Research On The Storage Model And Scheduling Algorithm For High Resolution Data Based On Cloud Computing Platform

With the launch of a high-definition survey satellite, our country is taking active measures to push the development of spatial information industry. Remote sensing data and its derivative data, which have their own intellectual property rights, are allowed to increase exponentially. In order to satisfy the high-data-concurrency needs from users and offer data sharing service, how to storage and manage data efficiently has become an important research direction in space information field.Cloud computing has advantages in high-performance computing, mass data storing, distributed application and data sharing service offering. But traditional storage model based on cloud platform drops comprehensive in supporting different kinds of remote sensing data, lacks good performance in resource distribution, and needs to improve the scheduling efficiency of data requirement. This paper designs a better data storage model, which can solve the problem above, and proposes improved task scheduling stratagem. The storage model and task scheduling stratagem have been put in to use, which reflect its availability and advantages.This paper analyzes and classifies the characteristics of multi-source heterogeneous remote sensing data. Then it set up storage model based on object-oriented and virtual memory under distributed environment. In addition, this paper applies improved Ant Colony Optimization to remote sensing data scheduling, which makes the scheduling algorithm with the advantages of self-adaption, quick response and high-efficiency. At last, this paper introduces the real application of the storage model and scheduling algorithm in this paper. There are three aspects of detailed studies:(1) In a distributed environment, this paper proposes a remote sensing data storage model based on cloud computing(RSDO). The model is set up according to the characteristics of remote sensing data: multisource, multiple space and time scales, big data, uniform distribution and virtual storage under cloud computing environment. This paper elaborates the construction method of this model and key elements. The followed experiment validates the reliability of RSDO.(2) This paper proposes a multi-task scheduling algorithm based on improved ant colony algorithm and QoS with characteristics of task requirement and resources load balance. The efficiency of task processing can be improved by this algorithm and it can also offer better service for users.(3) Both RSDO model and multi-task scheduling algorithm are applied to real high-resolution remote sensing data and application system. Currently, the system has entered the trial operation stage and obtained good performance, which verifies the feasibility of the storage model and scheduling algorithm in this paper.