摘要

Recent years of research on sensor networks have resulted in multi-scale processing techniques for sensor data mining able to reflect the dynamic nature of real-world context. However, few of these techniques provide a systematic view of the relationships between sensor data streams and correlated network behaviors. In this paper, an association model of inherent, data and network properties is presented and analyzed for a suite of event diffusion spotting applications. Based on the associated model, window-based in-network cooperation is conducted for sensitive event diffusion spotting. Experimental results verify the performance of our approach, and confirm the scalability of our association perspective of sensor properties in such event diffusion spotting networks.