Research And Implementation Of Load Balancing Strategy In Dispersed Computing

With the rapid development of Internet of Things technology,the Internet of Battlefield Things(Io BT)technology,which is applied in the military field,is rising rapidly.As the key technology of the transformation from the traditional battlefield operation mode to the information-based battlefield joint operation mode,Io BT has received extensive attention and research from the defense industry and academia.In the existing network architecture of Io BT,the cloud computing has become the key technology of mass information processing with its powerful performance,which is to ensure the real-time processing of battlefield information and the efficient execution of combat operations.The data generated by the ubiquitous and interconnected intelligent devices in the battlefield environment is increasing exponentially than ever before,and real-time battlefield data sharing and operational decision-making largely rely on the real-time data processing.However,cloud servers are often deployed in data centers that far away from Io BT users,the high latency will generated by the network transmission process,which is difficult to meet the operational requirements of real-time data processing in Io BT.In order to solve the above shortcomings of cloud computing when it is applied to the Io BT,this paper conducts research on the relevant theories of the dispersed computing paradigm,focusing on the analysis of the relevant characteristics of dispersed computing,utilizing the computing and storage resources that are close to the users and scattered on the network equipment of entire domain to perform business calculations in concert to provide real-time,agile and reliable information processing services for Io BT users by means of collaborative computing.At the same time,this paper deeply studies the load balancing strategy based on the generalized diffusion model in the dispersed computing.And on this basis,the dispersed computing simulation platform of the load balancing algorithm based on the generalized diffusion model is designed and implemented,and the feasibility of the dispersed computing paradigm is explored from the aspects of theoretical research and engineering implementation respectively.The main research contents are as follows:In the highly informatized battlefield environment,in view of the high business response latency that exists when cloud computing is applied to massive battlefield information interaction scenarios,a fog computing layer is built between the cloud center and users by using intelligent devices on the battlefield,and a “combat cloud-fog”network architecture is proposed.Based on this network architecture,considering the scattered locations and the weak computing power of equipment at the fog computing layer,the strategy of processing services in a collaborative manner of computing nodes is further proposed,the load balancing problem of the fog computing layer is studied,and the task processing latency minimum optimization problem in the fog network is built.And proposes a load balancing strategy based on generalized diffusion to solve the optimization problem.The simulation results show that the “combat cloud-fog”network architecture based on the generalized diffusion optimization load balancing algorithm can effectively reduce the mission response latency.On the basis of the research of load balancing strategy based on the generalized diffusion model,the dispersed computing simulation platform architecture is designed for battlefield computing service scenarios.And based on the Docker container and its network and orchestration technology,the network management module which can automatically generate the cluster network of battlefield equipment,the sub-business calculation module used for business processing and the front-end interface display module are successively realized.It uses the platform to execute the simulation task of the image processing,and verifys the feasibility of the dynamic load balancing based on the generalized diffusion model in the dispersed computing paradigm and the latency performance advantage when applied to battlefield computing service scenarios from the physical point of view.