A Practical Security Infrastructure for Distributed Agent Applications

Security is a vital feature for most real-world distributed applications. For applications using an agent-based middleware, comprehensive support for many security-related aspects can and should already be provided by the platform in a holistic manner. Moreover, security in practice does not only concern the enforcement of well-known security objectives like authenticity and con dentiality but also requires a simple yet e ective usage concept that renders it easy for developers to de ne their security requirements. In this paper a security concept and implementation for multi-agent systems is introduced, which focuses on external, i.e. inter-platform, security aspects. The solution encompasses a usage concept distinguishing security intents from realization details and allows service providers as well as service clients to impose security constraints on communication relationships. The security concept requirements have been elicited from and the results have been evaluated with a real-world company project in the area of distributed work ow management in business intelligence.