Publish/subscribe research has so far been mostly focused on efficient event routing, event filtering, and composite event detection. The little research that has been published regarding security in publish/subscribe systems has been tentative at best. This paper presents a model for secure type names, and definitions for type-checked, content-based publish/subscribe systems. Our model provides a cryptographically verifiable binding between type names and type definitions. It also produces self-certifiable type definitions that guarantee type definition authenticity and integrity. We also consider type management in a large-scale publish/subscribe system and present a way for delegating management duties to type managers by issuing SPKI authorisation certificates. We feel that secure names are a prerequisite for most other security related work with publish/subscribe systems.
Lauri I. W. Pesonen, Jean Bacon