Research On Some Key Technologies Of Efficient Spatiotemporal Big Data Services

Nowadays,big data has been recognized worldwide for its value and has gradually ascended to be the fundamental strategic resources of all the countries in the world.As an important part of big data,spatiotemporal big data acts as the basis for the integration,sharing,and service of other big data due to its ability to provide a unified spatiotemporal platform for the four-dimensional circumstance formed by the interweaving of all sorts of information resources in the three-dimensional time and space.Compared with the original geographical information and data,spatiotemporal big data differs greatly or even revolutionarily in many aspects like data volume,source acquisition,refresh cycle,structural pattern,and value density.This makes it far from being possible for the original means of geographical information service supply to meet the new requirements posed by the spatiotemporal big data service.It is,thus,urgently required to make some breakthroughs in the related technological aspects.The core objective of spatiotemporal big data service is to deliver the related resources in a more flexible,fast,and efficient way,so as to provide fundamental spatiotemporal support for the corresponding life,related and production activities of the users.In accordance with this objective,both domestic and foreign scholars have conducted studies in terms of data processing,data mining,and data visualization,but the studies concerning service architecture,efficient storage,spatial matching,and service distribution are still very lacking,while the corresponding service platforms are still in the stage of technological exploration.To cope with this requirement,this paper conducted a profound investigation into the efficient services of spatiotemporal big data from some key technologies and platform creation therein such as service architecture,storage,matching,and distribution.The subsequent study was then made based on the following aspects:(1)The design and implementation of a high-performance spatiotemporal big data service architecture.A well-created service architecture serves as the basis for spatiotemporal data to achieve an efficient service supply.The existing service architectures available for spatiotemporal data are mostly monolithic services-oriented structures which can be easily deployed and expanded in the early stage of development and service but will turn out to be difficult in the future when it comes to providing flexible expansion and update or efficient service response against the gradually rising service volume.In view of this problem,this paper designed and implemented an efficient service architecture for spatiotemporal big data on the basis of the microservice concept,constructed an efficient microservice API gateway that took into account the service features of spatiotemporal big data,made breakthroughs in the efficient API matching algorithm,spatial heat-based load balancing algorithm,and inter-service communication protocol,as well as solved the problems concerning service expansion,deployment,and slow service response of spatiotemporal big data in the original service architectures.A comparative experiment indicated that the service architecture for spatiotemporal big data proposed in this paper features more strengths and higher efficiency when assessed from the service performance,service load balancing,and inter-service communication ability.(2)The design and optimization of the storage and retrieval strategy for spatiotemporal data.The storage and retrieval efficiency of the spatiotemporal data of varying resources and structures continues to be the key link in distinguishing the present spatiotemporal big data services from the traditional ones.At present,the spatiotemporal information is mainly managed with relational databases that are provided with spatial management abilities,but plagued by such problems as difficult circulation,disunified storage,and low efficiency when applied to the storage and management of multi-source heterogeneous spatiotemporal big data.With the features and requirements of the current spatiotemporal data storage in mind,this paper designed and implemented the storage architecture and retrieval strategies for spatiotemporal big data in distributed database,revolutionized some core methods including NoSQL-database-based vectors,grid data storage and retrieval,and vector-grid integrated multi-level retrieval,as well as provided solutions to the problems concerning lowly-efficient spatiotemporal big data storage and management in the traditional relational database due to the distributed storage,dynamic growth,and index maintenance of such big data.A comparison with the traditional relational database has also proven that the distributed spatiotemporal data storage and retrieval strategy adopted in the NoSQL-database-based Cassandra has made an improvement in the query efficiency of varying spatiotemporal big data.(3)The enrichment of the methods of property unification and name-address extraction for name-and-address based spatiotemporal big data.After all sorts of spatiotemporal big data are efficiently stored,they should undergo a unified spatial matching before being able to provide different services of uniform spatial and temporal criteria.Since such data have their original spatial coordinates,when there are no standard spatial positioning attributes,it is hard to achieve rapid and efficient matching and mapping of them due to their distinct sources and spatial information features.As a result,in this paper,typical thematic data properties consistency tables were designed with multi-source public data of spatial property fields and various unstructured data containing implicit spatial information only as the targets.It expands the geographical name-and-address-extraction algorithm that relies on toponymical genes.The extraction,matching,and mapping of actual geographical names and addresses testifies that the spatiotemporal big data property unification and name-address extraction can effectively achieve spatial matching,and mapping of the spatiotemporal big data that lack the spatial coordinate information.(4)The design and implementation of an approach for rapidly constructing a tile-pyramid of time-series images.The distribution service,a critical link in the direct supply of spatiotemporal big data service for users,mainly concerns how the latest data resources can be provided at the fastest pace.Taking a massive set of time-series remote sensing images as an example,a highly-efficient distribution service would mean solving the issue of how image tiles can be provided and posted online much more quickly.The existing methods are plagued by the time-consuming and lowly-efficient construction process due to dissatisfactory algorithms and parallel strategies against the continuous rise in image format variety and image data quantity.In view of this,this paper designs and implements an approach to quickly construct a pyramid of massive time-series image tiles that transcends such core methods as time series tile segmentation,homogeneous tile blending,and multi-thread parallel strategy,while also eliminating such problems in the overall construction of an image-fusion-based time series tile pyramid as being very time-consuming and having low efficiency.Through a comparative experiment on the actual time-series image data tile-pyramid construction,it was confirmed that the approach proposed herein consumes less time but achieves better overall efficiency.Thus,the approach is quite feasible.(5)The erection of a service platform and the execution of actual deployment and operation.The efficient microservice architecture for spatiotemporal big data and the three key technologies proposed by this paper were executed using the core algorithm from the NewMap underlying software as the basis to form an overall systematic framework of the spatiotemporal big data platform.The core functional algorithms therein were implemented.The framework was then applied to the pilot construction project of the spatiotemporal big data in the Smart City,so as to verify the service performance of the key technologies proposed in this paper and its corresponding platform as well as shed some light for future studies.