Research And Application Of Computational Offloading Algorithm For Cloud-Edge Collaboration In Smart Agriculture

With the development of information technology,as the direction of national key support and the fundamental of the country,the wisdom of agriculture is an important field strongly supported by the state in recent years,and is also an important part of the national strategy of rural revitalization.However,with the deepening of agricultural wisdom,how to timely analyze and process all kinds of massive sensory heterogeneous information acquired during agricultural operations,and then provide real-time data support for wise decision making and control of various agricultural operations,is still an urgent problem to be solved in the process of agricultural wisdom.With the development and popularization of communication technologies such as 5G and the enhancement of various cloud,edge and end computing capabilities,a computing model based on cloud-edge-end collaboration provides the possibility of timely analysis and processing of massive heterogeneous information generated from various agricultural operation processes.Because of this,this thesis conducts an in-depth study on the offloading method of cloud-edge collaborative computing for smart agriculture,thus,the main work of the thesis is as follows.(1)The overall process of the cloud-side collaborative computing offloading method for smart agriculture is designed for the characteristics of farming operations,and the algorithm model is established in the cloud according to the actual farm environment,which alleviates the problems of heavy computational pressure and the real-time nature required for computational tasks in agricultural systems.By designing the task volume prediction model,response time model and power consumption model to model the overall computing process of farming operations from multiple perspectives to reduce the conflict rate of computational offloading results,and on this basis,DQN(Deep Q Network)algorithm is introduced to find the optimal solution to the mixed integer nonlinear optimization problem in the model.Experiments show that the model can better solve the dynamic conflict problem in farming operations,improve the offloading accuracy,and then improve the accuracy of cloud-side collaborative computation.(2)An RP-Double DQN(Re Experience Pool Double Deep Q Network)algorithm based on the DQN algorithm is proposed to solve two key problems in solving the offloading model of cloud-side collaborative computing for smart agriculture,namely: the introduction of the Double DQN algorithm based on the DQN algorithm,which alleviates the data instability and Q-value anomalies caused by the algorithm itself.a high-priority sample dynamic separation mechanism and an experience pool storage sampling mechanism are designed to mitigates the inefficiency of training the model with sample sampling in the experience pool and the slow convergence during iteration.Finally,its effectiveness is verified by experimental analysis.(3)Based on the above research and combined with the actual needs of farms for intelligent agricultural management system,we designed and implemented the intelligent agricultural management system based on cloud-side collaborative computing offloading with five modules,including user management module,information collection module,plot management module,cloud-side collaborative computing offloading module and farm machinery management module.It achieved good results.