Design And Implementation Of Remote Sensing Data Scheduling Model Based On Two Level Parallel In Clusters

The development of modern space remote sensing technology has made the sensor spatial resolution,spectral resolution and time resolution greatly improved, and significantly increase the amount of remotesensing image data, thus two main problems occured:(1) The speed of Remote sensing image’s processingis too slow.Hardware constraints makes people access to information from the image much more slowlythan the growth rate of the amount of image data in remote sensing image processing;(2) Further increasedue to the requirements of remote sensing images of high-performance、fast processing, The existingparallel scheduling algorithms can not meet the demand. to solve the above problems, this paper firstpresents a cluster-based between either of the two levels of parallel massive data scheduling model, basedon this, puts forward the task scheduling algorithm based on cluster between two levels of parallel, thealgorithm mainly consists of two level of scheduling algorithm:(1) the task scheduling algorithm based onpriority between clusters.(2) single computing nodes remote sensing algorithm based on the algorithm ofspanning tree is parallel scheduling policy. Therefore, this article research mainly from the following twoaspects: first. between the cluster load balancing strategy and task scheduling algorithm, and proposes atask scheduling algorithm based on priority, second, single computing node use multi cores and multithreads to speed up the remote sensing algorithm implementation. Finally, a model was proposed in thispaper, two algorithms were applied to the actual development of the system, this paper discusses the systemdesign and implementation of the process. The following is the paper work and contributions:(1) proposed Remote sensing data parallel processing model.Based on the research of commonparallel processing model,On the analysis of the massive remote sensing data parallel processing platformshould achieve which goal and basis, combining with the actual needs of remote sensing image processing,puting forward the cluster between two levels of parallel parallel the massive remote sensing dataprocessing model.(2) Cluster priority-based task scheduling algorithm. This is used in massive remote sensing data andprocess model of the first stage of parallel scheduling algorithm, through the analysis of several classic taskallocation algorithms, combined with the actual needs of remote sensing data processing,this paper,put forward a kind of algorithm suitable to task scheduling between cluster according to the performance ofreal-time computing nodes and load information, to produce the compute node of the priority queue, Afterfour steps dynamicing tasks to parallel computing node; The same time, each time you assign tasks,real-time adjustment of priority queue. Proposed the optimal operation of the task solving process, speed upthe task of solving speed, saving computing resources.(3) single computing nodes remote sensing algorithm based on the algorithm of spanning tree isexecuted in parallel algorithm, is the second-level scheduling algorithm used in remote sensing data andprocess models: based on the algorithm of split and dependency analysis, remote sensing products putforward the concept of the spanning tree algorithm, and on this basis, designed and implemented based onthe algorithm of spanning tree implementation methods of parallel algorithm, the proposed algorithm canlargely improve the compute nodes of CPU execution efficiency, shorten the execution time of the remotesensing algorithm.(4) The implemention of the system. Combining the actual business requirements of environmentalprotection and its own characteristics, in.Net development platform, to achieve the task parallel betweenbased on cluster and single internal computing node algorithm components of parallel parallel processingplatform-the massive remote sensing data of remote sensing products production subsystem, ecologicalenvironment platform after put into use after put into use, to solve the problem of remote sensing productsproduction, realized the remote sensing product management, scheduling, fast parallel processing, productdata, image display. Remote sensing products production is not complicated, simple and quick operation,and practicability of the system and the advantage of ease of use, from a certain extent, improve theproduction efficiency. Platform are the obvious economic benefit and social benefit. Economic benefit isobvious, in particular, through the use of the product production plan scheduling of production taskmanagement subsystem and remote sensing products parallel processing has greatly increased theproduction efficiency.