Abstract
The advent of the Internet of Things (IoT), with modern sensors and sensor-based devices, will significantly stimulate the development of context-aware applications. An effective means to extract higher-level contextual information from sensor data is distributed complex event processing (CEP), which facilitates the analysis of real-time data streams coming from heterogeneous and distributed sources. Considering that user context is inherently sensitive information, the preservation of privacy is critical once the processing of user context takes place over several (possibly malicious) devices, especially in collaborative scenarios. In this paper, we tackle this issue by introducing a trust-based approach for the placement and execution of CEP operators in a distributed environment. We propose a trust management model based on communication interactions among the users. Furthermore, we incorporate trust recommendations using a cosine-based similarity check in order to overcome collusion and on-off attacks. We developed a smartphone-based distributed CEP system called TrustCEP to evaluate our approach for trust management. Based on the evaluation of TrustCEP, we observe that our approach induces a minimal increase in average battery consumption compared to privacy-negligent approaches.