and computation resources allocation.Based on the BCD method,the MINLP mentioned above is solved approximately by combining the Lagrangian duality theory and greedy algorithm.A Charging-Offloading Resource Allocation Scheme is proposed and simulation results demonstrate that the task completion latency can be effectively reduced by proper device scheduling and resource management,ensuring the QoE of users.(2)As for the air communication network,in general,the trajectory of UAVs,the core of the air communication network,can not be determined in advance.Considering the problem of the system planning and scheduling,from the perspective of network operators and service providers,taking into account the task completion latency and system energy consumption,the optimization problem subjecting to devices connection,task execution,time-slot allocation and UAV trajectory design is presented.In order to comprehensively balance the UAV’s work efficiency and service performance,an edge-end integration algorithm,the Intelligent Offloading and Trajectory Design Scheme(IOTDS)is proposed by combining the Successive Convex Approximation(SCA)and Deep Q-Network,to solve the problem of the UAV working efficiency maximization.And,simulation results present the efficiency of the proposed algorithm.Further,considering the reality of the proposed scheme and algorithm,the working process and the example of industrial applications of the air-ground integration system are demonstrated in the end.