Distributed Detection of Complex Events on Streams of Linked Data

The Internet of Things has created the need for scalable, distributed detection of complex events across organizational boundaries. We present a RESTful architecture that enables distributed detection of complex events on streams of Linked Data. Our approach transforms declarative event patterns expressed in a DatalogMTL-based temporal logic formalism into a network of stream containers and reasoning agents that can operate across organizational boundaries. Key contributions include: (1) A modular architecture based on the Linked Data Platform for federated stream processing, (2) A method for transforming declarative patterns into executable components, (3) A formal model using Colored Stochastic Petri Nets to validate correctness and analyze performance, and (4) an implementation and experimental validation of our approach. Experimental results demonstrate that our system achieves high throughput through parallel processing while maintaining a predictable latency that scales linearly with program depth.