Global planning of geo-processing workflow in a distributed and collaborative environment

This paper presents a global planning based framework to construct a complex but optimal geo-processing workflow for integrating geospatial Web services, i.e. Web processing services in a distributed and collaborative environment. The geo-processing workflow is implemented using a Petri net model, in which the binding of services is modeled as a directed acyclic graph question. To screen the best fitted service among vast candidate services, both functional and QoS factors of services have to be considered. The selection over service functionality is simply achieved using OGC WPS application profile that describes services with metadata information, while the further selection through service quality mainly relies on QoS attributes using a global planning method that is discussed in details in this paper. In this work, the OGC compliant geospatial Web service is extended to support QoS and a tree-coded genetic algorithm (GA) is developed to determine the optimal composition of geospatial Web services. The workflow planning performance is finally analyzed through an example and well verified in comparison with integer linear programming (ILP).