Research Overview
This project studies optimal approaches for the storage, retrieval and dynamic updating of spatial data in a distributed networked environment. Spatial data is commonly used to describe, among other applications, geographic and geospatial information. The research focuses on ways to optimally organize, distribute, and replicate data to support efficient searchability and, as a consequence, allow for enhanced data availability. There is a strong need for research in this area as large stores of spatial data now exist, including satellite (e.g., RADARSAT) imagery, geographical mapping, aerial inspection (laser rader or LADAR) and weather observations, with new spatial data continually being gathered. Indeed, recent advances in sensor technology have led to an increase of several orders of magnitude in the amount of data collected. These large collections are typically hosted in a distributed manner, across several hosts in a networked environment, either locally or remotely. This results in a vast number of research questions that need to be addressed, chiefly:
- The design of spatial data structures and algorithms for geographic information systems that have as low a memory footprint as possible and still allow operations such as node additions and removals, node locking, and range queries to be performed efficiently.
- I/O aware spatial data structures that are sensitive to varying access costs across the memory hierarchy, including issues related to the cost of distributed, remote, and secondary storage.
- Caching strategies for distributed spatial data structures that coordinate paging of indexing structures in order to reduce external (network or disk) accesses given changes in user requests.
- Issues of consistency, replication, and recoverability in distributed massive spatial data structures.
- Data management and organization of geospatial data in ways that maintain efficiency of access, accuracy of data, as well as the ability to distribute the data.
- Design of data structures for storing massive amounts of data that can be readily distributed over a network.
