An agent-based architecture for distributed imagery and geospatial computing

Current architectures for imagery and geospatial computing are limited to (1) stand-alone systems that include all processing functionality in one location: data retrieval, analysis, and display or (2) distributed systems in the client-server paradigm. Such systems are used in the intelligence gathering, cartography, and resource management domains.; In the near future, the amount of imagery and geospatial data to be processed by these systems is expected to increase in a super-linear fashion. New architectural paradigms need to be developed to address this increase in data, as traditional middleware approaches are difficult to use in this domain for several reasons. First, the required resources to answer a particular request or query are not always known in advance. Second, time-sensitive imagery and geospatial systems need to be able to scale up to support widely distributed collaborators and information repositories. Third, queries posed to imagery and geospatial systems don't always have single, or even clearly defined answers. The central hypothesis of this thesis is that an agent-based approach to imagery and geospatial computing improves scalability by distributing data processing functionality throughout the network, and improves knowledge reuse as processing agents store prior processing results.; In this thesis, an architecture for distributed, agent-based imagery and geospatial computing is presented. The architecture supports agent description, communication, and composition. The architecture defines an ontology and agent communication language for imagery and geospatial computing. In addition, two novel contributions are made in the area of distributed multi-agent systems. The first is an agent discovery mechanism that is shown to scale to support thousands of agents. The second is a novel mobility mechanism that introduces a new factor to be included when calculating the costs and benefits of mobility. The architecture and underlying contributions are demonstrated through a Java/Jini prototype. The architecture demonstrates its usefulness in the domains of imagery and geospatial computing, data mining, and hypothesis processing.