Research On Offloading Strategy Of Multi-UAV Task For Transmission Line Inspection

As the transmission lines increase year by year and the inspection environment in some areas is very dangerous,intelligent multiple UAV inspection has been widely used in the operation and maintenance of transmission lines.However,when it comes to computing intensive tasks,the lack of computing power and battery capacity is obvious.At the same time,in order to ensure the quality of service(Qo S)requirements,a new Computing framework--mobile edge computing(MEC)proposed to sink the network control and storage capabilities to the Edge of the network.The computing tasks can be divided into local tasks and remote tasks by relative offloading decision mechanism.The remote tasks will be sent to the edge server(ES)with rich computing resources for execution,and some local tasks with small computation will be processed locally.Therefore,this thesis mainly studies the impact of mission offloading on energy consumption and delay under different UAV offloading strategies,and proposes offloading strategies for three different line inspection scenarios:(1)Mission offloading strategy for multiple UAVs-single ES scenarios.At present,aerial transmission lines are mainly inspected by multiple UAVs.Due to the weak computing power of the equipment carried by the UAV,the endurance is short.In the single ES scenario,this thesis proposes a task offloading strategy based on Deep reinforcement learning,proposes an end-to-end edge-cloud collaborative offloading architecture,establishes minimum delay optimization problem,constructs markov decision,and uses deep Q-network(DQN)to solve the optimization problem.Simulation results show that the offloading strategy has lower delay and higher reliability.(2)Mission offloading strategy for multiple UAVs-multiple ES scenarios.In view of this situation,this thesis proposes a offloading strategy based on Game Theory and reinforcement learning.This strategy establishes a non-cooperative game theory(GT)between UAVs,and defines the minimal cost function as the combination of energy expenditure and delay.At the same time,the existence of at least one Nash equilibrium(NE)is proved,and a distributed algorithm is proposed to solve the NE solution of both sides of the game.On this basis,A reinforcement Learning method based on stochastic learning automata(SLA)theory is proposed to effectively select edge servers for UAV.Simulation results show that compared with other offloading strategies,the offloading mechanism proposed in this thesis can significantly reduce uav energy consumption,system cost and network delay.(3)Mission offloading strategy for multi-UAV cooperative scenario.Multi-UAV cooperation is mainly proposed for unmanned aerial vehicles performing emergency fault inspection in remote mountainous areas.Due to poor communication links in remote areas,long-distance data transmission delay is high,and intensive computing tasks cannot be performed well.Aiming at this problem,this thesis proposes a cooperative offloading strategy(COS),which consists of two levels: the first level uses to determine whether a task can be executed locally;The second level is to offload a drone task with limited resources to a nearby drone for processing.Considering the time delay and energy consumption in the process of computation and transmission,an optimal resource allocation strategy based on threshold is developed to ensure system stability and meet Qo S requirements.Simulation results show that the UAV cooperative algorithm is effective,and compared with its offloading strategy,it can effectively reduce the delay and improve the system utility.