A Hybrid Peer-to-Peer Architecture for Global Geospatial Web Service Discovery

For any large-scale distributed system, both communication and data management distill down to the problem of resource discovery. Similarly, the geoweb needs a resource discovery service for users to find the relevant web services that providing data of interest. The Open GIS Consortium (OGC) recommends tackling the question by using Web Catalog Service (CS/W). However, the system provides only local knowledge. In this work we propose and implement a locality-aware peer-to-peer (P2P) based system for global geospatial web service discovery. The system is scalable because it operates on a cooperative model and has no single point of failure. We use spatial hash indexing to preserve spatial locality information while retaining the load balancing properties of the underlay P2P networks. We evaluate our implementation with both simulated and real world data sets. Our experiments show promising potential of the architecture in both performing spatial and keyword query for discovering geoweb services. Background and Relevance Efficient methods for geospatial data discovery and exchange still pose a major challenge for scientists. It is not unusual that scientists working on multidisciplinary Earth Science research have to spend more than 50% time and resources on locating and acquiring data and information, pre-processing and assembling them into analysisready form (Di & McDonald, 1999). The technology for information extraction and knowledge discovery is accordingly considered far behind the technology for data collection (Di et al., 2008). Open Geospatial Consortium (OGC) recommended the Web Catalog Service (CS/W) to tackle the issue (OGC, 2007). However, existing OGC architectures and implementations have the following issues: 1) Single points of failure: CS/W becomes a system performance bottleneck. When a CS/W portal ceases to function, users are not able to search for services in interest even though the services are functioning; 2) A priori knowledge of CS/W locations: users have to know the location of existing CS/W servers a priori to search for OGC web services (OWS). 3) Difficult to maintain and manage: data publishers have to find one or multiple CS/W portals to publish their resources. Without proper tools, management and maintenance can be difficult and challenging. We argue that using a peer-to-peer (P2P) approach can address the above-mentioned problems and enhance the existing CS/W systems. However, building a P2P system for geospatial web service discovery needs to consider the following unique settings: