The Massive DEM Data Management And Scheduling Method Based On Document NoSQL Database

Digital Elevation Model is a solid ground model, which is expressed by the numerical matrix. It is the most important and the most basic data that composed of 3D virtual reality environment.Traditional management methods usually use files or relational database to manage DEM. They usually use a layered and blocked strategy to save the DEM pyramid. In this way, the number of data blocks will increase dramatically and the access efficiency decreases dramatically in pyramid. It is difficult to achieve large-scale terrain real-time roaming. Faster data access speed can improve the speed of real-time terrain roaming. In the course of the terrain, the data in each demand is closed to each other in the space. If they are also closed to each other in the database, the access efficiency will be greatly improved. Therefore, it is important to maintain closed relationship between the closed blocks in the database. Data prefetching is one of the effective means to improve the speed of roaming. The methods for DEM data prefetching are based on the speed and direction of the viewpoint to predict the position of the point of view. When the viewpoint has no speed and direction, the data will not be obtained.Therefore, the method of DEM data storage, DEM pyramid encoding and DEM data prefetching method are studied in this paper.Research results are as follows:(1) A storage model based on nested document structure in pyramid is proposed. The DEM pyramid model is decomposed into small pyramid. And this thesis use nested document structure of NoSQL database to store the small pyramid. It can retain the nested relation-ship between the layers of the pyramid and improve the access efficiency.(2) A DEM pyramid encoding method is proposed to balance the relationship between the level structure of pyramid and the data blocks in the same layer. This thesis uses the Hilbert curve encoding every layer of pyramid and combines the Hilbert code with the level code into the code of the block. This code method can not only retain the tree structure of pyramid, but also retain the relationship between the data block of the same level in the pyramid. And it can improve the data access efficiency.(3) A data prefetching method based on moving direction transfer matrix is proposed. Map operation is represented as a map display view move and the map remove is recording by a three dimensional vector. Then a six order matrix is used to record the transition between two moving directions. Combined with the six order prediction matrix, it predicts the direction and distance of the next moving vector based on the previous moving vector. And then pre fetch the data.(4) Design and experiment of prototype system. The prototype system is established to verify the data organization, spatial index method and map data prefetching method. Experimental results show that the small pyramid overall storage method can effectively improve the efficiency of data access. The DEM pyramid coding method can improve the access speed of 50%. And the data prefetching method can improve the response speed of the system.