Managing real time spatial data in a database

Abstract

In a literature review, this work first examines the capabilities of databases for spatial data management in time-critical scenarios. Three different usages of the term real-time are found to be a valuable guidance for connections between system architecture and use case. These are hard real-time, soft real-time and a colloquial usage of the phrase. Different classes of Database Management System (DBMS) and their application for spatial data are characterized. Widely used databases that run on general purpose computers are found to be unsuited for hard real-time applications where predictability under time limits is required. For cases where timing constraints are less strict, techniques and software frameworks that are used in traditional Geographic Information Systems (GISs) might be used. But the potentially complex character of spatial data is argued to be a burden for predictability.In an experiment, a specific use case for a spatial database is explored. For disaster management, a server provides access to tiled orthophotos. Aperiodically, new images are uploaded to the server. Multiple clients are connected to the server and want to receive the image tiles if they are within their view port on a screen. A prototype for matching users’ area of interests and metadata about new tiles in an object-relational database is built. Together with a bidirectional communication channel via a WebSocket connection, the architecture is found to be a viable option for a spatial message filter that alerts clients for new orthoimages they are interested in. Based on the literature review, the architecture is classified as a non-real-time solution.