Options
2010
Conference Paper
Titel
Context adaptive service selection within a SOA for surveillance applications
Abstract
In the last years service oriented architectures (SOA) have been used to integrate newly developed and already existing surveillance resources such as sensors, databases, and information processing capabilities (e.g. Finland Network-Enabled Defence, or US Coast Guard's Long Range Identification and Tracking). The paradigm of SOA is to decouple the implementation of the overall application and utilized function implementations (services). This is realized by introducing a service registry instance where each service provider registers its functions together with the syntax of the invocation protocol. This allows for late binding: The creator of an application that will use such service resources does not have to know the specific service providers at system development time but delays the decision which implementation to use until run time when the application uses the service registry to find feasible service providers that can be utilized by a standardized protocol. Making an application on top of a SOA means to orchestrate the sequence of such service calls. Due to late binding SOA applications can be changed ad hoc. Service providers can be easily replaced without touching the orchestrated application. The advantage of this approach can turn into a weakness: Large SOAs can consist of a plethora of such service providers from internal as well as external sources. Thus service discovery (interactive and automated) may become a problem because the service registry has only syntactical information about service names and interfaces but knows nothing about performed function, constraints, or quality of the implementation. A solution to this problem is to enrich the representation of services in the registry semantically. This exposes service discovery processes to reasoning, automated matchmaking, and dynamic generation of orchestration. We have developed a new semantics aware service discovery component that fits into SOAs. It is based upon the SOAR rule engine. In a proof of concept a perimeter control application has been built with some simple service enabled sensors. The new service discovery component uses an ontological representation of attributes and capabilities of deployed sensors and a custom rule set that uses context information to deduce constraints of the current situation and proposes sensor services best suited to current task and context.