Research On Task Offloading Algorithm Of Edge Computing System For Intelligent Farming

With the intelligent development of agricultural Internet of Things,its application value in intelligent breeding technology has become increasingly prominent.In the intelligent breeding system,applications will produce many computation-intensive and delay-sensitive tasks,and the traditional task execution mode is not enough to meet users’ requirements on delay and cost.With the development of new technology,researchers find that mobile edge computing is an innovative mode to deal with this computing task.But on a mobile device can move any more users and server scenario,the high energy consumption performed by mobile edge device tasks is limited by capacity-constrained battery power.Due to factors such as spatiotemporal dynamics,competitive wireless communication,transmission noise interference,and resource competition in mobile devices,task offload strategy faces new challenges.The main research contents are as follows:(1)Combine existing edge computing application research.Considering user mobility and edge device mobility,the multi-user single-move and multi-server models are proposed in the third chapter,and the multi-user and multi-server all-mobile models are proposed in the fourth chapter.Mathematical modeling and theoretical analysis of the two models are also carried out.(2)This paper proposes an intelligent computing offloading strategy that adopts the quality of user experience cost and the offloading rate of computing tasks as performance indicators.The strategy is based on Lyapunov optimization and LODCO algorithm,and different execution modes are selected for each mobile device calculation request,which can gradually obtain the global optimal results of the system.In addition,the introduction of energy harvesting technology enables the high energy consumption of intelligent services to be effectively balanced with the battery capacity of mobile devices with limited capacity.The algorithm has low complexity,does not require much prior knowledge,and can be well adapted to more complex environments.(3)This paper considers the scenario that the portable mobile device is used as the edge computing server for task execution,and the application device can learn the offloading experience of its adjacent Io T devices during the offloading execution.On this basis,an adaptive learning algorithm based on Multi-armed Bandit(MAB)theory is designed to minimize the average offloading delay.The proposed algorithm works in a distributed manner and does not require frequent acquisition of equipment information.Through extensive simulation,it can be concluded that the algorithm achieves low delay performance with low learning regret.And has consciousness perception function to adapt to the reality dynamic environment.