Research And Implementation On Rapid Release Service Of Massive Remote Sensing Tiles Data

Currently,the application of remote sensing data has penetrated various fields.At the same time,all kinds of application software based on remote sensing tile data are constantly emerging,these software put forward higher requirements on the response speed of Sever for date request.Faced with massive remote sensing tile date on the TB level,how to improve the speed of release,is the current research hotspot.There is a homosexual problem when viewing high-resolution remote sensing tile data at the client: the application reacted with caton phenomenon,and has the interface blank phenomenon in sliding operation.By the analysis of the reasons,found the main reason was the Server has a slow response to request.Therefore,this paper focus on the server response policies.This paper studied and analyzes the related mechanisms of GIS,map slices and their publishing service such as WMS(Web Map Service)and WMTS(Web Map Tile Service).Because WMS adapt to the real-time dynamic mapping mechanism leading to Servers response to client request slow,while WMTS based on remote sensing tiles data has the faster response speed to client request,now more and more tile release severs use WMTS standards.However,the WMTS standards relies on the operating system's own caching mechanism,which leads to low hits and can't meet application requirements.Focus on the above problems,this paper takes the server memory cache replacement algorithm as the core,through the design of memory cache replacement strategy,consistence strategy and preload strategy,and has a deep research on response time of server and the cache hit ratio,to improve the response speed of the server.The main content of this paper contains the following two aspects:(1)Proposed a memory cache replacement algorithm based on Ordered Weighted Geometric Average Operator(OWGA)memory cache.This algorithm uses a sheet of tile data as the smallest access unit for memory,allocates different proportions for different attributes of the tile data according to their importance and calculates the OWGA value of each tile object.When the memory capacity reaches a threshold,replace memory object which has low OWGA values to save the recently and frequently request tile in memory and to reduce time-consuming file I/O operation,and thereby to improve the response speed.Finally,the algorithm is compared to the LUR algorithm and the LFU algorithm in the average response time and the hit ratio to verify its advantages and disadvantages.(2)Proposed memory cache preload strategy.According to the principle of locality: that is,the probability of the object being requested close to the object accessed will be increase,read tile data into memory in advance.Perform preload strategy according to the user's operating action direction and the distribution of tiles around the target tile.Finally,design experiment to compare the hit rate of the algorithm with the preload strategy.The experiment results show that the proposed memory cache replacement algorithm based on OWGA and related strategies,in random access mode,the server response time is shorter than the LRU algorithm and the LFU algorithm by about 3ms;at the same time combined with the preload strategy,the memory cache hit rate increased by about 30%.Finally,the algorithm and strategy proposed in this paper was applied to the application software mentioned above.When browsing tile dates tile browsing,the display effect shows more smoothly.