MRI: Development of a High-Performance Database Appliance for Geospatial Applications

点击翻译按钮获取中文摘要

Proposal #: CNS 08-21345PI(s): Rishe, Naphtali D. Christidis, Evangelos; Li, Tao; Rangaswami, RajuInstitution: Florida International University Miami, FL 33174-2516Title: MRI/Dev.: Dev. Of a High Performance Database Appliance for Geospatial ApplicationsProject Proposed:This project, developing a high-performance next-generation hardware/software instrument able to efficiently perform complex geospatial and other queries, enables close-to-storage complex operations related to geospatial data transformation and querying, including data search with information retrieval (keywords), structural (SQL), and geospatial criteria. This replicable instrument consists of a scalable storage and computation cloud comprising an array of processors, each equipped with solid-state disks (SSD) and mechanical disks. This coupling of processors to SSDs allows an order of magnitude in performance improvement for many geospatial problems. The work entails adapting existing spatial database management, querying, and interface systems to this new architecture and deploying open-standard APIs. The instrument will include a base-map comprising nationwide aerial photography, street vectors, demographics, and cadastre data. For completeness of the basemap, most of the data (40TB) already assembled, ameliorated, and mosaiced for the TerraFly project that will be ported into the instrument, requires expansion. The instrument facilitates several areas in computer science research, including query algorithms, storage systems, geographic information systems (GIS), and data mining, feature recognition, and visualization. Examples of enabled research on algorithms range from optimization of query filtering on keywords 'containment in the objects' description to enriching spatial query languages to incorporate IR constraints and creating complex personalized ranking functions that consider the shape, location, and textual description of objects. Many examples involve feature recognition, appropriately serving multiple computer science and disaster mitigation research areas. Moreover, the enablement of real-time multi-dimensional indexing, combined with the multitude of geospatial data, also facilitates applied research in disaster management (e.g., use of high-resolution hurricane impact models to predict the possibility of damage on a house-by-house basis).Broader Impacts: This project enables applications in environmental monitoring, transportation, education, public health, and safety. By providing efficient spatial and temporal management of data needed by many constituencies such as disaster management, the instrument broadly benefits society. Specifically, the instrument impacts computer science and disaster mitigation research nationwide. Once the instrument is successfully commercialized, it should also serve many critical government and business applications requiring high-performance querying of very large geospatial databases. Furthermore, many students in this Minority Serving University (MSI) are likely to be inspired to continue towards a graduate degree and/or benefit by continuing their careers in science and engineering.