This paper introduces a generic architecture for semantic complex event processing (CEP) over sensor data, as exemplified by a reference implementation system in development for the security and surveillance domain. The system gathers data from a number of sensor subsystems and classifies this data according to ontology-based situation models and rules. The output of the system is alerts about threat situations that supports human operators in deciding when and how to deploy security personnel to manage these threats. The novelty of the proposed approach lies in the use of modular ontology design patterns for system configuration, which enable non-technical users to rapidly configure the system for particular scenarios.